Osborne_16 Bit_Microprocessor_Handbook_1981 Osborne 16 Bit Microprocessor Handbook 1981
Osborne_16-Bit_Microprocessor_Handbook_1981 Osborne_16-Bit_Microprocessor_Handbook_1981
User Manual: Osborne_16-Bit_Microprocessor_Handbook_1981
Open the PDF directly: View PDF 
.
Page Count: 803
| Download | |
| Open PDF In Browser | View PDF |
OSBORNE
16-Bit
Microprocessor
Handbook
OSBORNE
16-Bit
Microprocessor
Handbook
OSBORNE/McGraw-Hili
Berkeley, California
Copyright © 1981, 1979, 1978, 1976, 1975 by
McGraw-Hili, Inc. All rights reserved.
Printed in the United States of America. No part of
this publication may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means,
electronic, mechanical, photocopying, recording or
Otherwise without the prior written permission of
the publishers.
Published by
OSBORNE/McGraw-Hili
630 Bancroft Way
Berkeley, California 94710
U.S.A.
For information on translations and book distributors
outside of the U. S. A. , please write
OSBORNE/McGraw-Hili at the above address.
123456789 KPKP 8987654321
ISBN 0-931988-43-8
Cover design by Marc Miyashiro.
Contributing Authors
The following persons have contributed
to the writing of sections of this book
in addition to its principal authors.
Bob Abramovitz
Janice K. Enger
Curtis A. Ingraham
Susanna Jacobson
Patrick L. McGuire
Contents
1.
2.
The National Semiconductor PACE and INS8900 1-1
PACE and INS8900 Microcomputer System Overviews 1-2
INS8900 and PACE Timing and Instruction Execution 1-11
The INS8900 and PACE Instruction Set 1-11
The PACE DP8302 System Timing Element (STE) 1-35
The PACE Bidirectional Transceiver Element (BTE) 1-36
Using Other Microcomputer Support Devices with the PACE and INS8900
Data Sheets 1-D1
The General Instrument CP1600 2-1
CP1600 Instruction Timing and Execution 2-10
The CP1600 Instruction Set 2-16
Support Devices that may be used with the CP1600
The CP1680 Input/Output Buffer (lOB) 2-30
Data Sheets 2-01
1-38
2-27
3.
The Texas Instruments TMS 9900, TMS 9980, and TMS 9440 Products
The TMS 9900 Microprocessor 3-2
TMS 9900 Timing and Instruction Execution 3-15
The TMS 9900 Instruction Set 3-35
The TMS 9980A and the TMS 9981 Microprocessors 3-44
The TMS 9940 Single-Chip Microcomputers 3-52
The TIM 9904 Four-Phase Clock Generator/Driver 3-67
The TMS 9901 Programmable System Interface (PSI) 3-70
The TMS 9902 Asynchronous Communications Controller 3-82
The TMS 9903 Synchronous Communications Controller 3-95
Data Sheets 3-D 1
4.
Single Chip Nova Minicomputer Central Processing Units
A Product Overview 4-2
CPU Logic and Instruction Execution 4-18
9440 Timing and Instruction Execution 4-24
The MicroNova and 9440 Instruction Sets 4-35
9440-Nova Bus Interface 4-77
Data Sheets 4-01
5.
The Intel 8086 5-1
The 8086 CPU 5-4
8086 Timing and Instruction Execution 5-30
The 8086 Instruction Set 5-47
The 8088 CPU 5-87
The Intel 8284 Clock Generator/Driver 5-91
The Intel 8288 Bus Controller 5-98
The 8282/8283 8-Bit Input/Output Latch 5-102
The 8286/8287 8-Bit Bidirectional Bus Transceivers 5-104
Some 8086 Microprocessor Bus Configurations 5-105
Data Sheets 5-D1
6.
The Zilog Z8000 Series 6-1
The Z8001 and Z8002 CPUs 6-3
Z8001 and Z8002 Timing and Instruction Execution
The Z8000 Instruction Set 6-36
Data Sheets 6-D1
ix
6-23
4-1
3-1
7.
The Motorola MC68000 7-1
MC68000 Pins and Signals 7-9
The MC68000 Instruction Set 7-38
Interfacing the MC68000 with 6800 Peripherals 7-44
Data Sheets 7-01
8.
2900 Series Chip Slice Products 8-1
The 2901. 2901 A. and 2901 B Microprocessor Slice 8-2
The 2903 Microprocessor Slice 8-40
The 2902 Carry Look-Ahead Device 8-87
The 2909 and 2911 Microprogram Sequencers 8-92
The 2910 Microprogram Sequencer 8-110
The 2930 and 2932 Program Control Units 22-125
Data Sheets 8-01
x
INTRODUCTION
This is one o'f two books that replace An Introduction to Microcomputers: Volume 2 - Some Real Microprocessors. That
volume went through several printings and in 1978 was printed loose-leaf. Six bimonthly updates to the loose-leaf version were published in 1979 and early 1980 to provide information on newly introduced microcomputer devices. The
loose-leaf version proved, however, to be quite unpopular with bookstores because of packaging and handling considerations. It also became more and more difficult to maintain a timely flow-of the bimonthly updates. For these
reasons, Volume 2 is being replaced by two bound paperback books: the Osborne 4 & 8-Bit Microprocessor Handbook
and the Osborne 16-Bit Microprocessor Handbook. Together these handbooks include all of the information that was
contained in Volume 2 and the six updates. All known errors have been corrected and new data sheets have been
added to the two handbooks. We have divided Volume 2 into two separate handbooks because the single-volume version would be over 1800 pages in length and rather difficult to bind. In addition, the devices lend themselves to this
grouping since the 16-bit microprocessors are generally much more powerful than the four- and eight-bit
microprocessors, and thus are directed toward different applications.
Volume 2 was part of a four-volume Introduction to Microcomputers series:
Volume 0 - The Beginner's Book was written for readers who know nothing about computers.
Volume 1 - Basic Concepts provides a detailed explanation of microprocessor concepts including number
systems, addressing modes, typical instruction sets, input/output techniques, and so on. The device descriptions in the 4 & 8-Bit Microprocessor Handbook and the 16-Bit Microprocessor Handbook assume that you
have a working knowledge of the general concepts presented in Volume 1, and we will occasionally make
references to material presented in Volume 1.
Volume 2 - Some Real Microprocessors, which is being replaced by these handbooks.
Volume 3 - Some Real Support Devices, which describes general support devices that may be used with
any microprocessor. Some dedicated support devices are the 4 & 8-Bit Microprocessor Handbook and the
16-Bit Microprocessor Handbook. We define a "dedicated" support device as one best used with its parent
microprocessor. We define a "general" support device as one that can be used with any microprocessor. We
will occasionally make reference in this book to some of the general support devices in Volume 3. When
designing a system based on one of the microprocessors described in this handbook, you should not automatically assume that the dedicated support devices described in this book are the only ones or the best
ones to use with a particular microprocessor: you should always check the functionally equivalent parts described in Volume 3
In addition to this Introduction to Microcomputers series, we have begun publishing other individual handbooks. The
first two handbooks of this series are: The 8089110 Processor Handbook, which includes the 8289 bus arbiter, and the
CRT Controller Handbook, which describes five LSI CRT controller devices. This individual handbook approach w'ill be
used in the future to maintain a convenient flow of detailed, objective information on new microprocessors and related
support devices.
SIGNAL CONVENTIONS
Signals may be active high, active low or active in two states. An active high signal is one Which, in the high
state, causes events to occur, while in the low state has no significance. A signal that is active low causes
events to occur when in the low state, but has no significance in the high state. A signal that has two active
states will cause two different types of events to occur, depending upon whether the signal is high or low; this
signal has no inactive state. Within this book a signal that is active low has a bar placed over the signal name.
For example, WR identifies a "write strobe" signal which is pulsed low when data is ready for external logic to
receive. A signal that is active high or has two active states has no bar over the signal name.
xi
TIMING DIAGRAM CONVENTIONS
Timing diagrams play an important part in the description of any microprocessor or support device. Timing
diagrams are therefore used extensively in this book. All timing diagrams observe the following conventions:
1)
A low signal level is equivalent to no voltage. A high signal level is equivalent to voltage present:
No voltage
2)
A single signal making a low-to-high transition like this:
A single signal making a high-to-Iow transition is illustrated like this:
high
4)
high
I
low
3)
Voltage present
I
\
low
When using two or more parallel signals exist. the notation:
r=
signals change
I
states that one or more of the parallel signals change level. but the transition (high-to low or low-to-high) is
unspecified).
5)
A three-state single signal is shown floating thus:
~-------~
Signal
floating
6)
A three-state bus containing two or more signals is shown floating thus:
_______r---~~---4~-----floating
7)
When one signal condition triggers other signal changes. an arrow indicates the relationship as follows:
Cond'bon ~
J
h".
Causes
change
here
xii
Thus a signal making a low-to-high transition would be illustrated triggering another signal making a high-to-Iow
transition as follows:
-~----A signal making a high-to-Iow transition triggering a bus change of state would be illustrated as follows:
c::::..x
8)
When two or more conditions must exist in order to trigger another logic event. the following illustration is used:
The,.
~
cond;tio",
)
cause
change
here
Thus a low-to-high transition of one signal occurring while another signal is low would be illustrated triggering a
third event as follows:
9)
When a single triggering condition causes two or more events to occur, the following illustration is used:
condition
This
~
causes
these
changes
xiii
Thus a low-to-high transition of one signal triggering changes in two other signal levels would be illustrated as
follows:
10)
All signal level changes are shown as square waves. Thus rise and fall times are ignored. These times are given in
the data sheets which appear at the end of every chapter.
INSTRUCTION SET CONVENTIONS
Every microcomputer instruction set is described with two tables. One table identifies the operations which
occur when the instruction set is executed, while the second table defines object codes and instruction times.
Because of the wide differences that exist between one instruction set and another, we have elected not to
use a single set of codes and symbols to describe the operations for all instructions in all instruction sets. We
believe any type of universal convention is like to confuse rather than clarify; therefore each instruction set
table is preceded by a list of symbols as used within the table alone.
A short benchmark program is given to illustrate each instruction set. Some comments regarding benchmark
programs in general are, however, in order. We are not attempting to highlight strengths or weaknesses of
different devices, nor does this book make any attempt to comparative analyses, since the criteria which make
one microcomputer better than another are simply too dependent on the application.
Consider an application which requires relatively high speed processing. The only important criterion will be program execution speed, which may limit the choice to just one of the microcomputers we are describing.
COMPARATIVE
ANALYSIS
Execution speeds of all of the microcomputers may, on the other hand, be quite adequate for a second application; in
this case, price may be the only overriding factor. In a third application, a manufacturer may have already invested in a
great deal of engineering development expense, using one particular microcomputer that was available in quantity earlier than any others; the advantages or disadvantages of using a different microcomputer, based on minor cost of performance advantages, will likely be overwhelmed by the extra expense and time delays involved with switching in
midstream.
And what about benchmark programs 7
There have been a number of benchmark programs in the literature, purporting to show the
strengths or weaknesses of one microcomputer versus another; individual manufacturers
have added to the confusion by putting out their own competing benchmarks, aimed at showing their product to
be superior to an immediate rival.
Benchmark programs are misleading, irrelevant and worthless for these reasons:
1)
In a majority of microcomputer applications, program execution speed, and minor variations in program
length, are simply overwhelmed by pricing considerations.
2)
Even assuming that for some specific application, program length and execution speed are important, trivial
changes in the benchmark program definition can profoundly alter the results that are obtained. This is one
point we will demonstrate in this book, while describing individual instruction sets.
3)
Benchmark programs are invariable written by the smartest programmers in an organization, and they take
an enormous amount of time to ensure programming accuracy and excellence. This is not the level at which
any user should anticipate "run of the mill" programmers working; indeed, a far more realistic evaluation of
a microcomputer's instruction set could be generated by giving an average programmer too little time in
which to implement an incompletely defined benchmark. This will more closely approximate the working
conditions under which real products are developed. Of course, defining the "average programmer," "too
little time" and an "incomplete specification" are all sufficiently subjective that they defy resolution.
xiv
We will demonstrate the capriciousness of benchmark programs via the following benchmark program:
Raw data has been input to a general purpose input buffer. beginning at IOBUF. This raw data is to be moved to
a permanent table. which may be partially filled; the raw data is to be stored in the data table starting with the
first unfilled byte. The benchmark may be illustrated as follows:
r-----....
TABLE
HOW THIS BOOK HAS BEEN PRINTED
Notice that text in this book has been printed in boldface type and lightface type. This has been done to help you
skip those parts of the book that cover subject matter with which you are familiar. You can be sure that
lightface type only expands on information presented in the previous boldface type. Therefore. only read boldface
type until you reach a subject about which you want to know more. at which point start reading the lightface type.
xv
Chapter 1
THE NATIONAL SEMICONDUCTOR
PACE AND INS8900
PACE was developed by National Semiconductor as a single-chip implementation of its multi-chip IMP-16.
Since it was the first 16-bit. single-chip microprocessor. PACE is the first 16-bit microprocessor described in
this book.
As might be expected of an early entry product. PACE had a number of problems - both in design and fabrication
technology - which limited its acceptance. Therefore the INS8900 was recently introduced by National Semiconductor. The INS8900 is a redesigned. NMOS PACE. with internal logic problems resolved.
In this chapter we will describe both PACE and the INS8900. Specifically. we will identify the problems faced by a
PACE user. which have been eliminated in the INS8900.
PACE and the INS8900 are 16-bit microprocessors because they handle data in 16-bit units. In many ways. however.
the internal architecture of PACE and the INS8900 have an 8-bit orientation; this is something you should keep in mind
while reading this chapter. because it does result in PACE and the INS8900 having program execution speeds that are
comparable to. rather than being significantly faster than. the 8-bit microprocessors we have described in earlier chapters.
The only current manufacturer for PACE and the INS8900 is:
NATIONAL SEMICONDUCTOR. INC.
2900 Semiconductor Drive
Santa Clara. CA 95050
There are agreements between Rockwell International and National Semiconductor and between Signetics and
National Semiconductor to exchange microcomputer technical information and to produce each other's products. At
the present time, neither Signetics nor Rockwell International has elected to second source PACE or the INS8900, and it
is extremely unlikely that they will since both PACE and the INS8900 are products with limited futures. The amount of
support that National Semiconductor provides is rapidly declining as newer, more powerful 16-bit microprocessors
enter the marketplace.
As shown in Figure 1-1, a typical PACE microcomputer will consist of a mixture of special-purpose PACE support
devices and standard devices. The PACE microcomputer devices described in this chapter consist of:
• The PACE CPU
• The System Timing Element (STE). which generates clock signals for PACE and the system.
• The Bidirectional Transceiver Element (BTE). which converts the MOS-Ievel PACE signals to TTL-level signals
for other devices. The BTE is 8 bits wide.
The INS8900 needs a clock generator; a 2 MHz crystal and a 74C04 inverter are recommended. Otherwise. there are no
special INS8900 support devices; in fact. you can easily use any NMOS support devices described in Volume 3
with the INS8900. Specifically. the STE and BTE devices cannot be used with the INS8900. because they provide
MOS-to-TTL signal level conversions for PACE
PACE requires +5V. +8V and -12V power supplies. The +8V is a substrate vcltage requirement of the CPU and can be derived from the +5V power using a few discrete components.
Therefore, a system can be implemented using only two primary power supplies: +5V and
-12V. The INS8900 also uses three power supplies: +12V. +5V and -8V.
PACE/INS8900
POWER SU PPL Y
EXECUTION
SPEED
The INS8900 uses a 500 nanosecond clock to provide typical instruction execution times in the range of 8 to 20
microseconds. PACE (tPC-16A/520D) uses a 750 nanosecond clock to provide typical instruction execution times in
the range of 12 to 30 microseconds.
1 ·1
Before making direct comparisons of these instruction execution times with those of other devices. however. note
carefully that because of the 16-bit architecture of PACE and the INS8900. it may take many instructions on another
microcomputer to perform the same operations as a single INS8900/PACE instruction.
MOS level signals are input and output by PACE. TTL level signals are input and output by the
INS8900.
PACE/INS8900
LOGIC LEVEL
P-channel silicon gate. MOS/lSI technology is used with PACE. N-channel MOS technology is
used by the INS8900.
PACE AND INS8900 MICROCOMPUTER SYSTEM OVERVIEWS
Figure 1-1
conceptually illustrates a PACE system. Figure 1-2 conceptually illustrates an INS8900 system.
As with any mini- or microcomputer system, the CPU outputs data, address, and control signals. In the case of
PACE and the INS8900, the data and address signals use the same bus lines; therefore, they are said to be
multiplexed.
Timing signals needed by PACE are generated by the System Timing Element (STE).
PACE signals are all MOS level; the STE therefore generates two sets of timing signals;
one set are MOS level for PACE, the other set are TTL level for external logic.
SYSTEM TIMING
ELEMENT (STE)
BIDIRECTIONAL
Since PACE signals are MOS level, Bidirectional Transceiver Elements (BTEs) must be
TRANSCEIVER
present to translate outgoing signals from MOS to TTL levels, and to translate incoming
ELEMENT (BTE)
signals from TTL to MOS levels. BTEs are quite indiscriminating in the signals they translate:
in either direction. any signal arriving at an input pin is faithfully reproduced at the corresponding output pin. Control signal options allow a BTE to operate bidirectionally. to drive output Signals only. or to
place both the MOS and TTL outputs in a high-impedance mode. Since the BTE is 8 bits wide. two BTEs operating
bidirectionally provide buffering for the 16-bit Address/Data Bus. A third BTE. operating in the drive-only mode. provides buffering for the PACE control signals (NADS. ODS. IDS. and Flags).
A complete TTL level bus is created by combining BTE outputs with the TTL level timing
signals output by the STE. Remember. though. that the 16 address/data lines are multiplexed.
External logic that can demultiplex these lines and that can respond to the PACE timing and control signal logic can connect directly to the TTL level address/data lines. For example. National Semiconductor provides
ROM and RAM devices with on-chip address latches: these devices can interface directly to the TTL level bus.
ADDRESS
If memory devices or I/O ports are used that cannot demultiplex the address/data lines. you must
LATCHES
provide separate logic to perform this function. No special PACE family devices are available for
AND
this purpose: however. standard logic devices can be used. For example. two hex flip-flop devices
and a quad flip-flop device would provide a latched 16-bit Address Bus. Two 8212 I/O ports could
DECODERS
also be used to latch the 16 bits of address information. The PACE Address Data Strobe (NADS)
signal can be used as the ClK input to the flip-flops or as the STB input to the 8212s. The PACE Address Data Strobe
(NADS) signal can be used as the ClK input to the flip-flops. In many systems this is the most effective approach since
a latched Address Bus allows you to use simpler address decoding techniques to generate memory chip enable and I/O
port select Signals.
Figure 1-2 illustrates an INS8900 microcomputer system. logic is quite elementary - and equivalent to that
which you would expect with any other microcomputer. Control Bus. Data Bus. and Address Bus lines are buffered
using INS8208 bidirectional buffers. These are National Semiconductor standard catalog devices. recommended by
National Semiconductor and illustrated in their literature: however. any other buffer would do equally well. The
Data/Address Bus is shown being demultiplexed by 8212s to create separate Data and Address Busses. This again is
straightforward logic.
1-2
INTERRUPT
AND JUMP
-
CONDITIONS
STE
PACE
()
"
ADDR/DATA
~
CONTROL
'4-.../
BTE
(2)
BTE
TTL LEVEL TIMING/CONTROL
:;:;
If
..
RAM WITH
ROM WITH
ADDRESS
ADDRESS
LATCHES
LATCHES
TTL LEVEL
ADDRESS/DATA
ADDRESS
LATCHES
AND/OR
DECODERS
I/O PORT
•••••• , ••4)
:i :.--"..
···· ...
~
8212,
8255
:
·: .:
: !
: i
"
MEMORY
WITHOUT
ADDRESS
LATCHES
TTL LEVEL
ADDRESS
DATA
~V,
~V,
DATA
SENSE
CONTROL
ADDRESS
LINES
LINES
LINES
LINES
IN
OUT
OUT
IN/OUT
Figure 1-1. A National Semiconductor PACE Microcomputer System
1-3
SENSE LINES IN
CONTROL BUS
INS8900
TWO
INS8208s
nnOi
74C04
TWO
INVERTER
INS8212s
ROM
RAM
PERIPHERAL
Figure 1-2. A National Semiconductor INS8900 Microcomputer System
INS8900 PROGRAMMABLE REGISTERS
The INS8900 (and PACE) has four 16-bit Accumulators and a 16-bit Program Counter; these registers may be illustrated as follows:
ACO
AC1
Secondary Accumulator
AC2
Secondary Accumulators
AC3
PC
and Index Registers
Program Counter
Primary Accumulator
Accumulator ACO may be likened to a primary Accumulator as described for our hypothetical microcomputer in
Volume 1.
Accumulator AC1 is a secondary Accumulator.
Accumulators AC2 and AC3 are equivalent to a combination of secondary Accumulators and Index registers.
Recall from Volume 1, Chapter 6 that an Index register differs from a Data Counter in that the Index register contents
are added to a displacement (which is provided by a memory reference instruction) in order to determine the effective
memory address.
The Program Counter serves the same function in an INS8900 system as it does in our hypothetical microcomputer described in Volume 1.
Figure 1-3 illustrates that part of our general microcomputer system logic which has been implemented in the
INS8900 microprocessor.
1-4
Clock Logic
Interface Logic
Interface Logic
Programmable
Timers
Read Only
Memory
I/O Ports
Figure 1-3. Logic of the INS8900 Microprocessor
INS8900 STACK
A Stack is provided on the INS8900 (and PACE) chip. The Stack is 16 bits wide and 10 words deep. The Stack is
not a cascade stack. as described in Volume 1. Chapter 6; rather. chip logic maintains its own Stack Pointer to identify
the next free Stack word. The Stack Pointer is automatically incremented and decremented in response to Push and
Pull operations. Stack Push and Pull operations are initiated by CPU logic during execution of Jump-to-Subroutine
(JSR) and Return-from-Subroutine (RTS) instructions. and during interrupt processing. to automatically save and
restore the Program Counter.
In addition. the Stack can be used for temporary storage of data or status information. There are instructions
which allow you to transfer words between the Stack and any Accumulator. or the Status and Control Flag register.
This capability can significantly reduce the number of memory accesses required (thus increasing system speed) and
can also reduce read/write memory requirements since intermediate values can be stored on trh-.e.S..t.a-.c..k_ _ _..
Whenever the Stack becomes completely filled or emptied. an Interrupt Request is
generated on the INS8900 chip. If you have enabled Stack Interrupts. program execution will
be suspended. allowing you to deal with the situation. A Stack Fu II condition will indicate that
it is time to dump data accumulated on the Stack out to read/write memory.
1-5
INS8900 AND
PACE STACK
INTERRUPTS
INS8900 AND PACE ADDRESSING MODES
Most INS8900 (and PACE) memory reference instructions use either direct or direct, indexed addressing. A few
instructions also offer indirect addressing and pre-indexed, indirect addressing. Refer to Volume 1, Chapter 6 for a
description of these addressing modes.
All memory reference instructions have the following object code format:
15 14 13 12 11 10 9 8
II I I I
f
7 6
5 4 3 2 1 0 ~ Bit No.
IxlRf I I I I I
-
f
l
II
-
....~.~,.
........~~r-~-y~~~....___
..~~......~/
'~
' - - - - - - - - - Address displacement
{ Addressing mode selection
J 00 = Base Page address
--------------- 11
PROCESSOR
DURATION
CYCLES
-:_
",:..
_ . - - - DRIVEN LOW EXTERNALLY
APPROX. 4 CYCLES
STALL
I
RESUME NORMAL OPN
~~f~I--~mm~
.. ~
.. ""'"
..._ _ _ _ _ _ _ _ _ _ _ _ _....J DRIVEN HIGH EXTERNALLY
NHALT
(OR USING INTL. PULLUP)
1-----::::
o
3 ClK CYCLES
~ 5+te
CYCLES
....I
....I
2: 4
CYCLES
~
CO NT
CONTINUE DRIVEN
EXTERNALLY
.....f - -.........--CONTINUE DRIVEN BY PACE - - - - -__........----~
(EXTERNAL CIRCUITS HIGH IMPEDANCE)
CONTINUE
DRIVEN
EXTERNAUY
~:
1. EXTERNAUY GENERATED TTL INPUTS
OVERRIDE PACE MOS OUTPUTS.
2, ~ CROSSHATCH INDICATES "DON'T
~
CARE" INPUT STATE.
3. te = DURATION OF EXTEND DURING
PACE I/O CYCLES. TIMING ASSUMES
NO OTHER EXTENDS AND NO SUSPENDS.
Figure 1-10. Timing Diagram for Processor Stall USing
NHALT and CONTIN Signals
But we must have a way of determining whether the CPU is going to be using the System Busses. There are
several methods of making this determination; we will conceptually examine each of them within the context of three
different DMA schemes:
1)
DMA block data transfers initiated by the CPU
2)
DMA block data transfers initiated by external logic
3)
Cycle-stealing DMA transfers
From a hardware point of view, the simplest method of implementing DMA in a PACE or
CPU
INITIATED
INS8900 system is to have the CPU initiate block transfers of data. Consider the following
DMA BLOCK
approach. The CPU will treat an external DMA controller as a peripheral device and will estabDATA TRANSFERS
lish initial conditions such as starting address, word count. and direction (memory read or
write). This information can be passed to the controller by treating its registers as memory
locations and using Store instructions to write into the registers. When the required information has been passed, the
CPU simply executes a Halt instruction. As we described earlier, when a Halt instruction is executed, the NHAL T
control output line from the CPU is driven low (7/8 duty cycle). This signal could thus be used by the DMA controller as an indication that the CPU will not be using the System Bus and the DMA transfer can begin. When the
transfer is completed, the DMA controller will use the CONTIN input to the CPU, as shown in Figure 1-9 , to
terminate the Halt instruction. Normal CPU operation will then resume.
1-16
DMA BLOCK
DATA TRANSFERS
INITIATED BY
EXTERNAL LOGIC
IN PACE AND
INS8900
SYSTEMS
Most microprocessor.s have a Bus Request input signal that can be used by external logic to request access to the System Busses. In a PACE or INS8900 system. the NHAL T input signal
can be used to force the CPU into a Processor Stall. as described earlier. and thus free
the System Busses for DMA operations. The Acknowledge Interrupt (ACK INT) pulse on
the CONTIN output line shown in Figure 1-10 is then equivalent to a Bus Grant signal.
and the DMA controller may begin the data transfer. When the transfer is complete. the
CONTIN line is used as a control input line to the CPU to terminate the Processor Stall.
CYCLE-STEALING
Cycle-stealing DMA operations typically transfer a single word via the System Busses during a
DMA IN PACE
brief interval when the CPU is not using the busses. With this method. CPU operations need
AND INS8900
not be stopped; instead. they are only slowed down slightly. or in some cases not affected at
all. In order to implement cycle-stealing DMA. external logic must have a way of detectSYSTEMS
ing those time intervals when the CPU will not be using the System Busses. There are' ' - - - - - - - -...
two ways that this can be accomplished with the INS8900 or PACE CPU. The first method involves the use of the EXTEND input signal to the CPU to suppress or suspend input/output operations; the second method uses a special technique to sense when the CPU is beginning an internal (non-I/O) machine cycle.
Earlier we described how to use the EXTEND input signal to lengthen the CPU input/output cycles. The EXTEND signal can also be used to prevent the CPU from beginning an I/O cycle. and
thus ensure that the System Busses will be available to external devices for DMA operations.
EXTEND USED
TO SUSPEND
INS8900 AND
PACE I/O
DURING DMA
OPERATIONS
Figure 1-11 illustrates both uses of the EXTEND signal. The CPU looks at the EXTEND input signal at internal clock phases T1 and T6. Notice that during I/O cycles the IDS or ODS signal goes
high at the beginning of T6 and low at the beginning of T1. If EXTEND is high during T6. then extra clock cycles are inserted after T8; this is the method that would be used to lengthen an I/O cycle. If EXTEND is high during T1. then extra clock cycles are inserted between T3 and T4; this is the method we would
use for DMA operations.
The trailing edge of IDS/ODS indicates that the CPU has just completed an I/O cycle and is therefore not using the
System Busses at this instant. By setting EXTEND high at this time. we suppress the beginning of another I/O cycle
while we use the busses for a DMA transfer.
Notice that we are merely lengthening the beginning of the machine cycle. and thus delaying that part of the machine
cycle where the CPU might begin I/O activity. We do not know whether the current machine cycle will be an internal
machine cycle or an I/O cycle. and we do not care. We have merely stolen the busses by slowing down the CPU.
I
.
I
750 nsec
~
1.5 j1.sec
I..
~
Internal
I
Clock Phase :T1 T2 T3 T4 T5 T6 T7 T8 EEl T1 T2 T3 E E T4 T5 T6 T7 T81T1 T2 T3 E
,
IDS/ODS
I
I
j...---""
CPU I/O CYCLE
CPU I/O CYCLE
CPU I/O CYCLE
EXTENDED ONE CLOCK
DELAYED ONE CLOCK
PERIOD
DELAYED TWO CLOCK
PERIODS
I
!.. .1
1.5 j1.sec
C .l
2.25 j1.Sec
Figure 1-11. Using PACE EXTEND Signal for Cycle-Stealing DMA
1-17
I
I
I
PERIOD
BUS
AVAILABLE
~I
E E E T4 T5 T61
There are two draw\Jacks inherent in the EXTEND method of cycle-stealing DMA. First. whenever we use the System
Busses for a DMA transfer. we slow down the operation of the CPU. Second. we must wait until the CPU has just completed an input/output cycle before we can perform the cycle steal. Since only about one-third of the CPU machine cycles are used for 110. this means that bus access for DMA will be quite limited. Both of these drawbacks can be eliminated if we can find some technique for determining when the CPU is performing an internal (non-I/O) machine cycle.
We could then use the System Busses any time that the CPU is not using them (which is more than 60% of the
time) and we could perform the 'DMA transfer without slowing down CPU operations. We shall now describe
just such a technique.
CYCLE-STEALING
We stated earlier in this chapter that the internal clock phases (T1 through T8) are not availaDMA DURING
ble to external logic. However. National Semiconductor data sheets include a figure that shows
INS8900 AND
circuits for internal drivers and receivers. A detailed examination of this figure reveals a very
PACE INTERNAL
interesting and useful fact: the JC13 (Jump Condition 13) pin on the CPU is intended as an input signal: but because of the way in which the receiver for this signal is designed. it also proMACHINE CYCLES
duces an output pulse on the JC13 pin during every machine cycle. The output pulse occurs
during T4 of each machine cycle. and we can use this fact to design a very efficient cycle-stealing DMA arrangement.
~~-------------------------------------------------- BUS REQUEST
(From DMA Device)
NADS
D
Q
~----------------------~~~BUsGRANT
(To DMA Device)
CLR
CLK
DIVIDE-BY -FOUR
CLR
NINIT
-----------------1........-------------'
TCLK--------------------------------------------------~
(From
STE)
Figure 1-12. Idealized Circuit for Cycle-Stealing DMA During INS8900 and
PACE Internal Machine Cycles
Figure 1-12 shows a circuit that uses the output pulse provided by JC13 to implement cycle-stealing DMA. Recall
that the CPU sends out a negative-going NADS pu Ise at T4 of every input/output cycle. This NADS signal is ANDed in
our circuit with an external device's DMA Bus Request and applied to the D input of a flip-flop. The JC13 output pulse.
which also occurs at T4. is inverted via a transistor and applied to the clock input of the flip-flop. Thus. if NADS is high
at T4 (indicating that the current CPU machine cycle is not an I/O cycle) the flip-flop will be set if there is a Bus Request
present. The output of this flip-flop is then used by external logic as a Bus Grant signal and the DMA transfer can be in-
1-18
itiated. Since we do not know whether or not the next cycle will be a CPU I/O cycle. we must terminate DMA activity on
the bus prior to the next T4 time. In Figure 1-12. this is accomplished using a divide-by-four counter.
The ClK input to the counter is a combination of the Bus Grant signal and the TClK signal which is available from the
PACE STE. This results in the timing shown in Figure 1-13. Notice that this scheme makes the bus available for about
7/8 of a machine cycle. or approximately 2.25 microseconds. If you refer back to Figure 14-10 you will notice that this
is about the same length of time as was obtained by using the maximum du ration of EXTEND. So. we have not increased the maximum time available for a DMA transfer. But. we have made two significant gains: DMA transfers can
occur more frequently. and these transfers do not slow down CPU operations.
We must add a final note of caution to the description of this otherwise straightforward DMA technique. There are
several critical timing paths in the idealized circuit shown in Figure 1-12. Both the JC 13 pulse and the NADS signal
occur at T4. although the trailing edge of NADS does occur slightly after the trailing edge of JC13. Therefore. the components used to provide ClK and D inputs to the flip-flop must be selected carefully to ensure that there is not a race
condition. Additionally. we have shown the Bus Grant signal being reset at the end of T3. Since the leading edge of
NADS occurs at T4. this timing relationship can be critical. However. if external devices such as address latches and
decoders use the trailing edge of NADS. this timing should present no problems.
T3
T4 T5 T6 T7
TB T1
T2 T3
T4 T5 T6 T7 TB
T1
T2 T3
T4 T5
NClK
(TClK*)
ClK
(TClK)
JC13
~~ ____________~r-\~
______________________
T1
NADS
BUS REO
BUS GRANT
~
2.25 fLsec for DMA Transfer
Figure 1-13. Timing for Cycle-Stealing DMA During INS8900 and PACE Internal Machine Cycle
THE INS8900 AND PACE INTERRUPT SYSTEM
The INS8900 and PACE CPUs have complete on-chip interrupt systems. Six separate levels of interrupts are
provided: one internal and five external interrupt request inputs. including a non-maskable input. Priority logic is
provided on the CPU. and all interrupts are vectored. thus eliminating any polling requirements. Because of the
various ways in which interrupts can be initiated. and also because of a few problems that exist in the PACE interrupt system. we will divide our description of the system into three parts:
1)
2)
3)
Low priority external interrupts
Internal (Stack) interrupts
Non-maskable (Level 0) interrupts
But first. let us take an overview of the INS8900 and PACE interrupt system.
1-19
IRO
INT
ENABLE
lEN
INTERRUPT
LEVEL 0
INTERRUPT
NHALT
"
(TO CPU'S INTERNAL
CONTROL CIRCUIT)
REQUEST
(lRO)
.-.
""
J
STACK FULL OR
EMPTY INT REQ
(INTERNAL TO PACE)
,
s
IR1
f-
t-H
R
i'""
~
IE1
4
~
NIR2
)-[>0.....
.-
IR2
R
~
IE2
(
NIR3
....
~
}[>0i'""
j
S
.f
PRIORITY
ENCODER
~
....
-
)
S
IR3
R
IE3
NIR4
)-£>0....
S
IR4
R
IE4
(
NIR5
)-1>0i'""
S
IR5
R
IE5
Figure 1-14. Internal View of INS8900 and PACE Interrupt System
1-20
INTERRUPT
POINTEA
ADDRESS
Figu re 1-14 depicts the interrupt log ic that is contained on the CPU. The highest priority in-·
terrupt request is the non-maskable Level 0 interrupt request, which is initiated using
the NHALT control input to the CPU. The lowest priority interrupt request is NIR5.
I NS8900
AND PACE
INTERRUPT
PR.IORITIES
The Stack Interrupt and each of the four lower-priority external interrupt requests can be
ENABLING AND
individually enabled or disabled by setting or clearing associated bits (lE1 - IE5) in the
DISABLING
Status and Control Flag register. Notice in Figure 1-14 that these bits are shown as providINS8900 AND
ing the 'R' input to a latch. The'S' input to each of these latches is the actual interrupt request
PACE INTERRUPTS
line. The significance of this is rather subtle. It means that an interrupt request need not supply
a continuous low level until it is acknowledged. Instead, any pulse exceeding one PACE clock
period will set the associated interrupt request latch: this allows narrow timing or control
pulses to be used as interrupt request inputs. Note, however, that the 'R' input to the latches overrides the'S' input.
Therefore, if the individual Interrupt Enable flag is reset. it not only prevents the latch from being set by interrupt requests, it will also clear a previously latched request that mayor may not have been serviced. If this logic is not clear to
you, you shou Id study the characteristics of the RS flip-flop.
A master interrupt enable (lEN) flag is also provided in the Status and Control Flag register. lEN must be set true
to allow any of the latched interrupt requests to be recognized by the CPU.
The CPU checks for interrupts at the beginning of every instruction fetch. If an interrupt request is
present (and enabled!. the instruction fetch is aborted, the contents of the Program Counter are
pushed onto the Stack, and the master interrupt enable (lEN) is set low. The CPU then loads the
Program Counter with the address vector for your interrupt service routine and executes the instruction contained at that address. (We'll describe the address vectors in the next paragraph.)
The interrupt request just described requires a total of 28 clock cycles from the time the interrupt is
CPU until the time when the first instruction of your interrupt service routine begins execution.
Memory locations 000216 through 000816 are used as pointer locations or address vectors.
You load each of these locations with the starting address of the interrupt service routine for each
interrupt as follows:
MEMORY LOCATION
2
3
4
5
6
7
8
INTERRUPT POINTER FOR
Stack Interrupt
NIR2
NIR3
NIR4
NIR5
Level 0 Program Counter Pointer
Level 0 Interrupt Origin
l
r
INS8900 AND
PACE
INTERRUPT
RESPONSE
recognized by the
INS8900
AND PACE
INTERRUPT
POINTERS
Special
case
The level 0 interrupt is a special case which we will describe on its own. But first let us look at interrupts in
general.
When the CPU responds to an interrupt. it loads the Program Counter with the contents of memory locations 2 through
6, depending on the specific level of interrupt that is being acknowledged. Control is thus vectored to the proper service routine. Suppose, for example, memory location 4 contains the value 2A3016. If an interrupt request occurring at
pin NIR3 is acknowledged, then during the acknowledge process the contents of the Program Counter are saved on the
Stack, following which the value 2A3016 is loaded into the Program Counter. Had the value 472816 been in memory
location 4, then 472816 would have been loaded into the Program Counter instead of 2A3016. Thus. whatever memory address is stored in the memory location associated with the interrupt being acknowledged. this address will be
loaded into the Program Counter, becoming the starting address for the specific interrupt service routine to be executed.
INS8900
As part of the interrupt response we've just described. the CPU sends out a low-going pulse on
AND PACE
the CONTIN line. Refer back to Figure 1-10 and associated text for a description of the ACK
INTERRUPT
INT pulse. The last instruction executed by your interrupt service routine must be a Returnfrom-Interrupt (RT!) instruction. This instruction sets lEN high to re-enable interrupts. then
ACKNOWLEDGE
AND RETURN
pulls the top of the Stack into the Program Counter. This returns program control to the point
FROM INTERRUPT
where it was interrupted. The RTI instruction does not clear the internal Interrupt Request
latch; therefore your interrupt service routine must reset the latch (using a Pulse Flag instruction), or the same interrupt request will still be present after the RTI instruction has been executed. Once the latch has
been cleared. it can then be re-enabled for subsequent interrupt requests.
1-21
The interrupt sequence does not save the contents of any registers except the Program Counter. If
the program that was interrupted requires that the contents of CPU registers be saved and then
restored. your interrupt service routine must perform these operations.
The CPU's response to a Stack interrupt is as described for external interrupts. However. the interrupt request is generated internally by the CPU chip; it can be caused either by a Stack Full or a
Stack Empty condition. Remember that the 10-word Stack is part of the CPU chip. It consists of an
internal RAM and a pointer that can address Stack words 0 through 9. A Stack Empty interrupt request is generated whenever the pointer is at 0 and a Pull instruction is executed. A Stack Full interrupt request occurs when the pointer is at 7 (eight entries on the Stack) and a Push instruction
is executed to fill the ninth word. The tenth word of the Stack will then be used as part of the interrupt response to store the Program Counter contents. Unless you intend to extend the Stack out
into main memory. your application program will not require a Stack Empty or Full interrupt. These
error conditions and can be avoided by careful programming.
SAVING
INS8900 AND
PACE CPU
REGISTERS
DURING
INTERRUPTS
INS8900 AND
PACE STACK
INTERRUPTS
interrupts become
If your program is treating the Stack Empty and Stack Fu II interrupts as error conditions. then you can disable Stack interrupts. in which case the full ten words of the Stack are available for nested interrupts and subroutines. Of course.
this means that a Stack Full or Empty condition. should it occur. will become an undetected error. with unpredictable
consequences.
PACE
When using PACE. but not the INS8900. there is an additional reason for not using the Stack interrupt capability unless you really need it. PACE has an internal circuit problem that can cause
STACK
improper interrupt response. If a Stack interrupt request occurs at the same time as an NIR3
INTERRUPT
or NIR5 interrupt request, the Stack interrupt address vector will be incorrectly accessed
PROBLEMS
from location 0 instead of location 2. The solution recommended in PACE literature is to load
both of these locations with the Stack interrupt vector. This apparently straightforward solution is complicated by the
fact that location 0 also happens to be the initialization address; whenever the CPU is initialized. the first instruction executed is the one that is contained in location O. Thus. the word in location 0 must serve a dual purpose:
1)
It serves as an instruction whenever the CPU is initialized.
2)
It serves as an address vector if a Stack interrupt occurs at the same time as NIR3 or NIR4.
Here's an example. The object code for a Copy Flags to Register (CFR) instruction is 040016. So. if locations 0 and 2
both contain a value of 040016 the problem is solved. Your Stack interrupt service routine would have to begin at
memory address 040016. but you would be correctly vectored to that address regardless of whether or not the interrupt error we've just described occurs. On initialization. the first instruction executed would be the CFR instruction: this
is not a very useful initialization instruction. but at least no damage is done.
For a fuller discussion of this interrupt problem and the solution. refer to PACE literature. Also keep in mind that
the problem has been fixed in the INS8900.
The non-maskable (Level 0) interrupt cannot be disabled and differs from the other interrupt levels both in the
way it is initiated and in the way the CPU responds to it.
The Level 0 interrupt request is initiated using the NHALT control input signal in comINS8900
bination with the CONTIN input line. Figure 1-15 shows the timing relationships betAND PACE
ween NHAL T and CONTIN that are required to initiate the non-maskable interrupt. If you
NON-MASKABLE
(LEVEL 0)
compare this figure with Figure 1-10. you will notice that the Level 0 interrupt request and
the Processor Stall begin in exactly the same way; NHAL T is driven low by external logic and
INTERRUPT
held low for some time after a low-going pulse (ACK INT) has been sent out on the CONTIN
line. The only difference between the two operations is towards the end of the timing sequence. For a Processor Stall.
NHALT is allowed to return high whi.le CONTIN is still high; for a Level 0 interrupt. the CONTIN line must be driven low
by external logic before the NHAL T line is allowed to go high. This critical timing sequence is the only way that the CPU
has to differentiate between a Processor Stall and a Level 0 interrupt. Notice that this Level 0 interrupt timing sequence
never requires external logic to drive CONTIN high. Therefore. if you're using the CONTIN line for any of its other multiple functions (including the ACK INT output pulse) you can merely tie CONTIN to ground and Use NHALT to initiate the
Level 0 interrupt.
The response of the CPU to the Level 0 interrupt is subtly different from its response to
INS8900
other interrupts. These subtle differences are related to the slightly different purpose of a nonAND PACE
maskable interrupt versus a normal program interrupt request. A non-maskable interrupt is
LEVEL 0
typically used only when there is a catastrophic error or failure (such as loss of power) or to impleINTERRUPT
ment a control panel for program development or debug purposes. Both of these uses require that
RESPONSE
an asynchronous. unplanned program termination have a minimum effect upon system status;
that is. you want to leave behind a picture of the system as it looked immediately before the program termination occurred.
1-22
o
I--- >
11
+'e
cycusi®
CYCUS1
~ 8 +'e
~DRIVENLOWEXTERNALLY
~U
.,-
NHALT
DRIVEN HIGH EXTERNALLY
(OR USING INTERNAL PULLUP)
I
I
5 CLOCK CYCLE MIN.
..J
W
~
$.
3 CLK
CYCLES
~
~
~
EXTERNALLY
~
-
I
~
CONTINUE DRIVEN EXTERNALLY
I
I
J
CONT
I
CONTINUE DRIVEN BY PACE
- I
5 + 1e CYCLES
\
~
EXECUTION
-
I
• ACK .•
APPROX.2 1/2+te0~
I
>
.. So 15 + 21e CYCLES0
__ J
CLOCK CYCLES
CONTINUE
DRIVEN
0)
INTERRUPT RESP. TIME
o
I
-
EXECUTION SUSPENDED
I
- I
INTERRUPT SERVICE STARTS
-
NOTES:
1. EXTERNALLY GENERATED TTL INPUTS
OVERRIDE PACE MOS OUTPUTS
2. ~ CROSSHATCH INDICATES "DON'T
~ CARF'INPUT STATE.
3. te = DURATION OF EXTEND DURING PACE
I/O CYCLES. TIMING ASSUMES NO OTHER
EXTENDS AND NO SUSPENDS
Figure 1-15. Initiating INS8900 and PACE Level 0 Interrupt
Using NHALT and CONTIN Signals
Remember that other levels of interrupts store the contents of the Program Counter or the Stack and reset the lEN flag
in the Status and Control Flag register. This sequence obviously alters the "picture" of the CPU, since both Stack contents and Status and Control Flag register contents are changed. To avoid this, the Level 0 interrupt response by the
CPU uses an external memory location to store the contents of the Program Counter. Memory location 000716 holds
the address of the memory word where the Program Counter will be stored. The contents of the Status and Control Flag
register are unaltered. CPU internal circuitry resets an "IRO INT ENABLE flag to prevent another interrupt from being
recognized (refer to Figure 1-16). but this is not discernible to you. After the Program Counter has been saved in the
designated memory location, the instruction contained in memory location 000816 is executed: this is the first instruction of your Level 0 interrupt service routine. Suppose, for example, that memory location 000716 contains the value
FF0016. Following a Level 0 interrupt request. the Program Counter contents will be stored in location FFO016. Following the Level 0 interrupt acknowledge, the actual instruction stored in memory location 000816 is executed.
Note that the Level 0 interrupt acknowledge sequence has not altered anything within the CPU that is discernible to
you or to a program: the Stack, Accumulators, and Status and Control Flag register are all unchanged. Additionally,
avoiding use of the Stack ensures that there will not be a Stack overflow - and in consequence a Stack interrupt will
not be generated by this interrupt response sequence.
RETURN FROM
The normal Return-from-Interrupt (RTIl instruction that must be executed at the end of your interrupt service routine causes the Program Counter to be restored from the Stack Since the Level 0
PACE LEVEL 0
interrupt sequence does not utilize the Stack to store the Program Counter, a different techINTERRUPT
nique must be used to return control to the interrupted program. First you must execute a Set
Flag (SFLG) or Pulse Flag (PFLG) instruction, referencing bit 15 in the Status and Control Flag register. This bit always
appears to be set to a '1', but must be referenced in this case to enable lower levels of interrupts. Next you must ex-
1-23
ecute a Jump Indirect (JMP@) through the location pointed to by the contents of memory location 000716 to restore
the original Program Counter contents.
PACE
LEVEL 0
If a Level 0 interrupt occurs within the 12-clock-cycle period following the recognition of
INTERRUPT
any other interrupt, PACE will either perform a Processor Stall (which we described earlier)
PROBLEMS
or PACE will execute the Level 0 interrupt - but using the wrong pointer address. In short.
you don't know what might happen under these circumstances. There is a solution for this problem. It requires that external logic allow NHALT to be applied to the PACE CPU only while the NADS signal is present.
provided no Acknowledge Interrupt (ACK INT) has occurred since the last NADS pulse. ACK INT is accompanied by a
negative-going pulse on the CONTIN line. Sound complicated? It is.
PACE. but not the INS8900. has some Level 0 interrupt circuit problems.
The circuit shown in Figure 1-16 is reproduced from PACE literature and solves the problem we've just described. We
won't attempt to describe here how this circuit solves the problem. Note that this circuit only takes care of Level 0 interrupt problems; if you also want to use NHAL T and CONTIN to cause a Processor Stall. you must design additional external logic.
Once again, we must advise that these interrupt system problems exist in PACE CPU chips. The INS8900 has
none of these problems.
THE INS8900 AND PACE INSTRUCTION SET
Table 1-1 summarizes the INS8900 and PACE instruction set.
The primary memory reference instructions have typical minicomputer addressing modes. These instructions will also
be used as I/O instructions. since external devices are identified via selected memory addresses . ...._ _ _ _ _'"
In Table 1-1 . "direct addressing options" means the instruction can reference memory using any
of the direct or direct indexed addressing options described earlier.
"Indirect addressing options" similarly specifies any of the indirect addressing options described
earlier.
INS8900
AND PACE
DIRECT
ADDRESSING
OPTIONS
Both Branch and Skip instructions are provided. and each differs significantly from the philosophies described in Volume 1, Chapter 6.
There are 16 conditions that can cause a Branch, as shown in Table 1-3 . Notice that three of the conditions are determined by external inputs JC 13. 14. and 15. If a Branch-on-Condition is true. then the displacement which is added to
the Program Counter is an 8-bit signed binary number as described in Volume 1. Chapter 6.
There are three varieties of Skip-on-Condition instructions. SKNE. SKG and SKAZ compare the contents of an Accumulator to a memory location which is addressed using direct or direct indexed addressing. Based on the results of
the comparison. the instruction following the Skip mayor may not be executed. These three instructions are therefore
combined Skip and Memory Reference instructions.
ISZ and DSZ identify a memory location using direct or direct indexed addressing; the contents of the addressed
memory location are incremented (ISZ) or decremented (for DSZ); if after the increment or decrement operation the
memory location contains a 0 value. then the Skip is performed.
The AISZ instruction adds an 8-bit. signed binary number to the contents of an Accumulator; if the result is 0, a Skip is
performed.
These Skip instructions will be very familiar to minicomputer programmers. and on most microcomputers are
equivalent to a secondary Memory Reference or Immediate Operate instruction. followed by a Branch-on-Condition instruction.
1-24
v
LEVEL 0 INTERRUPT REQUEST
NOTE: If the Level 0 Interrupt request has not
already been reset to a logic '1' level
before lACK goes to a logic T, then
lACK should be used to reset the request signal.
'/27476
FFl
10S----0I
Q ....- - - - - -...
8094
NHALT
-
lK
CLR
0
'/27476
7404
'/.74L74
FF3
FF2
NAOS
Q
CP
Q
CP
PACE
K
-
CLR
SET
'V'/V-O
lK
+5
CLR
0
'/274L74
FF4
lACK (normally '0')
Q
74L08
CONTIN
ODS·
INIT·
Figure 1-16. Circuit to Prevent Conflicts Between PACE Level 0
Interrupts and Lower Priority Interrupts
1-25
The following symbols are used in Table 1-1 :
ACO
Accumulator 0
C
Carry status
CC
4-bit Condition Code described in Table 15-3
D
Any Destination register
DATA8
8-bit binary data unit
DISP(X)
Direct or indexed addressing operands as explained in the text.
@DISP(X)
Indirect addressing operands as explained in the text.
EA
The effective address generated by the specified operands.
f
4-bit quantity selecting a bit in the Flag Word.
FW
Flag Word described in the text.
lEN
Interrupt Enable status
I
A 1-bit unit determining whether LINK is included in the shift/rotate.
L
Link status
n
Seven bits determining how many single bit shift/rotates are performed.
o
Overflow status
PC
Program Counter
S
Any Source register
ST
Top word of on-chip Stack.
Any register of the Accumu lator: ACO. AC 1. AC2 or AC3
x < y.z >
Bits y through z of the quantity x. For example. r< 7.0> is the low-order byte of the specified register.
[ ]
Contents of location enclosed within brackets. If a register designation is enclosed within the brackets.
then the designated register's contents are specified. If a memory address is enclosed within the brackets.
then the contents of the addressed memory location are specified.
£[ ]]
Implied memory addressing; the contents of the memory location designated by the contents of a register.
A
Logical AND
V
Logical OR
.:v-
Logical Exclusive-OR
Data is transferred in the direction of the arrow.
Data is exchanged between the two locations designated on either side of the arrow.
Under the heading of STATUSES in Table 1-1 . an X indicates statuses which are modified in the course of the instruction's execution. If there is no X. it means that the status maintains the value it had before the instruction was executed.
1-26
Table 1-1. INS8900 and PACE Instruction Set Summary
STATUSES
TYPE
MNEMONIC
zw
>
II:
c(
~ --
II:
::;
W
~
0
;;
ii: > Z
Q.
II:
c(
OPERAnON PERFORMED
BYTES
c
0
r.DISP(X)
2
[rl-[EA]
LD
O,@DISP(X)
2
Load any Accumulator, direct addressing options.
[ACO]-[EA]
ST
r,DISP(X)
2
Load Primary Accumulator, indirect addressing options.
[EA]-[rl
ST
O,@IDISP(X)
2
Store any Accumulator. direct addressing options.
[ EA]-[ACO]
LSEX
O,DISP(X)
2
Store Primary Accumulator, indirect addressing options.
[ACO]-[ EA](sign extended)
0
:!:
W
:!:
Load a signed byte into Primary Accumulator; extend sign bit into high order byte. Direct
addressing options.
w _
N
-..J
>
II:
c(
o
oz
W
ADD
r.DISP(X)
2
X
X
[r]-[r]+[EA]
DEC A
O,DISP(X)
2
X
X
Add to any Accumulator. direct addressing options.
[ACO]-[ ACO]-:-l EA]-.- [C)
SUBB
O,DISP(X)
2
X
X
AND
O,DISP(X)
2
OR
O,DISP(X)
2
r,DATA8
2
W
I-
c(
II: II:
~ ~
W
Z II:
o >
o II:
~ ~
L
LD
w
0
OPERAND(SI
0
>
II:
0
!
w:!:
:!:-
_LI
Add decimal with Carry to any Accumulator, direct addressing options.
[ACO]-[ACO] - [EA].+lC]
Subtract from Primary Accumulator with borrow, direct addressing options.
[ACO]-[ACO]/\ [EA]
AND with Primary Accumulator, direct addressing options.
[ACO]-[ ACO] V [EA]
OR with Primary Accumulator, direct addressing options.
[r< 7,0>]- DATA8 (sign extended)
Load immediate into any Accumulator. DATA8 is an 8-bit signed binary value. The sign bit
W
I-
C
W
JMP
DISP(X)
2
is propagated through 8 high order bits.
[PC]-EA
:!:
~
JMP
@DISP(X)
2
Jump by loading the effective difect address into the Program Counter.
[PC]-EA
e(
Jump by loading the effective indirect address into the Program Counter.
...
-_._--
------
---------
- - - - - - _ . _ - --
Table 1-1. INS8900 and PACE Instruction Set Summary (Continued)
STATUSES
TYPE
MNEMONIC
OPERAND(SI
OPERATION PERFORMED
BYTES
C
JSR
DISP(XI
JSR
@OISP(X)
2
I&IQ
1-1&1
~i
~~
2
~O
-g
1&1
CAl
r.DATAB
2
1-1&1
co
L
[STl-[ PCl
[PC]-EA
Jump to subroutine direct. As JMP direct. but push old Program Counter contents onto
Stack.
[STJ-[ PC]
[PC]-EA
Jump to subroutine indirect. As JMP indirect. but push old Program Counter contents onto
Stack.
[rl-[ r] +DATAB (sign extended)
Complement contents of any register. then add immediate data.
c( .....
~
0
Q~
1&11&1
~o.
~O
SOC
CC.DISP
2
If CC true: then [PCJ- EA
Branch on CC true. as defined in Table 14-3.
SKNE
r.DISP(X)
2
~i2~
SKG
O.DISP(X)
2
>01&1
SKAZ
O.DISP(X)
2
If [rl " [EA]: then [PC]-[ PC] + 1
Skip if any Accumulator not equal.
If [ACO] > [EA]: then [PC]-[PC]+ 1
Skip if Primary Accumulator greater.
If ([ ACO] /\ (EA]) = 0: then [PC]-[ PC] + 1
Skip if AND with Primary Accumulator is zero.
Zz
00
iSE
ZQ
c(Z
a:0
II:IU
1&1
U
Z
1&1
a:o.i=
~U) .....
a:ZI&I
Oc(5!l
~
1&1
~
Table 1-1. 'INS8900 and PACE Instruction Set Summary (Continued)
STATUSES
TYPE
MNEMONIC
OPERAND(S)
BYTES
C
o
OPERATION PERFORMED
L
2
ISZ
DISP(X)
DSZ
DISP(X)
AISZ
r.DATA8
[rl-[ r]+DATA8
If [r] =0; then [PC]-[PC]+l
Add immediate to any Accumulator. Skip if zero. DATA8 is an 8·bit signed binary immediate data value.
a:www
a:
RCPY
S.D
[D]-[S]
SS~
ww
RXCH
S.D
Move contents of any Accumulator (S) to any Accumulator (D).
[D]--[S]
>~w!!:
[EA]-[ EA]+,
If [EA] = 0; then [PC]-[ PC]+ 1
Increment memory. skip if zero.
[EA]-[ EA] - ,
If [EA] = 0; then [PC]-[ PC] + 1
Decrement memory. skip if zero.
a:21-~
OWc(U)
::!ia:a: O
~~~~
a:
~~!!:
Q~~
wwO
::!ia.Z
~Oc(
...,
~
CD
1-1->
Exchange contents of any Accumulators.
a: a:
w
RADD
S.D
a:w
wa.
RADC
S.D
RAND
S.D
RXOR
S.D
SHL
SHR
ROL
r.n.1
r.n.1
~
.a:
1-0
~a:
~w
2
x
x
x
x
[D]-[S]+[D]
Binary add any Accumulator to any Accumulator.
[D]-[S1+[D]+[C]
Binary add with Carry any Accumulator to any Accumulator.
[D]-[S] 1\ [D)
AND any Accumulator with any Accumulator.
[D]-[S] ¥[D]
Exclusive-OR any Accumulator with any Accumulator.
WI-
a:U)
(;
w
a:
a:w
~~
U)a:
(;~
~O
ROR
r.n.'
r.n.1
2
x
2
2
X
X
X
Shift any Accumulator left n bits. Simple if 1
Shift any Accumulator left n bits. Simple if 1
As SHL. but rotate.
As SHR. but rotate.
= 0; through Link if
= O.
1
through Link if ,
=
1.
= 1.
Table 1-1. INS8900 and PACE Instruction Set Summary (Continued)
STATUSES
TYPE
MNEMONIC
OPERAND(S)
OPERATION PERFORMED
BYTES
C
PUSH
r
L
[ST]-[r]
2
Push any Accumulator contents onto Stack.
[ST1-[FW]
Push flags onto Stack.
[r1-[ST]
2
PUSHF
PULL
0
r
2
~
u
~
PULLF
2
X
X
X
en
XCHRS
....
RTS
Pull top of Stack into any Accumulator.
[FW]-[ST1
r
2
Pull top of Stack into flags.
[ST1--[r]
DISP
2
Exchange contents of any Accumulator with top of Stack.
[PC]-[ ST1 + DISP
Return from subroutine. Move sum of DISP and top of Stack to PC. DISP is an a-bit signed
binary number.
iN
o
I~
:I
RTI
DISP
2
[PC]-[ ST]+ DISP
[IEN]-1
II:
II:
w
Return from interrupt. Like RTS. but enable interrupts.
I-
~
en
:::I
Ic(
CFR
r
2
CRF
r
2
SFLG
f
2
PFLG
f
2
I-
en
HALT
2
[r1-[FWI
X
X
X
Copy flags to any Accumulator.
[FW]-Er1
Move any Accumulator contents to flags.
[FW1-1
Set flag I to 1. (f= 0 to 15).
[ FW < f>]- 1 for lour clock periods
Pulse flag f (invert flag status for four clock periods). (f = 0 to 15).
Halt
The following symbols are used in Table 1-2:
aa
Two bits choosing the destination register.
bb
Two bits choosing the Index register
cccc
Four bits choosing the Condition Code. See Table 1-3.
ee
Two bits choosing the source register.
ffff
Four bits selecting a bit in the Flag Word.
One bit determining whether Link is included in a shift or rotate.
nnnnnnn
Seven bits determining how many single bit shifts or rotates are performed.
pp
8-bit signed displacement
00
Eight bits of immediate data
x
A "don't care" bit
xx
A "don't care" byte
Table 1-2. INS8900 and PACE Instruction Set Object Codes
INSTRUCTION
OBJECT CODE
ADD
r,DISP(X)
1110aabb
AISZ
r,DATA8
0111"108a
MACHINE CYCLES
BYTES
TOTAL
INTERNAL
INPUT
2
4
2
2
2
5/6
4/5
1
2
4
2
2
2
5/6
4/5
1
2
5
4
1
2
4
3
1
2
4
3
1
2
7
5
2
2
7/8
4/5
2
-
1
OUTPUT
pp
GG
AND
O,DISP(X)
101010bb
pp
BOC
CC,DISP
0100cccc
pp
CAl
r,DATA8
011100aa
GO
CFR
f
CRF
f
DECA
O,DISP(X)
DSZ
DISP(X)
000001aa
XX
000010aa
XX
100010bb
pp
101011bb
1
pp
HALT
()()()()()()xx
2
ISZ
DISP(X)
XX
100011bb
pp
2
7/8
4/5
2
JMP
DISP(X)
000110bb
2
4
3
1
2
4
2
2
2
5
4
1
2
5
3
2
2
4
2
2
2
5
2
3
2
4
3
1
2
4
2
2
2
4
2
2
pp
JMP
((/lOISP(X)
100110bb
pp
JSR
DISP(X)
000101bb
pp
JSR
~IJOISP(X)
100101bb
pp
LD
r,DISP(X)
1tOOaabb
pp
LD
O,OOISP(X)
101000bb
pp
LI
r,DATA8
010100aa
00
LSEX
O,DISP(X)
101111bb
pp
OR
O,DISP(X)
10100tbb
pp
1-31
1
Table 1-2. INS8900 and PACE Instruction Set Object Codes (Continued)
MACHINE CYCLES
INSTRUCTION
OBJECT CODE
BYTES
TOTAL
INTERNAL
INf:»UT
2
6
5
1
PFLG
f
oollffff
PULL
r
01100laa
XX
2
4
3
1
000looxx
XX
011000aa
2
4
3
1
2
4
3
1
2
4
3
1
OUTPUT
Oxxxxxxx
PULLF
PUSH
r
XX
S,D
0000llxx
XX
0011101aa
2
4
3
1
S,D
eexxxxxx
011010aa
2
4
3
1
S,D
eexxxxxx
010101aa
2
4
3
1
2
4
3
1
2
5+3n
4+3n
1
PUSHF
RADC
RADD
RAND
RCPY
S,D
ROL
r,n,1
eexxxxxx
010111aa
eexxxxxx
ool000aa
r,n,1
nnnnnnni
00loolaa
2
5 + 3n
4+3n
1
RTI
nnnnnnni
011111xx
pp
2
6
5
1
RTS
100000xx
2
5
4
1
RXCH
PP
011011aa
2
6
5
1
2
4
3
1
2
5
4
1
ROR
S,D
eexxxxxx
010110aa
eexxxxxx
RXOR
S,D
SFLG
f
SHL
r,n,1
oollffff
1xxxxxxx
oo1010aa
2
5+ 3n
4+3n
1
r,n,1
nnnnnnni
oo1011aa
2
5+3n
4+3n
1
O,DISP(X)
nnnnnnni
101110bb
2
5/6
3/4
2
2
7/8
5/6
2
SHR
SKAZ
SKG
O,DISP(X)
PP
100lllbb
pp
"
SKNE
r,DISP(X)
1111aabb
PP
2
5/6
3/4
2
ST
r,DISP(X)
1101aabb
PP
2
4
2
1
1
ST
O,@\DISP (X)
101100bb
pp
2
4
1
2
1
SUBB
O,DISP(X)
1ool00bb
2
4
2
2
2
6
5
1
PP
XCHRS
r
000lllaa
XX
·AII instructions may take additional cycles if Extend Read and Extend Write are implemented.
1-32
Table 1-3. Branch Conditions for INS8900 and PACE BOC Instruction
Condition
Code (CC)
Mnemonic
0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100
1101
1110
1111
STFL
REOO
PSIGN
BITO
BIT1
NREOO
BIT2
CONTIN
LINK
lEN
CARRY
NSIGN
OVF
JC13
JC14
JC15
Condition
Stack Full (contains nine or more words).
(ACO) equal to zero (see Note 1).
(ACO) has positive sign (see Note 2).
Bit 0 of ACO true.
Bit 1 of ACO true.
(ACO) is nonzero (see Note 1).
Bit 2 of ACO is true.
CONTIN (continue) input is true.
LINK is true.
lEN is true.
CARRY is true.
(ACO) has negative sign (see Note 2).
OVF is true.
JC 13 input is true (see Note 3).
JC 14 input is true.
JC15 input is true.
NOTES:
1.
If selected data length is 8 bits, only bits 0 through 7 of ACO are tested.
2.
Bit 7 is sign bit (instead of bit 15) if selected data length is 8 bits.
3.
JC13 is used by INS8900 and PACE Microprocessor Development System and is not accessible
during prototyping.
THE BENCHMARK PROGRAM
For PACE, our standard benchmark program adopts this modified form:
LOOP
LD
LD
RCPY
LD
ST
AISZ
AISZ
DSZ
JMP
RCPY
ST
2,IOBUF
O,@TABLE
0,3
0,0(2)
0,0(3)
2,1
3,1
10CNT
LOOP
3,0
O,@TABLE
LOAD I/O BUFFER ADDRESS INTO AC2
LOAD ADDRESS OF FIRST FREE TABLE BYTE
MOVE TO AC3
LOAD NEXT BYTE FROM I/O BUFFER
STORE IN NEXT TABLE BYTE
INCREMENT AC2
INCREMENT AC3
DECREMENT I/O BUFFER LENGTH. SKIP IF ZERO
RETURN FOR MORE BYTES
MOVE AC3 CONTENTS TO ACO
RESTORE ADDRESS OF FIRST FREE TABLE BYTE
In order to take advantage of INS8900 and PACE indirect addressing, three memory locations are reserved on page 0 as
follows:
10BUF
holds the beginning address of the I/O buffer.
TABLE
holds the address of the first free byte in the permanent data table.
10CNT
holds the number of data words in the I/O buffer.
1-33
Memory, as organized for the benchmark program will look like this:
-; :::~ I~;; I}
Memory
Data on Base Page
IOCNT - - -.
~
. 0012
0013
' ' S-Start
i
0014 =
;
I
I
•
I
•
~Start
I
YYYV
of I/O .....
of
Data Table
I
~"'t .......... T_
of Data
Suppose the benchmark program rules arbitrarily require that a displacement be stored in the first word of the data table, and that this displacement be added to the address of the first word of the data table in order to compute the address of the first free data table word:
~
DISP
~
Fnt data table
Rm .... data
wom
tab~
Now the instructions:
LD
RCPY
O,@TABLE
0,3
LOAD ADDRESS OF FIRST FREE TABLE BYTE
MOVE TO AC3
must be replaced by these instructions:
LD
LD
RADD
3,TABLE
0,0(3)
0,3
LOAD BEGINNING ADDRESS OF DATA TABLE
LOAD DISPLACEMENT TO FIRST FREE TABLE WORD
ADD DISPLACEMENT TO AC3
The new displacement must be restored to the first data table word. The instructions:
RCPY
ST
3,0
O,@TABLE
MOVE AC3 CONTENTS TO ACO
RESTORE ADDRESS OF FIRST FREE TABLE BYTE
1-34
won!
.
must be replaced by these instructions:
LD
CAl
RADD
RCPY
LD
ST
O,TABLE
0,1
0,3
3,0
3,TABLE
0,0(3)
LOAD BEGINNING ADDRESS OF DATA TABLE IN ACO
FORM TWOS COMPLEMENT
SUBTRACT ACO FROM AC3 TO FORM DISPLACEMENT
MOVE DISPLACEMENT TO ACO
LOAD BEGINNING ADDRESS OF DATA TABLE IN AC3
SAVE DISPLACEMENT IN FIRST FREE TABLE WORD
Forcing an INS8900/PACE programmer to conform to programming logic suited to some other microcomputer's instruction set only proves that the two microcomputers have different instruction sets.
THE PACE DP8302 SYSTEM TIMING ELEMENT (STE)
The STE is a very elementary clock device used with PACE, but not with the INS8900; it accepts inputs from an
external crystal and generates the MOS clock signals for PACE, plus a pair of TTL-level clock outputs that can
be used for synchronizing system operations. Figure 1-17 illustrates the pin assignments of the STE.
1
Xl
X2
EXTC
TCLK
TCLK*
GND
PIN NAME
-
-
16
15
2
3
4
7
14
13
12
11
10
8
9
5
STE
DP8302
6
VCC
CK
CLK
NCLK
-
DESCRIPTION
Xl. X2
External crystal connections
CLK. NCLK
CK.NCK
TCLK. TCLK*
EXTC
Damped MOS clocks to PACE
Undamped MOS clocks to PACE
LCK. LCK*
TTL clocks to microcomputer system
Extemal oscillator option
Non-overlap capacitor connection
VCC·VGG
Power and Ground
VGG
NCK
LCK
LCK*
TYPE
Input
Output
Output
Output
Input
Figure 1-17. DP8302 System Timing Element (STE) Pins and Signals
The frequency of the MOS clocks output by the STE is one-half the input crystal frequency. The
STE is designed to operate with a 2.6667 MHz crystal. The MOS clock frequency is thus 1.3333
MHz which results in a clock period (tp) of 750 nanoseconds (tp = 1/f); this is the optimal clock
period for the PACE CPU.
Two pairs of MOS clock outputs are generated by the STE; NCLK/NCLK* and NCK/NCK*. The first pair of outputs
contain a 25 n series of damping resistor; typically, these outputs will be used in circuit board layouts where the STEto-PACE interconnect lines are less than two inches. The other MOS outputs, NCK and NCK*, are undamped. and you
can select some other value of series damping resistors that might be better suited for your particular board layout.
In addition to the +5V and -12V power supplies typically needed with MOS devices, the
PACE CPU has a third power supply requirement: a substrate bias voltage (Vaa) of +8V
must be applied to the CPU chip. Since it is unlikely that any other devices in your microcomputer system would require this voltage level. the need for a third external system power source
can be eliminated by providing a voltage converter circuit. Figure 1-18 shows a circuit that
generates the required Vaa voltage level; the circuit requires only a few components and uses
one of the STE's TTL clock outputs as a 'charge pump' for the circuit.
1-35
GENERATING
THE PACE
SuaSTRATE
BIAS
VOLTAGE
PACE
23
(+8V)
STE
VBB
..
O.lf1-F
TCLK-
7
lN914
....r.
I
-:::~
lN914
1-
t:; ~ LM103
0
3V
'"'
.
C)
+5V
Figure 1-18. Circuit to Generate Substrate Bias Voltage (VBB) for PACE CPU
THE PACE BIDIRECTIONAL TRANSCEIVER ELEMENT (BTE)
The DP8300 BTE is an 8-bit device that provides an interface between the PACE MOS-Ievel signals and the
TTL-level signals required by other devices in a microcomputer system (the BTE is not used in INS8900
systems). If you refer to Figure 1-1 at the beginning of this chapter. you will see that a typical PACE microcomputer
system requires three BTEs: two are used to buffer the CPU's 16 address/data lines. and the third is used as a TTL
driver for the CPU's control signal outputs (NADS. ODS. IDS. F11 - F14).
Figure 1-19 shows the pin assignments for the BTE.
MBI/O 00
MBI/O 01
..
MBI/O 02
MBI/O 03
MBI/004
MBI/O 05
..
1
24
2
23
3
4
22
21
5
20
19
6
7
BTE
18
MBI/O 06
8
17
MBI/O 07
9
10
16
WBO-
11
12
GNO
PIN NAME
15
14
13
- -
--
..
-
--'"
-
--
VCC
BOI/O 00
BOI/O 01
BOI/O 02
BOI/O 03
BOI/O 04
BOI/O 05
BOI/0 06
BOI/O 07
CEl
CE2-
STR-
DESCRIPTION
TYPE
MBIIO 00 - 07
MOS Bus Data Lines
Input/Output
Input/Output
Input
BOI/O 00 - 07
TTL Bus Data Lines
CE1. CE2-.
STR-. WBO-
Mode Control Signals
VCC·GNO
+ 5V Power. and Ground
Figure 1-19
BTE Signals and Pin Assignments
1-36
Table 1-4 summarizes the operating modes of the BTE.
BTE MODE
WBD* is the main mode control signal; when this signal is low, the other control signals are ig~~~T:~L
nored and the BTE simply converts the MOS signals from the CPU into TTL-level output signals.
The TTL outputs have a high fan-out capability and can service up to thirty 50 milliampere loads.
The BTE used to buffer the PACE control signals normally operates continuously in this 'drive-only' mode (Mode
1) and is kept in this mode by simply connecting the WBD* signal to ground.
The BTEs used to buffer bidirectional (address/data) lines must be switched back and forth between Modes 1
and 2; Mode 1 is used for CPU data output and Mode 2 for CPU data input. The simplest way of accomplishing this
is to continuously enable the CE1, CE2*, and STR* controls by connecting them to appropriate logic levels (+5V or
ground) and then use the WBD* signal for directional control. For example, in a PACE system, the IDS signal from the
CPU could be used as the input to WBD*. During a PACE data input cycle, IDS will go high at the appropriate portion of
the cycle and place the BTE in Mode 2; IDS is low at all other times and the BTE will operate in Mode 1.
Table 1-4. PACE BTE Truth Table
CONTROL INPUTS
MODE
MODE DESCRIPTION
#
CE1
CE2*
STR*
WBD*
1
X
X
X
0
Receive MOS signals and
drive TTL signals
2
1
0
0
1
Receive TTL signals and
drive MOS signals
0
0
0
1
0
1
0
1
1
1
0
1
3
4
X
X
1
1
Outputs in
high-impedance
state
On positive-edge transition
of STR*, latch into Mode 2
or 3 as determined by state
of CE1 and CE2*
X = don't care
+5V
15
CEl t - - -...
BTE
105-----"'
BUS GRANT _ _ _..._ _ _ _~l-4_t CE2*
STR*
13
t---~
Figure 1-20. Signal Connections to Control BTE in a DMA System
1-37
In a DMA or multiprocessor we will need to use BTE Mode 3 to place the BTE outputs in a high-impedance state
and thus free the System Busses for use by other devices. In such a system an externally generated Bus Grant signal could be used to place the BTE in Mode 3. Figure 1-20 illustrates one method of doing this: whenever the BUS
GRANT signal is high. the BTE is in Mode 3. At other times the IDS signal operates as we've just described to switch the
BTE back and forth between Modes 1 and 2.
The fourth BTE mode uses a negative-to-positive transition on the STR* input to latch the state of CE1 and
CE2*, and then places the BTE in either Mode 2 or Mode 3. This latch mode function might be useful when the BTE
is used as a simple input buffer. For example. in a system with multiplexed address/data lines (such as PACE). address
outputs could be applied to CEl and CE2*. and an address strobe signal (such as NADS) connected to STR*. Then.
when the BTE is selected by the appropriate address bits. the trailing edge of the strobe signal will gate TTL data
through the BTE and apply the data to the MOS lines of the CPU. When the BTE is not selected (addressed). its outputs
will be in the high impedance state (Mode 3).
USING OTHER MICROCOMPUTER SUPPORT DEVICES
WITH THE PACE AND INS8900
The INS8900 CPU has numerous control signals which allow general purpose microcomputer support devices to
be included in an INS8900system.
Let us see how 8080A support devices might be used with the INS8900 CPU. First, we'll take an overview of
the general CPU-to-device interface that all the 8080A family of devices expect.
All of the 8080A family devices require that address information (or enabling/select signals derived from the address lines) be valid during the data transfer (read/write) portion of an input/output cycle. Recall that the INS8900
data lines are multiplexed: at the beginning of an input/output cycle. the data lines are used to output address information; the address information is then removed and the data lines are used for the actual input or output of data during
the latter portion of the I/O cycle.
Thus, the first thing we must do to interface the INS8900 to an 8080A family device is
to demultiplex the INS8900 address/data lines. There are several different approaches
that we can use to accomplish the required demultiplexing.
DEMULTIPLEXING
THEINS8900
ADDRESS/DATA
LINES
The most obvious way is to use D-type flip-flops or data registers with the INS8900
NADS signal as the clock pulse. Here are some of the standard 7400 family devices that might be used:
·7475 Double 2-Bit Gated Latches with Q and Q Outputs
·7477 Double 2-Bit Gated Latches with Q Output Only
• 74100 Double 4-Bit Gated Latches
·74166 Dual 4-Bit Gated Latches with Clear
·74174 Hex D-Type Flip-Flops with Common Clock and Clear
·74175 Quad D-Type Flip-Flops with Common Clock and Clear
Some of these devices require that the NADS signal be inverted to provide the necessary clocking Signal. Remember.
though. that PACE address information is valid during both the leading edge (high-to-Iow transition) and trailing edge
(Iow-to-high transition) of NADS; this generally simplifies the demultiplexing operation.
In many systems you will not need to latch all16 bits of address information since it would be an unusual application that required all of the 64K of address space that this provides. There will usually be some tradeoff between system
address requirements (how many system devices require a latched Address Bus) and the type and amount of address
decoding required. When a fully latched Address Bus is provided. then simpler nonlatched address decoders can be
used. In fact. often address bits can then be used directly as device select signals. or simple AND/OR gate combinations can perform the decoding.
The alternative method of demultiplexing the address/data lines is to use address decoding devices that are
clocked by the NADS signal and provide latched outputs. These latched outputs can then be used as the
device/chip select Signals during I/O cycles.
Many systems will use some combination of a fully latched Address Bus and simple or latched address
decoders. In the discussions that follow, we will not generally describe in detail the method used to obtain the
required addressing or select/enabling signals, since the method used is so dependent on the particular system
that you are designing.
1-38
Once the INS8900 address/data lines have been demultiplexed, the only major conINS8900 CONTROL
siderations we are left with are to ensure that the input/output control signals are of
SIGNAL POLARITY
the proper polarity, and to verify that there are no timing problems. We will see that
CONSIDERATIONS
generally the INS8900 I/O control signals must be inverted to operate with the 8080A
family of devices. although the 8212 offers us a choice of using the IDS and ODS signals. in either their original or inverted form.
Now we will provide a few specific examples of how devices from the 8080A family can be used with the
INS8900 CPU.
In our first example the 8212 I/O Port is used as a simple input port by the INS8900 CPU.
The interconnections required are shown in the following figure:
DOO
THE 8212 USED
AS A SIMPLE
INPUT PORT IN
AN INS8900
SYSTEM
DIO
Data to
INS8900 CPU
Data from
external logic
(System Bus)
D07
a Ds1
Derived from - - - -. .
Address Lines
DI7
8212
1 0 S - - - -......... DS2
(from INS89(0)
Tie MD to Ground. Now STB clocks
latches and
NADS
(from INS89(0)
STB
DSi.
DS2 enable buffers.
MD
cur
--
NINIT-------~
Here, the INS8900 Address Strobe signal (NADS) is inverted and used as the STB input to the 8212. Since MD
is tied to ground, the STB signal clocks the data into the 8212: this will occur every time the INS8900 performs
an input/output cycle, but the latched data will only be placed on the System Bus when the 8212 is selected.
We accomplish device selection by applying a negative-true decoded address signal to the OS 1 input and then
using the INS8900 IDS strobe signal as the DS2 input. Now, whenever the proper address is decoded, the IDS
signal will cause the data that was previously latched by NADS to be placed on the System Bus for input to the
INS8900. The timing would look like this:
NADS
STB
DIO - DI7
OS2 (IDS)
r----,
000- 007
Latched data output
onto System Bus
1-39
Notice that the data from external logic will be latched whenever NADS occurs. The actual selection of the 8212 and
input of the latched data to the INS8900 might not occur for quite some time. Frequently, this arrangement will be
completely acceptable. If not. then an input-with-handshaking arrangement. which we will describe next. might provide a better solution.
Before we proceed to our next example. let us make one more general comment about interfacing devices to
the INS8900 CPU.
The INS8900 is a 16-bit microcomputer: it can transfer 16 bits of parallel data in a single input or output cycle.
All of the other devices that we will be discussing are 8-bit devices. Frequently. you may not need the full
width of the 16-bit Data Bus when transferring data between the CPU and external logic. In these cases, you
can simply connect the data lines to/from the support device to the less significant data lines (DO - 07) of the
INS8900 System Bus, as we have shown in our first example. Masking of the unused, more significant data bits
would then be handled under program control.
When you are going to utilize the full 16 bits of the Data Bus. you merely connect two 8-bit devices in parallel.
as described in more detail for the CP1600 in Chapter 2. One device would be connected as we've already described; the data lines of the other device would then be connected to the more significant bits (08 - 015) of
the System Bus. All other connections to the two devices (device select signals. strobe signals. etc.) would be
identical.
In this example. we will use the 8212 interrupt request signal INT to establish an input
port with handshaking. The connection diagram is very similar to our first exam.ple:
000
010
Data to
INS8900 CPU
··
·
(System Bus)
Derived from
IDS
-'"
-
,... Dsi
8212
STB
e xternal logic
External logic strobes
d ata into latches
OS2
(from INS8900)
toINS8900
Interrupt or
0 ata from
017
007
Address Lines
THE 8212 USED
IN AN INS8900
SYSTEM FOR
INPUT WITH
HANDSHAKING
Tie MO to Ground. Now STB clocks
latches and 'i5s'l OS2 enable buffers
MO
INT
Jump Condition
1
Input Pin
Here. the device select signals are the same as in our first example. However, instead of using the INS8900
NADS signal to clock data into the latches, we will require external logic to input the STB signal when it has
data ready. When the data has been latched. the 8212 will output the INT signal. which will be used as the input to one of the INS8900 CPU interrupt request lines (NIR2 - NIR5) or Jump Condition inputs (JC13 - JC15).
The CPU,will then execute a service routine program that will include an instruction to read the data from the input
port. This instruction will send out the input port's address, thus generating the DS 1 signal. and then gate the latched
1-40
data onto the System Bus when the IDS signal is generated. When the latched data is read out of the 8212. the INT signal returns high to complete the transaction. This sequence is summarized by the following timing diagram
Data latched by
external logic
010 - 017
STB
----'
OS2 (lOSl
DOO - 007
..
..------------
--------..~~
............--....
Interrupt request or
Jump condition input
.. ..........-
~~--
,
~
onto System Bus
to INS8900 CPU
Using the 8212 as an output port in an INS8900 system requires a simple reversal of the
connections we have described in the two preceding examples. and we will now use the
ODS (Output Data Strobe) signal from the INS8900 instead of the IDS signal.
010
THE 8212 USED
AS AN OUTPUT
PORT IN AN
INS8900 SYSTEM
000
Data from
Data to external
logic
INS8900 CPU
(System Busl
017
007
DsT
8212
STB
OS2
Select Signal
MO
M
derived from
Address Lines
-
~
ODS
(from INS8900)
to INS8900
interrupt lines
or JC inputs
1-41
••
•
:
______J•
Select signals generated
by external logic
To external logic
When the output port's address is sent out and decoded from the Address Bus, one input to the AND gate is enabled.
The ODS signal then goes high to generate the STB signal and latch the contents of the system Data Bus into the 8212.
This will cause the INT signal to go low and inform external logic that data has been loaded into the output port. The
external logic will then generate the DS1 and DS2 signals to gate the data out of the latches. When the data has been
gated out. the TNT signal will return high. This low-to-high transition could be used as an interrupt request or jump condition input to an INS8900 to enable output of new data. Notice that if we continuously enable the 8212 outputs
by tying CST to ground and DS2 to +5V, then whenever the INS8900 loads a new data word into the latch, it
will be immediately output to external logic. This approach may be more advantageous in some applications.
Although the 8255 Programmable Peripheral Interface (PPl) is a more complicated
device than the 8212, interfacing the 8255 to an INS8900 CPU is no more complicated
(from a hardware point of view) than the INS8900-to-8212 interfaces we've described.
This is due to the programmability of the 8255; mode control is performed by your program instead of by hardwired signals. Let us look at an example to illustrate this point:
8255 PPI
DEVICES
USED IN
AN INS8900
SYSTEM
DO
To/From
INS8900 CPU
(System Bus)
07
Decoded Select - - - - -....,..;;.11
signal derived
CS
from Address Bus
8255
From latched {
Address Bus
From
INS8900
CPU
------1
l
To/From
Extemal Logic
AO
A1
iDS
AD
ODS
WR
RESET
NINIT
The CS signal selects the 8255 and this signal would typically be the output of an address decoder. The AO and
A 1 inputs select one of the three I/O ports (A, B or C) or the 8255 Control registers. The RD and WR control signals are obtained by simply inverting the IDS and ODS signals from PACE. A generalized timing diagram for input/output operations would look like this:
NADS
CS·AO·A1
~
_ _ _ _ _.. Select Device and Port Select
~
IDS (ODS) _ _ _ _ _ _ _ _ _ _
m~
?
~
Data transferred
1-42
If two· 8255s are used in parallel to provide 16-bit I/O ports. there is one special consideration beyond the general rules that we discussed earlier. Recall that mode control of
the 8255 is accomplished by writing data into one 8-bit Control register within the
device. When wired in parallel. one 8255 would be connected to bits 0 - 7 of the system
Data Bus. and the other 8255 would be connected to bits 8 - 15. Therefore. when we
send out a 16-bit control word from the INS8900 CPU to establish the desired mode of
operation. the upper and lower bytes of the word must be identical.
From a hardware point of view. interfacing either of these devices to an INS8900 CPU is
no different than interfacing an 8255 PPI to the INS8900. All we need to do is invert the
IDS and ODS signals from the CPU to obtain RD and WR (or lOR and lOW) signals. and
provide chip select and latched address bits for input to the devices. All other interfacing
and usage considerations are software functions and are described in Chapter 4. We will
not describe them here since those portions of the device descriptions apply regardless
of the CPUbeilig used.
TWO 8255
DEVICES USED
FOR 16-BIT
I/O PORTS
WITH INS8900
THE 8251
USART AND 8253
PROGRAMMABLE
COUNTERITIMER
USED IN INS8900
SYSTEMS
We will conclude our discussion of the use of 8080A devices in INS8900 systems by
INS8900 AND
comparing INS8900 System Bus signals with those of 8080A systems. This comparison
8080A SYSTEM
BUSSES
will be a useful guide for interfacing any 8080A device to an INS8900 system. Table
COMPARED
1-5 is a summary of INS8900 System Bus signals and the corresponding signals available in 8080A systems. Two separate columns are provided for 8080A signals: the first applies strictly to the 8080A CPU; the right-hand column refers to the Signals present in a typical three-chip 8080A
system consisting of the CPU, a~ 8228 System Controller, and an 8224 Clock Generator and Driver.
Since we have already discussed these signals in preceding paragraphs, we won't perform an item-by-item analysis of
the table. Nonetheless, there are a few signals in this table that do need additional explanation.
We have included the INS8900 BPS signal in the I/O Control Signal group although it is not the type of signal you
would normally classify within this group. However. you will recall that when the BPS input is high. the INS8900
operates in a Base-Page-Split mode; base page then consists of the top 128 words of memory and the bottom 128
words of memory. In our earlier discussion of the BPS Signal. we described how this mode can be used to simplify addressing of I/O devices. If you refer back to that discussion, you will see that by doing a little address decoding we
can come up with a signal that will tell us when the INS8900 is addressing an I/O device (as opposed to memory).
Let us call this decoded signal '1/0 Device' (100). Now, we can combine this decoded signal with IDS and ODS as
shown below to generate signals equivalent to the 8080A liaR and tlow signals.
IDS
-----------~--""
~--------------------VOR
1/00 - - - - - - - - - - _
OOS----------------~---L
__~
K>----------------------I/OW
And if we invert the 1/00 signal we can generate the 8080A MEMR.and MEMW signals.
10S-...;....---------,.--""""
K>-----------------------~
1 / 0 0 - - -.....
OOS~-----------------_1L-
1-43
__~
~---------------MEMW
One other portion of Table 1-5 requires some explanation. Notice that we have not drawn a line to separate the
I/O control signals from the DMA-Related Signals. We've done this intentionally because there is some overlapping of functions with some of these signal,. For example, the INS8900 EXTEND 'signal can be used either to extend
I/O cycles or to suspend I/O to allow DMA operations. We've also compared the INS8900 NHAl T output signal to the
8080A WAIT signal. This comparison is valid if limited to the CPU Halt state initiated in either system by a Halt instruction. However, in 8080A systems the WAIT signal is also an acknowledgement to the READY or RDYIN input signals.
There is no comparable EXTEND acknowledgement signal in PACE systems.
6800 SUPPORT
The 6800 family includes many devices that might be useful in INS8900 systems, UnforDEVICES NOT
tunately. all of these devices have one common requirement which effectively makes them
COMPATIBLE
incompatible for use in an INS8900 system. That requirement is enabling input signal E which
WITH INS8900
should more accurately be described as a synchronizing signal. In 6800 systems. E is usually
generated by ANDing one of the primary system clock signals (<1>2) with the Valid Memory
Address signal (VMA) from the 6800 CPU. The clock period of the resulting E signal can be no less than one microsecond. The clock signals (ClK and NClK) used in PACE systems. however. cannot have a clock period greater than 850
nanoseconds. and therefore cannot be used to simulate the 6800 2 signal. Therefore. we cannot recommend using
6800 family devices in an INS8900 system.
Table 1-5. Comparing INS8900 System Busses to 8080A System Busses
INS8900
SYSTEM
SIGNALS
SYSTEM BUS
8080A
CPU
SIGNALS
8080A SYSTEM
(CPU. 8228. 8224)
SIGNALS
Bidirectional
Data Bus
000 - 015
(16 Bits)
00- 07
(8 Bits)
DBO - DB7
(8 Bits)
Address Bus
000 - 015
Address information
must be demultiplexed
from Data Bus
AO-A15
AO-A15
Control Bus
I/O
Control
Signals
NADS
Strobe signal used
by external logic
to demultiplex
add'ress from
Data Bus
-
-
IDS
DBIN
MEMR and 1I0R
ODS
WR
M"EMW 'and iiOw
EXTEND
READY
NHALT (output)
WAIT
RDYIN
WAIT
NHALT and
CONTIN il"lputs
CONTIN
tACi<: INT o~tput)
HOLD
HOLD
HLDA
HLDA
NIR2 - NIR5
CONTIN
(ACK INT output)
INT
DO and SYNC
-
BPS.
DMARelated
Signals
-
-
Interrupt
Signals
-
INTE
Non-maskable
Interrupt
(CONTIN and
NHALT inputs)
BUS EN
INT.
INTA
INTE
-
RESIN
Initialize
NINIT
RESET
Jump Condition
Inputs
JC13-JC15
-
-
Control Flag
Outputs
F11 - F14
-
-
1-44
DATA SHEETS
This section contains specific electrical and timing data for the following devices:
PACE CPU
INS8900
PACE STE
PACE BTE
1-01
PACE CPU
r - -;;,;;;.;- -- - - - - ;;;;v;;'--'
~~
v.
v..
~
~.~
.. ,,:'·1~
""~~
~ .......
v.
-r-r
'DS
I
I
I
I
.,.IT
(un.ol
v.
-L
~
'.
FI1-Fl.
l
,I
v..
4leu:
Y rl
IV..
'.
~
l
I
~
:
:v..-f
:::i~r-l
.'M
I
~
L _____
~'!!·~~
III·'"·,,,·,,CT''''
r . ,• •
_____ ..J
i f
FIGURE 4. PACE Driver and Receiver Equivalent Circuits
external clock timing
PACE requires non·overlapping true and complemented
clock inputs as shown in Figure 5. Refer to Electrical
Characteristics for timing specifications.
where:
tp' CLOCK PERIOD
'NOVA' 'NOVB' CLOCK NONOVERLAP
'WCLK • 'WNCLK = CLOCK WIDTH
FIGURE 5. External Clock Timing
We reprint data sheets on pages 1-02 through 1-017 by permission of National Semiconductor Corporation.
1-D2
PACE CPU
In DMA or multiprocessor systems it may be desirable
to prevent I/O operations by PACE when the bus is in
use by another device. This may be done by using the
EXTEND signal immediately following an IDS or ODS
as shown in Figure 10. Alternatively. the extend timing
of Figure 9 may be used. as the extend function occurs
independent of whether there is an I/O operation. that
is. whenever the internal clock phase 6 occurs.
For systems utilizing memories with access times greater
than 2 clock periods it may be desirable to use the
EXTEND input to lengthen the I/O cycle by multiples
of the clock period. Timing for this is shown in Figure 9.
In the case of either input or output operations. the
extend. should be brought true prior to the end of
internal phase 6. The timing shown in Figure 9 will
provide the minimum extend of one clock period. Holding EXTEND true for n additional clock periods
longer will cause an extension of n + 1 clock periods.
CLOC
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
00::.. ~POWE".OCLOC'SST"LE
~
.'.IT~~
~'~l_____ll~.========~,,~cL~o~c'~'E~R,~oo~s========~.1
L
••os~
~~~'~2--------------------FIGURE 6. Initialization Timing
.OOREa
DATA
~~
,~~
'ACE _____....;O!.!!uT!!:.u!!!TS!.!.~CT~'VE~____~F=t~~oDiiuy;TPuiiTTSS;HiiG'GHiH'Mi"'~EO~••~CE~=======ti.=_
--.+1 I-toc
TRANSISTOR ON
-h. __ 1;:1
OUTPUT
1_1DC
TRA::~::
TRANSISTOR Off
_-'-IN'UT'UffEADISA'LED~MOA'AVAll~
I::;!
-41
- i ~'oo
'os ...,,'-...._
....,"" .... _............
TRANSISTOR OFF
~~...d?@&J
I~
1@J"'f:J:J...+,1=.0..,..--------
1...... cI. . , . . . .....""' .... ,.,trtftCl ••1y.tMy".nDlftlllai.... IIf...Hy
·vl. . . . . . . . .
·V.-2.lSVlttlmt.... d'IIIIC"''',nput
Figure 7. Address Output and
Data Input Timing
AOOR£SS
OATA
au::;!
:-100
~:~
OOS V'UO
....._________~
OATAVAlIO--~~----
--:11-_.00_ _ _ _ _ _ __
FIGURE 8. Data Output Timing
1-03
PACE CPU
EXTRA CLOCK
CYCLEISIOUE
I
I
ClOCKP:;lS:~~~
INTERNAL
TO EXTEND
ClK_~~~~\ADOO"::: -+_.......JI%1t:.:1
HADS
PACE
OUTPUTS
PACE PUllU'
TRANSISTOR
INPUT
DATA
ACTIVE
E%t
a
HIGH IMPEDANCE
1.:21
OFF
~
ON
DATA
DISAIIlED
VALID
OUTPUT
DATA
I?I
ODS/IDS
I?I
t::a.
DATA VALID
~
EXTEND
I-IEH
1%1-
---j __ tes
~~
-I-'ES
FIGURE 9. Extend I/O Signal Timing
absolute maximum ratings
All Input or Output Voltages with
Respect to Most Positive Supply
Voltage (VaB)
Operating Temperature Range
electrical characteristics
+0.3V to -21.5V
(TA = o°c to +70°C, vss = +5V ±5%, vGG
PARAMETER
I
_65°C to +150°C
300°C
Storage Temperature Range
Lead Temperature (Soldering, 10
seconds)
CONDITIONS
= -12V
I
±5%, vaa
MIN
I
= vss
MAX
I
+ 3V ±0.5V)
UNITS
OUTPUT SPECIFICATIONS
000-015, F11-F14, OOS, lOS, NAOS (These are
open drain outputs which may be used to drive
OS3608 sense amplifiers, or may be used with pulldown resistors to provide a voltage output.)
Logic "1" Output Current (Except F11-F14)
Logic "1" Output Current, F11-F14 (Note 7)
Logic "0" Output Current
VOUT = 204V
VOUT = 204V
VGG"';; VOUT:« VSS
NHALT, CONTIN (Low Power TTL Output.)
Logic "1" Output Voltage
Logic "0" Output Voltage
lOUT
lOUT
= -650p.A
= 300p.A
-1.0
-0.7
-5.0
-5.0
±10
rnA
rnA
p.A
004
V
V
2.4
INPUT SPECIFICATIONS
000-015, NIR2-NIR5, EXTENO, JC13-JC15,
CONTIN, NINIT, NHALT (These are TTL
compatible inputs.) (Note 2)
Logic "1" Input Voltage
Logic "0" Input Voltage
Pullup Transistor "ON" Resistance
(000-015) (Note 3)
Pullup Transistor "ON" Resistance
(all others)
Logic "0" Input Current (000-015)
Logic "0" Input Current (NHAL T, CONTIN)
Logic "0" Input Current (all others)
Capacitance, Input and Output (except clocks)
BPS (This is a MOS Level Input.) (Note 4)
Logic "1" Input Voltage
Logic "0" Input Voltage
Logic "1" Input Current
CLK, NCLK (These are MOS Clock Inputs)
Clock "1" Voltage (Note 5)
Clock "0" Voltage
Input Capacitance (Note 6)
Bias Supply Current
VGG Supply Current
VSS Supply Current
VSS-1
VSS-7
VIN = VSS-1V
VIN
= VSS-1V
kD
-1.8
-12
-3.6
20
rnA
rnA
rnA
pF
VSS-1
VGG
VSS+0.3
VSS-7
100
V
V
JJ.A
VSS-1
VGG
30
VSS+0.3
VGG+ 1
150
100
40
85
V
V
pF
JJ.A
rnA
rnA
VIN = VSS-1V
1-D4
V
V
kD
5
VIN=Oo4
VIN=Oo4
VIN=Oo4
VIN = VSS, fT = 500 kHz
VBB = VSS +3.0V
tp = .65p.s, T A = 25°C
tp = .65p.s, T A = 25°C
VSS+0.3
VSS-4
7
PACE CPU
EXTRA CLOCK
CYCLE!SIDUE
l--~
,
,
INTERNAL
CLOCK PHASE
FIGURE 10. Suspend I/O Signal Timing
TIMING SPECIFICATIONS (See Figures 5 to 10 for additional timing information.)
ClK, NClK (See Figure 5) (Referenced to
10% and 90% Amplitude)
Rise and Fall Time (t r , tf)
Clock Width (tw ClK tw NClK)
Clock Non-Overlap (tNOVA, tNOVS)
Clock Period (t p )
EXTEND
Individual Extend Duration
Extend Setup Time (tES) (Note 10)
Extend Hold Time (tEH) (Note 13)
Propagation Delay (tOO)
NHAlT, CONTIN (Note 9)
NADS, IDS, ODS, 000-015 (Note 8)
000-015
Input Setup Time (tOS) (Note 11)
Hold Time (tOH) (Note 12)
Turn-on or Turn-off Time of Pullup
Transistor (toe) (Note 13)
Fll-F14 Pulse Flag (PFlG) Pulse Width
NINIT Initialization Pulse Width
NIR2-NIR5 Input Pulse Width to Set latch
10
300
5
.65
50
375
.8
2
ps
ns
ns
200
100
ns
ns
100
20
Cl = 20 pF
VOUT= 2.4V
200
ns
ns
ns
o
150
4tp -300
8
1
ns
ns
ns
ps
4tp +300
ns
clock periods
clock periods
Note 1: Maximum ratings indicate limits beyond which permanent damage may occur. Continuous operation at these limits is not intended
and should be limited to those conditions specified under de electrical characteristics.
Note 2: Pullup transistor provided on chip (See Figure 4.)
Note 3: Pullup transistors on JC13, JC14, JC15 are turned on one out of 8 clock intervals. Pullup transistors on 000-015 are turned on
during last clock period of Input Data Strobe (IDS). Other pullup transistors are on continuously when in data input mode.
Note 4: Pulldown transistor provided on chip.
Note 5: Clamp diodes and series damping resistors may be required to prevent clock overshoot.
Note 6: Capacitance is not constant and varies with clock voltage and internal state of processor.
Note 7: For VSS ;;. VOUT ;;. 2.0V output current is a linear function of VOUT.
Note 8: Delay measured from valid logic level on clock edge initiating change to valid current output level
Note 9: Delay measured from valid logic leilel on clock edge initiating change to valid voltage output level.
Note 10: With respect to rising edge of NClK. (See Figure 9 and 10.)
Note 11: With respect to falling edge of elK. (See Figure 7.)
Note 12: With respect to the valid "0" level on the falling edge of Input Data Strobe (IDS). (See Figure 7.)
Note 13: With respect to valid logic level of appropriate clock.
1-05
INS8900
Absolute Maximwn Ratings
Voltage at Any Pin with Resepct to
Most Negative Supply (VSS) . . . . . . . . . . . . . . . . -0.3 V to +20V
Operating Temperature Range . . . . . . . . . . . . . . . . O°C to +70°C
Storage Temperature Range. . . . . . . . . . . . . . . . -65°C to +150°C
Lead Temperature (soldering, 10 seconds) . . . . . . . . . . . . +300°C
Electrical Characteristics
(TA = o°c to +70°C, VSS=
ov, Voo = +12V ± 5%,
VCC = +5V ± 5%, VSS
Conditions
Parameter
Symbol
= -8V ± 5%)
Min
Max
Units
OUTPUT SPECIFICATIONS
VOH
VOL
000-015, Fll-F14, OOS, IDS, NAOS
(These are low-power Schottky·compatible push-pull outputs.)
Logic "1" Output Voltage
Logic "0" Output Voltage
lOUT = -500}JA
lOUT = 900}JA
2.4
VOH
VOL
NHAL T, CONTIN (low-power Schottky outputs)
Logic "1" Output Voltage
Logic "0" Output Voltage
lOUT = -250}JA
lOUT = 600}JA
2.4
0.4
V
V
0.4
V
V
INPUT SPECIFICATIONS
VIH
VIL
IL
IlL
IlL
000-015, NIR2-NIR5, EXTENO, JC13-JC15, NINIT,
CaNTIN, NHALT (low-power Schottky inputs)
Logic ''1'' Input Voltage
Logic "0" Input Voltage
Input Leakage Current (except NHAL T, CaNTIN, JC13-JC15)
Logic "0" Input Current, NHALT, CONTIN (Note 2)
Logic "0" Input Current, JC13-JC15 (Note 2)
VIH
VIL
IIH
SPS (This is an MOS level input.)
Logic "1" Input Voltage
Logic "0" Input Voltage
Logic "1" Input Current (Note 3)
VCIL
VCIH
CIN
CLKX (This is an MaS level input.)
Clock "0" Voltage
Clock "1" Voltage
Input Capacitance
100
Average Supply Current (VOO) (Note 4)
tp = 500 ns, T A = 25°C
100
ICC
Average Supply Current (Vec) (Note 4)
tp = 500 ns, T A = 25°C
10
mA
IBB
Average Supply Current (VBB)
VSB=-8V
-200
}JA
2.4
-1.0
VSS ~ VIN ~ VCC + 1
VIN=0.4V
VIN = 0.4V
VIN = 13.6V
1-06
VCC+ 1
+0.8
40
-7.0
-3.0
V
V.
}JA
mA
mA
VOO-l
VOO+ 1
-1.0
+0.8
750
V
V
}JA
-1.0
+0.8
VOO-l
VOO+ 1
20
V
V
pF
mA
INS8900
Tining Specifications
Symbol
Parameter
Conditions
Min
Max
Units
5
30
ns
tCLK, tNCLK
CLKX
Rise and Fall Times (Note 5)
(Referenced to 10% and 90% amplitude)
Clock Period
Pulse Width (Referenced to 50% amplitude)
500
tp/2 - 5%
650
tp/2 + 5%
ns
ns
2
tES
tEH
EXTENO
Individual Extend Duration
Extend Setup Time (Note 6)
Extend Hold Time (Note 6)
70
120
ps
ns
ns
tOOl
Propagation Oelay
NHALT, CONTIN (Note 7)
200
ns
200
ns
t r , tf
tp
NAOS, IDS, ODS, 000-015 (Note 7)
tD02
CL = 40pF,
1 low-power Schottky load
CL=40pF,1INS820810ad
000-015
Input Setup Time (Note 6)
Hold Time (Note 8)
50
0
tFW
Fll-F14 Pulse Flag (PFLG) Pulse Width
4tp - 300
tNW
NINIT Initialization Pulse Width
8
tp
tlRW
NIR2-NIR5 Input Pulse Width to Set Latch
1
tp
tDS
tOH
ns
ns
4tp + 300
Note 1: Maximum ratings indicate limits beyond which permanent damage may occur. Continuous operation at these limits is not
intended and should be limited to those conditions specified under DC electrical characteristics.
Note 2: NHAL T, CONTIN, and JC13-JC15 logic "0" input currents specified when the internal chip loads are putting out a logic "1."
Note 3: Pull-down transistor provided on chip.
Note 4: Supply currents measured with 40 pF and I NS8208 loads.
Note 5: Clamp diode and series damping resistor may be required to prevent clock overshoot.
Note 6: Measured with respect to appropriate valid logic level of CLKX.
Note 7: Delay measured from valid logic level on CLKX edge initiating change to valid output voltage level.
Note 8: With respect to the valid "0" level on the falling edge of Input Data Strobe (I DS>'
Note 9: Typical load circuit:
INS8900
RL = 3.6k (3.3k for testing)
CL=40pF
VREF = 1.72V
_ VREF
Note 10: Typical output delay versus load capacitance CL
for load circuit in Note 9:
Note 11: Typical VDD supply current versus temperature.
125.0
250 ~
100.0 ~
200
~
C
oS
150
'"'"
I!> 25'C
50.0 ~
100 ~
50
-~
75.0
r-
-
I
I
I
120
140
I
I
25.0
I
lI
160
180
100 (ns)
200
0
220
25
I
50
TEMP I'C)
1-07
.
75
100
ns
INS8900
Tming Waveforms
~
jJ:tNCLK:f.!l-tCLK-!
J
ttl- -ltrl-
Figure 1. External Clock Timing (CLKXI
POWER
AID
CLDCICS
POWER AND CLDCK$STABLE
I-
IIIIT
I
INW
• CLOCK PE'1!,DS MINIMUM~
I-
~--u
16-24 CLOCK PERIODS
L
IADS~
~: ~~~----------------------------------------------Figure 2. Initialization Timing
IITERIAL
CLOCKPNASE
CLKX
ADDIIE.
DATA
IADI
j--tDD
'=
'ACE _________
DUTPUT
I--tDD
~o~u~~~U~n~A~~~IV!E·~________~~::::::::::::~~~~~~~;:::::::::::=t~__~____
II
_ I I P u r IUFFER DIUILED
~~7-_t~DD~________________~~
IDI ______________________________
..
~
·VII MUST IE AT THE CORRECT LOGIC LEVEL AT THIS TIME.
10TE: .. IOU ARE REFEREICED TO VALID LOGIC LEVELS ON CLOCK IIPUT. IITERNAL CLOCK PHASES ARE SHOWW
FOR REFEREICE OILY; THEY ARE lOT AVAILAILE EXTERIALL Y.
Figure 3. Add.... Output and Data Input Timing
1-08
IDD
INS8900
lining Wavefonns
(continued)
INTERNAL
CLOCK PHASE
M':~~~~_\'-----'_I~t~
t§j
NADS
OU~!~! ~~ ~~
~
\---1
~@J----++-~~-tDD---++--11-----t-FI
\
-J§__f.::::::::::::::::~DruAUTiAVVAlAL~ID~::::::::::::::=ta~~~
______________________
m
~
-Ii
I-toD
VALID
ODS _ _ _ _ _ _ _ _ _ _ _ _ _....I:l§Jo;:.]
I-tDD
Figure 4. Data Output Timing
i
INTERNAL
CLOCK PHASE
EXTRA CLOCK,
CYCLES DUE
TO EXTEND
E
E
CLKX
ADDRESS
DATA-1_ _ _
~
NADS
PACE
OUTPUTS
-I__________.....;.;:AC:.;T.;.;IV.;.E_________....:.:JI
INPUT_~-----~~~L------~~~~~~~~~~~~~~~~~~~~~~-DATA
OUTPUT
DATA
DISABLED
-1_----------------------........:4
ODS/IDS _1-_____________________........:4
Figure 5. Extend I/O Signal Timing
INTERNAL
CLOCK PHASE
CLKX
DATA
NADS
IDS/ODS
EXTEND __
~~~~~~~~~ri~~~~~~~~------------------------Figure 6. Suspend I/O Signal Timing
1-09
_____
INS8900
Tming Wavefonns (continued)
r->II+leCYCLES~1
...
••
1-----DR1VEN LOW EXTERNALL Y - - - - - - 1 1 - - ,,:,.;':;~.~,::,"n:::.:.'<";;',,: 'n::.:-::L.~:
NHALT
.. 3 CLK
CYCLES
=1
1-
5 CLOCK CYCLES MIN
INTERRUPT RE5I' TIME
I--- .. 15 + 2 Ie
I---
CYCLES~I
~
CONTIN
NOTES:
1. EXTERNAllY GENERATED TIL INPUTS OVERRIDE PACE OUTPUTS.
2.
CROSSHATCH INDICATES "DON'T CARE"INPUT STATE.
@Ie' DURATION OF EXTEND DURING PACE 110 CYCLES TIMING ASSUMES NO OTHER EXTENDS AND NO SUSPENDS.
III
Figure 7. Relative Timing for Level-O Interrupt Generation
I
--t- PROCESSOR STAll DURATIDN-l- RESUME NORMAL O P N ~
I---- APPROX 4 CYCLES
.
EXECUTION -I--EXECUTION SUSPENDED
> 11 + Ie CYCLES-=----l
@
L.._ _ _ _ _ _ _ _ _ _::::::::::::::~t--- D(~I:~~I:~~N~~~t~~~~~
NHALT
1-;.
5 + Ie CYCLES@..!--;' 4 CYCLES---l
CONTIN
~
--1
@
I
I
!--APPRDX 2% + Ie CLOCK CYCLES
CONTINUE DRIVEN ......._ - - - - CONTINUE DRIVEN BY PACE _ _ _ _ _ _ -CDNTINUE DRIVEN_
EXTERNAll Y
(EXTERNAL CIRCUITS HIGH IMPEDANCE)
,
EXTERNALL Y
I
NOTES:
1. EXTERNAllY GENERATED TIL INPUTS OVERRIDE PACE OUTPUTS.
2. ~ CROSSHATCH INDICATES "DON'T CARE"INPUT STATE.
@Ie+ DURATION OF EXTEND DURING PACE 110 CYCLES TIMING ASSUMES NO OTHER EXTENDS AND NO SUSPENDS.
Figure 8. Relative Timing for Processor Stall
The architecture of the INS8900 (shown in Figure 9)
features a number of resources to minimize system program and read/write storage, increase throughput, and
reduce the amount and cost of external support hardware. Principal resources that allow these efficiencies to
be achieved include:
Four 16-bit general purpose working registers available
to the user reduce the number of memory load and store
operations associated with saving temporary and intermediate results in system memory.
An independent 16-bit status and control flag register
automatically and continuously preserves system status.
The user may operate on its contents as data, allowing
masking, testing, and modification of several bit fields
simultaneously.
A ten-word U6-bid last-in, first-out (LIFO) stack
inherently decreases response time to interrupts while
eliminating both program and read/write system storage
overhead associated with storing stack information
outside the microprocessor chip.
Stack full/stack empty interrupts are provided to facilitate off-chip stack storage in those applications where
additional stack capacity is desirable.
A six-level vectored priority interrupt system internal to
the chip provides automatic interrupt identification,
eliminating both program storage overhead and the time
normally required to poll peripherals in order to identify
the interrupting device.
Three sense inputs and four control flag outputs allow
the user to respond directly to specific combinations of
status present in the microprocessor-based system, thus
eliminating costly hardware, program overhead, and
throughput associated with implementing these functions over the system data bus.
A comprehensive set of input/output control signals
provided by the internal control logic simplifies interfaces to memory and peripherals and allows flexible
control of I NS8900 operations.
Single-phase 2.0 MHz clock input is easily generated with
a minimum of external components.
1-010
PACE STE
recommended crystal specifications
• AT-cut crystal
•
2.6667 MHz ± 0.1%, fundamental
mode
• 5 mW maximum
• 150 n maximum series resistance
timing diagram
tTDz
TClKO
TClK
tPll
-tPHZ
911%
NClK
OR
NCK
ClK
OR
CK
90%
911%
-tpw
TIMES FOR NCLK, NCK, ClK, AND CK MEASURED AT 111% AND 911%
Figure 2.
1-011
PACE STE
absolute maximum ratings [1]
operating conditions
Supply Voltage (Vee) . . . . . . . . . . . . . . . . . . . 7.0 V
(VGG) . . . . . . . . . . . . . . . . . -15.0 V
Input Voltage . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 V
Storage Temperature . . . . . . . . . . . . -65°C to +150°C
lead Temperature (soldering, 10 seconds) ..... 300°C
dc electrical characteristics
Min.
Max.
4.75
5.25
-11.40 -12.6
+70
0
Supply Voltage (Vee!
(VGG)
Temperature
Units
V
V
°c
(Notes 2 and 3)
Conditions
Parameter
Min.
Typ.
Max.
Units
OUTPUT SPECIFICATIONS:
T ClK, T ClK* (TTL Clocks)
VOH logic "1" Output Voltage
Vee =4.75V
IOH=-1mA
VOL logic "0" Output Voltage
los Output Short Circuit Current
Vee = 4.75V
10L = 32mA
3.65
4.25
V
0.4
V
-55
mA
0.25
(Note 4), Vee = 5.25 V, Vo = 0
-33
-10
CK, NCK, ClK, NClK
-100~A
VOH logic "1" Output Voltage
IOH =
VOL logic "0" Output Voltage
Vee= 4.75V
VGG=-11.4V
Vee = 5.25V
4.5
Vee - 0.9
I
V
10L = 100~
VGG + 0.1
VGG+ 0.25
I
10L = 5mA
VGG + 0.2
VGG
I
V IN=2.4V
+0.5
V
V
INPUT SPECIFICATIONS:
EXTC
V IH
logic "1" Input Voltage
IIH
logic "1" Input Current
V IL
logic "0" Input Voltage
IlL
logic "0" Input Current
VeLAMP
Input Clamp Diode
2.0
I
V
40
~
1.0
mA
VIN = 5.5V
0.8
V
Vee = 5.25V
V IL = 0.4V
-0.9
-1.6
mA
Vee = 4.75V
IlL = -12mA
-0.8
-1.5
V
20
30
mA
-40
-55
mA
POWER SUPPLY CURRENT
lee
Supply Current from Vee
IGG Supply Current from VGG
ac electrical characteristics
Symbol
Vee = 5.25V
VGG = -12.6V
Crystal Freauency at 2.6667 MHz
I
A = O°C to +70°C, Vec - VGG = +17V ± 5%
Limits
Parameter
Min. Typ. Max.
tNOV1' tNov2
tpw
Non-Overlap Time
5
MOS Clocks Pulse Width (NClK, ClK, NCK, CK)
300 320
tR
MOS Clocks Rise Time (NClK, ClK, NCK, CK)
tF
MOS Clocks Fall Time (NCLK, ClK, NCK, CK)
tPH1' tPH2
TTL Clocks to MOS Clocks High Level Delay
tPL1' tPL2
TTL Clocks to MOS Clocks Low Level Delay
tTD1' tTD2
TTL Clock to TTL Clock Delay
tSTART
Time Delay from Last Power Applied to MOS Clocks Stabilized
-40
-25
12
Units
Test
Conditions
ns
See Note 5
ns
See Note 5
40
ns
See Note 5
40
ns
See Note 5
40
ns
See Note 5
80
ns
See Note 5
25
ns
See Note 5
100
ms
See Figure 7
Not..:
1. "Absolute Maximum Ratings" are those values beyond which the safety of the device cannot be guaranteed. They are not meant to imply
that the devices should be operated at these limits. The table of "Electrical Characteristics" provides conditions for actual device operation.
2. Unless otherwise specified, minImax limits apply across the O°C to +70°C temperature range and VCC = 4.75 V to 5.25 V, VGG = -11.4 V
to -12.6 V power supply range. All typical, are given for Vce = 5.0V, VGG = -12V, and TA = +25°C.
3. All currents into device pins are shown as positive; currents out of davice pins are shown as negative. All voltages are references to ground
unless otherwise noted.
4. Only one output at a time should be shorted.
5. The test conditions for measuring AC parameters are shown in Figures 2 and 3, with C1 = C2 = 60 pF, C3 = 80 pF, CNOV = 60 pF. Load
conditions for MOS clocks and TTL clocks are shown in Figures 4 and 5. Including probe and jig capacitance, CL 1 = 20 to 80 pF, and
CL2 =40pF.
1-012
PACE STE
test conditions
TCLK*, TCLK LOAD
NCLK, NCK, CLK, CK LOAD
OUTPUT UNDER
TEST
VCC
L".
OUTPUT
UNDER
TEST
RL = 390n
r'u"
Cl = C2 = 60pF, C3 = 80pF, CNOV = 60pF*
* ALL CAPACITD RS ARE ±5%
Figure 5.
Figure 4.
Figure 3.
typical characteristics
TYPICAL NON·OVERLAP TIME VS.
NON-OVERLAP CAPACITOR
80
VC~=5 J
VGG=-12V
CLI = 80pF
TA=25°
./V
0
/
0
0
0
0
V
25
V
50
V
75
VCC
VGG
l/
NCLK,
CLK,
NCK,
OR CK
100
125 150 175 200
..--ISTART_
CNOV(Pf)
NDN·OVERLAP CAPACITANCE
Figure 7.
Figure 6.
1-013
PACE BTE/S
absolute maximum ratings (Note 1)
Supply Voltage
7V
Input Voltage (All Inputs Except MBIIO Input Active)
5.5V
Output Voltage
5.5V
±10mA
MaS Bus Input Current
--65°C to +150°C
Storage Temperature
Lead Temperature (Soldering, 10 seconds)
300°C
dc electrical characteristics
recommended operating conditions
MIN
4.75
0
Supply Voltage (VCC)
Temperature (T A)
UNITS
V
°c
MAX
5.25
+70
(Notes 2 and 3)
PARAMETER
TTL BUS PORT (BDI/O 00-07)
I
CONDITIONS
I
MIN
I
TYP
I
MAX
VIH
Logical "1" Input Voltage
VIL
Logical "0" Input Voltage
VOH
Logical "1" Output Voltage
WBO* = O.SV,
MBI/O = 0.5 mA
10H = -5.2 mA
VOL
Logical "0" Output Voltage
WBO* = O.BV,
10l = 20 mA
0.25
0.4
MBI/O = 100pA
10L = 50 mA
0.4
0.5
2.0
VCC-l.l
2.4
UNITS
V
O.S
10H=-1 mA
I
V
V
VCc-O·S
3.7
V
V
V
lOS
Output Short Circuit Current
WBO* = O.BV, MBI/O = 0.5 mA,
IIH
Logical "1" Input Current
WBO* = 2V, VIH = 2.4V
SO
pA
II
Input Current at Maximum
WBO* = 2V, VIH = 5.5V,
1
mA
/J. A
-10
-35
-75
mA
VOUT = OV, VCC = 5.25V, (Note 4)
Input Voltage
VCC = 5.25V
IlL
Logical "0" Input Current
WBO* = 2V, VIL = 0.4V
-10
-250
VCLAMP
Input Clamp Voltage
WBO* = 2V, liN = -12 mA
--0.2
-1.5
100
Outputllnput Bus Oisable Current
WBO* = STR* = 2V, BOliO = 0.4V
V
-BO
SO
/J. A
-5.0
0.10
mA
5.0
mA
O.S
V
to 4V, VCC = 5.25V
MOS BUS PORT (MBI/O 00-07)
10
Logical "0" Input Current
WBO* = O.BV, 10L(TTL) = 50 rnA,
VOL::; 0.5V, (Note 5)
11
Logical "1" Input Current
WBO* = O.BV, 10H(TTL) = -1 rnA,
0.50
VOH ~VCC-l.1V, (Notes 5 and 6)
Vo
Logical "0" Input Voltage
WBO* = O.BV, 10L(TTL) = 50 rnA,
VOL ::;0.5V
Vl
Logical "1" Input Voltage
WBO* = O.BV, 10H(TTL) = -1 rnA,
2.0
1.5
V
2.4
3.3
V
VOH ~ VCC - 1.1V
VOH
Logical "1" Output Voltage
WBO* = CEl = BOliO = 2V,
IOH(MOS) = -1 mA, CE2* =
STR* = O.BV
VOL
Logical "0" Output Voltage
WBO* = CEl = 2V, 10L(MOS) =
5 rnA, CE2* = STR* = BOliO = O.BV
lOS
Output Short Circuit Current
WBO* = CEl = BOliO = 2V,
0.2B
-7
-15
0.5
V
-45
mA
VCC = 5.25V, VOUT = OV,
STR* = CE2* = O.BV, (Note 4)
VCLAMP
Input Clamp Voltage
liN = -12 rnA
100
Outputllnput Bus Oisable Current
MBI/O = O.4V to 4V, VCC = 5.25V
-BO
-1.5
V
BO
/J.A
CONTROL INPUTS (WBD*, CE1, CE2*, STR*)
VIH
Logical "1" Input Voltage
VIL
Logical "0" Input Voltage
2.0
V
IIH
logical "1" Input Current
VIN = 2.4V
20
pA
11
Input Current at Maximum
VIN = 5.5V
1.0
mA
O.S
Input Voltage
1-014
V
r
PACE BTE/S
--
dc electrical characteristics
PARAMETER
(Continued)
(Notes 2 and 3)
I
CONTROL INPUTS (WBD*, CE1, CE2*, STR*)
I
CONDITIONS
MIN
I
TYP
I
MAX
I
UNITS
(continued)
IlL
Logical "0" Input Current
V,N
VCLAMP
Input Clamp Voltage
liN
= O.4V
-250
-400
JiA
= -12 rnA
--0.85
-1.5
V
= 5.25V
70
110
rnA
POWER SUPPLY CURRENT
ICC
Power Supply Current
VCC
Note 1: "Absolute Maximum Ratings" are those values beyond which the safety of the device cannot be guaranteed. They are not meant to
imply that the devices should be operated at these limits. The table of "Electrical Characteristics" provides conditions for actual device operation.
Note 2: Unless otherwise specified, minImax limits apply across the O°C to +70°C temperature range and the 4.75V to 5.25V power supply range.
All typicals are given for VCC = 5V and T A = 25°C.
Note 3: All currents into device pins are shown as positive, out of device pins are negative. All voltages are referenced to ground unless otherwise
noted.
Note 4: Only one output at a time should be shorted.
Note 5: The MBI/O Input Characteristic Graph illustrates this parameter and defines the regions of guaranteed logical "0" and logical ", .. outputs. See equivalent input structure for clarification. When the MBI/O input is loaded with a high impedance source (open), the TTL output will
be in the logic "0" state.
Note 6: The maximum MOS bus positive input current specification is intended to define the upper limit on guaranteed input clamp operation.
At higher input currents (up to the absolute maximum rating) clamp operation is not guaranteed but TTL bus logic state is valid and no device
damage will occur.
Note 7: In most applications the MOS bus data lines are higher impedance and more sensitive to noise coupling than TTL bus lines. Conservative
design practice would dictate routing MOS bus lines away from high speed, low impedance TTL lines and MOS clock lines or providing a ground
shield when they are adjacent.
ac electrical cha racteristics
VCC
= 5V
±5%, T A
PARAMETER
= o°c to
+70°C
CONDITIONS
MIN
TYP
MAX
UNITS
DATA TRANSFER SPECIFICATIONS
Receiving Mode (BOliO Bus to MBIIO Bus)
WBD*
= 3V,
RL = 1
kn,
= 15 pF,
CL
(Figures 4 and 6)
= CEl = OV,
STR* = CE2* = 3V,
CL = 50 pF, R L = 100 n,
Driving Mode (MBI/O Bus to
WBD*
BOliO Bus)
tpdO
17
40
tpd1
20
40
ns
tpdO
40
60
ns
tpd1
40
60
ns
ns
(Figures 3 and 5)
TRANSCEIVE~
MODE SPECIFICATIONS
Select Bus
tDS
Chip Enable Data Set-Up
(Figure 1)
45
tDH
Chip Enable Data Hold
(Figure 1)
0
ns
tES
Set-Up
(Figure 1)
0
ns
23
ns
TTL Data Bus (BOlIO 00-07)
= 100 n, (Figure
tBO 00
Bus Data Output Disable
CL
=5
tBO OE
Bus Data Output Enable
CL
= 50 pF,
tBO IE
Bus Data Input Enable
(Figure 1)
30
ns
tBO 10
Bus Data Input Disable
(Figure 1)
30
ns
pF, RL
RL
1)
= 100 n, (Figure
5
1)
20
50
ns
25
80
ns
MOS Data Bus (MBI/O 00-07)
tMB 00
MOS Bus Output Disable
CL
= 15 pF,
RL
= 1 kn.
(Figure 1)
tMB OE
MOS Bus Output Enable
CL
= 15 pF.
RL
= 1 kn.
(Figure 1)
tMB 10
MOS Bus Input Disable
(Figure 1)
55
ns
tMB IE
MOS Bus Input Enable
(Figure 1)
20
ns
15
50
100
ns
50
100
ns
Select Bus
tCLR
Clear Previous Chip Enable
I
I
(Figure 2)
1-015
I
25
I
50
I
ns
PACE STE/S
switching' time waveforms and ac test circuits
STR*
C E Z · - - - - -.......
CEI-------'
WID'
TTL DATA
IUS
--------------+-
.MOS DATA
BUS
FIGURE 1
STR'~1I%
CEZ' 3V
CEIOV----------------~-----------------------------------------
t.L.:---1
WID'
TTL DATA
BUS
----------+~
MUS OATA
BUS
----------~
FIGURE 2
VCC=SV
INPUTIOV~
SV
SV
WAVEFORM
(NOTE I)
O.'PF
OV
INPUT
PdltpdO
VOH
OUTPUT
~
1.5V
VOL
100
T
OUTPUT
I.SV
ALL DIODES
ARE IN31114
FIGURE 3. BOlIO Bus
*This input network simulates the actual drive characteristic of the PACE outputs
FIGURE 5. MBIIO to BOlIO ac Loads
VCC =sv
INPUT3V~
1.5V
I.SV
WAVEFORM
(NOTE 1)
OV
'tpdO
VOH
~UTPUT
~
1.5V
VOL
I.SV
ALL DIODES
ARE IN3064
FIGURE 4. MBIIO Bus
FIGURE 6. BOlIO to MBIIO ae Loads
Note 1: Freq = 1 MHz, duty cycle = 50%, tR = tF :s:. 10 ns (refer to Figures 5 and 6).
Note 2: All capacitance values include probe and jig capacitance (refer to Figures 5 and 6).
1-016
PACE BTE/S
typical performance characteristics
MOS Bus Input Characteristic
MOS Bus Input Characteristic
MOS Bus Voltagl! Threshold
0.8
0.6
C
Vee = 5V
STANDARD
OUTPUT
LOAj'NG
C
oS
oS
I-
z
0.4
IZ
02
u
IC
IC
IC
IC
=
u
l...=
,
---
TA =1 70'e
=
...=
I-
TA = ode
z
z
f-TA ~ 25'e
'I
I
I
I
I
o L-...l-.....~::t:::=:L____L-...J
-0.2
0.5
1.5
1.0
2.0
2.5
3.0
o
0.5
1.0
INPUT VOLTAGE (VI
MOS BUS
1.5
2.0
2.5
MOS Bus Current Threshold
2.0
:::>
I-
1.8
'"
I
0.5
1.1
220
1.2
1.3
1.4
30 1--1--1r--~f-H'----!---l
TA =25'e
20
1.0
:::>
I
Vee 15V
STANDARD
OUTPUT
LOADING
1.5
1.6
OL--J.L.oCLI...-_I...----'I....----I
o 0.1 02 0.3 0.4 0.5
1.7
INPUT VOLTAGE (V)
ITLBUS
INPUT CURRENT (IlA)
MOS BUS
High-Level Output Voltage
vs Output Current TTL Bus
lOW·lEVEl OUTPUT VOLTAGE (V)
TIL BUS
Output Current vs Output Voltage
High Impedance State TTL Bus
Vee = 5V
TA = 25'C
1.7
40 r--r--1i--1i---J'Hi----l
I
-
1.5
I-
!'I..
1.6
2.5
~
200
1.5
50
I
TA = 70'e
to
lBO
1.4
Low-Lavel Output Voltage
vs Output Current TTL Bus
TA = o'e
~
160
1.3
INPUT VOLTAGE (VI
MOS BUS
3.0
140
1.2
TTL Bus Threshold Characteristic
3.5
120
3.0
INPUT VOLTAGE (V)
MOS BUS
Low-Lavel Output Voltage
vs Output Current MOS Bus
10
c
oS
IZ
'"I'\.r-...
en
:::>
II:
II:
=
U
I:::>
~
'\
I\.
o
-5
-10
O'--_ILU._L...----'L...----'L...----I
o
-10
-3~
-20
-40
-1
HIGH LEVEL OUTPUT CURRENT (rnA)
TTL BUS
4.0
3.5
to
~
...=
3.0
>
2.5
I-
2.0
'"
1.5
r""\.
""
I-
Output Current vs Output
Voltage High Impedance
State MOS Bus
:::>
;
%
to
i:
Vee ~ 5V
TA=25'C-
,
10
c
oS
IZ
IC
IC
:::>
I\.
I:::>
'\.
1.0
'"=
I-
r\.
0.5
o
0.1
-5
'\.
o
-4
-8
-12
-16
-20
-2
HIGH LEVEL OUTPUT CURRENT (rnA)
MOS BUS
OUTPUT VOL TAGE (V)
MOS BUS
1-017
0.2
0.3
0.4
lOW lEVEL OUTPUT VOLTAGE (V)
MOS BUS
OUTPUT VOLTAGE (V)
ITlBUS
High-Level Output Voltage
vs Output Current MOS Bus
~
o
10
0
0.5
Chapter 2
THE GENERAL INSTRUMENT CP1600
The CP1600 and the TMS 9900 were the first two NMOS 16-bit microprocessors commercially available. Even
a superficial inspection of the CP1600 shows it to be more powerful than the National Semiconductor PACE (or 8900).
yet the CP1600 is not widely used. This is because General Instrument does not support the CP1600 tothe extent
that National Semiconductor originally supported PACE. or most manufacturers support their 8-bit microprocessors.
General Instrument's marketing philosophy has been to seek out very high-volume customers; General Instrument supports low-volume customers only to the extent that this support would not require substantial investment on
the part of General Instrument.
From the viewpoint of the low-volume microprocessor user. General Instrument's marketing philosophy is unfortunate.
The CP1600 is an ideal microprocessor for the more sophisticated video games that are appearing. and its rich instruction set and capable architecture make it an ideal choice for data processing terminals and home computer systems.
However. due to its limited support. potential low-volume CP1600 customers are likely to choose another equally capable product.
Three CP1600 parts are available, differentiated only by the clock speeds for which they have been designed.
The CP1600 requires a 3.3 MHz. two-phase clock and generates a 600 nanosecond machine cycle time.
The CP1600 requires a 4 MHz. two-phase clock and generates a 500 nanosecond machine cycle time.
The CP161 0 requires a 2 MHz. two-phase clock and generates a 1 microsecond cycle time.
In addition to the CP1600 microprocessors themselves, the CP1680 Input/Output Buffer (lOB) is described in
this chapter. Additional support devices for the CP1600 may be found in An Introduction to Microcomputers:
Volume 3 - Some Real Support Devices.
The sole sou rce for the CP 1600 is:
GENERAL INSTRUMENT
Microelectronics Division
600 West John Street
Hicksville. New York 11802
There is no second source for the CP1600. General Instrument has a policy of discouraging second sources for its
product line.
The CP1600 is fabricated using NMOS ion implant LSI technology; the device is packaged as a 40-pin DIP.
Three power supplies are required: +12V. +5V and -3V.
THE CP1600 MICROCOMPUTER SYSTEM OVERVIEW
Logic of our general microcomputer system which has been implemented by the CP1600 CPU is illustrated in
Figure 2-1.
Observe that the CP1600 requires external logic to create its various timing and clock signals.
Some bus interface logic is shown as absent because a number of devices must surround the CP1600; these include:
1)
An address buffer. since data and addresses are multiplexed on a single 16-bit bus.
2)
Buffer amplifiers to provide the power required by the type of memory and I/O devices that will normally be connected to a CP1600 CPU.
3)
4)
A one-of-eight decoder chip to create eight individual control Signals out of three controls output by the CP 1600.
A one-of-sixteen multiplex chip to funnel sixteen external status Signals into the CP1600 if using external
branches.
2-1
Were you to compare Figure 2-1 with an equivalent figure for a low-end microprocessor such as the SC/MP (which is
described in Chapter 3 of the Osborne 4 & 8-Bit Microprocessor Handbook{Osborne/McGraw-Hill, 1980), the CP1600
might appear to offer fewer logic functions; but within the functions it does provide, the CP1600 provides considerably
more logic and program execution capabilities. Where low-end microprocessors choose to condense, onto a single
chip, simple implementations of different logic functions, high-end products such as the CP1600 choose to provide
more devices - with greater capabilities on each device.
Clock Logic
o
Interface Logic
Programmable
Timers
I/O Ports
Figure 2 -1. Logic of the CP 1600 CPU and CP 1680 1/0 Buffer
2-2
CP 1600 CPU
CP1680 I/O Buffer
CP1600 PROGRAMMABLE REGISTERS
The CP1600 has eight 16-bit programmable registers. which may be illustrated as follows:
RO
R1
R2 } Deta Counters
R3
R4 } Data Counters with
auto-increment
R5
Stack Pointer
R6
Program Counter
R7
General Purpose registers
The way in which the registers illustrated above are used is unusual when compared to other microcomputers described in this book. All eight 16-bit registers can be addressed as though they were general purpose registers;
however. only Register RO has no other assigned function. We may therefore look upon Register RO as the Primary Accumulator for this CPU.
Registers R1. R2. and R3 serve as general purpose registers. but may also be used as Data Counters.
In addition to serving as general purpose registers. R4 and R5 may be used as auto-incrementing Data Counters.
Memory reference instructions that identify Register R4 or R5 as holding the implied memory address will cause the
contents of Register R4 or R5 to be incremented - after the memory reference instructions have completed execution.
Registers R6 and R7. in addition to being accessible as general purpose registers. also serve as a Stack Pointer and a
Program Counter. respectively.
Having the Stack Pointer accessible as a general purpose register makes it quite simple to maintain more than one
Stack in external memory; also. you can easily address the Stack as data memory using the Stack Pointer as a Data
Counter.
Having the Program Counter accessible as a general purpose register can be useful when executing various types of
conditional branch logic.
While having the Stack Pointer and the Program Counter accessible as though they were general purpose registers
may appear strange. this is a feature of the PDP-11 minicomputer - and is a very powerful programming tool.
CP1600 MEMORY ADDRESSING MODE
The CP1600 addresses memory and 1/0 devices within a single address space.
When referencing external memory. you can use direct addressing. implied addressing. or implied addressing
with auto-increment.
Direct addressing instructions are all two or more words long. where the second or last
word of the instruction object code provides a 16-bit direct address.
CP1600 DIRECT
ADDRESSING
CP 1600 direct addressing instructions are complicated by the fact that CP 1600 program
memory is frequently only 10 bits wide. That is to say. even though the CP1600 is a 16-bit microprocessor. its instruction object codes are only 10 bits wide. If program memory is only 10 bits wide. then direct addresses will only be 10
bits wide. A 1O-bit direct address will access the first 1024 words of memory only.
2-3
Were you to implement a 16-bit wide program memory. then you could directly address up to 65.536 words of memory; however. six bits of the first object program word for every instruction in program memory would be wasted. This
may be illustrated as follows:
Program
Memory
15
o
10-9
~
Bit Number
Three memory
....1------"'"'7"- reference
instructions
that specify
direct addressing
Six unused
bits in each·
of these
memory locations
Two single
word instructions
Instructions that reference memory using implied addressing identify general purpose
Register R1, R2, or R3 as containing the implied address.
A memory reference instruction which identifies Register R4 or R5 as providing the external
memory address will always cause Register R4 or R5 contents to be incremented following the
memory access; thus you have implied memory addressing with auto-increment.
CP1600
IMPLIED
ADDRESSING
Memory reference instructions that specify implied memory addressing via Register 1, 2, 3, 4, or 5 can access
8-bit memory. An SDBD instruction executed directly before a valid memory reference instruction forces the memory
reference instruction to access memory one byte at a time. If implied memory addressing via Register 1. 2. or 3 is
specified. then the same byte of memory will be accessed twice. For an instruction that loads the contents of data
memory into Register RO. this may be illustrated as follows:
Memory
}
SOBD
Rol
Program memory
MVI R1.RO
yy
PPQQ
2-4
XXYY
}
Data memory
If Register R4 or R5 provides the implied memory address for the instruction which follows an SDBD instruction. then
the implied memory address is incremented twice. and two sequential low-order bytes of data are accessed. For an instruction which loads data into Register RD. this may be illustrated as follows:
Memory
RO
R5
Program memory
...
~--~---
I
PP
Data memory
"
The SDBD instruction may also precede an immediate instruction. Now the immediate data will be fetched from the
low-order byte of the next two sequential program memory locations. This may be illustrated as follows:
Memory
MVII XXYY.RO
Without the preceding SDBD instruction. an immediate instruction will access the next single program memory word
to find the required immediate data. Ten or more bits of immediate data will be accessed. depending on the width of
program memory words.
The CP1600 has no Stack reference instructions such as a Push or Pull; rather, a variety of
CP1600
STACK
memory reference instructions can identify Register R6 as providing the implied address.
When Register R6 provides the implied address. it is treated as an upward migrating Stack
ADDRESSING
Pointer. When a memory write operation specifies Register R6 as providing the implied memory
address. Register R6 contents will be incremented following the memory write. A memory read instruction that
specifies Register R6 as providing the implied memory address will cause the contents of Register R6 to be decremented before the read operation occu rs.
An unusual feature of the CP1600 is the fact that a variety of secondary memory reference instructions can also
reference memory via the Stack Pointer. When these instructions are executed. Register R6 contents are decremented before the memory access occurs - as though a Pull operation from the Stack were being executed.
Logically. Register R6. the Stack Pointer. is being handled as though it were a Data Counter with post-increment and
pre-decrement.
2-5
Jump instructions use direct memory addressing. Jump instructions are all three words long. The direct address is
computed from the second and third memory words as follows:
98765432
0
0
0
0
0
x
X
A
A
A
B
B
B
B
B
0
A
B
0
0
1
0
0
A
A
y
y
B
B
B
B
--
--
-
~
JR or JSR
Word 2
Word 3
AAAAAABBBBBBBBBB Jump address (binary)
are enable/disable bits for interrupts
xx identify the register where the return address will be stored for JSR
xx and yy are described in detail in Table 2-4.
yy
You can enable or disable interrupts whenever you execute a Jump or Jump-to-Subroutine instruction.
The only difference between a Jump instruction and a Jump-to-Subroutine instruction is that the Jump-to-Subroutine
instruction saves the Program Counter contents in Register 4, 5, or 6. The two high-order bits (xx) or the second Jumpto-Subroutine object code word specifies which of the three registers will be used to hold the return address.
Jump-to-Subroutine instructions, like the Jump instruction, allow direct memory addressing only.
CP1600 STATUS AND CONTROL FLAGS
The CP1600 CPU has four of the standard status flags; in addition, it has some unusual control signals.
These are the four standard status flags:
Sign (5), This status is set equal to the high-order bit of any arithmetic operation result.
Zero (Z). This status is set to 1 when any instruction's execution creates a zero result. The status is set to 0 for a nonzero
result.
The Carry (C) and Overflow (0) statuses are standard carry and overflow, as described in Volume 1.
Four control signals (EBCAO - EVCA3) are output during a Branch-on-External (BEXT) instruction. The'se four signals are output to reflect the low-order four bits of the SEXT instruction's object code. External logic receives these four
signals and (depending on their state), mayor may not return a high input via ESC!. If ESC I is returned high, then the
SEXT instruction will perform a branch: if ESCI is returned low. then the SEXT instruction will cause the next sequential
instruction to be executed. The four control signals ESCAO - EBCA3 therefore provide the CP1600 with a means of testing 16 external conditions.
CP1600 CPU PINS AND SIGNALS
CP1600 CPU pins and signals are illustrated in Figure 2-2.
DO - 015 is a multiplexed Address and Data Bus. Given a total of 40 pins in a package, CP1600 designers have been
forced to share 16 pins between addresses and data. Three control signals, BDIR, BC1, and BC2, identify the traffic
on the Address/Data Bus. External logic (one MSI chip) must decode these three signals to create eight control
signals, as summarized in Table 2-1.
Remaining signals may be divided into four groups: timing, status/control, interrupt, and OMA.
Two timing clock signals are required: <1>1 and <1>2. These are complementary clock signals which may be illustrated
as follows:
<1>1
--1
\
I
\
I
<1>2
\ " , , _ _~I
\
I
\~_.,Jr
2-6
L
ESCI
1
MSYNc'
2
BCl
BC2
3
4
BDIR
5
6
015
014
013
012
9
10
011
40
39
CPl600
11
12
13
14
010
09
OS
00
01
07
15
16
D6
17
06
04
18
19
03
20
CPU
PCiT
GNO
«111
«112
38
37
36
YeO
35
34
33
VBB
VCC
BOROY
32
31
STPST
30
29
HALT,
28
27
26
25
24
23
22
21
BUsRQ
BUSAK
iNiR
iNTRM
TCI
EBCAO
EBCAl
EBCA2
EBCA3
02
Pin Name
Description
Type
00-015
Data and Address Bus
Bus control signals
Clock signals
Master Synchronization
External branch condition address lines
Tristate. Bidirectional
BDIR. BCl. BC2
cl»1. «112
MSYNC
EBCAO - EBCA3
ESCI
PCiT
External branch condition input
Program Counter inhibit/software
interrupt signal
BDROY
WAiT
STPST
CPU stop or stan on high-to-Iow transition
Halt state signal
Interrupt request lines
Terminate current interrupt
HALT
iNTR. iNTRM
TCI
ijSiiQ
Bus request
External bus control acknowtedge
"BUSAK
VB8.VCC.VOO.GNO
Output
Input
Input
Output
Input
Input
Input
Input
Output
Input
Output
Input
Output
Power and Ground
Figure 2-2. CP1600 CPU Signals and Pin Assignments
MiYNC is a somewhat unusual signal. as compared to other microcomputer clock signals in this book. Following
powerup. MSYNC must be held low for at least 10 milliseconds. On the subsequent riSing edge of MSYNC. logic internal to the CP1600 CPU will synchronize the <1>1 and <1>2 clock signals to start a new machine cycle. Most of the CPU
devices we have described in this book use a reset signal. or have internal powerup logic which performs this clock
synchronization.
Now consider the status and control signals.
First of all. there are the four control outputs which we have already described: EBCAO - EBCA3. There is one conditional Branch instruction (BEXT) which will only branch if a high signal is input via EBCI. When the BEXT instruction is executed. the low-order four BEXT instruction object code bits are output via EBCAO - EBCA3. External
logic is supposed to decode these four signals by whatever means are appropriate - and thence determine whether
ESCI should be input high or low. A high input. as we have just stated. will result in a branch: a lowinput will cause the
next sequential instruction to be executed.
In reality. there is no connection within CP1600 CPU logic between the EBCI input and the four EBCAO - EBCA3 outputs. So far as external logic is concerned. the execution of a BEXT instruction is identified by signal levels output and
maintained on the EBCAO - EBCA3 outputs. while the EBCI input determines whether a branch will or will not occur.
How external logic chooses to determine whether EBCI will be set high or low is entirely up to external logic. The only
vital function served by ESCAO - EBCA3 is to identify the instant at which a BEXT instruction is executed.
Another unusual control signal provided by the CP1600 is PCIT; this is a bidirectional Signal. When input low. this
Signal prevents the Program Counter from being incremented following an instruction fetch. This Signal is also output
as a low pulse following execution of a software interrupt instruction. Instruction timing separates the active input and
2-7
active output of this signal: providing external logic adheres to timing requirements, a conflict between input and output logic will never arise.
BDRDY is equivalent to the WAIT signal we have described for a number of other microcomputers. BDRDY is input low by any external logic which requires more time in order to respond to an 1/0 access. Recall that the CP1600
uses a single address space to reference memory or I/O devices. The 'SDR]jy signal causes the CPU to enter a Wait
state for as long as B"i5'RI)V is being input low: however, during the Wait state CPU logic is not refreshed. Thus a Wait
state cannot last for more than 40 microseconds, or the contents of internal CPU locations will be lost.
STPST, a Halt/Reset input, is an edge-triggered signal. When external logic inputs a high-to-Iow transition via STPST,
the CPU will complete execution of any interrupt instruction, then will enter a Halt state and output HALT high. If a
non-interruptable instruction is being executed, then the Halt state will not being until completion of next interruptable
instruction's execution. The Halt state will last until external logic inputs another high-to-Iow STPST transition, at
which time the Halt output will be returned low and normal programming execution will continue. Execution of the
HL T instruction also causes the CP1600 to enter a Halt state, as described above.
Let us now look at interrupt signals.
The CP1600 has two interrupt request inputs -INTR and INTRM. INTR has higher priority than INTRM. INTR cannot be disabled. Typically, TNiI1 will be used to trigger an interrupt upon power failure or other catastrophes.
The interrupt acknowledge signal is created by external logic which must decode the BC1, BC2, and BDIR signals, as shown in Table 2 -1. Observe that there are, in fact. two interrupt acknowledge signals: the first (I NT AK)
acknowledges the interrupt itself, while the second (DAB) is used as a strobe for external logic to return an interrupt address vector. The interrupt sequence is described later in this chapter.
The CP1600 has two additional interrupt-related signals which are unusual when compared to other microcomputers
described in this book.
TCI is output high when an End-of-Interrupt instruction is executed. This signal makes it easy for external log ic to
generate interrupt priorities which extend across the execution of an interrupt service routine.
Table 2-1. CP1600 Bus Control Signals
BC1
BC2
BDIR
SIGNAL
FUNCTION
0
0
0
NACT
The CPU is inactive and the Datal Address Bus is in a high impedance state.
0
0
1
BAR
A memory address must be input to the CPU via the DatalAddress
Bus.
0
1
0
lAB
Acknowledged external interrupt requesting logic must place the
starting address for the interrupt service routine on the Address Bus.
0
1
1
DWS
Data write strobe for external memory.
1
0
0
ADAR
This signal identifies a time interval during which the DatalAddress
Bus is floated, while data input on the Data Bus is being interpreted
as the effective memory address during a direct memory addressing
operation.
1
0
1
DW
The CPU is writing data into external memory. DW will precede
DWS by one machine cycle.
1
1
0
DTB
This is a read strobe which external memory or I/O logic can use in
order to place data on the Datal Address Bus.
1
1
1
INTAK
This is an interrupt acknowledge signal. It is followed by lAD which
is a strobe telling the external logic which is being acknowledged to
identify itself by placing an address vector on the DatalAddress Bus.
2-8
MC
T1
T2
MC
T3
T4
T2
T1
T3
T4
BC 1,BC2,BDIR
OO-D15
---51~-+-----l~-
Undefined
state
preceding
data output
Data
Output
Data
Input
Figure 2-3. CP1600 Machine Cycles and Bus Timing
BAR
NACT
MC2
MC1
T3
T2
T1
T4
T1
T2
T3
DTB
MC3
T4
T1
T2
T3
<1>1
BC1
BC2
~
____ __ __________-+______________J
~
~
BDIR
DO-D15---~
Instruction
address out
Instruction
object code in
Figure 2-4. CP1600 Instruction Fetch Timing
2-9
T4
INSTRUCTION FETCH
,
BAR
MCl
I
I
MEMORY READ
NACT
MC
I
I
DTB
M~
I
I
I
Tl: T2 :T3:-T4 Tl: T2 :T3: T4 T1\ T21T3
I
I
I
Instruction
address out
I
I
I
"
BAR
MCl
NACT
lT4 Tl\ T2: T3; T4 Tl,:T21:T31'T4
I
I
I
I
I
I
Instruction
object code in
I
I
I
I
NACT
MC2
I
DTB
MC3
I
I
I
I
I
I
I
I
I
I
I I
I
Tli T2, T3, T4 Tl'T2, T3 ,T4
,
'"
Data address out
Data in
Figure 2-5. CP 1600 Timing for Memory Read I nstruction with Implied Memory Addressing
CP1600 INSTRUCTION TIMING AND EXECUTION
CP1600 instructions are executed as a sequence of machine cycles. Each machine cycle has four clock periods,
as illustrated in Figure 2-3. Machine cycles are identified by their cycle number and by the levels of the BC1. BC2.
and BDIR signals. Each of the eight level combinations is given a name. taken from Table 2-1. This name becomes the
name of the machine cycle. Thus in Figure 2-4. and in subsequent instruction timing illustrations. each machine cycle is identified by a signal name from Table 2-1.
Figure 2-3 shows general case timing for data output or input on the Data/Address Bus. In between data input or output operations the bus is floated.
CP1600 MEMORY ACCESS TIMING
Figure 2-4 illustrates instruction fetch timing for a CP1600 instruction's execution. Three machine cycles are required. During the first mac.hine cycle an address is output. Nothing happens during the second machine cycle: it is a
"time spacing" machine cycle that routinely separates two CP1600 Bus access machine cycles. The object code for the
accessed instruction is returned during the third machine cycle.
Figure 2-5 illustrates timing for the simplest memory read instruction's execution. In this .case the data memory
address is taken from one of the CPU registers. There is no difference between timing for the three machine cycles of an
instruction fetch or a data memory read. As illustrated in Figure 2-5. a simple memory read instruction's execution
consists of two three-machine cycle memory read operations. separated by a spacing no operation machine cycle.
2-10
MEMORY WRITE
INSTRUCTION FETCH
BAR
MCl
I
I
DTB
MC3
NACT
MC2
,
I
I
I
I,
BAR
MCl
NACT
I
I
1
I
I
I.
NACT
MC2
I
I
DWS
MC4
DW
MC3
I
I
I
I
I
I
I
I
.,
I
I
I
"
I
I
I
I
,I I I ,I
I I
I
Tl :T2:T3:T4 Tli T2,T3.T4 Tl, T2, T3. T4 Tl, T2, T3,T4 Tl,T21 T3 1T4 Tl, T2, T3, T4 Tl, T2. T3, T4 Tl, T21 T3.T4
,
,.
I
I
I
I
I
'I
•
I
I'
I I
I
I
I
I
Data out
Data address out
Instruction
object code in
Instruction
address out
Figure 2-6. CP1600 Timing for Memory Write Instruction with Implied Memory Addressing
Figure 2-6 illustrates timing for a simple CP1600 memory write instruction execution. Data is output for two
machine cycles. giving external logic ample time to respond to the data output. External logic uses the DWS machine
cycle as a write strobe.
Any memory reference instruction that specifies direct memory addressing will require one three-clock-period machine
cycle to fetch each word of the instruction object code: an NACT clock period will separate each machine cycle. After
the first instruction fetch machine cycle. an ADAR-NACT clock period combination will be inserted in the second (and
third. if present) instruction fetch machine cycle. During an ADAR clock period. BC1 is high. while BC2 and BDIR are
low. No other control signals are active. Thus. for a two-word memory read or memory write instruction that
specifies direct addressing, the following clock periods and machine cycles will be required for instruction execution:
Direct Addressing
Memory Read
Machine Cycles
Direct Addressing
Memory Write
Machine Cycle
BAR }
Fetch first instruction
{ BAR
NACT "~-------object code word ------t.~ NACT
DTB
DTB
~NACT
NACT ..4.-----Spacing machine c y c l e - - - - - -...
~~~~}
NACT
DTB
instruction------t.~{ ~~~~
..'4.----Fetch second
object code word
NACT .......- - - - - Spacing machine
BAR }
Memory read
NACT .....---machine cycle
DTB
cycle------.~
NACT
DTB
NACT
Memory write
{ BAR
machine cycle---__
.~ NACT
DW
DWS
2-11
BAR
NACT
NACT
<1>2
BC1
-----------------+------------
BC2
BOIR
BOROY
Figure 2-7. CP1600 Wait State Timing
THE CP1600 WAIT STATE
The CP1600 has a Wait state equivalent to those described for other microcomputers in this book. External logic that
requires more time to respond to an access must input BDRDY low before the end of the BAR machine cycle. during
which an address is output and the device is selected. Timing is illustrated in Figure 2-7.
If you examine Figures 2-4, 2-5 and 2-6. you will see that an address is output during a BAR machine cycle to initiate any external device access. The BAR machine cycle is always followed by an NACT machine cycle; in the middle of
T1 during this NACT machine cycle, the CP1600 samples BDRDY. If ~ is low. then a sequence of NACT machine
cycles occurs. In the middle of T4 for every NACT machine cycle, the CP1600 samples BDRDY again. Upon detecting
BDRDY high, the CP1600 resumes instruction execution with a DTB machine cycle.
A Wait state must last for less than 40 microseconds, since the CP1600 is a dynamic device.
THE CP1600 HALT STATE
The CP1600 has a Halt state which may follow execution of the Halt instruction, or may be initiated by external
logic.
When the Halt instruction is executed. then. following the instruction fetch machine cycle, the HALT signal is output
high and a sequence of NACT machine cycles is executed.
External logic initiates a Halt state by making the STPST input undergo a high-to-Iow transition. Following execution of
the next interruptable instruction. a Halt state begins. The HALT signal is output high and a sequence of NACT
machine cycles is executed.
A Halt state, whether it is initiated by execution of a Halt instruction or by a high-to-Iow transition of STPST. must be
terminated by a high-to-Iow transition of STPST. This will cause the Halt state to end at the conclusion of the next
NACT machine cycle. Timing for a Halt state which is initiated and terminated by STPST may be illustrated as follows:
STPST{L-_ _l
HALT
~
~~-------------------------------------------~~--------
Next interruptable
instruction's /
execution
ends here
~ \.
y
)
1
\
HALT STATE
2-12
Next NACT machine
cycle ends here
ThePerFsignal as an input inhibits CP1600 Program Counter increment logic. Thus. external
logic can input PCIT low - in which case the same instruction will be continuously re-executed
until PC IT goes high again. However. PCIT should only change levels while the CPU has been
halted. Thus. PCIT and STPST should be used together as follows:
CP1600
PCIT
SIGNAL
PCIT REQUEST
STPST
PCIT
CP1600 INITIALIZATION SEQUENCE
The CP1600 is initialized by inputting the MSYNC signal low for a minimum of 10 milliseconds after power is
first applied to the CPU. '
MSYNC must make a low-to-high transition. marking the end of the initialization. on a rising edge of the «1>1 clock signal. On the next rising edge of «1>1. instruction execution will begin. This may be illustrated as follows:
I
I
Ii:
I T1 : T2 I T3 I T4 I
<1>,
MSYNC
When instruction execution begins. interrupts are disabled. The following sequence of machine cycles is executed:
NACT
lAB -4--- Read Datal Address Bus and load into Program Counter
NACT
NACT
NACT
BAR-4-- Output Program Counter contents to fetch first instruction
NACT
DTB
etc
During the lAB machine cycle. external logic must supply a 16-bit address at DO - 015. Your external logic must provide this address. which in the simplest case may be 0000 by grounding the bus. or FFFF16 by tying it to +5V following
a startup.
The address which is input at lAB is output at BAR. initiating program execution.
CP1600 DMA LOGIC
CP1600 DMA logic is quite standard. When external logic wishes to transfer data under DMA control, it inputs
BUSRQ low. At the conclusion of the next interruptable instruction's execution, the CPU floats the
Data/Address Bus and enters a Wait state, during which a sequence of NACT machine cycles is executed.
BUSAK is output low at the beginning of the first NACT machine cycle.
The NACT machine cycles that occur during a DMA operation refresh the CPU. NACT machine cycles that occur
during a'Wait state do not refresh the CPU. This means that any number of NACTmachine cycles can occur during a
DMA break. while a Wait state must be shorter than 40 microseconds.
The DMA break ends when external logic inputs BUSRO high again. BUSRO is sampled during T1 of every DMA NACT
machine cycle. When BUSRO is sampled high. two additional NACT machine cycles are executed. then BUSAK is output high and normal program execution resumes.
DMA timing is illustrated in Figure 2-8.
2-13
Last machine cycle
of an interruptable
instruction's
execution
I
I
I
I T4
I
T11
I
NACT
NACT
n
I
I
I T2 I T3 I T4
,I
I
NACT
I
n I
I
T3 I T4
I
I
I
I
I
I
I
NACT
I
I
T11 T2 I T3 I T4
I
1
n
I
I
I
T2 I T3
I
I
BAR
I
I T4
1
BC 1,BC2,BOIR
-------------i~~~~----------~~------~~~~------------~--BUSRQ
BUSAK
Figure 2-8. CP1600 DMA Timing
INTAK
I
I
ow
NACT
I
I
I
I
I
I
I
I
I
I
I
I
OWS
I
'I
I
I.
I
NACT
I
I
I
"
I
NACT X 4
lAB
I
I
I
I
I
I
I
I
I
I
I
I
BAR
T1, T2: T31 T4 T11 T2 I T31T4 nIT2 I T31T4 TlI T2 JT3 1T4 T11 T21 T31T4 TlI T21 T31T4 T11 T2 I T4
1 I
I
I
I
I
1 I
I
I
I
I
I
I
I
I
I
1
I
<1>1
<1>2
\
BC1)
BC2
J
\
\
BOIR}
~~--a
\
I
t
Output
Stack
Pointer
\
I
,
I
0
t
Current Program
Counter contents
written to memory
stack
II
External
logic inputs
starting
address
for interrupt
service routine
Figure 2-9. CP1600 Interrupt Service Routine Initialization
2-14
Start executing
interrupt
service routine
INSTRUCTION EXECUTE/FETCH
INSTRUCTION FETCH
BAR
MCl
DTB
MC3
NACT
MC2
,
I
I
"
I
I'
I
I
I
I
I
I
t
NACT
MC2
BAR
MCl
NACT
I
I
I
I
I
DTB
MC3
I
I
I
I
I
I
I
Tl :T2: T31T4 Tll T2: T3:T4 T11 T21 T3:T4
Tl:T2:T3:T4 T1:T2:T3:T4 Tll, T2 ~I T3:T4 Tl:I T2 lT3 ;T4
I
I
I
I
I
I
I
Instruction
address out
I
I
I
I
Next instruction
address out
TCI instruction
object code in
Next instruction
object code in
Figure 2-10. CP1600Timing for TCllnstruction's Execution
THE CP1600 INTERRUPT LOGIC
The CP1600 uses a vectored interrupt processing system.
External logic requests an interrupt by inputting a low signal at either the INTR or INTRM pins.
Following the execution of the next interruptable instruction. the CP1600 acknowledges the interrupt by pushing
Register R7 contents (the Program Counter) onto the Stack; then the CP1600 outputs 111. followed by 010 at BC1.
BC2. and BOIR. External logic must respond by placing 16 bits of data on the Data/Address Bus. These 16 bits of data
will be loaded into Register R7. the Program Counter. thus causing program execution to branch to an interrupt service
routine dedicated to the interrupt. Timing is illustrated in Figure 2-9.
The PCIT signal is output low following execution of a software interrupt instruction (SIN). This is the only microcomputer described in this book which allows external logic to respond to a software interrupt in this fashion. Allowing external logic to respond to a software interrupt only makes sense when you anticipate your product being used in a
minicomputer-like environment. Typically. the software interrupt will interface to logic of a front panel or console.
When an SIN instruction is executed. a one-machine cycle low PC IT pulse is output.
You may. if you wish. end an interrupt service routine by executing a Terminate Current Interrupt (TCI) instruction. in
which case the TCI signal will be output high.
Timing for TCI is given in Figure 2-10.
Following an interrupt acknowledge. the interrupt service routine must execute instructions in order to disable interrupts and save the contents of registers on the Stack. The exception is Register R7. the Program Counter. which is automatically pushed onto the Stack following an interrupt acknowledge.
External logic is entirely responsible for any type of interrupt priority arbitration which may occur. and for the generation of the interrupt vector address which must be input following an interrupt acknowledge.
2-15
It is quite easy to generate signals equ iva lent to other microcomputer system busses from the CP1600 System Bus.
Therefore, you can use parts described in Volume 3 to handle CP1600 interrupt requirements.
THE CP1600 INSTRUCTION SET
The CP1600 instruction set is relatively straightforward. Addressing modes, which we have already described, are simple, and instructions are typical of those we have seen and described for other microcomputers. Unusual features relating to addressing modes available with individual instructions are summarized in Table 2-2, which describes the
CP1600 instruction set.
If you have never programmed a PDP-11 minicomputer, then you should pay particular attention to programming techniques that result from the Stack Pointer and Program Counter being accessed as general purpose
registers.
A wide variety of Register Operate instructions allow you to compute data and load the result directly into Register R7,
the Program Counter. In effect these become computed Jump instructions.
The ability to manipulate Register R6, the Stack Pointer, as though it were a general purpose register means that it is
easy to maintain a number of different Stacks in external read/write memory.
The Jump-to-Subroutine instruction has a minicomputer flavor to it. Rather than saving the return address on the
Stack, Register R7 contents are moved to General Purpose Register R4 or R5. A number of minicomputers will save a
subroutine return address in a general purpose register in this fashion. The problem with this logic is that you must execute an additional instruction within the subroutine to save the return address on the Stack if you are going to use
nesting subroutines. If you are passing subroutine parameters, however, this is an excellent arrangement for the Jumpto-Subroutine instruction places the address of the parameter list directly in a Data Counter with auto-increment. We
have described the concept of parameter passing in Volume 1, Chapter 7.
Note that the CP 1600 instruction set lacks a logical OR.
In Tables 2-2 and 2-4, instruction length is given in terms of "words" rather than "bytes", as we have done in previous chapters. Since only the lower 10 bits of the CP1600 object code are presently used, system configurations need
not have the full 16-bit word size. Hence a "word" may be 10 to 16 bits wide, depending on the implementation.
The following notation is used in Table 2-2:
ADDR
One word of direct address
condCondition on which a branch may be taken. Table 1-3 lists all 14 branch conditions.
One word of immediate data.
DATA
DISP
One word displacement. See Table 2-4 for location of sign bit.
External branch condition.
E
EBCAO-3
The external branch condition address lines: EBCAO, EBCA 1, EBCA2, and EBCA3.
EBCI
The external branch condition input line.
LABEL
A 16-bit direct address, target of a Jump instruction. See Table 2-4 for the bit format.
PC IT
The software interrupt output line.
RB
General Purpose Register R4, R5, or R6.
RD
One of the general purpose registers, used as a destination for operation results.
RM
One of the general purpose registers used as a Data Counter, R4 or R5, if specified, is auto-incremented
after the memory access. R6 is incremented after a write, and decremented before a read.
RR
General Purpose Register RO, R1, R2, or R3.
RS
One of the general purpose registers, used as the source of an operand.
Statuses:
S the Sign status
C the Carry status
Z the Zero status
o the Overflow status
The following symbols are used in the STATUSES column:
X the status flag is affected by the operation
a blank means the status flag is not affected
the operation clears the status flag
1 the operation sets the flag
2 the Overflow flag is affected only on 2-bit shifts or rotates
°
2-16
SW
The Status Word. whose bits correspond to the condition of the status flags in the following way:
3 2 1 0 ....
IsI z laic I
Bit No
Status Word
When the status word is copied into a register. it goes to the upper half of each byte:
[SW]
When the status word is loaded from a register. it comes from the upper half of the lower byte:
~11-5----------------8·1-7--~::::rr~-4~3------~ol ~~3~~---o~1
[RS]
x
(.2)
[SW]
Bits y through z of the Register x. For example. R7 < 15.8 > represents the upper byte of the Program
Counter
Indicates that the operand ".2" is optional
[[ ]]
A low pulse
Contents of location enclosed within brackets. If a register designation is enclosed within the brackets.
then the designated register's contents are specified. If a memory address is enclosed within the brackets.
then the contents of the addressed memory location are specified.
Implied memory addressing: the contents of the memory location designated by the contents of a register.
A
Logical AND
-¥±
Logical Exclusive-OR
[ ]
Addition or subtraction of a displacement. depending on the sign bit in the object code.
Data is transferred in the direction of the arrow.
2-17
Table 2-2. CP1600 Instruction Set Summary
sTAtuses
WORDS
MVI
ADDR,RD
2
MVI@
RM,RD
1
[RD]-[ADDR]
Load register from memory, using direct addressing.
[RD]-[[RM))
~~ffi
MVO
RS,ADDR
2
Load register from memory, using implied addressing.
[ADDR]-[RS]
a::Za::
MVO@
RSJIM
1
O~II.I
MNEMONIC
S Z
C
0
::::0(,)
>~z
~Q~
ILc(
~
....
co
OPERATION PE~FORMED
OPERAND(S)
TYPE
Store register to memory, using direct addressing.
[[RM))-[RS]
Stbre register to memory, using implied addressing. If RS=R4, R5, R6 or R7, then RS=RM Is not
supported.
2
X X
X
X
[RD]-[RD)+ [ADDR]
RM,Rt)
1
X X
X
X
Add memory contents to register, using direct addressing.
[RD]-[RD]+ [[RM))
SUB
ADDR,RD
2
X X
X
X
SUB@
RM,RD
1
X X
X
X
Add memory contents to register, using implied addressing.
[RD]-[RD] - [ADDR]
Subtract memory tontents from registEir, using direct addressing.
[RD]-[RD] - [[RM]]
CMP
ADDR,RS
2
X X
X
X
Subtract memory contents from register, using implied addressing.
[RS] - [ADDR]
CMP@
RM,RS
1
X X
X
X
>
AND
ADDR,RO
2
X X
Compare memory contents with registers, u$ing direct addressing. Only the status flags are
affected.
[RS] - [(RM))
Compare memory contents with register's, using implied addressing. Only the status flags are
affected.
[RD]-[RD] A [AOOR]
Q
AND@
RM;RD
1
X X
AND memory contents with those of register, using direct addressing.
tRD]-[RD) A [[RM))
XOR
ADDR,RD
2
X X
AND memory contents with those of register, using implied addressing.
[RD]-[RD]-¥- [ADDR]
XOR@
RM,RD
1
X X
Exclusive-OR memory contents with those of register, using direct addressing.
[RD]-[ RO].y. [[ RM))
ADD
ADDR,RD
11.1
(,)
ADD@
11.1
Z
11.1
a::
u.
11.1
a::
>
a::
0
~
11.1
:E
Q
Z
c(
g
II!
c(
z
0
(,)
w
en
Exclusive-OR memory contents with those of register, using implied addressing.
Table 2-2. CP1600 Instruction Set Summary (Continued)
STATUSES
S Z C
OPERATION PERFORMED
TYPE
MNEMONIC
OPERAND(S)
WORDS
w
MVII
DATA,RD
2
[RD]~DATA
MVOI
RS,DATA
2
Load immediate to specified register.
[[R7] + l1~[AS]
~
0
0(
Q
w
~
Store contents of specified register in immediate field of MVOI instruction, This is only possible if
program memory is reed/write memory (rather than ROM).
~
ADDI
w
DATA,RD
2
X X
X
X
[RD]~[RD]
~
a:
w
SUBI
DATA,RD
2
X X
X
X
CMPI
DATA,RS
2
X X
X
X
Subtract immediate data froni specified register.
[RD]-DATA
ANDI
DATA,RD
2
X X
[RD]~[RD] A DATA
XORI
DATA,RD
2
X X
[RD]~[RDl¥DATA
A.
0
w
~
Compare immediate data with contents of specified register. Only the status flags are affected.
0(
t;-J
is
CD
~
!
+ DATA
Add immediate to specified register.
[RD]~[RD] - DATA
0(
I
w
~
AND immediate data with contents of specified register.
I
Exclusive-OR immediate data wit., contents of specified register.
J
LABEL
3
A.
JR
AS
1
...,
JSR
RB,LABEL
3
[RB]~[R7l; [R7]~LABEL
B
DISP
2
[R71~[R7] + 2±DISP
Bcond
DISP
BEXT
DISP,E
[R7]~LABEL
Jump to given address.
:i!
X X
I
[R7]~[RS]
Jump to address contained in specified register.
:;)
Jump to given address, saving Program Counter in A4, R5, or AS.
Branch relative to Program Counter contents.
z Z
o
2
If cond is true, [A7]~[ A71 + 2±DISP
2
Branch relative on given condition; otherwise, execute next sequential instruction.
EBCAO-3 ~E;
0
~ E
Z Q
0( Z
a: 0
III 0
If EBCI=l, [R7]~[R7]+2±DISP
Branch relative if external condition is true.
I
Table 2-2. CP1600 Instruction Set Summary (Continued)
STATUSES
TYPE
MNEMONIC
OPERAND(S)
WORDS
S Z
C
MOVR
RS,RD
1
X x
a:'"
",'"
... C
en a:
ADDR
RS,RD
1
XX X
X
c,,1L
SUBR
RS,RD
1
X
X
X
X
CMPR
RS,RD
1
X X
X
X
ANDR
RS,RD
1
X X
XORR
RS,RD
1
X X
-'"
"'0
lIFe
a:Z
~C
!!~
&!o
a::I
OPERATION PERFORMED
0
[RD]-[RS]
Move contents of source register to destination register.
[RD]-[RS]+ [RD]
Add contents of specified registers.
[RD]-[ RD] - [RS]
Subtract contents of source register from those of destination register.
[RD]- [RS]
Compera registera' contents. Only the status flags ara affected.
[RD]-[RD] A [RS]
AND contents of specified registers.
[RD]-[RD].y.[RS]
Exclu,ive-OR contents of specified registera.
II-)
N
o
CLRR
RD
1
0
1
[RD]-[RD] V [RD]
TSTR
RS
1
X
X
Clear specified register.
[RS]-[RS]
INCR
RD
1
X X
DECR
RD
1
X X
Test contents of specified register.
[RD]-[RD]+l
Incremant contents of specified register.
[RD]-[RD] - 1
a:
COMR
RD
1
X X
Decrement contants of specified ragister.
[RD]-[RD]
C;
a:
NEGR
RD
1
X X
X
X
ADCR
RD
1
X X
X
X
SLL
RR(.2)
1
X X
'"~
'"0
IL
'"en...
'"
Complement contents of specified ragister (ones complement).
[RD]-[RD] + 1
Nagate contents of specified register (twOI complement).
[RD]-[RD]+ [e)
Add Carry bit to lpecified regilter contentl.
115 + - - 01+-0
[RR]
Shift logical left one or two bitl, clearing bit 0 (and bit 1 if Ihifting twice).
Table 2-2. CP1600 Instruction Set Summary (Continued)
STATUSES
TYPE
MNEMONIC
OPERAND lSi
WORDS
RLC
RRI.2)
1
S Z
C
0
x
X
2
X
OPERATION PERFORMED
L(D:(@]H15 -
oiJ
[RR]
Rotate left one bit through Carry, or rotate 2 bits left through Overflow and Carry.
SLLC
RRI.2)
1
X X
X
2
~[[JH15'-
0 .... 0
[RR]
Shift logical left one bit into Carry, clearing bit 0, or shift left two bits into Overflow and Carry,
Q
w
clearing bits 0 and 1.
:;)
N
N
.....
z
i=
z
SLR
RRI.2)
1
x x
~ a'
0 ..... 15
[RR]
Shift logical right.one or two bits, clearing bit 15 (and bit 14 if shifting twice).
0
9
w
~
00(
SAR
RRI.2)
1
a:
w
"0
a:
X X
dj
Shift arithmetic right one or two bits, copying high order bit.
w
~
II)
C;
w
a:
.0'
[RR]
RRC
RRI.2)
1
X X
X
2
4ciirm~15
..
o~
[RR]
Rotate right one bit through Carry, or rotate. two bits right through Overflow and Carry.
SARC
RRI.2)
1
X X
X
2
oK(2]M:D
dE
11,g D.
[RR]
Shift arithmetic 'right one bit into Carry,. or two bits into Overflow and Carry.
SlNAP
RRI.2)
1
X X
[RR]
Swap bytes of register once, or twice.
Table 2-2. CP1600 Instruction Set Summary (Continued)
STATUSES
TYPE
MNEMONIC
:.t
y
c(
I-
m
llL
::I
a:
a:
w
I-
~
':"
N
N
CI)
::I
S Z
C
WORDS
PSHR
RS
1
Separate mnemonics for MVO@ RS,R6.
PULR
RD
1
Separate mne~nics for MVI@R6,RD.
SIN
(2)
1
PCiT'- lS
Software interrupt.
Enable interrupt syetem.
Disable interrupt system.
Terminate current interrupt.
Jump to given address and enable interrupt system.
Jump to given address and disable interrupt system.
Jump to given address, saving Program Counter in R4, R5 or R6, and enable interrupt system.
Jump to given address, saving Program Counter in R4, R5 or R6, and disable interrupt system.
1
1
EIS
DIS
TCI
1
JE
JD
LABEL
LABEL
3
3
JSRE
JSRD
RB,LABEL
RB,LABEL
3
3
GSWD
RD
1
RSWD
RS
1
X X
X
I-
m
X
[RD<15,12»-[SW); [RD<7.4»-[SW)
Place Status Word in upper half of each byte of the specified register. RD may be RO, Rl, R2 or
R3.
[SW]-[RS<7,4»
CLRC
1
0
Load Status Word from bits 7 through 4 of the specified register.
[C)-O
SETC
1
1
Clear Carry.
[C)-I
NOPP
NOP
HLT
SDBD
2
1
1
1
c(
I-
OPERATION PERFORMED
0
OPERAND IS)
Set Carry.
(2)
No Operation.
Halt after executing next instruction.
Set double byte data mode for next instruction, which must be of one of the following types;
Primary or sacondary I/O or memory reference
Immediate or immediate operate
If implied addressing through R I, R2, or R3 is used, the same byte will be accessad twice; addressing through R4, R5, or R7 will give bytes from the addressed location and that addressed after
auto-increment. Direct addressing and Stack addressing are not allowed in double byte mode.
Table 2-3. CP1600 Branch Conditions and Corresponding Codes
MNEMONIC
C
LGT
BRANCH CONDITION
OBJECT CODE
DESIGNATION
C= 1
Carry
(logical greater than)
c=o
No Carry
(logical less than)
0001
OV
0= 1
0010
NOV
0=0
PI..
s=o
Plus
5 = 1
NC
liT
1001
Overflow
1010
No overflow
MI
0011
1011
Minus
ZE
EO
NZE
NEQ
LT
GE
LE
GT
USC
ESC
Z = 1
Zero (equal)
Z =0
Nonzero (not equal)
5VO=1
Less than
5 .... 0=0
Greater than or equal
ZV(5 .... 0) = 1
Less than or equal
ZV(5 .... 0) =0
Greater than
C.... S = 1
Unequal sign and cany
C .... 5 =0
Equal sign and cany
0100
1100
0101
1101
0110
1110
0111
1111
The following notation is used in Table 2-4:
Where ten digits are shown. they are the ten low-order bits of a 10 to 16-bit word. (Word size depends on the system
implementation.) Where four digits are shown. they represent the hexadecimal notation for an entire word (10 to 16
bits).
bb
Two bits indicating one of the first three general purpose registers:
00 = RO
01 = R1
10 = R2
cccc
ddd
Four bits giving the branch condition. as shown in Table 2-3.
Three bits indicating a destination register. RD:
000 = RO
001 = R1
010 = R2
011 = R3
100 = R4
101 = R5
110 = R6
111 = R7
Four bits giving the external branch condition. E. Control signals EBCAO-EBCA3 reflect the state of these four
bits.
One word of immediate data (10 or 16 bits)
eeee
1111
2-23
mmm
Three bits indicating a Data Counter Register RM:
000 = RO
001 = Rl
010 = R2
all = R3
100 = R4
101 = R5
110 = R6
111 = R7
m
One bit indicating the number of rotates or shifts:
one bit position
1 two bit positions
p
One bit of immediate address
a
P
One hexadecimal digit (4 bits) of immediate address
rr
Two bits indicating one of the first four general purpose registers:
00 = RO
01 = Rl
10 = R2
11 = R3
sss
Three bits indicating a source register. RS:
000 = RO
001 = Rl
010 = R2
011 = R3
100 = R4
101 = R5
110 = R6
111 = R7
z
Sign of the displacement:
add the displacement to PC contents
1 subtract the displacement from PC contents
a
In the "Machine Cycles" column. when two numbers are given with one slash between them (e.g .. 7/9). execution time
depends on whether or not a branch is taken. When two numbers are given. separated by two slashes (such as 81/11).
execution time depends on which register contains the implied address.
THE BENCHMARK PROGRAM
For the CP1600 our benchmark program may be illustrated as follows:
LOOP
MVII
MVII
MVI@
MVII
MVI@
MVO@
DECR
BNZE
MVO@
IOBUF.R4
TABLE.Rl
Rl.R5
CNT.R2
R4.RO
RO.R5
R2
LOOP
R5.Rl
LOAD THE I/O BUFFER STARTING ADDRESS INTO R4
LOAD THE TABLE STARTING ADDRESS INTO Rl
LOAD ADDRESS OF FIRST FREE TABLE WORD INTO R5
LOAD WORD COUNT INTO R2
LOAD NEXT DATA WORD FROM 10BUF
STORE IN NEXT TABLE WORD
DECREMENT WORD COUNT
RETURN IF NOT END
RETURN ADDRESS OF NEXT FREE TABLE BYTE
This benchmark program makes very few assumptions. The input table IOBUF and the data table TABLE can have any
length. and can reside anywhere in memory. The address of the first free word in TABLE is stored in the first word of the
TABLE.
2-24
Table 2-4. CP1600 Instruction Set Object Codes
INSTRUCTION
OBJECT CODE
WORDS
MACHINE
CYCLES
ADCR RD
ADD ADDR.RD
000010lddd
1011000ddd
1
2
10
ADD@ RM.RD
ADDI DATA.RD
PPPP
1011mmmddd
1011111ddd
1
2
8/ /11
8
MOVR RS.RD
MVI ADDR.RD
PPPP
oo10sssddd
10 1OOOOddd
ADDR RS.RD
AND ADDR.RD
1111
oo11sssddd
lll0000ddd
1
2
6
MVI', RM.RD
10
..
AND@ RM.RD
6
1
ANDI DATARD
1110111ddd
2
ANDR RS.RD
B DISP
1111
0110sssddd
l000z00000
1
6
2
7/9
2
7/9
2
7/9
BEXT DISP.E
CLRC
PPPP
looozOcccc
PPPP
looozleeee
PPPP
CLRR'RD
CMP ADDR.RS
0006
0111dddddd
1101000sss
6
10
CMP@ RM.RS
CMPI DATA.RS
PPPP
1101mmmsss
1101111sss
1
2
8//11
8
CMPR RS.RD
COMR RD
1111
0101sssddd
00000llddd
1
1
DECR RD
00000IOddd
1
6
6
6
0003
0002
0000lloorr
1
1
INCR RD
J LABEL
0004
MVOI RS.DATA
NEGR RD
NOP (2)
NOPP
4
0000
000000lddd
MVO RS.ADDR
MVO@ RS.RM
1
GSWD RR
HLT
MVII DATA.RD
8/ /11
8
1
2
DIS
EIS
JSRE RB.LABEL
pppp
IllOmmmddd
Beond DISP
INSTRUCTION
1
1
PSHR RS
PULR RD
RLC RRI.2)
RRC RRI.2)
RSWD RS
SAR RRI.2)
SARC RRI.2)
SDBD
SETC
SIN (2)
4
SLL RRI.2)
SLLC RRI.2)
4
SLR RRI.2)
6
JD LABEL
0004
JE LABEL
0004
PPPP
JR RS
JSR RB.LABEL
1
0004
bbppppppOO
3
7
12
3
12
PPPP
JSRD RB.LABEL
0004
bbppppppl0
PPPP
2-25
8/ /11
2
8
2
11
1
2
9
9
1111
loolooosss
PPPP
lOOlmmmsss
lOO1111sss
1111
0000 1OOddd
000011010m
l000z0 1000
PPPP
loolllOsss
1010110ddd
0001 0 IOmrr
ooo1110mrr
0000111sss
ooo1101mrr
ooo1111mrr
0001
0007
000011011m
0001 00 1mrr
1
6
1
6
2
7
1
9
11
1
1
1
1
1
1
1
1
1
6/8
6/8
6
6/8.
6/8
4
4
6
1
8/ /11
lloolllddd
1111
0100sssddd
2
8
1
6
ooolOOOnrr
1
1
6/8
1
6//7
10
0005
00 1Ossssss
llll000ddd
PPPP
llllmmmddd
2
XORI DATA.RD
l111111ddd
2
8/ /11
8
XORR RSiRD
1111
0111sssddd
1
6
XOR@ RM.RD
oo10sss111
1
1010111ddd
PPPP
lloommmddd
XOR ADDR.RD
l1ppppppOl
PPPP
1010mmmddd
6/8
10
SUBR RS.RD
SWAP RRI.2)
TCI
TSTR RS
12
6/ /7
10
1
SUBT DATA.RD
3
1
2
2
SUB@ RM.RD
PPPP
12
ooollOOmrr
l100000ddd
6
12
3
SUB ADDR.RD
12
3
0004
bbppppppOl
ooo1011mrr
1
l1ppppppl0
MACHINE
CYCLES
6/8
6/8
3
PPPP
WORDS
1
1
4
llppppppOO
OBJECT CODE
1
4
8080A
CP1600
System Bus
Signals
DO
015
System Bus
Signals
-· ---
..
..-
-
-.
···
··· ..
-· -
...-
4
MUX
--
-
.-..
Latched
Address
Buffer
Latched
Data
Buffer
- --
4
BAR
.~
.
~
-==
--
OTB
.,...
..,
OWS
......
....,
.
----
BC1
-
BC2
-..
BOIR
lAB
1 of 8 Decoder
-
INTAK
"""
AOAR
ON
NACT
INTR
INTRM
--
--
-
;;t
-
40
-
~
STSTP
,....
L?r
---'
4
TCI
EBCAO
EBCI
...
Figure 2-11. CP1600 to 8080A Bus Conversion
2-26
A15
DO
07
!
High-order
Low-order
07
byte
-INT
..
-
.--
INT
BUSEN
HOLD
-
ROYIN
-..
byte
DO
.-.-
-----------------------------------------------------------------~~--------...
HALT
EBCA3
--
: ::
,....
BUSRQ
MSYNC
..
:=
AO
WAIT
RESET
SUPPORT DEVICES THAT MAY BE USED WITH THE CP1600
A CP1600 microcomputer system with any significant capabilities will use support devices of some other
microprocessor. Parallel 110 capability is available with the CP1680, (described next!. but priority interrupt logic, DMA
logic, and serial I/O logic, to mention just a few common options, may need additional support devices. Fortunately, it
is quite easy to generate an 8080A-compatible system bus from the CP1600 system bus. Logic is illustrated in
Figure 2-11.
The CP1600A is the fastest version of the CP1600 CPU; it runs with a 500 nanosecond machine cycle. The CP1600
machine cycle is equivalent to an 8080A clock period. Since the standard 8080A clock period is also 500 nanoseconds,
no speed conflicts will arise.
The bus-to-bus interface logic illustrated in Figure 2 -11 is self-evident with the exception of bus demultiplexing logic.
The CP1600 Data/Address Bus is shown buffered by a demultiplexing buffer that is connected to two latched buffers.
One of the latched buffers accepts the demultiplexer outputs only when a valid address is being output as identified by
BAR high. The second latched buffer may be a bidirectional latched buffer. or it may be two unidirectional latched
buffers. Three latching strobes are required: DTS, lAB, and DWS.
DTB and lAB are data input strobes. DTB strobes data input that is to be interpreted as data, while lAB stroves data input that is to be interpreted as an address. So far as external logic is concerned, both of these signals are simple data input strobes. We could therefor~ generate a single data input strobe as the OR of DTB and lAB. When this data input
strobe is high, information on the 8080A System Bus side of the latched data buffer must be input to the buffer; this
data must simultaneously be transmitted to the multiplexer.
DWS is the data output strobe. When high, this signal must strobe data from the multiplexer to the latched data buffer;
this latched data must immediately appear at the 8080A System Bus side of the latched data buffer.
Since the CP1600 uses a 16-bit Data Bus, you will probably have to generate two external device data busses; a highorder byte bus and a low-order byte bus. All external devices that transmit or receive parallel data must be present in
duplicate. For example, were 8255 parallel interface devices to be present the following connections would be required:
.-
---:
-=
-=
--
DO
PB high
PC high
--- -
- ..-- -
07
08
I~
~
015
•
AO
A1
A2
-
I,
60
-- ..
-- --
-. AD
·.....
·.--..
-:..
·.--
h
-
PA high
WR
..
Device
Select
Logic
I'
,.. ,
07
WR
8255
PPI
-
t
RD
AO
A1
CE
DO
--
..-
--
-
-
----
-
2-27
WR
1m
AO
A1
CE
8255
PPI
A15
07
-- .-
-- -....
-- ..--
-
--- -..
PA low
PBlow
PC low
The CP1600 and MC6800 system busses are singularly incompatible. You should not attempt to use MC6800 support
devices with the CP1600.
ro
IMSKO
00
01
02
03
04
05
06
07
CKi
PCLR
PDO
PD1
P02
P03
P04
PD5
PD6
PD7
1
2
3
4
5
6
7
40
39
38
37
36
35
34
8
9
10
11
12
13
14
15
16
17
18
19
20
33
32
31
30
29
28
27
26
25
24
CP1680
lOB
INTRQ
IMSKI
BC1
BC2
BOIR
CE
ERROR
VCC
GNO
VOD
PE
AR
PD15
PD14
PD13
P012
PD11
PD10
PD9
PD8
23
22
21
Pin Name
Description
Type
DO - 07
PDO - P015
BOlA. BC 1, BC2
CK1
CPU Data/Address Bus
Peripheral I/O Port
Bus Control signals
Clock signal
Chip Enable
I/O handshake control
I/O handshake control
Interrupt request
Terminate current interrupt
Daisy chain priority
Daisy chain priority
Error interrupt request
Reset
Power, Ground
Bidirectional, tristate
Bidirectional
Input
Input
Input
Output
Input
Output
Input
Input
Output
Input
Input
"IT
PE
AR
INTRQ
Tel
IMSKI
IMSKO
ERROR
PCLR
VCC, VDO' GND
Figure 2-12. CP1680 lOB Signals and Pin Assignments
2-28
=rt-01~~1
, I II I - . ,
..J
.. DATA READ
,
:~~S
EBCI~
.~DO
EBCAO....-EBCA 1""4--EBCA2........EBCA3........PCIT ..........
BDRDY~
STPST~
N
I
N
co
BUSRO----.,
BUSAK""4--HALT........INTRM~
.,
T
,
T
•
••
D7
"D8
I
.1 1
11
liT
••
11
liT,
11
liT
:~~~
1:.,dIII II III .!lIfh II III .!lIftT II III~ic
[
o~I•• "I
~
DEVICE
SELECT
]
uw
CKI
Ta--
IINTR:
B&!~~
Bel
PCLR
PC 1680
-
lOB
-
IMSKI
IMSKO
t·; ii +
o
LO
0..
~
o
Figure 2-13. A CP1600-CP1680 Microcomputer Configuration
~
a:a:w
04:0..
a:
ffi
--
THE CP1680 INPUT/OUTPUT BUFFER (lOB)
The CP1680 lOB is a parallel I/O device designed specifically for the CP1600 CPU. This device provides a single
16-bit parallel I/O port, which may optionally be configured as two 8-bit I/O ports. Primitive handshaking control
signals are available with the parallel I/O logic. Elementary interval timer and prioritized interrupt logic is also
provided.
Figure 2-1 also illustrates that part of our general microcomputer system logic which has been implemented on
the CP1680 lOB.
The CP1600 lOB is packaged as a 40-pin DIP. It requires two power supplies. +5V and +12V. All inputs are TTL compatible. The device is implemented using N-channel MOS technology.
Figure 2-13 illustrates a CP1600 microcomputer system with three CP1680 lOB devices in the configuration.
CP1680 lOB PINS AND SIGNALS
The CP1680 lOB pins and signals are illustrated in Figure 2-12. We will summarize these signals and the functions they serve before examining device operations in detail.
Let us begin by looking at the interface between the CP1680 lOB and the CP1600 CPU.
DO - 07 provide an 8-bitparallel Data/Address Bus via which all communications between the CPU and lOB occur. This bus must connect to the low-order eight bits of the 16-bit CPU Datal Address Bus.
The three bus control signals, BC1, BC2, and BDIR, connect the CP1680 to the CP1600 as illustrated in Figure
2-13. The CP1680 lOB decodes these three bus control signals internally.
A clock input is required by the CP1680. This clock input (CK1) is used by internal logic to determine when BC1,
BC2, and BDIR are valid. CK1 must have the following wave form:
,
I
I
I'
T1 I T2 I T3 I T4
I
I'
T1
I
I T2 I T3 I T4
I
I
I
CK 1 must be derived from the CP1600 clock signals by external logic.
Let us now look at the interface between external logic and the CP1680 lOB.
POO - PD15 provide a 16-bit parallel I/O port which can optionally be configured as
CP1600 I/O
two 8-bit I/O ports. While POO - P015 are in theory bidirectional. these pins are more acPORT PIN
curately described as pseudo-bidirectional. This is because when a zero has been written
CHARACTERISTICS
to one of these pins. the output can sink 1.6 mA for an output voltage of +O.5V. External
logic will have a hard time overcoming this sink in order to pull the pin high. In contrast. when a 1 is written to one of
these pins. the output sources just 1OOJ.J.A at +5V. External logic will have little problem sinking 100J.J.A in order to pull
a pin low. Therefore. you should output a 1 to any pin that is subsequently to receive input data. External logic will then
leave the pin high when inputting 1. while pulling the pin low to input O.
The handshaking control signals which link the CP1680 lOB with external logic are PE and AR. PE is a control signal
which is output by the CP1680. and AR is a control signal which is input to the CP1680.
Now consider CP1680 interrupt signals.
An interrupt request is transmitted to the CP1600 CPU via INTRQ. The CPU acknowledges the interrupt via the
INTAK combination of BOIR, BC1, and BC2. TCI must be output low by the CPU at the end of the interrupt service routine. This signal is required by CP1680 interrupt logic. which uses the low TCI pulse in its priority arbitration.
as described later in this chapter.
2-30
Interrupts may be generated by conditions internal to the CP 1680. or by a low input at ERROR. The ERROR input is
reserved for error conditions detected by external logic.
IMSKI and IMSKO are interrupt priority input and interrupt priority output signals, respectively. These signals are
used to generate daisy chain interrupt priorities between CP1680 lOB devices. as illustrated in Figure 2 -13. We will
describe CP1680 interrupt priorities in more detail later in this chapter.
'MCLR is the master reset control input for the CP1680. This Signal
must be input low for at least 10 milliseconds in
order to reset the CP1680 lOB.
CP1680 ADDRESSABLE REGISTERS
The CP1680 has eight addressable locations, which may be illustrated as follows:
Control
Data. low
PDO - PD15
Data. high
DO - 07
Timer. low
Timer, high
I/O interrupt
vector
Timer interrupt
vector
Error interrupt
vector
These eight addressable locations are all 8-bit registers; they are addressed using the first eight addresses in a 256-address block. as follows:
Register
Address
Control
Data buffer. low-order byte
Data buffer. high-order byte
Timer. low-order byte
Timer. high-order byte
I/O interrupt vector
Timer interrupt vector
Error interrupt vector
2-31
o
1
2
3
4
5
6
7
The actual 256 addresses will be identified by the eight high-order CP1600 Data/Address Bus lines. which will be used
to create CP1680 device select logic. This device select logic creates CE(the chip enable signal); it may be illustrated
as follows:
---
-..:.
.-
-::
---
,.
,.
I
XXXXXXXX
causes CE
low
D~
•
07
08
·•
015
r
DO - - - 07 at CP1680
~
xxxxxxxx
OOOOOY Y Y
Valid CP1680 addresses
T
T
' " - - - - - May be 000,001,010,011, 100, 101, 110, 111
{May have any 8-bit pattern that device select logic
has been designed to create CE low in response to.
THE CP1680 CONTROL REGISTER
We will summarize the individual bits of the CP1680 control register before describing the operations they control.
Here are CP1680 Control register bit assignments:
6
5
4
o
3
11111 fT I
j
j
I
j
j
j
J' ,
~BitNo.
1
r-
CP 1680 Control register
Parallel I/O active }
{01 _ Parallel I/O inactive
This is called the
Ready bit.
PE=Ready
ERROR input signal level held here
0- PDO-PD15 configured as two 8-bit ports
1 - PDO-PD15 configured as one 16-bit port
{
parallel I/O and Error interrupts
{o1 -- Disable
Enable parallel I/O and Error interrupts
timer interrupts
{o1 -- Disable
Enable timer interrupts
Disable clock logic
{o1 -- Enable
clock logic
Parity of 08-015 byte} 0 = even parity
Parity of 00-07 byte
2-32
1 = odd parity
Bit 0 is always the complement of the PE control output. This bit may be interrogated by the CPU. If parallel data
transfer interrupts are disabled. this allows the CPU to poll on status when monitoring parallel data transfers. PE signal
levels are illustrated in Figures 2-14 and 2-15.
Bit 1 reflects the level of the ERROR input. If parallel data transfer interrupt logic is disabled. then the Error interrupt
logic is also disabled. thus. the CPU must also examine the Error status bit when polling the CP 1680.
Bit 2 determines whether PDO - PD15 will act as a single 16-bit I/O port. or as two 8-bit I/O ports. This is only important
when outputting data.
Control register bits 3 and 4 are used to enable and disable parallel data transfer and Error interrupt logic. and timer interrupt logic.
Control register bit 5 is used .to enable and disable CP1680 interval timer logic. If this bit is O. the interval timer will not
decrement.
Bits 6 and 7 report the parity of the high-order byte and low-order byte for data that is input or output via PD~ - PD15. 0
indicates even parity while 1 indicates odd parity.
All Control register bits may be written into or read. You should be very careful when setting or resetting individual bits
not to simultaneously modify other Control register bits. This means you should use a three-instruction sequence with
an AND or OR mask to set or reset any Control register bit. For details see Volume 1. Basic Concepts.
CP1680 DATA TRANSFER OPERATIONS
The CPU inputs and outputs data via the CP1680 lOB by executing MVI and MVO instructions, respectively.
The CPU must access the CP1680 in byte mode. since an 8-bit Data/Address Bus (DO - 07) connects the CPU and the
CP1680 lOB. Whether the I/O port PD~ - PD15 is configured as a single 16-bit port or as two 8-bit ports has no bearing
on the fact that the CPU must access the CP1680 in byte mode.
The most efficient way of accessing the CP1680 is by using the SDBD instruction with implied memory addressing. Consider data input. If PD~ - PD15 is configured as two 8-bit I/O ports and you wish to access just one of
these I/O ports. then you can use implied memory addressing via R1. R2. or R3. We may illustrate input from the highorder byte of I/O Port PD8 - PD15 as follows:
Register 01
RO
4F
DO - 07
R1
2E
---.--
02
Register 02
CP1600
CPU
CE
2-33
PDQ - P07
If PDO - PD15 are configured as two 8-bit I/O ports or ·as a single 16-bit I/O port. and you want to read both I/O ports.
then you should use the SDBD instruction with implied memory addressing via R4 or R5. This may be illustrated as
follows:
PDO - PD7
RO
R4
CP1600
CPU
CE
2E
generates
CE=O
Control register bit 2 configures PDO - PD15 as a single 16-bit I/O port or as two 8-bit I/O ports.
Given the fact that MVI and MVO instructions (in byte mode) should be used to access the CP1680. when should these
accesses occur?
The answer is that the PE and
AR signals
control event sequences.
Consider parallel data input, as illustrated in Figure 2-14.
When the CPU is ready to input data in resets the
Control register READY bit low. This forces the PE
output high - - - -_ _ _ _ _ _ _ __.
Extemal logic uses PE high to trigger data transfer
to the PD1680. Extemal logic signals the end of
data input by inputting AR low - - - - -
PE
INTAK
Figure 2-14. PD1680 Handshaking with Data Input
2-34
When the CPU is ready to receive data. it resets Control register bit 0 to 0; this forces the PE control signal high.
When external logic senses PE high. it must transmit data to the PD~ - PD15 I/O port. At this point it makes no
difference whether pins have been configured as two 8-bit ports or as a single 16-bit port. External logic will pu II to
ground selected high pins. while leaving other high pins alone. When external logic has completed data input. it signals the fact by inputting AR low. It is the high-to-Iow transition of the AR control input which indicates the presence of
new data for the CPU to read. When Ali makes its high-to-Iow transition. PE also makes a high-ta-Iow transition. and
Control register bit 0 is set to 1. If interrupts have been enabled. then an interrupt is requested via INTRO. Figure 2-14
assumes that interrupts have been enabled; therefore INTRQ is shown making a high-to-Iow transition.
The CPU will acknowledge the interrupt request. as described earlier in this chapter. by outputting INTAK via BC1.
8C2. and BOIR. Logic internal to the CP1680 uses INTAK to reset INTRO high again.
There are many ways in which external logic can determine when to set AR high again. In Figure 2-14 we show external logic using PE to set A11 high. Clearly. when PE makes a low-to-high transition. the CPU must have acknowledged
AR low; therefore external logic can now set AR high. Now that AR is high again. external logic can input new data. An
alternative scheme would be for external logic to constantly hold AR low. using the level of the PE output to determine
when new data could be transmitted. When PE is high. external logic will transmit new data to the CP1680 once. As
soon as it transmits new data. external logic will strobe the data with a short. high AR pulse. then wait for PE to go low
and high again before inputting more data. This may be illustrated as follows:
CPU ready
for input
CPU is
ready
again
for input
Extemal
logic inputs
data
Extemal
logic inputs
data
Data output handshaking is illustrated in Figure 2 -15.
DO - 07
L
:hen CPU outputs data, PE is automatically set
.~.nal.
Extemallogic uses PE high as a "valid data ready"
Aft., """,;,. th;, / ' "
Ali low
\
PE
INTAk
Figure 2-15. P01680 Handshaking for Data Output
2-35
The most important point to note is that there is no control bit which specifies data input mode or data output
mode. Thus, the signal sequences we described for data input and those we are about to describe for data output occur automatically; the input or output mode is purely a function of CPU and external logic interpretation.
Whenever the CPU outputs data to the PD 1680, the arrival of data forces PE output high. If PDO - PD 15 has been configured as two 8-bit ports, then the arrival of a single data byte to either port will cause PE to be output high. If PDO PD15 is configured as a single 16-bit 1/0 port then PD will not be output high until two bytes of data have been
received from the CPU by the PD1680.
Once PE is output high, nothing more happens until external logic responds. External logic cannot tell by the simple inspection of any control signals whether a data input operation or a data output operation is in progress. It is up to you,
when designing your system, to dedicate CP1680 devices to input or output: or you must generate your own identification logic in the event that a CP1680 lOB is bidirectional. In Figure 2-15 we simply assume that external logic knows
data is to be read, and knows whether the data is 16 bits or 8 bits wide. Furthermore, if the data is 8 bits wide, external
logic must know which 8 bits to read. In any event. when external logic has completed its undefined operations, it must
input AR low. The high-to-Iow transition of AA forces PE low again. and if interrupts are enabled, an interrupt will be requested via INTRO. When the CPU acknowledges the interrupt by outputting INT AK via BC 1, BC2. and BDIA. the
PD1680 uses the INTAK pulse to reset INTRQ high.
The method used by external logic to reset AR high again is undefined. In Figure 2-15, we show PE going high as the
trigger which external logic uses to reset AJ1 high. This is clearly a viable scheme; PE will not go high again until fresh
data has been output at which point it is safe to assume that the CPU knows prior data has been read by external logic.
It would be equally viable for external logic to hold AR continuously low, transmitting a short high pulse whenever it
reads data. This may be illustrated as follows:
PE
CPU has
output
data
CPU has
output
more data
logic has
read data
Because there are no control signals which identify the PD1680 operating in input mode or output mode, there
is no straightforward scheme for handling bidirectional data transfers with a single PD1680 device.
THE CP1680 INTERVAL TIMER
The CP1680 has very elementary interval timer logic. A 16-bit Timer register, addressed as two separate 8-bit locations, decrements once every eight CK1 pulses, providing the timer has been enabled. You enable and disable timer
logic via Control register bit 5. As a separate event timer interrupts may be disabled via Control register bit 4. If timer
interrupts are enabled, then when the timer decrements to 0, an interrupt request will occur. (Timer interrupt logic is
described with other CP1680 interrupt logic later in this chapter.) If timer interrupts are not enabled, then the timer itself is effectively disabled, since you cannot test any timer status flag to see if the timer timed out: nor can you accurately read the contents of the Timer registers on the fly, since there is no protection against reading timer contents
while it is in the process of being decremented.
The only timer programmable option you have is to load an initial value before the timer is enabled. The timer
has no buffer; therefore, once it times out it begins decrementing again, if still enabled, beginning with the
value FFFF16. This may be illustrated as follows:
f
~XXXX*S*CKl
t
Load TImer
starting
value XXXX
and start
TImer
,
~I"
Time intervals
t
FFFF*S*CKl
Time out.
Restart
,
~I"
Time out.
Restart
2-36
t
FFFF*S*CKl ~
t
TIme out.
Restart
The only accurate long time intervals you can compute are exact multiples of FFFF16 • 8 • CK 1.
The CP1600A uses a 4MHz two-phase clock. which generates a 500 nanosecond cycle time. Thus. CK1 equals 500
nanoseconds. and long CP1600A time intervals must be an exact multiple of 262.144 milliseconds - the time it will
take for the cou nter to decrement from FFFF16 to 0000.
The CP1600 uses a 3.3MHz two-phase clock. which generates a 600 nanosecond cycle time; therefore. long time.intervals must be exact multiples of 314.572 milliseconds.
The CP161 O. which runs on a 2MHz two-phase clock and has a one microsecond cycle time. will compute long time intervals that are exact multiples of 524.288 milliseconds.
You cannot attempt to generate clock periods that are multiples of shorter time intervals by loading some initial value
into the timer following each time out an unknown amount of time will elapse between the interval timer interrupt occurring and being acknowledged. The length of this unknown period of time will depend on the number of non-interruptable instructions which may be executing in sequence when the interrupt request first occurs. plus any higher
priority interrupts which may exist. Therefore. if you load an initial value into the timer. it should be to compute an isolated time interval only. Here is an appropriate instruction sequence:
MVI
ANDI
MVO
MVII
MVO
MVII
MVO
MVI
ADDI
MVO
10B.RO
CFH.RO
RO.IOB
2AH.RO
RO.IOB+3
34H.RO
RO.JOB+4
10B.RO
30H.RO
RO.IOB
;INPUT CONTROL REGISTER CONTENTS
;ZERO BITS 4 AND 5
;RETURN TO CONTROL REGISTER
;TRANSMIT LOW-ORDER TIMER
;INITIAL BYTE
;TRANSMIT HIGH-ORDER TIMER
;INITIAL BYTE
;LOAD PRIOR CONTROL REGISTER CONTENTS
; SET BITS 4 AND 5
;START TIMER
The instruction sequence above begins with three instructions that load the CP1680 Control register contents into
Register RO. Bits 4 and 5 are zeroed. then the result is returned to the Control register. Thus. the timer and timer interrupts are disabled. We do not bother with an SDBD instruction. Since the data source is eight bits wide. only the loworder byte of Register RO will be significant. This being the case. we can use an 8-bit immediate AND mask to modify
Register RO contents before returning the low-order byte to the Control register.
Next. we load the initial timer value. one byte at a time. into Register RO. Each byte is written out to the appropriate half
of the Control register. Once again we do not need to use the SDBD instruction. Since an 8-bit data path connects the
CPU to the 1680 lOB. only the low-order byte of Register RO will be significant during the data output.
Finally. we start the timer by loading Control register contents into Register RO. setting bits 4 and 5 to 1 and writing
back the resu It.
When you write into the Timer registers. you clear any timer interrupt requests which may at that time be pending.
CP1680 INTERRUPT LOGIC
A CP1680 lOB will generate an interrupt request by outputting a low signal at 1NiliQ if anyone of these three
conditions occurs:
1)
A low input at ERROR. External logic can request an interrupt via the CP1680 using the ERROR input.
2)
The AR handshaking control input makes a high-to-Iow transition. This is illustrated in Figures 2-14 and 2-15.
3)
The Interval Timer decrements from 1 to O.
Recall that there are two separate interrupt enable/disable control bits in the Control register. One control bit applies to
the Interval Timer. while the other control bit applies to both the AR handshaking and ~ interrupts.
Interrupt priorities among the three sources within a single CP1680 lOB are as follows:
ERROR highest
handshaking
Timer lowest
AR
When more than one CP1680 lOB is present In a CP1600 microcomputer system, then daisy chain priority is implemented using the MSKI input signal and the MSKO output signal. Signal connections are illustrated in Figure
2 -13. The manner in which interrupt priorities are handled by the CP1680 is a little unusual.
Two or more CP1680 devices may combine their interrupt request signals. which are wired ORed and input to the
CP1600 via INTRO. The CP1600 acknowledges an interrupt via the INTAK combination of BC1. BC2. and BDIR. We de-
2-37
scribed this process earlier in the chapter. All CP1680 devices simultaneously receive the INTAK combination;
however. a CP1680 which is acknowledged raises its IMSKO signal high. causing it to become the IMSKI input to the
next CP1680 in the daisy chain. Any device that receives a high IMSKI input ignores the interrupt acknowledge. Thus.
only the highest priority. interrupt requesting CP1680 device in the daisy chain will process the interrupt acknowledge.
However. it takes a finite amount of time for IMSKO high signals to propagate as IMSKI signals. and thus ripple through
the daisy chain. Consequently. a maximum of eight CP1680 devices may be present in the daisy chain. A ninth device
will receive its IMSKI high signal too late and will respond to an interrupt acknowledge.
CP1680 lOB devices maintain their interrupt priority status until they receive a high TCI pulse. At that time. prior interrupt priorities are reset at all devices. and new priority arbitration begins. Thus. when using CP1680 lOB devices. you
are required to end all interrupt service routines by executing a TCi instruction.
Note that if one CP1680 lOB has more than one active interrupt request (for example. an ERROR interrupt request and a
timer interrupt request), then this internal interrupt priority will take precedence over the daisy chain interrupt priority.
That is to say. the ERROR interrupt request will be acknowledged and serviced first. After the next TCI instruction is executed. the timer interrupt request will be serviced before any interrupt request from a lower priority CP1680 device is
acknowledged.
Every CP1680 device has three 8-bit Interrupt Vector registers. one dedicated to each of the three interrupt
sources. These three Interrupt Vector registers were illustrated earlier in the chapter. Following an interrupt
acknowledge. when the lAB combination appears at BC1, BC2. and BDIR, the contents of the Interrupt Vector
register for the highest priority active interrupt will be returned to the CPU. Interrupt acknowledge timing is illustrated in Figure 2-9. At the interrupt service location a Jump-to-Subroutine instruction will probably be stored.
Since the Jump-to-Subroutine object code is three words long. a maximum of 85 interrupts can be origined in the first
256 words of memory. This is more than sufficient. since only eight CP1680 devices with 24 interrupts can be supported in a single daisy chain.
2-38
DATA SHEETS
This section contains specific electrical and timing data for the following devices:
• CP1600 CPU
• CP1600A CPU
·CP1610CPU
• lOB 1680 I/O Buffer
2-01
CP1600·CP1600A·CP1610
IUS TIMING DIAGRAM
JS CONTROL
DO-DI5
1m
BAR
@
NACT
~
@
FlOAT
+--+
UNDEFINED
x::x
IIACT
@
@
BAR
-=x
FLOAT
+--+
OUTPUT PC+I TO
FETCH DISPLACEMENT
INPUT
IEXT INSTRUCTION
~STABLE AS LONG AS ADDRESS IS STMILE ;
I
EBCI: •
@
....
OUTPUT
PROGRAM COUNTER
E3~3A!XXXXXXX
DTB
DON'T CARE
I---tAI
~
•
O-DOth CARE-;
.......
VWD INPUT
THROUGHOUT TSI
TYPICAL INSTRUCTION SEQUENCE
~IJlCYCLE~
~~A~A~~~A~R~~
tcv----t
I
tBI-l
~ -i
I 1
00-015 :
~
CHANGING FROM
FLOAT IIOOE TO
OUTPUT IIODE
B~
~s
OUTPUT CHAHGING FROM
VALID OUTPUT IIOOE TO
FLOAT IIOOE
I
t-tB2
1
,...-------i
~-----~
W
INPUT
INSTRUCTION
OR DATA
OPERANO
BRANCH ON EXTERNAL CONDITION INSTRUCTION
Data sheets on pages 2-02 through 2-06 reprinted by permission of General Instrument Corporation,
2-02
CP1600
ELECTRICAL CHARACTERISTICS (CP1600)
Maximum Rating.·
"Exceeding these ratings could cause
permanent damage to these devices.
Functional operation at these conditions is
not implied-operating conditions are
specified below.
Vo o, Vce, GNO and all other input/output voltages
with respect to Vaa . . . . . . . . . . . . -0.3V to +18.0V
Storage Temperature . . . . . . . . . . . . -55 0 C to +150 0 C
Operating Temperature . . . . . . . . . . . . . O°C to +70°C
Standard Condition.: (unless otherwise noted)
VBa= -3V±10%, 0.2mA(typ) ,2mA(max.)
Operating Temperature (T A)=O° C to +70° C
Voo=+12V±5%, 70mA(typ), 110mA(max.)
Vce=+5V±5%, 12mA(typ) , 2SmA(max.)
Characterlltlc
DC CHARACTERISTICS
Clock Inputl
High
Low
Logic Inputs
Low
High (All Lines except BOROY)
High (Bus Data Ready Line
See Note)
Logic Outputl
High
Low (Data Bus Lines 00-015)
Low (Bus Control Lines,
BC1,BC2,BOIR)
Low (All Others)
Sym
Min
Typ··
Max
Units
VIHC
VILC
10.4
0
-
Voo
0.6
V
V
VIL
VIH
0
2.4
-
0.65
Vcc
V
V
VIHa
3.0
-
Vee
V
VOH
VOL
2.4
Vee
-
-
-
0.5
V
V
IOH = 1oo~
IOL = 1.6mA
VOL
VOL
-
-
-
-
0.45
0.45
V
V
IOL = 2.0mA
IOL = 1.6mA
t1/>2, tl/>2
120
-
Condltlonl
AC CHARACTERISTICS
Clock Pulse Inputs, 1/>1 or 1/>2
Pulse Width
-
ns
-
ns
t12, t21
0
Clock Period
tcy
0.3
-
2.0
",s
Rise & Fall Times
tr,tf
-
-
15
ns
tms
-
-
30
ns
tBO
-
-
120
ns
t BF
t B1
tB2
-
50
0
10
-
-
ns
ns
ns
t
DC
-
-
120
ns
t BU
tro
trw
-
150
200
300
tOE
tAl
-
-
-
-
150
400
ns
ns
C1/>1,C1/>2
-
20
30
pF
CIN
-
-
-
6
5
12
10
pF
pF
Co
-
8
15
pF
Skew (1/>1, 1/>2 delay)
Malter SYNC:
Delay from I/>
DO-D15 BUI Signall
Output delay from 1/>1
(float to output)
Output delay from 1/>2
(output to float)
Input setup time before 1/>1
Input hold time after 1/>1
Bu. Control Signall
BC1,BC2,BDIR
Output delay from 1/>1
BUSAK Output delay from 1/>1
TCI Output delay from 4>1
TCI Pulse Width
EBCA output delay from BEXT
input
EBCA wait time for EBCI input
-
-
-
ns
ns
ns
CAPACITANCE
1/>1, 1/>2 Clock Input capacitance
Input CapaCitance
00-015
All Other
Output CapaCitance
00-015 in high impedance state
1 TTL Load & 25 pF
TA=+25°C; Voo=+12V; Vee = +5V;
VBB =-3V; tl/>1 t ct>2 = 120ns
'"Typical values are at +25°C and nominal Voltages.
NOTE:
The Bus Data ReaOY(BOROY) line is sampled during time period TSI after a BAR or AOAR bus control signal. BOROY must
go low requesting a wait state 50 ns before the end of TS1 and remain low for 50 ns minimum. BOROY may go high
asynchronously. In response to BOROY, the CPU will extend bus cycles by adding additional microcycles up to a maximum
of 40 ",sec duration.
2-03
CP1600A
ELECTRICAL CHARACTERISTICS (CP1600A)
Maximum Ratings·
"Exceeding these ratings could cause
permanent damage to these devices.
Functional operation at these conditions is
not implied-operating conditions are
specified below.
Voo, Vee, GNO and all other inpuVoutput voltages
with respect to Vaa . . . . . . . . . . . . -0.3V to +18.0V
Storage Temperature . . . . . . . . . . . . -55°C to +150°C
Operating Temperature . . . . . . . . . . . . . O°C to +70°C
Standard Conditions: (unless otherwise noted)
Voo=+12V±5%, 70mA(typ) ,14OmA(max.)
Vee=+5V±5%, 12mA(typ) ,25mA(max.)
Characteristic
OC CHARACTERISTICS
Clock Inputs
High
Low
Logic Inputs
Low
High (All Lines except BDROY)
High (Bus Oata Ready Line
See Note)
Logic Outputs
High
Low (Oata Bus Lines 00-015)
Low (Bus Control Lines,
BC1,BC2,BDIR)
Low (All Others)
VBa= -3V±10%, 0.2mA(typ) , 2mA(max.)
Operating Temperature (TA)=O°C to +70°C
Sym
Min
Typ·"
Max
Units
VIHe
VILe
10.4
0
-
Voo
0.6
V
V
VIL
VIH
0
2.4
-
0.65
Vee
V
V
VIHB
3.0
-
Vee
V
VOH
VOL
2.4
-
Vee
-
0.5
V
V
10H = l001lA
10l= 1.6mA
VOL
VOL
-
-
0.45
0.45
V
V
10l= 2.0mA
10l= 1.6mA
t.p2, t.p2
95
t12, t21
0
0.25
-
-
-
Condillons
AC CHARACTERISTICS
Clock Pul.. Inputs, .p1 or .p2
Pulse Width
Skew (411, 412 delay)
-
ns
-
-
ns
Clock Period
tcy
-
2.0
~s
Rise & Fall Times
MMterSYNC:
Delay from 41
00-015 Bus Signals
Output delay from 411
(float to output)
Output delay from .p2
(output to float)
Input setup time before 411
Input hold time after 4>1
Bus Control Signals
BC1,BC2,BDIR
Output delay from 411
tr, tf
-
-
15
ns
tms
-
-
30
ns
BUSAK Output delay from 411
TCI Output delay from 4>1
TCI Pulse Width
EBCA output delay from BEXT
input
EBCA wait time for EBCI input
tBO
-
-
95
ns
t BF
t B1
tB2
-
50
0
10
-
-
-
-
ns
ns
ns
t
DC
-
-
t BU
tTO
tTW
-
-
150
200
300
tOE
tAl
-
-
150
400
ns
ns
C4>l,C4I~
-
20
30
pF
CIN
-
-
-
6
5
12
10
pF
pF
Co
-
8
15
pF
-
200
-
-
ns
ns
ns
ns
CAPACITANCE
.pl, .p2 Clock I nput capacitance
Input Capacitance
00-015
All Other
Output Capacitance
00-015 in high impedance state
1 TTL Load & 25 pF
TA=+25°C; Voo=+12V; V ec =+5V;
VBB=-3V; t'4>l t tP2 = 120ns
""Typical values are at +25°C and nominal voltages.
NOTE:
The Bus Oata ReaOY(BOROY) line is sampled during time period TSI after a BAR or ADAR bus control signal. BOROY must
go low requesting a wait state 50 ns before the end of TSl and remain low for 50 ns minimum. BDROY may go high
asynchronously. In response to BDRDY, the CPU will extend bus cycles by adding additional microcycles up to a maximum
of 40 ~sec duration.
2-04
CP1610
ELECTRICAL CHARACTERISTICS (CP1610)
Maximum Ratings·
·Exceeding these ratings could cause
permanent damage to these devices.
Functional operation at these conditions is
not implied-operating conditions are
specified below.
Vo D , Vee, GNO and all other input/output voltages
with respect to VBB • • • • • • • • • • • • -0.3V to +18.0V
Storage Temperature . . . . . . . . . . . . -55° C to +150° C
Operating Temperature . . . . . . . . . . . . . O°C to +70°C
Standard Conditions: (unless otherwise noted)
Voo=+11V±5%. 70mA(typ). 110mA(max.)
Vee=+5V±5%. 12mA(typ) .25mA(max.)
Characteristic
DC CHARACTERISTICS
Clock Inputs
High
Low
Input current
Logic Inputs
Low
High (All Lines except BOROY)
High (Bus Data Ready Line
See Note)
Logic Outputs
High
Low (Data Bus Lines 00-015)
Low (Bus Control Lines.
BC1.BC2,BOIR)
Low (All Others)
VBB= -3V±10%. 0.2mA(tYPI .2mA(max.)
Operating Temperature (TA)=O°C to +70°C
Sym
Min
Typ·'
Max
Units
VlHe
VILe
10.0
0
-
-
V
V
mA
Conditions
-
-
Voo
0.6
15
VIL
VIH
0
2.4
-
0.65
Vee
V
V
VIHB
3.0
-
Vee
V
VOH
VOL
2.4
-
Vee
-
0.5
V
V
10H = 1001tA
10L= 1.6mA
VOL
VOL
-
-
-
-
0.45
0.45
V
V
10L= 2.0mA
10L= 1.6mA
VIHC
= Voo-1
AC CHARACTERISTICS
Clock Pulse Inputs, eIl1 or eIl2
-
ns
-
-
ns
0.5
-
.2.0
p's
-
-
15
ns
tms
-
-
30
ns
tBO
-
-
200
ns
t BF
t B1
tB2
-
0
10
50
-
-
ns
ns
ns
t
DC
-
-
200
ns
t BU
tTO
tTw
-
150
200
300
-
ns
ns
ns
tOE
tAl
-
-
150
400
ns
ns
Cq>1, C4>2
-
20
30
pF
-
Pulse Width
tell2. tell2
250
Skew (eIl1. eIl2 delay)
t12, t21
0
Clock Period
tcy
Rise & Fall Times
Master SYNC:
Delay from ell
00-015 Bus Signals
Output delay from .p1
(float to output)
Output delay from .p2
(output to float)
Input setup time before eIl1
Input hold time after eIl1
Bus Control Signals
BC1,BC2,BOIR
Output delay from .p1
tr, tf
BUSAK Output delay from .p1
TCI Output delay from eIl1
TCI Pulse Width
EBCA output delay from BEXT
input
EBCA wait time for EBCI input
-
-
CAPACITANCE
eIl1, eIl2 Clock Input capacitance
Input capacitance
00-015
All Other
Output CapaCitance
00-015 in high impedance state
1 TTL Load & 25 pF
,
TA=+25°C; Voo=+12V; Vee = +5V;
VBB =-3V; t'4>1 t eIl2 = 120ns
-
-
6
5
12
10
pF
pF
Co
-
8
15
pF
CIN
·'Typical values are at +25°C and nominal voltages.
NOTE:
The Bus Data ReaOY(BDROY) line is sampled during time period TSI after a BAR or ADAR bus control signal. BDRDY must
go low requesting a wait state 50 ns before the end of TS1 and remain low for 50 ns minimum. BOROY may go high
asynchronously. In response to BDROY, the CPU will extend bus cycles by adding additional microcycles up to a maximum
of 40 p'sec duration.
2-05
1081680
ELECTRICAL CHARACTERISTICS
Maximum Ratings·
Voo and Vee and all other input/output voltages
with respect to GNO ........................................... -0.3Vto+18V
Storage Temperature ..................................... -55° C to +150° C
Operati ng Tem peratu re ....................................... 0° C to +70° C
'Exceeding these ratings could cause
permanent damage. Functional operation of
this device at these conditions is not
implied-operating ranges are specified
below.
Standard Conditions (unless otherwise noted)
All voltages referenced to GNO
Voo = +12V ± 5%
Vee = +5V ±5%
Operating Temperature (T A) = 0° C to +70° C
Symbol
Min
Typ··
Max
Unit
High
Vihc
2.4
-
Voo
V
Low
Vilc
0
-
.5
V
High
Vih
Vii
2.4
-
Vee
.65
V
Voh
Vol
2.4
Vce
-
-
V
loh
-
.5
V
101
tpc
tcl
tcr,tcf
0.4
70
-
4.0
-
I1S
-
-
-
10
ns
-
6
12
pF
Yin
5
10
pF
Yin
-
8
15
pF
Characteristic
Condition
DC CHARACTERISTICS
Clock Input:
Logic Inputs:
Low
High
Logic Outputs:
Low
0
V
= 100pA
= 1.SmA
AC CHARACTERISTICS
Clock Inputs
CK1
Clock period
Clock width
Rise & Fall times
ns
= 25°C,
Voo = +12V,
Vee = +5V)
CAPACITANCE (TA
Input Capacitance: 00-07
Cin
All others
Output Capacitance:
Cout
=OV
= OV
"Typical values are at +25° C and nominal voltages.
TIMING DIAGRAM
H~
/
wo
-U
-'J
tef
BOIR
BC2.BC1
..
H
H H
~
t~e
----ll+:----ll--- ter
I tel
..
/
u
U
I
/
-+l
'"-_ _ _ _ _ __
toe
Note: CK1' not drawn to scale.
I+-
CIRCUIT DESCRIPTION
This circuit is designed to provide all the data buffering and
control functions required when interfacing the Series 1600
Microprocessor System to a simple peripheral device. Data is
transferred to and from the peripheral on 16 bidirectional lines.
each of which can be considered to be an input or output. The
transfer of information with the CP1600 is accomplished via an 8bit highway, the 16-bits being transferred as two 8-bit bytes. the
register addresses are assigned CP1600 memory locations. as
follows (N is an arbitrary starting address):
Register Address
N
N+1
N+2
N+3
N+4
N+5
N+6
N+7
2-06
Description
Control Register
Data Register Low Order 8-bits
Data Register High Order 8-bits
Timer Low Order 8-bits
Timer High Order 8-bits
Peripheral Interrupt Address Vector
Timer Interrupt Address Vector
Error Interrupt Address Vector
Chapter 3
THE TEXAS INSTRUMENTS TMS 9900,
TMS 9980, AND TMS 9940 PRODUCTS
The TMS 9900 was the first 16-bit microprocessor that could compete effectively in the minicomputer market. In fact.
the TMS 9900 is a one-chip implementation of the TM 990 series minicomputer Central Processing Units.
The TMS 9900 is packaged as a 64-pin DIP; it generates signals for a 15-bit Address Bus and a separate 16-bit Data
Bus, whereas other 16-bit microprocessors multiplex their Data and Address Busses. The TMS 9980 series
microprocessors are 40-pin DIP versions of the TMS 9900; in order to reduce pin counts, the TMS 9980 series
microprocessors access external memory via an 8-bit Data Bus and 14-bit Address Bus. The TMS 9940 is a one-chip
microcomputer containing a subset of the TMS 9900 Central Processing Unit. together with on-chip memory and realtime clock logic.
The TMS 9900 product line has for some time been one of the enigmas of the microprocessor industry. Even a
casual examination of the TMS 9900 instruction set shows that from the programmer's viewpoint. this microprocessor
was at least two years ahead of its time. While it may have had problems competing in high-volume, simple applications, it was certainly the microprocessor of choice for data processing-type, program-intensive applications, yet it was
not widely used in these markets.
The reason for this lack of acceptance has been poor support from Texas Instruments.
Texas I nstru ments initially offered little support for the TMS 9900 because this microprocessor was designed as a lowend product of the TM 990 minicomputer series. That is to say, customers were expected to develop products around
the TM 990 minicomputers; then, if they chose to, they could build production models around the TMS 9900
microprocessor. This development path did not call for extensive TMS 9900 support. In all probability, Texas Instruments was caught by surprise by the buoyancy of the microprocessor market - as a market in its own right. Certainly,
if Texas Instruments had given the TMS 9900 the same level of support that Intel gave the 8080A. we would see entirely different microprocessor product distributions today. But the TMS 9900 and its derivative products are powerful
enough that the belated support they are now receiving from Texas Instruments will give the product line a reasonable
share of future markets.
Texas Instruments now provides full support for the TMS 9900 microprocessor line.
TMS 9900 support devices are designed specifically for the TMS 9900 and can be used with the TMS 9900,
'TMS 9980, or TMS 9940 products. The following devices are described:
The
The
The
The
TIM 9904 Clock Generator
TMS 9901 Programmable System Interface
TMS 9902 Asynchronous Communications Controller
TMS 9903 Synchronous Communications Controller
Texas Instruments is the primary manufacturer for all of the TMS 9900 series products. TMS 9900 series products are handled out of the following Texas Instruments office:
TEXAS INSTRUMENTS, INC.
P.O. Box 1443
Houston. Texas 77001
Second sources for the TMS 9900 family are:
AMERICAN MICROSYSTEMS, INC.
3800 Homestead Road
Santa Clara. California 95051
SMC MICROSYSTEMS CORP. (TMS 9980 series only)
35 Marcus Blvd.
Hauppage, N.Y. 11787
3-1
THE TMS 9900 MICROPROCESSOR
The TMS 9900 is manufactured using N-channel silicon gate MOS technology. It is packaged as a 64-pin DIP. Three
power supplies are required: -5V. +5V. and +12V.
Using a 3 MHz clock. instruction execution times range between 3 and 10 microseconds.
A TMS 9900 FUNCTIONAL OVERVIEW
Figure 3-1 illustrates that part of our general microcomputer system logic which is implemented by the TMS 9900
CPU.
The most important features of Figure 3-1 are:
• The absence of programmable registers
• The presence of significant interrupt handling logic
• The presence of serial-to-parallel data conversion logic
• The absence of I/O port interface logic
Clock Logie
Accumulator
Registerlsl
Programmable
Timers
I/O Ports
Figure 3-1. Logic of the TMS 9900 CPU
3-2
Let us first consider the manner in which the TMS 9900 handles programmable registers.
TMS 9900 PROGRAMMABLE REGISTERS
Within the logic of the TMS 9900 itself. there are just three 16-bit programmable registers: a Program Counter. a
Workspace register. and a Status register.
The Program Counter and Status register are straightforward. The Program Counter contains the address of the
next instruction to be executed. The Status register maintains various statuses. which we describe later in this chapter.
The Workspace register is a unique and powerful programming feature of the TMS 9900. This register identifies
the first of sixteen 16-bit memory locations which act as .16 General Purpose registers. This may be illustrated
as follows:
lS-BIT MEMORY
Special Functions
LOCATION
Any memory
addresses
HIGHORDER
BYTE
~~
xxxx
WP
.,
I
xxxx
~~--------~~~------~,
LOWORDER
BYTE
I
--...IIi!.
l'
RO
xxx x + 2
xxxx + 4
Rl
xxx x + 6
xxx x + 8
R3
R4
xxx x + A
xxxx + C
xxxx+ E
R2
I
:
R5
RS
R7
R8
I
J
xxxx+ 10
xxxx + 12
I
xxx x + 14
I
R9
Rl0
:
Rll
I
i
•
R12
R13
R14
I
R15
xxx x +
xxxx +
xxxx +
xxxx +
16
18
lA
lC
xxxx + lE
RO cannot be an Index register .
Shift instruction will seek shift
count in low-order four RO bits if
instruction object code specifies
o shifts.
:
Subroutine return address or XOP eftective
address
CRU Bit address
Save old WP
Save old PC
Save old ST
Some of the 16 registers serve special functions, as defined by the text on the right-hand side of the illustration
above. For the moment. do not attempt to understand these special functions. They are described later in the chapter.
In TMS 9900 microcomputer systems, external memory consists of 16-bit memory words.
Each 16-bit memory word has its own memory address. Within the TMS 9900 CPU,
however, memory is addressed as a sequence of 8-bit locations. For this to occu r. the CPU
3-3
TMS 9900
MEMORY
ADDRESSES
generates an internal 16-bit memory address; the high-order 15 bits of the internal memory address create the external memory addresses. This may be illustrated as follows:
This 16-bit address is created
by program logic to address 65536 bytes
(
................ ..............
~
~~~
~
These 15 address bits are output
to access 32768 external, 16-bit memory words
I
I
j~
13 12 11
I
j ~
I
A ......_ _ _ _ _ _ _ _ _)__
~SB
~5 14
I
j
I
~~
.
~
10
.
I
9
-.
I
~
8
I
•
- -
7
6
11
j, .
~
5
I
~
4
2
.-
1
Byte Discrimination Bit
.Y 0 = Even Byte
Lts!1 1 = Odd Byte
I-0
Bit No.
I I I 1 J Memory Address I nside the CPU
1
j~
3
~
j
tL
AO
(LSB)
A1
A2
A3
A4
A5
A6
A7
External Address Bus
A8
A9
A10
A11
A12
A13
A 14 (MSB)
When designing hardware around the TMS 9900. you will implement external memory as 16-bit words. which are addressed by a 15-line Address Bus. That is to say. 32.768 16-bit words may be addressed.
But when you are programming the TMS 9900 you will visualize memory as 65.536 bytes. addressed by a 16-bit address. An even byte address will access the low-order byte of an external 16-bit memory word, while an odd
memory address will access the high-order byte of an external 16-bit memory word.
Any 16 contiguous words of read/write memory may serve as the current 16 general purpose registers for the
TMS 9900.
You may have as many sets of 16-bit registers as you wish, limited only by the size of implemented memory.
If you are using more than one set of 16-bit registers, then at any time just one set of 16-bit registers can be
selected. The WP register identifies the first of the 16 contiguous memory locations serving as the current 16
general purpose registers.
Each of the 16 general purpose registers may be used to store data or addresses. Thus. each general purpose register
may serve as an Accumulator or as a Data Counter.
Registers R11 through R15 are used as special Pointer storage buffers; we will be describing the way in which
these registers are used as the chapter proceeds.
Having 16 general purpose registers in read/write memory, rather than in the CPU, is the single most important
feature of TMS 9900 architecture. The advantage of having 16 general purpose registers located anywhere in
read/write memory is that you can have many sets of 16 general pu rpose registers. For example. following an interrupt
acknowledge. you no longer need to save the contents of general purpose registers - all you need to do is save the
contents of the Program Counter. the Workspace register and the Status register. and that is done automatically by
TMS 9900 interrupt handling logic. By loading new values into the Program Counter and the Workspace register. you
3-4
can begin executing a new program. accessing 16 new memory words general purpose registers.
which will be treated as a new set of 16
The disadvantage of having 16 general purpose registers in read/write memory is that no TMS 9900 microcomputer system can be configured without read/write memory; and if you are going to use many different sets of 16bit registers. then you are going to require a significant amount of read/write memory. Furthermore. you lose the speed
associated with executing register-to-register operations; there are no source and destination locations left in the CPU.
Every register access becomes a memory access.
TMS 9900 literature refers to the process of switching from one set of general purpose
registers to another as a context switch. This terminology reflects the complete change of program environment that results from the switch.
TMS 9900
CONTEXT
SWITCH
Special instructions allow you to perfqrm a forward context switch or a backward context switch.
During a forward context switch. you load new values into the Workspace register and Program Counter. while
simultaneously saving the old Workspace register. Program Counter. and Status register contents in the new General
Purpose Registers R13. R14. and R15.
A backward. or reverse context switch loads the current contents of General Purpose Registers R13. R14. and R15 into
the Workspace register. Program Counter. and Status register. respectively. thus returning you to your previous set of
general purpose registers.
You can perform context switches as often as you like and whenever you like. For example. a very effective way of
using context switching is to group data into contiguous memory words which you can identify as a register set. Upon
entering a subroutine. you can perform a context switch which automatically creates all necessary initial data and address values in appropriate general purpose registers. This may be illustrated as follows:
MEMORY
WORDS
RO
Rl
R2
R3
R12
R13
R14
R15
Data and parameters
used by subroutine are
stored here by the calling
program before calling
the subroutine
RO
~
Rl
R2
R3
R4
Rl0
3-5
Arbitrary
Memory
As illustrated above. when you perform a forward context switch, the current Program Counter
contents. Status register contents. and WP register contents are saved in what will become the
new Registers R13. R14 and R15 .. respectively. Here is the exact sequence in which events occur:
TMS 9900
FORWARD
CONTEXT
SWITCH
1)
The new WP register contents are loaded into the CPU and held in temporary storage.
2)
The current Status register contents are written out to the memory location which will become the new Register
R15.
3)
The current Program Counter contents are written out to the memory location which will become the new Register
R14.
4)
The current WP register contents are written out to the memory location which will become the new Register R13.
5)
The new WP register contents. which were held in temporary storage. are moved into the WP register.
6)
The new value is loaded into the Program Counter
Thus. when a forward context switch is performed. an audit trail ensures that program logic knows the exact machine
state at the instant of the forward context switch.
When a backward context switch occurs, the contents of the current General Purpose
registers R13, R14, and R15 are loaded into the WP register, the Program Counter, and the
Status register, respectively. Thus. program logic returns to the location of the forward context
switch
TMS 9900
BACKWARD
CONTEXT
SWITCH
TMS 9900 MEMORY ADDRESSING MODES
The TMS 9900 provides these four methods of addressing memory:
1)
2)
3)
4)
Direct memory addressing
Direct, indexed memory addressing
Implied memory addressing
Implied memory addressing with auto-increment
The way in which the TMS 9900 implements these four memory addressing modes is exactly as described in Volume 1.
Chapter 6. The important point to note is that the TMS 9900 looks upon its address space as consisting of 32.768 16bit memory words which are addressed using 15. rather than 16. Address Bus lines: yet programs compute all addresses as 16-bit words. This logic was described earlier.
Direct memory addressing instructions provide the memory address in the second word
of an instruction's object code:
MSB
15
TMS 9900
DIRECT
ADDRESSING
LSB
14
13
12
11
10
9
8
6
5
4
2
o
'--BitNo.
I I I II I I II I I I III I I
\.------JL
f
Instruction Object Code
Byte ;dontme,
""og~,ed
by CPU logl,
....
- - - - - - - - - - - - - - - - - Direct address output via Address Bus
Direct, indexed memory addressing instructions provide a base address in the second
object code word, but they also identify a general purpose register whose contents are
to be added, as a signed binary number, to the base address. Again. the low-order bit of the
computed address is not output via the Address Bus. but is interpreted by CPU logic as a byte
identifier.
TMS 9900
INDEXED
ADDRESSING
General Purpose Register RO cannot be specified as an index register.
Direct. indexed addressing is very useful in a TMS 9900 microcomputer system. It allows you to address the previous
set of general purpose registers. following a context switch. without knowing where the previous registers were. Suppose you want to access the contents of the memory word which was being used as General Purpose Register R5
3-6
before you switched to your current set of general purpose registers. Recall that the previous Workspace register contents are stored in your current General Purpose Register R13. You could thus address the previous General Purpose
Register R5. without knowing where this general purpose register may have been. by using direct. indexed addressing
as follows:
Instruction
Base Address
Read/Write
Memory
ARBITRARY
MEMORY
ADDRESSES
~
HighOrder
Byte
LowOrder
Byte
(
RO
I
XXXX
Rl
I
XXXX
R2
!
XXXX
+4
R3
1
XXXX
+6
R4
+2
I
XXXX
R5
I
XXXX
+S
+A
R6
R7
j
XXXX
+C
Previous
:
XXXX
+E
General
RS
!
XXXX
+ 10
Purpose
R9
Rl0
Rll
I
XXXX
registers
i
XXXX
+ 12
+ 14
I
XXX X
R12
i
XXXI(
+ 16
+ 18
R13
I
XXXX
+ lA
R14
i
XXXX
R15
I
XQX
+ lC
+ lE
I
I
1
RO
yyyy
yvyy + 2
Rl
R2
I
"R3
R4
I
YVYV + 4
yyvy + 6
yvyy +8
~
R5
I
yyyy+A
R6
i
yvyy +C
Current
General
I
yyyy + E
yyyy + 10
yvyy+ 1'2
vyyy + 14
yyyy + 16
yyyy + 18
yvyy+ lA
yvyy + lC
i
YVYY + lE
:
R7
RS
I
R9
Rl0
I
Rll
I
I
R12
R13
R14
R15
I
Xl.;
I
xx
Purpose
registers
I
An implied memory addressing instruction will specify one of the 16 current general purpose registers as providing the effective memory address.
TMS 9900
IMPLIED
If you specify implied memory addressing with auto-increment, then the contents of the
ADDRESSING
identified general purpose register will be incremented after the memory access has
been performed. If the instruction specifies a byte operation. the register contents will be incremented by one; the
register contents will be incremented by two after a full-word operation.
3-7
Six object code bits identify the data memory addressjng option selected by any TMS 9900 instruction that accesses
data memory. The six object code bits are interpreted as follows:
T
R
~~
'-v--'~
L {
0000 through 1111 select the general purpose register to be accessed during the memory address computation
00 - Not a memory reference instruction. The selected register is
accessed directly.
01 - Implied memory addressing
10 - Direct addressing if register RO is selected.
Direct, indexed addressing otherwise.
{
11 - Implied memory addressing with auto-increment
Two-address instructions will include 12 memory addressing option bits:
MSBr--p__~~~~~~~~~~~~~~~~~~__~-.LSB
I
I
.1
Two-address instruction object code
'"-v-' ~ '"-v-' ~
TO
RD
TS
RS
~~
Destination
address
Source
address
Some instructions allow a source to be anywhere in memory. but the destination must be a general purpose register.
These object codes include TS. RS. and RD. but not TO.
TMS 9900 Jump instructions use program relative. direct addressing. These are one-word
instructions. where the low-order byte of the instruction object code provides an 8-bit. Signed
binary value. which is added to the incremented contents of the Program Counter. This is
straightforward program relative. direct addressing.
TMS 9900
PROGRAM
MEMORY
ADDRESSING
TMS 9900 I/O ADDRESSING
As compared to other microcomputers described in this book. the TMS 9900 has unusual I/O logic. In addition to addressing I/O devices as memory locations. you can address a separate I/O field of up to 4096 bits. Texas Instruments' literature refers to this field as the "Communications Register Unit" (CRU).lf you are programming a TMS
9900 microcomputer system that has already been configured by Texas Instruments. then it is justifiable to look upon
the Communications Register Unit as a form of I/O port. If you are building your own interface to a TMS 9900 CPU. then
instructions that are supposed to access the Communications Register Unit in reality Simply make alternative use of
part of the Address Bus in conjunction with three control signals: CRUCLK. CRUIN. and CRUOUT
There are two classes of TMS 9900 CRU instructions. The first class accesses individual bits (or signals). while
the second class accesses bit fields that may be between 1 and 16 bits wide.
There are three single-bit CRU instructions; they set. reset. or test the identified CRU bit. This is equivalent to setting. resetting. or testing an external signal or single I/O port bit. When a bit is to be set or reset. the new level is output
via CRUOUT. and a CRUCLK pulse indicates that valid data is on the CRUOUT line. When the condition of a bit is to be
input or tested. then external logic is required to return the level of the tested bit via CRUIN.
3-8
A CRU bit instruction outputs a 12-bit address which is computed as follows:
Instruction Object Code.
r-----------------~~--------------~\
MSB 15 14 13 12 11 10 9
8
II
7
6
5
4
3
o
2
LSB
I II I I I IX I y Iy I y Iy I y I y Iy I
General Purpose Register R12
~~--~~~'=~~~~~~-----------~----------~
r \
MSB 15 14 13 12 11 10 9 8
II
7
6
5
4
3
2
o
LSB
I Izlzlzlzlzlzlzlzlzlzlzlzi
x
+
X X X X Y Y Y Y Y Y y
Z Z Z Z Z Z Z Z Z Z Z Z
L S u m "",,om .. effec.;'. CRU "jd""
X, Y and Z represent any binary digits
The 12-bit address is output on the 12 lower-order address lines; the three higher-order address .lines are all 0 to
designate a CRU address.
Now during the execution of a CRU bit instruction, the address which is output is supposed to be a bit address - that
is, an address identifying one bit in a possible 4096-bit fieJd. So far as external interface logic is concerned. the address
can be interpreted in any way. However. data output will' occur via CRUOUT only; data is input via CRUIN, and
stored in the Equal bit of the Status register.
There are two multi-bit CRU instructions: one. LDCR. transfers data from an addressed memory location to any addressed CRU bit field. The other. STCR. transfers data from an addressed CRU bit field to any addressed memory location. Anywhere from 1 to 16 bits of data may be transferred by the LDCR and STCR instructions Instruction object
codes are interpreted as follows:
MSB
8
7
II I III III
......
R
T
15 14 13 12 11 10 9
-
~
~
6
LSB
.-A-.~
5
4
3
2
1
o·
I I I I I I I
Bit No,
Multi-bit CRU Instruction
-~
~~
L{These four bits identify the general purpose register which is to be
used in the memory address computation 0000 = RO to 1111 =
R15,
/ 00 - Register is the memory location
01 - Implied memory addressing via address in the register
10 - If Register RO is selected, then direct memory addressing is
I
specified; the direct address is in the next program memory
'------.<
word. If any register other than RO is selected, then direct, inI
dexed addressing is specified. The contents of the selected
register are added to the contents of the next program memory word.
11 - Implied memory addressing with auto-increment
L . . - - - - - - - - - - C R U bit field length (0 is interpreted as 16)
L -_ _ _ _ _ _ _ _ _ _ _ _ jOOll00 = LDCR
1001101 = STCR
3-9
The source/destination memory location is identified as it would be for any memory reference instruction.
The address of the first CRU bit is specified by Register R12. For a multi-bit CRU instruction, the CRU bit address is incremented for each succeeding bit access, but the incremented address is held in a temporary storage location. The
contents of Reg ister R12 are not incremented
Thus, mUlti-bit CRU instructions may transfer anywhere from 1 to 16 bits between any memory location and any CRU
bit field. Note that memory must be divided into 16-bit words, each of which has identified bit boundaries, but
there are no equivalent bit boundaries in the CRU bit field. That is to say, any CRU bit may be identified via Register
R12 as the first bit in a multi-bit field, while the length of the multi-bit field is identified by the instruction object code.
This may be illustrated as follows:
CRU
MSB
15 14 13
LSB
12 11
10
9
8
7
6
5
4
3
2
0
1
Ixlxlxlxlxlxlxlxlxlxlxlxl
R12
r
MSB
15 14 13 12
Start of CRU
Bit Field
LSB
11
10
9
8
7
6
5
4
3
2
t, I
y
0
<
y
I
t
t
I
CRU Instruction
Object Code
,
If YYYY is 0000, the CRU bit field is assumed to be 16 bits in length.
3-10
End of CRU
Bit Field
When bits are transferred from a memory location to a CRU bit field. the contents of the memory location are not
actually modified. but the transfer occurs as though bits had been right shifted out of the memory location. Bits
arriving within the addressed CRU bit field are stored in sequential CRU bit locations with ascending addresses. This
may be illustrated as follows:
CRU
Data Memory
/
~
~
a
1
Lowest CRU Bit
Address
a
xlxlxlxlxl1 11101011101111101110
1
1
\
\
a
1
"~
a
0
--
1
1
Highest CRU Bit
Address
Eleven bits have been transferred in the illustration above. If eight or fewer bits are transferred from a general purpose
register. only the more significant byte is accessed:
MSB
LSB
15 14 13 12 11 10 9 8 7 6 5 4 3 2
a
I x 1x 1X 11 1a 1a 11 11 1y 1y 1y 1y 1y 1y 1y 1y I
General Purpose Register
CRU
Lowest CRU Bit Address
a
a
Highest CRU Bit Address
Our illustration shows a transfer of five bits.
3-11
If eight or fewer bits are transferred from a memory location, then the memory address will be considered a byte address rather than a word address; that is, the transfer will be from the low-order bits of the addressed byte, which may
be either the upper or lower byte of a 16-bit memory word. Thus you can access the lower byte of a general purpose
register by addressing it as a memory location.
A data transfer from the CRU to data memory occurs as the exact logical reverse of the illustration above, except
that high-order bits of the destination data memory word are zeroed if unfilled. This may be illustrated as follows:
CRU
Data Memory
Lowest CRU Bit Address
o
o
o
Unused, Therefore Reset
Highest CRU Bit Address
As with data transfers from memory to the CRU, if eight or fewer bits are transferred, only a byte will be affected. This
will be either the addressed memory byte:
CRU
Data Memory
_ _- - - - - 1 0
Lowest CRU Bit Address
o
o
Highest CRU Bit Address
These Bits Reset to 0
3-12
or the high-order byte of a general purpose register:
LSB
MSB
15 14 13 12 11 10 9 8 7 6 5 4 3 2 0
I 0 1010 11 11 101 I 01X I X 1X 1X I X 1X 1x'ix I General Purpose Register
~j
•
These Bits.
Reset to
0
-
"--
V
This Byte Unaffected
./
CRU
--
0
1
0
1
1
Lowest CRU Bit Address
Highest CRU Bit Address
TMS 9900 STATUS FLAGS
The TMS 9900 CPU has a 16-bit Status register which may be illustrated as follows:
o
6
7
15 14 13 12 11 10 9
2
3
4
5
8
8
9
6
10 11 12 13 14 15 .....--TMS 9900 Bit Number
4
3
o . - Our Bit Number
Status register
L-.._ _ _ _ _ _ _
Unused
' - - - - - - - - - - - - - XOP instruction executed
' - - - - - - - - - - - - - - Parity status
- - - - - - - - - - - - - - Overflow status
- - - - - - - - - - - - - - - Carry status
' - - - - - - - - - - - - - - - - - - Equal condition
.....- - - - - - - - - - -_ _ _ _ _ _ Arithmetic Greater Than condition
- - - - - - - - - - - - - - - - - - - Logical Greater Than condition
The low-order four bits of the Status register represent an interrrupt mask which identifies the level of interrupt
which is currently enabled. As the 4-bit interrupt mask would imply. 16 levels of interrupt are allowed. We will describe
interrupt processing later in this chapter.
The X status is set to 1 while an XOP instruction is being executed. This instruction allows you to perform a software interrupt - as described later in this chapter.
The P, 0, and C are standard Parity, Overflow and Carry statuses.
The Equal status (=) identifies a condition that currently exists, as the result of the execution of a previous instruction, that will cause a Branch-if-Equal instruction to branch. A CRU bit to be tested also gets stored in the
Equal status.
The Logical Greater Than and Arithmetic Greater Than statuses are set or reset following arithmetic. logical. or data
move operations. A Logical Greater Than treats the source data as simple, unsigned binary numbers. An
Arithmetic Greater Than interprets the operand as signed binary numbers.
TMS 9900 CPU PINS AND SIGNALS
Figure 3-2 illustrates the pins and signals of the TMS 9900 CPU.
Being a 64-pin DIP. the TMS 9900 can afford to have separate Address and Data Busses.
3-13
VBB
HOLD
VCC
WAIT
MeMEN
mAo
We
READY
CRUCLK
HOLDA
REsET
VCC
IAQ
4>1
A14
A13
013
A12
012
A11
011
A10
010
A9
09
A8
08
07
A7
A6
06
05
A5
(MSB)
(LSB)
015
014
4>2
(LSB)
A4
A3
A2
04
03
A1
AO
01
cI>4
Vss
02
00
(MSB)
Vss
VOO
4>3
DBIN
ICO
CRUOUT
CRUIN
IC1
IC2
iNTREa
IC3
Pin Name
Description
Type
AO - A14
DO - 015
4>1.4>2.4>3. <1>4
Address Bus
Data Bus
Clock Signals
Memory Enable
Instruction Fetch
Data Bus In
Write Enable
Memory Ready
Wait State Indicator
I/O Clock
Serial I/O Out
Serial I/O In
Interrupt Request
Interrupt Code
DMA Request
Hold Acknowledge
Load Interrupt
Reset
Power and Ground reference
Tristate.
Tristate.
Input
Tristate.
Output
Tristate.
Tristate.
Input
Output
Output
Output
Input
Input
Input
Input
Output
Input
Input
MEMEN
IAQ
DBIN
WE
READY
WAIT
CRUCLK
CRUOUT
CRUIN
iN'i'Rffi
ICO - IC3
HOLD
HOLDA
LOAD
RESET
VBB. Vee. VDD. Vss
(MSB)
(LSB)
output
bidirectional
output
output
output
Figure 3-2. TMS 9900 Signals and Pin Assignments
3-14
Pins AO - A 14 provide the 15-bit Address Bus. Note that Texas Instruments' literature numbers bits and pins
from left to right; therefore, address line AO represents the most significant address bit, where as address line
A 14 represents the least significant address bit.
DO - 015 provide a 16-bit bidirectional Data Bus. Once again. DO represents the most significant data bit in Texas
Instruments' literature.
Remaining signals may be divided into bus control, interrupt control, and timing.
External logic must provide four clock signals, cp1, <1>2, cp3, and <1>4. These are provided by the TIM 9904. described
later in this chapter.
Any memory access operation begins with an address being output via the Address Bus. The TMS 9900 CPU identifies a stable address on the Address Bus by outputting MEMEN low.
If the memory access operation is an instruction fetch, the IAQ is output high.
If the memory access is a read, then the TMS 9900 outputs a high level via DBIN. Memory interface logic must interpret the high DBIN level as a signal to place data on the Data Bus.
If the memory access is a memory write, then the TMS 9900 CPU outputs a low pulse via WE. Memory interface
logic must use the low WE pulse to signal that valid data is on the Data Bus. and to store it in the addressed memory
location. WE low does not last as long as DBIN high.
When external logic cannot respond to a memory access in the available time, it requests a Wait state by inputting READY low. The CPU acknowledges by outputting WAIT high.
CRUCLK, CRUIN, and CRUOUT are three signals used to implement single-bit or serial data transfers via the
CRU interface.
CRUOUT is used to output bits of data to the I/O devices. and CRUIN is used to retrieve input data from the I/O devices.
CRUCLK is active during output operations only. and defines when data bits on CRUOUT are valid.
Let us now look at interrupt control signals.
There is a single interrupt request input, INTREQ, which must be held low by any external device requesting an
interrupt. External devices identify themselves via control signals ICO - IC3. Thus. an interrupt request must be
accompanied by the appropriate input at ICO - IC3.
Observe that there is no interrupt acknowledge signal.
For DMA operations, external logic requests access to the System Bus by inputting HOLD low. The CPU
acknowledges the Hold request by outputting HOL'DA high.
LOAD is a nonmaskable interrupt.
RESET is a typical system Reset signal. However. TMS 9900 Reset logic uses the device's interrupt capabilities;
therefore. we will describe the Reset operation in detail when discussing TMS 9900 interrupt capabilities in general.
TMS 9900 TIMING AND INSTRUCTION EXECUTION
TMS instructions execute as a sequence of machine cycles, each of which contains two clock periods. Clock
periods are timed by four clock signals, cp1, cp2, <1>3, and cp4, as illustrated in Figure 3-3. Note that <1>2 is the first
phase of each clock period. and that <1>1 is the last phase.
The simplest instruction execution machine cycle is an internal operations cycle. No external
bus signals are active during this machine cycle. and no memory or I/O access occurs. Timing for
an internal operations machine cycle will consist of two clock periods, as illustrated in
Figure 3-3.
TMS 9900
INTERNAL
OPERATIONS
MACHINE
CYCLE
MEMORY ACCESS OPERATIONS
TMS 9900 memory access operations may consist of a memory read or a memory write. An instruction fetch is
a minor variation of a memory read.
Figure 3-4 illustrates memory read machine cycle timing.
MEMEN goes low at the beginning of any memory access machine cycle and stays low for the entire machine cycle.
3-15
a..~"'---CLOCK
I
I.....
PERIOD 1 - - -.......... t-----CLOCK PERIOD 2 - - -.....1
OSC
cf>2
cf>3
--------------~
cf>4
Figure 3-3. TMS 9900 Clock Periods and Timing Signals as
Generated by the TIM 9904
ONE MACHINE CYCLE
I
CLOCK PERIOD 1
~
CLOCK PERIOD 2
I
cf>1
cp3
---tt-
oJ
CP4
--+I+----.J
MEMEN
DBIN
AO-A14
00-015
ADDRESS OUT
INPUT MODE
INPUT MODE
CPU READS DATA
Figure 3-4. A TMS 9900 Memory Read Machine Cycle
3-16
DBIN goes high at the beginning of the memory read machine cycle and stays high for the entire machine cycle. External logic can therefore use fVi'E'MEiij low as a memory address indicator while DBIN high identifies the read operation.
A memory address is output stable on the Address Bus for the entire machine cycle.
a
The Data Bus operations during memory read machine cycle represent the only unusual characteristics of the
machine cycle. Input data needs to be stable during the <1>1 high pulse of the second clock period. However. the Data
Bus is connected to input logic for the entire memory read machine cycle and for a portion of the next machine cycle.
Thus. during a memory read machine cycle. external logic cannot access the Data Bus to perform direct memory access. or any other operations. on the assumption that the Data Bus is free until Data I n becomes stable. Moreover. since
the Data Busis held by data input logic of the CPU during the next machine cycle. a memory read machine cycle cannot be followed by a memory write machine cycle. A memory read machine cycle must be followed by an internal
operations machine cycle. or by another memory read machine cycle,
The only difference between an instruction fetch machine cycle and a memory read machine cycle is the fact that during an instruction fetch machine cycle. IAQ is output high. along with DBIN. for the duration of the machine cycle .
....I - - - - - - - - - O N E MACH I NE CYCLE --------~
CLOCK PERIOD 1
CLOCK PERIOO 2
cp1
cp2
CP3 ___-+_.1
cp4
____
~--------J
MEMEN
WE
AO-A14
ADDRESS OUT
00-015
OATA OUT
----~--------------------------------------------~~----~
Figure 3-5.
A TMS 9900 Memory Write Machine Cycle
Memory write machine cycle timing is illustrated in Figure 3-6, In this illustration. we see that data is output stable on the Data Bus for the entire duration of the memory write machine cycle. The Data Bus is not held by output logic
beyond this single machine cycle. Thus. no restrictions are placed on the type of machine cycle which can follow a
memory write machine cycle. Even though data output is stable for the entire memory write machine cycle. the write
3-17
enable strobe WE does not go low until close to the end of the first clock period. In many cases it is easier to use NOT
DBIN as a write control signal. Here is the necessary logic:
WRITE
DBIN
READ
TMS 9900 instruction execution machine cycle sequences are not always self-evident; therefore, let us look at
some memory reference examples.
Memory address computations make machine cycle sequences quite complex. particularly for two-operand instructions. Fortunately. the exact machine cycle sequences are rarely of any consequence to you as a programmer or logic
designer. The eventual number of machine cycles required to execute an instruction (and therefore its execution time)
is important.
Generally stated. instruction execution proceeds as follows:
1)
The instruction object code is fetched.
2)
3)
The first operand address is computed.
The second operand address (if there is one) is computed.
4)
5)
Any operation that may be required is performed.
If a result is generated. it is returned to the second operand address.
TMS 9900
INSTRUCTION
EXECUTION
SEQUENCES
Let us look at operand address computations using the ADD instruction (A) as a general example. First consider the instruction in its simplest form - where the contents of one reg ister are added to the contents of another register:
A
Cycle
1
2
3
4
5
6
7
Type
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
READ
READ
READ
WRITE
Figure
3-4
3-3
3-4
3-3
3-4
3-3
3-9
R1.R2
Function
Fetch instruction object code
Decode instruction
Fetch R1 contents
Fetch R2 contents
Add R1 and R2 contents
Store sum in R2
Now consider the same instruction's execution. but using implied memory addressing for the first operand:
A
Cycle
1
2
3
4
5
6
7
8
9
Type
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
READ
READ
READ
READ
Figure
3-4
3-3
3-4
3-3
3-4
3-3
3-4
3-3
WRITE
3-5
*R1.R2
Function
Fetch instruction object code
Decode instruction
Fetch R1 contents
Use R1 contents as a memory address (implied addressing)
Fetch contents of implied address location
Fetch R2 contents
Add data fetched in cycles 5 and 7
Store sum in R2
3-18
If the second (destination) operand uses direct addressing. here is the machine cycle sequence:
A
Cycle
1
2
3
4
5
6.7.8
9
10
11
12
13
Type
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
Figure
READ
3-4
3-3
READ
3-4
3-3
READ
3-4
3-3
READ
3-4
3-3
READ
3-4
3-3
WRITE 3-5
*R1.@LABEL
Function
Fetch instruction object code
Decode instruction
Fetch R1 contents
Use R1 contents as a memory address
Fetch contents of implied address location
Fetch the second instruction object code word; it holds the direct address
Fetch contents of directly addressed memory word
Add words fetched in cycles 5 and 11
Store sum in directly addressed memory word
Indexed. direct addressing results in the following sequence:
A
Cycle
1
2
3
4
5
6
7
8
9
10
11
12
13
Type
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
READ
Figure
3-4
3-3
READ
3-4
READ
3-3
3-4
READ
3-4
*R1.@LABEL(5)
Function
Fetch instruction object code
Decode instruction
Fetch R1 contents
Use R1 contents as a memory address
Fetch contents of implied address location
3-3
Fetch the second instruction object code word; it holds the direct address
3-3
READ
3-4
3-3
READ
3-4
3-3
WRITE
3-5
Fetch R5. the Index register contents
Add direct address and index
Fetch contents of memory word addressed by cycle 10 addition
Add memory words fetched in cycles 5 and 11
Store sum in memory word addressed by cycle 10 addition
If the first operand-implied address specified an auto-increment. we must add one more machine cycle:
A
Cycle
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Type
MEMORY
ALU
MEMORY
ALU
MEMORY
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
READ
READ
Figure
3-4
3-3
3-4
3-3
WRITE
READ
READ
3-5
3-4
3-3
3-4
*R1 +.@LABEL(5)
Function
Fetch instruction object code
Decode instruction
Fetch R1 contents
Increment fetched R1 contents
Write incremented R1 contents back to R1
Fetch contents of implied address location
Fetch the second instruction object code word; it holds the direct address
3-3
READ
3-4
3-3
READ
3-4
3-3
WRITE
3-5
Fetch R5. the Index register contents
Add direct address and index
Fetch contents of memory word addressed by cycle 11 addition
Add memory words fetched in cycles 5 and 12
Store sum in memory word addressed by cycle 11 addition
MEMORY SELECT LOGIC
MEMEN discriminates between memory and 1/0 accesses. It is therefore very important that MEMEN low be a
necessary component for any memory select.
You can map 1/0 into the memory space of the TMS 9900. This is true of any microprocessor. Memory addresses that
select 1/0 devices will. of course. also require l\ii'E'MEN low as a contributor to I/O device select logic.
3-19
MEMEN as a contributor to select logic may be illustrated as follows:
1__
-
~
~
CRU
SELECT
LOGIC
~~
..
··
·-..
j~
..
·• --
,
0_
SELECT TRUE ONLY IF
MEMEN IS HIGH AND
A12-A14 ARE 000
..-
MEMEN
AO
(LSB)
A11
A12
11.14
(MSB)
r
MEMORY
AND
MEMORY
MAPPED
I/O SELECT
LOGIC
··--··
··..
SELECT TRUE
ONLY IF
MEMEN
IS LOW
The three high-order address lines. A 12. A 13. and A 14. are not used to address CRU bits. When addressing a CRU bit.
these lines are all low. They are not low during execution of externally defined I/O instructions; therefore. A 12. A 13.
and A 14 low must be a prerequisite for any CRU bit select.
TMS 9900 I/O INSTRUCTION TIMING
All TMS 9900 I/O instructions transfer serial data via the Communication Register Unit (CRU). (This excludes I/O which
is addressed as TMS 9900 memory space.)
There are four types of TMS 9900 I/O instructions. They are:
1)
Data input. Anywhere from 1 to 16 bits of data may be transferred from the CRU bit field to memory.
2)
Data output. This is the simple reverse of data input. Anywhere from 1 to 16 bits of data may be output from
.
memory to the CRU bit field.
3)
Bit test. Any bit in the CRU bit field may be tested. The tested bit is input and stored in the Equal bit of the Status
register. Thence. condition branch instructions can be used to test the bit level.
4)
Externally defined I/O in8tructions~ These instructions generate I/O control signals. but they transfer no data.
Timing for CRU output and input machine cycles is illustrated in Figures 3-6 and 3-7, respectively. Each of
these figures shows two bits of data being transferred. (You should not attach any special significance to this fact; depending on the instruction being executed. anywhere from 1 to 16 bits may be transferred.) CRU machine cycles are
executed contiguously. one per bit.
Every CRU 1/0 instruction will require a memory reference machine cycle. together with one or more CRU machine cycles. For example. when an STCR instruction is executed to input data from the CRU to the CPU. the following
machine cycle sequence will occur:
Cycle
Type
2
Figure
Function
MEMORY READ
3-4
Fetch I nstruction Code
ALU
3-3
Decode Instruction
}
a Cycles, where 0 ~ a ~ 4
Obtain Destination Address
3+a
MEMORY READ
3-4
4+a
ALU
3-3
5+a
MEMORY READ
3-4
Fetch R12
6+a
7+a
ALU
ALU
3-3
Compute CRU Starting Address and Prepare
Control Signals
i Cycles
CRU IN
3-7
8+a + i
9+a + i
ALU
ALU
3-3
Fetch Destination Memory Word Contents
}
Load CRU Bits in Tempprary Register
}
r Cycles
Input i CRU Bits
Fill Upper Bits of Byte or Word With Zeroes
If i >8, r = 15 - i; if i ~ 8, r = 7 - i
10 + a + i + r to
12+a+i+r
ALU
3-3
Prepare to Store Memory Word
13+a+i+i'
MEMORY WRITE
3-5
Output Assembled Word to Memory Location Whose
Contents Were Fetched in Machine Cycle 3 + a
r M A C H I N E CYCLE'
1CLOCK PERIOD
'I
CLOCK PERIOO 2
MACHINE CYCLE 2 - - 1
CLOCK PERIOO ,
1CLOCK.PERIOD21
1
cJ>2
cJ>3
---to-'
cJ>4
AO-A14
CRUIN
CRU READS BIT
CRU READS BIT
Figure 3-7. Two TMS 9900 Input-from-CRU Machine Cycles
An LDCR instruction outputs a sequence of 1 to 16 data bits to a CRU bit field. Here is the LDCR instruction
machine cycle sequence:
Cycle
1
2
Type
MEMORY READ
ALU
a Cycles where 0~a~4
3+a
MEMORY READ
4+a
to
ALU
7+a
8+a
MEMORY READ
9+a
ALU
i Cycles CRU OUT
10+a+i ALU
Figure
3-4
3-3
3-4
Function
Fetch instruction object code
Decode instruction
Obtain sou rce address
Fetch source memory word contents
3-3
Prepare for data transmission
3-4
3-3
3-6}
3-3
Fetch R12
Compute CRU starting address
Output i bits to CRU
Machine cycle to conclude instruction
}
The SBO and SBZ instructions set or reset an addressed CRU bit; in essence. these instructions output one data
bit. Here is the machine cycle sequence via which the bit output occurs:
Cycle
1
2
3
4
5
6
Type
MEMORY READ
ALU
ALU
MEMORY READ
ALU
CRU OUT
Figure
3-4
3-3
3-3
3-4
3-3
3-6
Function
Fetch instruction object code
Decode instruction
Decode instruction
Fetch R12
Compute CRU address
Output to addressed CRU bit
The TB instruction inputs one CRU bit; its timing is identical to the SBO and SBZ instructions. except that
machine cycle 6 is a CRU IN machine cycle.
3-22
The Address Bus is used in an unusual way during a CRU machine cycle. As we have already stated, the CRU bit
field is 4096 bits wide - addressed by 12 of the 15 Address Bus lines. The three high-order Address Bus lines are
used to identify I/O control instructions, as defined in Table 3-1. We can conclude from Table 3-1 that when
MEMEN is high and the three high-order Address Bus lines are all low, an I/O transfer is occurring. Otherwise, one of
five externally defined I/O control instructions is being executed. There are dedicated functions for these five I/O controls in TM 990 minicomputer systems; these are shown in Table 3-1. But to anyone who is simply building a
microcomputer system around a TMS 9900, these five I/O states are undefined. Thus, Figure 3-8 illustrates TMS
9900 systems' bus utilization during both CRU operations and externally defined 1/0 operations. If CRU SEL and
MEMEN are high, CRU Select logic will be active.
Externally defined instructions output 0 on the 12 low-order Address Bus lines, AO - A 11; in addition, CRUCLK
pulses are output as part of the instruction executions.
CRUCLK is an active CRU output strobe only. This Signal pulses high whenever a valid level is present on the
CRUOUT Signal line. There is no pulse for CRUIN. External logic must generate its own strobe if it is needed, by combining MEMEN high with a valid bit pattern on the Address Bus.
CRU instructions that test the level of a bit are, to external logic. no different from CRU input instructions. External logic
is required to return. via CRUIN the level of the selected bit The fact that the CPU interprets this input as status, rather
than data, is immaterial to external logic.
THE WAIT STATE
Additional Wait State clock periods may be inserted between clock periods 1 and 2 of any memory access machine cycle. Timing is illustrated in Figure 3-9. At the rising edge of <1>1 of clock period 1, the CPU samples the READY input
signal. If this signal is low, then the next clock period is a Wait clock period. During a Wait cycle, the WAIT output signal is high; all other output signals hold the levels they had during clock period 1.
A Wait State can last for any number of clock periods. During the ct>1 high pulse of every Wait clock period, the CPU
samples the level of the READY input As soon as READY is sampled high, the Wait State ends. The next clock period
becomes clock period 2 of the machine cycle, and the memory operation is completed.
Table 3-1. High-Order Address Bus Line Used by TMS 9900 I/O Instructions
Instruction
Mnemonic
LDCR
SBO
SBZ
STCR
TB
IDLE
RSET
CKOF
CKON
LREX
Instruction
Type
Output
Output
Output
Input
Test (Input)
Control
Control
Control
Control
Control
A14
0
0
0
0
0
0
0
1
1
1
(MSB)
A13 A12
0
0
0
0
0
1
1
0
1
1
3-23
0
0
0
0
0
0
1
1
0
1
Function
Output data to CRU
Set CRU bit to 1
Reset CRU bit to 0
Input data from CRU
Input CRU bit to Equal status bit
Enter HALT condition
Reset the Interrupt mask
Real time clock on
These are
Real time clock off
TM 990 uses.
Execute bootstrap
Instructions
are undefined
in a TMS 9900
system.
I
I
-..
..
--
.
::
-
-
7
6
~
::
5
4
3
2
--
~
~
-
1
01""-
A14
A13
A12
A11
AO
CRUCL.K
CRUOUT
CRUIN
MEMEN
LREX
CKON
CKOF
UNUSED
RSET
HALT
UNUSED
CRU SEL
t
--•
•••
CRU
SELECT
LOGIC
CRU SELECT
SIGNALS
...
-
If CRU SEL and "M'EiVi'EN are high, CRU Select logic will be active.
Figure 3-8. TMS 9900 System Bus Utilization During I/O Operations
....t----------- ONE MACHINE CYCLE
CLOCK PERIOD 1
I
WAIT
CLOCK PERIOD
I
WAIT
CLOCK PERIOD
¢1
¢2
¢3 _ _ _-,
¢4 _ _ _ _....J
READY
WAIT
Figure 3-9. The TMS 9900 Wait State
3-24
-------------t~
CLOCK PERIOD 2
THE HOLD STATE
The TMS 9900 has a typical microcomputer Hold State, used to enable direct memory access operations. External logic initiates a Hold State by inputting FK>m low. At the beginning of the next non-memory reference machine
cycle. the CPU floats its Address and Data Busses. together with the DBIN. 'M"tMm and wt control signals. HOLDA is
output high as a Hold Acknowledge. Timing is illustrated in Figure 3-10.
(NON-MEMORY
ICLO~~~~~IOD 'I
q,1 \
HOLO
n. ._-
HOLO
I
CLOCK PERIOO'
I
rill}..oJ
q,3
--+~
HOLOA
---~~------
00-015
AO-A14, WE,
MEMEN,DBIN
}
_-"_'"
---
-------~~---------
~------------~
Figure 3-10. TMS 9900 Hold State Timing
The Hold State lasts until external logic raises HOLD high again.
It is up to external logic to perform all operations associated with a DMA transfer. The CPU simply floats the
System Bus in response to a Hold request.
The only nonobvious aspect of Figure 3-10 is the fact that Data Bus timing, during normal instruction execution.
differs from other System Bus signal timing. Figure 3-10 highlights this fact by showing the Data Bus floating at
the beginning of the first HOLD clock period. while other signals float earlier in the preceding clock period. This is not a
particularly significant event. The entire System Bus is floating once the HOLD clock period has begun. However. the
actual tristate condition for any signal begins at that point in the preceding clock period when the signal is no longer
being driven by current operations.
THE HALT STATE
The TMS 9900 IDLE I/O instruction generates a Halt State. When this instruction is executed. the CPU suspends all
program ~ion and internal operations. You must terminate the Idle condition with an interrupt request or a low
LOAD or RESET input. (['(5Ai) and RESET are treated as interrupts as we will describe soon.)
The TMS 9900 CPU does not relinquish the System Bus while halted. That is to say. after an IDLE instruction has
been executed. no System Bus lines are floated.
3-25
The IDLE instruction is usually executed when program logic requires that the CPU wait for an interrupt. or when external logic is computing a real-time interval - which will be terminated with an interrupt request.
You can, if you wish, initiate a DMA transfer by executing an IDLE instruction. In order to do this, you must
create a HOLD request from the Address Bus output characteristic of the IDLE instruction's execution. This may
be illustrated as follows:
--..
A14
A13
A12
A14 (MSB)
~13
A12
:: CRUCLK
-
CRUCLK
C)
+ 5V
.~
:~
I
PRE
-
---0
D
"--
i'""\
Q
~
Q
~
CK
"~
7474
HOLD
erR
t
HOLD
END HOLD
-
~
HOLDA ------------------------------------------------------~~.. HOLDA
As illustrated above, the combination of 010 on the three high-order Address Bus lines, along with the CRUCLK pulse,
identifies the IDLE instruction. Since the process of floating the System Bus will remove the conditions which generated a Hold request. these conditions are used to clock a flip-flop. Thus, external logic which receives the Hold
acknowledge signal and takes control of the System Bus must subsequently reset the Hold request flip-flop in order to
remove the Hold condition. That is to say, program logic can begin a Hold state within a Halt state, but it cannot
end this combination. Two steps are needed to terminate a Hold within a Halt. The Hold request must be
removed, then an interrupt request must follow to terminate the Halt.
TMS 9900 INTERRUPT PROCESSING LOGIC
The TMS 9900 has complex and capable interrupt processing logic. Sixteen levels of external interrupt are
available. Sixteen software interrupts are also available. Fifteen of the sixteen external interrupts are maskable; the
nonmaskable interrupt has highest priority and is the system Reset interrupt. There is, in addition, a non-maskable Load
interrupt. External interrupts may be summarized as follows:
Lc5AD
RESET
Maskable
Levels of
External
Interrupt
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
Priority
0
}
Non-maskable, Equal Highest
Priority Interrupts
2
3
4
5
6
7
8
9
10
11
12
13
14
15
3-26
Lowest Priority Interrupt
External logic identifies the priority of its interrupt request via the ICO. IC1. IC2. and IC3 inputs. as follows:
ICO
IC1
IC2
IC3
o
o
o
o
o
o
o
o
0
0
0
0
1
1
1
1
o
o
0
1
1
0
1
3
0
4
1
0
1
5
1
1
1
1
1
1
0
0
0
0
1
0
1
0
1
1
1
1
1
o
o
1
1
1
1
1
1
1
o
o
o
o
1
0
1
0
1
Priority
Should not be input by external logic - highest external
1
2
6
7
8
9
10
11
12
13
14
15 lowest external
Software interrupts are executed via the XOP instruction. There are, in addition, instructions that parallel the
RESET and LOAD interrupts. We will describe these instructions in due course.
Each one of the external interrupts has two dedicated memory words via which vectoring is
TMS 9900
enabled following an interrupt acknowledge. Figure 3-11 illustrates the memory map assoINTERRUPT
ciated with interrupt vectoring. The memory addresses in Figure 3-11 are byte addresses as
VECTOR MAP
seen by the programmer. Remember. the low-order bit of the address shown in Figure 3-11 is not
output on the Address Bus; therefore. you must divide the memory addresses shown in Figure 3-11 by 2 in order to
generate the address which will be seen by external memory.
The memory words dedicated to interrupt vectoring. as illustrated in Figure 3-11 . can be read-only memory.
read/write memory, or any combination of the two. Obviously, read-only memory will be used in applications that have
dedicated interrupt service routines for specific interrupt requests. Read/write memory might be used in minicomputer-type applications where the interrupt response will depend on the application being serviced.
Interrupt masking and priorities apply only to external interrupt requests. Interrupt masking priorities cannot be
applied to software interrupts (the XOP instruction). Since program logic must generate the software interrupt program logic can equally be relied on to know which software interrupt is to be executed, and whether the software interrupt is allowed by current program logic. That is to say, from the programmer's viewpoint a software interrupt is simply
the consequence of an XOP instruction's execution; you, as a programmer. can include an XOP instruction anywhere in
a program. within or outside an interrupt service routine. XOP instructions might be used in response to error conditions, or to call any frequently used subroutines.
Let us begin by looking at the way in which external interrupts are processed.
Any external device wishing to request an interrupt must pull the INTREQ input low while simultaneously placing a 4-bit code at the ICO - IC3 inputs. The CPU will acknowledge the interrupt, provided that its priority, as
identified by the ICO - IC3 inputs, is enabled. The interrupt will be acknowledged at the conclusion of the currently executing instruction. The BLWP and XOP instructions are exceptions; for the integrity of program logic,
they demand that the next sequential instruction be executed. Therefore, ifan interrupt request occurs while either of
these two instructions is being executed, the interrupt will not be acknowledged until this instruction and the next instruction have been executed.
3-27
MEMORY
MEMORY WORD CONTENT
ADDRESS,.
AREA DEFlNmoN
0000
WP LEVEL 0 INTERRUPT
0002
PC LEVEL 0 INTERRUPT
0004
WP LEVEL 1 INTERRUPT
0006
PC
003C
WP LEVEL 15 INTERRUPT
003E
PC LEVEL 15 INTERRUPT
0040
WP XOPO
0042
PC XOP 0
LEVE~
1 INTERRUPT
NTERRUPT VECTORS
XOP SOFTWARE TRAP VECTORS
I
007C
WP XOP 15
007E
PC XOP 15
0080
·••
GENERAL MEMORY AREA
MAY BE ANY
GENERAL MEMORY FOR
COMBINATION OF
~.DATA.AND
PROGRAM SPACE
OR WORKSPACE
WORKSPACE REGISTERS
••
•
FFFC
WP LOAD FUNCTION
FFFE
PC LOAD FUNCTION
Figure 3-11. TMS 9900 Memory Map
When an Interrupt Is acknowledged, the following machine cycles are executed:
Cycle
1
2
3
4
5
6
7
8
9
10
11
Type
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
Figure
3-3
READ
3-4
3-3
WRITE 3-5
3-3
WRITE 3-5
3-3
WRITE 3-5
3-3
READ
3-4
3-3
Function
Move new WP register contents from vector word to temporary storage
Store status in new R15
Store ICO - IC3 levels in four low-order Status bits
Store incremented PC in new R14
Store old WP register contents in new R13
Fetch new PC contents from vector word
Fetch new WP contents from temporary storage
Vector words are illustrated in Figure 3-11.
3-28
TMS 9900
At the conclusion of the interrupt acknowledge sequence listed above. the priority of the
NESTED
acknowledged interrupt request. less one. is recorded in the four low-order Status register bits.
INTERRUPT
Thus. subsequent interrupt requests will be acknowledged only if their priority is higher than that
PRIORITIES
of the interrupt being serviced. That is to say. whenever an interrupt request occurs. CPU logic
compares the levels input at ICO - IC3 with the levels present in the four low-order Status register
bits. If ICO - IC3 is not greater than the mask. then the interrupt request will be acknowledged. If ICO - IC3 is higher.
then the interrupt request will not be acknowledged. Thus. In the normal course of events. TMS 9900 interrupt
priority logic disables all interrupts of equal or lower priority than an acknowledged interrupt. while leaving higher priority interrupts enabled. Priorities are maintained for the duration of the interru~t service routine. This is illustrated in the following figure. which you should read in the sequence
@-
® -® -© -
o
Int."upts whh
~
5
® -® -® :
11
priorities 5,8 and 11~~ / - - ,
occur simultaneously
/
\1(3\
®
.,
Interru!'t ~Ith,
PriOrity 5 ,
acknowledged
/
//
/
/
\\:::Y
Interrupt 7. having highest
priority of three pending
interrupts (7,8 and 11) will
immediately be acknowledged
Main Program
@
Interrupt with
priority 7 occurs
and is denied
Interrupt service routine 5
completes execution
@
I nterrupt service
routine 2 executes
The interrupt priority arbitration logic of the TMS 9900 is exceptional among microcomputers. Most microcomputers
arbitrate priorities at the instant interrupts are being acknowledged. and once an interrupt has been acknowledged. all
interrupts are disabled. That is to say. interrupt priorities apply only during the acknowledge process. In contrast. the
TMS 9900 maintains interrupt priorities for the duration of the interrupt service routine. as illustrated above.
The net effect of the interrupt response steps illustrated above is to perform a context switch while disabling all interrupts that have the same priority as the acknowledged interrupt. or that have a lower priority.'
There are some very important and nonobvious advantages to initiating an interrupt service routine with a context switch.
Since the 16 new memory locations that will be used as general purpose registers may lie anywhere in read/write
memory. you can store parameters that will be used by the interrupt service routine. in advance of the interrupt. in
those memory locations that are ultimately to serve as general purpose registers for the duration of the interrupt service
routine.
You can. if you wish. modify the interrupt priority scheme that will control nested interrupts. As we have already
stated. if you do nothing about interrupt priorities. then any interrupt service routine may be interrupted by a higher
priority external interrupt. but not by an external interrupt that has the same priority or a lower priority.
If you wish to eliminate nested interrupts entirely. then the first instruction executed within an interrupt service routine
must be an LlMI 0 instruction (Load Interrupt Mask Immediate). which clears the four low-order Status register bits.
thus disabling all maskable interrupts. A RESET or LOAD interrupt - or a level a external interrupt request - will still
be acknowledged; these should be alarm conditions and not part of the normal interrupt logiC of any microcomputer.
You can execute variations of the LlMI instruction to increase or decrease the levels of priority that will be masked for
the duration of any interrupt service routine (or for that matter. any subsequent instruction within the interrupt service
routine) can load appropriate data into the four low-order bits of the Status register. thus changing the priority level at
which all subsequent interrupt requests will be disabled.
3-29
All interrupt service routines should end with an RTWP (Return Workspace Pointer) instruction. The RTWP instruction performs a reverse context switch. which puts the central processing unit back to the logical environment
which was interrupted. Observe that since the Status register is also saved during a forward context switch. the return
instruction will restore whatever level of interrupt priorities existed at the instant the interrupt was acknowledged. You
can. of course. modify the contents of General Purpose Registers R13. R14. and R15 in the course of an interrupt service routine's execution. This allows program logic to alter the conditions that will be restored when the return instruction executes a reverse context switch.
The TMS 9901 PSI, which we describe later in this chapter, provides multiple interrupt handling for TMS 9900
series CPUs. If your system does not include a TMS 9901, then external hardware required to support multiple
interrupts in a TMS 9900 microcomputer system will not be as straightforward as the software response.
First of all. we must cope with the fact that if more than one interrupt request occurs
TMS 9900
simultaneously. then there will be competition on the INTREO input. but there will also be
MULTIPLE
competition at the four priority inputs. ICO - IC3. Resolving competition on the·INTREQ input is
INTERRUPT
no problem; you can wire-OR interrupt requests from many devices to create the CPU input.
HARDWARE
But your external logic must make sure that only the highest priority combination of ICO - IC3
CONSIDERATIONS
appears at the TMS 9900 inputs. One method of doing this is to use latched decoders that
create a 4-bit output corresponding to the highest level input. provided that the decoder is enabled by a latching signal. This may be illustrated functionally as follows:
( +5V
9900
-
INTREQ
I,
TMS
-
_ACO
ENABLE
-
-
*u
....
-
i i
i<1)
*n
....
-
~)
if
(HI GHEST PRIORITY)
INT 1
-::.
AC1
AC2
~~
>
-:
DECODER
_AC3
-=
--
INT 15
(L OWEST PRIORITY)
In the illustration above. 15 external interrupt requests are input to a decoder. These interrupt requests are high true.
The 15 interrupt requests are buffered. inverted. and wire-ORed to create the master interrupt request INTREO. which
is input to the CPU. This master interrupt request also enables the decoder. That is to say. when the enable input to the
3-30
decoder is high, the four outputs, ICO - IC3 will be low. When the enable input to the decoder is low, ICO - IC3 will output a 4-bit value as follows:
o
....
U
U
C"I
(")
U
U
~
o 0 0 0
o
o
o
0
o
000
o
0
o
o
o
o
o
W
~
00
o ....
m....
N
(")
It)
~
~
~
~
~
~
~
~
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
0
T
*
*
*
*
*
*
*
* * * *
* * *
* * * *
* * *
o 0
* *
000
*
0
o
o
'It
~
o
o
0
(")
~
~
~
....
'It
It)
~
~
z z z z z z z z z z z z z z z
o 0 0 0 0 0 0 0 0 0 0 0 0 0 0
000
o
o
o
C"I
0
o
o
o
0
0
0 0
000
o
o
o
o
o
o
0
0
0
0
0
0
0
000
0
0
000
000
0
0
0
0
0
0
0
0
000
0
~
0
0
0
0
0
0
0
0
0
0
0
0
0
000
0
0
0
0
000
0
000
* REPRESENTS A "DON'T CARE" BIT
* * *
* * *
* * *
* * *
* * * *
* * * *
* * * *
0
0
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
0
0
0
* * * *
* * *
000
* *
0 0 0 0
*
0
0
0
000
0
0
0
0
0
*
*
*
*
*
*
*
*
*
* *
0
*
*
*
*
*
0
If you do not use the TMS 9901, Texas Instruments suggests the following circuit to accomplish priority encoding:
+5V
u
--
..L
~
- ..-
,..
EI
GS
A2
7 .....
6 .....
74148
Ioiio.. __
(TIM
A1
---
9907)
....
~
.:
iNTi
•
(HIGHEST
PRIORITY)
•
•
-
AO
EO
,
TMS
ICO
9900
)
EI
~ GS
IC1
IC2
7
~
~
"'"
~.-
(TIM
~
Ica
A2
=--
-
•
•
INT 15 (LOWEST
PRIORITY)
External logic must maintain its interrupt request until it receives its own specific interrupt acknowledge. This
need is obvious, since an interrupt request may be denied for a long time while higher priority interrupts are being serviced.
The problem is that the TMS 9900 has no interrupt acknowledge signals.
Interrupt acknowledge signals can be generated in one of two ways:
1)
By using CRU bit instructions to set and reset external flip-flops that create interrupt acknowledge signals.
2)
By decoding appropriate addresses on the Address Bus
Figure 3-12 illustrates two possible configurations that will allow CRU bit set and reset instructions to generate interrupt acknowledge signals. The logic in Figure 3-12A generates a short interrupt acknowledge pulse.
CRUOUT becomes the input to a flip-flop which is decoded to generate CRU select signals. The CRU bit select and
"M'E'M'E'N are gated to the flip-flop's Clear input. Therefore, when CRU bit "n" is selected, CLR is removed and CRUOUT
can be clocked through. A set bit (SBO) instruction switches the flip-flop on. As soon as the flip-flop address is removed
at the end of the CRU I/O machine cycle, the flip-flop is cleared, thus terminating the interrupt acknowledge pulse.
The logic illustrated in Figure 3-12A requires that you execute an SBO instruction at the beginning of every interrupt
service routine in order to generate an interrupt acknowledge. You could require every interrupt service routine to control the length of the interrupt acknowledge pulse by executing an SBZ instruction to terminate the pulse. Figure
3-12B shows logic to implement this scheme. When the flip-flop is selected by the appropriate CRU address, CRUCLK
will clock CRUOUT to INT ACK n. At other times, CRUCLK will merely clock the flip-flop's output through, thus making
no change. In this way, only SBO and SBZ instructions which address INT ACK n can set or reset the flip-flop.
Figure 3-13 illustrates generation of an interrupt acknowledge signal by identifying specific addresses on the
Address Bus. Following any interrupt acknowledge, specific memory locations will be accessed, as identified in Figure
3-11 ,in order to fetch the new values for the Program Counter and WP register. Figure 3-13 shows a very simple
scheme whereby Address Bus lines are combined with MEMEN low to generate high pulses for the duration of a valid
address. That is to say, the interrupt acknowledge signal will last for one machine cycle - the time that the valid address exists on the Address Bus.
External logic which requested an interrupt removes its interrupt request and priority signals upon receiving an
interrupt acknowledge.
3-32
..
..
A14
A13
A12
A11
:
AD
-
;;., AO (LSB)
T
'M"EMEN
CRU
ADDR
....
~
~
· .... t
CRUBIT
ADDRESS
DECODE
A14 (MSB)
A13
A12
A~1
i\.1'EM'"Erij
9 + 5V
.. ~
n SELECT
::
..-
PRE
-
CRUOUT
.
CRUCLK
0
D
INT ACK n
CK.
7474
Q
CLR
bJ
A) Logic to create a short I NT ACK n pulse
-a
n SE LECT - - - 4.....
r---------t-~ D
Ot-....- _.. INT ACK n
r-------~~>CK
+5V.
CRUOUT
CRUCLK
B) Logic to have a programmed INT ACK n pulse length'
Figure 3-12.
A14
A5
A4
A1
MEi\iiEN
A TMS 9900 Interrupt Acknowledge Pulse Generated Using an SSO Instruction
·•·•-l-
····
·
•
.-
I, .... r
..
A5
A4
A1
i\.1'EM'"Erij
..............
Y
.00000000
INTERRUPT n
SELECT
LOGIC
I
A14
INTERRUPT ADDRESS
n SELECT
!...J
-..
INT ACK n
Figure 3-13. TMS 9900 Interrupt Acknowledge Generated by Decoding Valid Addresses
3-33
THE TMS 9900 RESET
You reset the 9900 microcomputer system by inputting a low RESET signal. This signal must remain low for at
least 3 clock periods. When the low RESET signal is removed, the following machine cycle sequence is executed:
Cycle
1
2
3
4
5
6
7
8
9
10
11
12
13
Type
ALU
ALU
ALU
MEMORY READ
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
Figure
3-3
3-3
3-3
3-4
3-3
3-5
3-3
3-5
3-3
3-5
3-3
3-4
3-3
WRITE
WRITE
WRITE
READ
Function
Prepare for Level
a interrupt
Fetch new WP register contents from memory word 000016 to temporary
storage
Store Status register contents in new R15
Store Program Counter contents in new R14
Store old WP register contents in new R13
Fetch new Program Counter contents from memory word 000'16
Load WP register from temporary storage
Thus. program execution begins with a program whose starting address is stored in memory word 1. The starting address for the 16 general purpose registers is stored in memory word O.
The TMS 9900 has a Reset instruction (RSET). In reality. this instruction resets only the interrupt mask in the Status
register; it also outputs a code on the Address Bus. as identified in Table 3-1 and illustrated in Figure 3-8. TM 990
minicomputer systems use this signal to generate a program-initiated Reset. If you are designing your own TMS 9900based microcomputer system. you are free to use the RSET instruction in any way.
THE TMS 9900 LOAD OPERATION
The LOAD input to the TMS 9900 is a non-maskable, highest priority interrupt. Load must be input low for at
least one instruction's duration. Since the length of an instruction can vary, you must use the IAQ signal to control the LOAD input pulse width. Texas Instruments' literature recommends the following circuit:
+5V
C)
..
~
~ 0
lAO
.
PRE
PRE
0
CK 7474
Q
TMS
CLR
9900
__ i:O'AD
"'-
-
J
~
D
Q
t--
~CK 7474
a
CLR
I
EXTERNAL LOAD
-
3-34
The CPU checks LOAD at the end of each instruction's execution.
After a valid LOAD input has been acknowledged. the following machine cycle sequence is executed:
Cycle
1
2
3
4
5
6
7
8
9
10
11
Type
ALU
MEMORY READ
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
MEMORY
ALU
WRITE
WRITE
WRITE
READ
Figure
3-3
3-4
3-3
3-5
3-3
3-5
3-3
3-5
3-3
3-4
3-3
Function
Input new WP register contents from memory word 7FFE 16 to temporary
storage
Store in new R15
Store incremented Program Counter contents in new R14
Store old WP register contents in new R13
Input new Program Counter contents from word 7FFF16
Load WP register from temporary storage
There are two differences between Reset and Load. First. the RESET input provides a true hardware reset. synchronizing internal operations. as well as a level 0 interrupt; LOAD provides only a non-maskable interrupt. Second. the Reset vector in bytes 0 through 3. while the Load vector is in bytes FFFC16 through FFFF 16.
In TM 990 minicomputer systems. the LREX instruction is frequently used as a software load. Output due to
LREX is identified in Table 3-1 and Figure 3-8. In a TMS 9900 microcomputer system. you can use the LREX
signal in any way.
THE TMS 9900 INSTRUCTION SET
The TMS 9900 instruction set is extremely powerful when compared to any 16-bit microprocessor described in
this book. When you consider that the TMS 9900 was first manufactured in 1976•. the power of this instruction
set becomes more impressive.
With regard to instructions described in Table 3-2 • some explanations are required.
The ABS instruction converts the contents of a memory location to their absolute value. That is to say. this instruction
assumes that the memory location contains a signed binary number. If the number is positive. nothing happens. If the
number is negative. the twos complement of the number is taken.
A number of instructions act on specific bits within source and destination memory words. These include the SOC.
SOCB. SZC. SZCB. COCo and CZC instructions. In the OPERATION PERFORMED column of Table 3-2. the word
"corresponding" means that the source word bits are affected only if selected by the destination word bit pattern. For
example. the SOC instruction will be interpreted as follows:
Source:
Destination:
Here are the new destination contents.
After SOC:
This is equivalent to an OR operation.
The SOCB instruction is identical to the SOC instruction. except that only one byte is affected. This may be any memory byte or the high-order byte of a general purpose register.
TheSZC instruction may be illustrated as follows:
Source:
Destination:
After SZC:
3-35
This is equivalent to complementing the source operand and then ANDing the two operands. The SZCB instruction is
identical to the SZC instruction. except that only one byte is affected.
The cac instruction compares Source Register 1 bits with general purpose register bits that happen to be in the same
bit positions. If all corresponding general purpose register bits are also 1. then the Equal status is set. Matches are not
significant in bit positions if the source register bit is O.
The CZC instruction operates in the same fashion as the cac instruction. except that those source memory word bits
that are 0 become significant. That is to say. if every source memory worcf 0 bit has a corresponding Workspace
register 0 bit. then the Equal status is set. Matches are not significant in bit positions if the source register bit is 1.
The BLWP instruction is a subroutine call accompanied by a context switch. The operand memory address identifies
the first of two memory words within which the new WP register and Program Counter contents will be stored.
The BLWP instruction is remarkably powerful. The subroutine call and passing parameters to the subroutine become a
single operation. The memory words that are to serve as subroutine general purpose registers can be used as general
data memory locations prior to the subroutine call. Thus. the subroutine finds its registers pre-loaded with data when it
starts executing.
The RTWP instruction should be used to return from a subroutine that is called by the BLWP instruction.
One-bit position arithmetic shifts may be illustrated as follows:
Right Shift
Left Shift
1011010110100110
1011010110100110
~"""""""~
""""""""
Lost 0 1 1 0 1 0 1 1 0 1 0 0 110 O~
1 1 0 1 1 0 1 0 1 1 0 1 0 0 1 1 Lost
Inserted
A one-bit-position logical right shift may be illustrated as follows:
1011010110100110
,~"""""'",
o10
I nserted
'!I
1 1 0 1 0 1 1 0 1 0 0 1 1 Lost
A one-bit right rotate (Shift Right Circular) may be illustrated as follows:
15 14 13 12 11
10
9
8
7
6
5
4
3
2
0
You can specify any number of bits. from 1 to 15. as the number of bit positions for any TMS 9900 shift or rotate instruction. If you specify 0 for the bit count. then the actual bit count is taken from the four low-order bits of general purpose Register RO. If these four low-order bits are 0000. then the bit count is assumed to be 16.
3-36
The following symbols are used in Table 3-2:
AG
C
CNT
CRUA
d
DATA4
DATA16
DISP
EQ
G
LG
OP
OV
PC
R
Rxx
S
ST
WP
x
[ )
/
A
V
-¥-
Arithmetic Greater Than status
Carry status
4-bit count field
CRU base address from R12
Destination memory word. There are five possible options for the destination memory word. They are
represented by these combinations of addressing modes:
Workspace Register D
Implied through Workspace Register D
Direct address
Direct. indexed address
Implied through Workspace Register D. auto-increment Workspace Register D
4-bit data unit
16-bit data unit
8-bit signed displacement
Equal status bit of Status register
Both the AG and LG statuses
Logical Greater Than status
Odd Parity status
Overflow status
Program Counter
Any of the 16 Workspace registers
Workspace register. For example. R15 is Workspace Register 15
Source memory location. Addressing options identical to destination memory location
Status register
Workspace Pointer register
Bits y through z of the quantity x. For example. ([ S) * [R)) < 31.16> represents the high-order word of
the product of the contents of the Source Register S and the Workspace Register R.
Contents of location enclosed within brackets. If a register designation is enclosed within the brackets.
then the designated register's contents are specified. If a memory address is enclosed within the brackets.
then the contents of the addressed memory location are specified.
Mu Itiplication
Division
Logical AND
Logical OR
Logical Exclusive-OR
Data is transferred in the direction of the arrow
Under the heading of STATUSES in Table 3 -2. an X indicates statuses which are modified in the course of the instruction's execution. If there is no X. it means that the status maintains the value it had before the instruction was executed.
Byte-operand instructions will affect half of a 16-bit memory word. If the word is accessed as a general purpose
register, then only the high-order byte will be affected. If the word is accessed as non-register memory, then
the byte affected is determined by the least significant bit of the 16-bit address: 0 selects the high-order byte;
1 selects the low-order byte.
3-37
Table 3-2. TMS 9900 Instruction Set Summary
STATUSES
TYPE
MNEMONIC
OPERAND(S)
BYTES
OPERATION PERFORMED
G
g
EQ
C
OV
OP
LDCR
S,CNT
2
X
X
X*
STCR
O,CNT
2
X
X
X*
SBO
DISP
2
SBZ
DISP
2
TB
OISP
2
X
[CRUA]-[ S ]
Transfer the specified number of bits from source memory word to the CRU.
[D ] - [ CRUA]
Transfer the specified number of bits from the CRU to destination memory word.
[CRUA + OISP]- 1
Set bit in CRU to 1.
[CRUA + DISP] - 0
Set bit in CRU to O.
If [CRUA+DISP] =0, then [EQ]=l;orelse [EQ]=O
Test bit in CRU.
w
»(J
a:a:z
MOV
S,D
2
X
X
MOVB
S,O
2
X
X
COw
~~a:
-ww
g:~~
Co)
fA
00
C
a:
w
A0
A
S,D
2
X
X
X
X
AB
S,O
2
X
X
X
X
X
X
X
X
X
X
S
S,D
2
SB
S,O
2
X
X
S,D
S,O
2
2
X
X
X
w
C
CB
w
XOR
S,R
2
X
X
MPY
S,R
2
>
a:
0
::E
a:
~
w
I
X
16-bit add contents of source memory word to contents of destination memory word.
[O]-[S]+ [0]
a-bit add contents of source memory byte to contents of destination memory byte.
[D]-[D]- [S]
X
i
a-bit subtract contents of source memory byte from contents of destination memory byte.
Set status flags based on 16-bit comparison of source and destination memory word contents.
Set status flags based on a-bit comparison of source memory byte contents and destination
memory byte contents.
[R]-[S]¥ [R]
2
INC
0
2
X
X
X
X
INCT
0
2
X
X
X
X
Increment contents of memory word by 1.
[0]-[0]+2
DEC
0
2
X
X
X
X
Increment contents of memory word by 2.
[0]-[0]-1
X
>
a:
0
z
(/J
Decrement contents of memory word by 1.
*OP status is affected only if between 1 and a bits are transferred.
I
I
16-bit subtract contents of source memory from contents of destination memory word.
[D]-[D]- [S]
S,R
C
0
w
16-bit move contents of source memory word to destination memory word.
[D]-[S]
DIV
::E
(J
i
I
Multiply the contents of source memory word by contents of Workspace Register R. Store most
significant word of result in R. Store least significant word of result in Workspace Register R + 1.
[R]-([R,R + 11/ [S]Xquotient)
[R + 11-( [R,R + 111 [S]Xremainder)
Divide the 32-bit quantity represented by R (high-order word) concatenated with R + 1 (low
order) by the contents of the source memory word. Store the quotient in R, the remainder in
R + 1 and set overflow if quotient will exceed 16 bits.
[O]-[D]+ 1
>
a:
0
i
I
[O]-[S]
Exclusive-OR contents of source memory word with Workspace Register R.
[R]-[([S]* [R])<31,16 >]
[R+ 1l-[([S]*[R])<15,O>]
Z
w
a:
w
II.
w
!
[D]-[S]+ [D]
X
X
~
(J
I
a-bit move contents of source memory byte to destination memory byte.
iii
~
X
I
I
Table 3-2. TMS 9900 Instruction Set Summary (Continued)
STATUSES
TYPE
OPERAND(S)
MNEMONIC
OPERATION PERFORMED
BYTES
G
EQ
X
X
C
OV
DECT
D
2
CLR
D
2
SETO
D
2
INV
D
2
X
X
NEG
D
2
X
X
X
X
ABS
D
2
X
X
X
X
0
~
SWPB
D
2
-
SOC
S,D
2
X
X
saCB
S,O
2
X
X
SZC
S,D
2
X
X
SZCB
S,D
2
X
X
COC
S,R
CZC
S,R
2
LI
R,DATA16
4
LWPI
DATA 16
4
is
X
OP
[D]-[D]-2
X
Decrement contents of memory word by 2.
[D]-OOOO
Clear the ~~stination memory word.
[D]-FFFF
Set all bii~ of memory word.
[D]-(O]
1&1
=:)
Z
i=
Z
0
9
W
~
c(
Ones complement the destination memory word.
,[D]-['D.J+l
Twos complement the destination memory word.
[D]-I [D]I
Take the absolute (unsigned) value of the destination memory word's contents.
a:
1&1
A.
0
>
a:
[D<15,8>]--[0<7,O>]
Exchange the high and low bytes of the memory word.
If [S]=l,then [D]-l
Set the bits in the destination memory word that correspond to ls in the source memory word
1&1
~
Col
W
co
1&1
CJ
Z
1&1
a:
1&1
u..
1&1
a:
>
a:
0
~
X
for 8 bits.
If [S]=l, then [D]-P
Clear the bits in the destination memory word that correspond to ls in the source memory word
1&1
~
>
a:
c(
Q
Z
0
U
1&1
X
CI)
X
for all 16 bits.
If [S]="l, then [O]-l
Set the bits in the destination memory word that correspond to ls in the source memory word
X
for all 16 bits.
If [S]=l, then [D]-O
Clear the bits in the destination memory word that correspond to ls in the source memory word
for 8 bits.
If for all [S]=l, [R]=l,then [EQ]-l
If the bits in the Workspace Register R that correspond to the set bits in the source memory
word are all ls, set the EQUAL status.
If for all [S]=l, [R]=O, then [EO]=l
If the bits in the Workspace Register ~t correspond to set bits in the source memory word
are all Os, set the EQUAL status.
1&1
~
c(
is
1&1
:::E
~
-~--
-
--
-
X
X
[R]-DATA16
Load immediate to Workspace Register R.
[WR]-DATA16
Load immediate to Workspace Pointer Register, WR.
Table 3-2. TMS 9900 Instruction Set Summary (Continued)
STATUSES
TYPE
w
MNEMONIC
OPERAND(S)
BYTES
OPERATION PERFORMED
G
EO
C
OV
OP
Set the status flags based on 16-bit comparison between contents of Workspace Register Rand
CI
R,DATA16
4
X
X
11-
AI
R,DATA16
4
X
X
I-
ANDI
R.DATA 16
4
X
X
ORI
R,DATA16
4
X
X
B
S
2
[PC]+-[S]
JMP
D(SP
2
Branch unconditional to address in Source memory word.
[PC]+-[ PC] + DISP
Branch unconditional.
BL
S
2
BLWP
S
2
le(
immediate data.
[R]-[R] + DATA 16
II:
w
0
w
e(
0w
~
X
X
Add immediate to Workspace Register R contents.
[R]+-[R] A DATA 16
AND immediate with Workspace Register R contents.
[R]+-[ R] V DATA 16
OR immediate with Workspace Register R contents.
;:!
11-
~
..,
::;)
2
II:
[R111-[PC]+1
[PC]-[S]
Branch to subroutine at address in source memory word.
[R13]-[WP]
[R14]+-[ PC]
[R15]+-[ST]
[WP]+-[S]
W::;)
21Col
~
~~
::;)Q
~2
......
1IIe(
::;)
0
e(
(,)
2
RTWP
X
X
X
X
X
[PC]-[S+2]
Branch to subroutine whose address is stored in source memory word + 1. Perform context
switch to RD address contained in source memory word.
[WP]-[R13]
[PC]+-[R14]
[ST]-[R15]
Perform a backward context switch.
JEQ
DISP
2
JNE
DISP
2
JGT
DISP
2
JlT
DISP
2
2
JHE
DISP
2
Branch on arithmetic greater than.
If [AG]=O and [EO]=O; then [PC]+-[PC] +DISP
Branch on arithmetic less than.
If [lG]=1 or [EO]=1; then [PC]+-[PC] +DISP
:z:
(,)
JH
DISP
2
Branch on logical greater than or equal.
If [LG]=1 and [EO]=O; then [PC]+-[PC]+DISP
Jl
DISP
2
JLE
DISP
2
2
0
~
is
2
0
(,)
0
2
~
III
If [EO]=1; then [PC]+-[PC] +DISP
Branch if equal.
If [EO]=O; then [PC]-[PC] +DISP
Branch if not equal.
If [AG]=1; then [PC]-[PC] + DISP
Branch on logical greater than.
If [lG]=O and [EO]=O; then [PC]-[PC] +DISP
Branch on logical less than.
If [EO]=1 or [LG]=O; then [PC]+-[PC]+DISP
Branch on less than or equal.
Table 3-2. TMS 9900 Instruction Set Summary (Continued)
STATUSES
TYPE
MNEMONIC
OPERAND(S)
OPERATION PERFORMED
BYTES
G
EQ
C
OV
OP
JNC
DISP
2
If [C]=O; then [PC]-[PC]+DISP
ZZe
OOw
JNO
DISP
2
Branch on carry reset.
If [OV]=O;then [PC]-[PC]+DISP
z°j::
JOC
DISP
2
JOP
DISP
2
~Ei
~ZZ
00
m U !:!
Branch on overflow reset.
If [C]=l;then [PC]-[PC]+DISP
Branch on carry set.
If [OP]=l;then [PC]-[PC]+DISP
Branch on odd parity set.
a: W
W l-
I- 0(
~ ~
~ 0
SLA
SRA
2
X
X
X
X
X
X
X
X
2
SRL
R,CNT
R,CNT
R,CNT
SRC
R,CNT
2
2
STST
R
2
STWP
R
2
L1MI
DATA4
4
XOP
S,R
2
X
X
X
X
Arithmetic shift the Workspace Register R left the specified number of bits.
Arithmetic shift the Workspace Register R right the specified number of bits.
Logical shift the Workspace Register R right the specified number of bits.
Rotate the Workspace Register R right the specified number of bits.
X
[R]-[ST]
. Store the Status register into Workspace Register R.
[RJ-[WP]
. Store the Workspace Pointer into Workspace Register R.
w
~
01ZA.
0(;:)
0a:
;:)IZ:
I-w
0(1I-Z
0-
[SR<3,O>] -oATA4
Load immediate data into the interrupt mask bits of the Status register.
X
[R13]-[WP]
[R14]-[PC]
[R15]-[ST]
[Rll]-[S]
[WP]-[40,. +(4*[R)))
[PC]-[41,. +(4* [R)))
Perform a context switch. This is the software interrupt.
W
I-
X
S
2
Execute the instruction represented by the data in the source location. If that instruction has immediate operand words, those words must be located directly after the X instruction. The instruction [S] will affect the status flags but its fetch will not cause IAQ to go high.
2
CPU enters Halt state.
CPU clears interrupt mask and outputs 001 on three high-order Address Bus lines.
011 on three high-order Address Bus lines.
;:)
U
W
)(
W
IDLE
>
:le
o(w
ZZ
a:jL
WW
t<0
W
RSET
CKOF
CKON
LREX
110 out on three high-order Address Bus lines.
101 out on three high-order Address Bus lines.
111 out on three high-order Address Bus lines.
THE BENCHMARK PROGRAM
For the TMS 9900, our benchmark program may be illustrated as follows:
LOOP
BLWP
MOVE
CONTEXT SWITCH TO APPROPRIATE REGISTERS
MOV
DEC
JNE
RTWP
@IOBUF(R11.*R2+
R1
LOOP
LOAD NEXT INPUT WORD IN NEXT TABLE WORD
DECREMENT COUNT
RETURN FOR MORE
RETURN FROM SUBROUTINE
Let us look at how our benchmark program can collapse to just five instructions.
We assume that there is some set of 16 General Purpose registers within which we store the word count and the address of the first free word in TABLE. We illustrated this idea when describing context switching earlier in the chapter.
Observe that Register R1 contains the word count and is therefore used as an Index register. while Register R2 addresses the first free word in TABLE. Note that the contents of Register R2 are incremented automatically when the
next byte is loaded into the table.
The BLWP instruction will branch to the program which performs the required data move. but simultaneously it loads
the Workspace register with the appropriate initial address. We do not need to load any initial addresses or word
counts into registers. since we have adopted the memory space where this data is stored to serve as our General Purpose registers.
After the move has been completed. we do not have to update any counters or pointers. because they were updated
"in situ". All we have to do upon completing the move is store the contents of the current General Purpose Registers 13
and 14 to the Workspace register and Program Counter.
The following notation is used in Table 3-3:
aa
bb
Two bits determining the addressing mode for the destination memory word
Two bits determining the addressing mode for the source memory word
cccccccc
dddd
8-bit Signed address displacement
Four bits used with aa to determine the destination memory word
eeee
rrrr
4-bit count field
Four bits choosing the Workspace register
Four bits used with bb to determine the source memory word
16 bits of immediate data
ssss
xx
If either aa or bb is 1°2. and the corresponding register specified is 02. then an additional 16-bit direct memory address
word. used in computing the effective memory address of the operand. will follow the instruction.
If aa and bb are 1°2. and both corresponding register specifications are 0. then two additional 16-bit direct memory addreSSing words will follow the instruction: the first will be used in computing the source address; the second will be
used in computing the destination address.
3-42
Table 3-3.
INSTRUCTION
TMS 9900 Instruction Set Object Codes
OBJECT CODE
BYTES
CLOCK
INSTRUCTION
PERIODS·
OBJECT CODE
BYTES
CLOCK
PERIODS·
A
S,D
1010aaddddbbssss
2
DISP
000lll00cccccccc
2
S,D
1011aaddddbbssss
2
14-30 (1)
14-30 (1)
JOP
AB
LDCR
S,CNT
00ll00eeeebbssss
2
8/10(15)
22-52 (11)
ABS
0
0000011101aadddd
2
12-20 (6)
LI
R,DATA16
000000100000mr
4
12 (19)
AI
R,DATA16
00000o 1000 lOrrrr
4
14(17)
0000001100000000
4
16(21)
ANDI
R,DATA16
00000o 100 100rrrr
4
14(17)
XX
0000001111100000
2
6(14)
B
S
2
8-16 (7)
0000001011100000
4
10(20)
BL
XX
LlMI
XX
00000 1000 1bbssss
XX
DATA4
LREX
LWPI
DATA 16
MOV
S,D
l100aaddddbbssss
2
14-30 (1)
1101aaddddbbssss
2
14-30 (1)
00lllOrmbbssss
2
0000010100aadddd
2
52-60 (2)
12-20 (5)
S
0000011010bbssss
2
12-20 (9)
BLWP
S
00000 1OOOObbssss
2
26-34 (10)
C
S,D
l000aaddddbbssss
2
14-30 (1)
MOVB
S,D
CB
S,D
100 1aaddddbbssss
2
S,R
S,D
000000101000mr
4
14-30 (1)
14 (18)
MPY
CI
NEG
0
ORI
R,DATA16
0000001111000000
2
6 (14)
XX
CKON
CKOF
CLR
0
COC
S,R
CZC
S,R
DEC
0
DECT
0
DIV
S,R
IDLE
0000001110100000
2
6(14)
2
10-18 (5)
RSET
RlWP
00 1OOOrmbbssss
2
10-18 (1)
5
00l00lrrrrbbssss
2
14-22 (1l
00000ll000aadddd
2
14-22 (5)
00000ll00laadddd
2
10-18 (5)
00 1111 rrrrbbssss
2
10-18 (3)
0000001101000000
2
0000010110aadddd
INV
JEQ
JGT
14(17)
000000100110mr
XX
00000l00llaadddd
0
0
0
INC
XX
6 (14)
0000001101100000
2
0000001110000000
2
14 (8)
5,0
0110aaddddbbssss
2
14-30(1)
SB
S,D
0111 aaddddbbssss
2
14-30 (1)
SBO
DISP
00011101cccccccc
2
12 (13)
SBZ
DISP
00011110cccccccc
2
12(13)
SETO
0
00000lll00aadddd
2
10-18 (5)
6 (14)
SLA
R,CNT
0000 10 1Oeeeerrrr
2
14-52 (16)
16-124 (5)
SOC
S,D
l110aaddddbbssss
2
14-30(1)
0000010111 aadddd
2
10-18 (5)
50CB
S,D
1111 aaddddbbssss
2
14-30 (1)
0000010101 aadddd
2
10-18 (5)
SRA
R,CNT
00001000eeeerrrr
2
14-52 (16)
DISP
00010011cccccccc
2
10-18 (15)
SRC
R,CNT
00001011eeeemr
00010101cccccccc
2
8/10 (15)
SRL
R,CNT
0000100 leeeemr
2
2
14-52 (16)
DISP
JH
DISP
00011011cccccccc
2
8/10 (15)
STCR
D,CNT
001101eeeeaadddd
2
42-60 (12)
JHE
DISP
00010100cccccccc
2
8/10 (15)
STST
R
000000101100mr
2
8(23)
JL
DISP
000 11010cccccccc
2
8/10 (15)
STWP
R
000000101010mr
2
8 (22)
JLE
JLT
DISP
000 100 1Occcccccc
2
8/10 (15)
SWPB
0
0000011011aadddd
2
10-18 (23)
DISP
000l000lcccccccc
2
8/10 (15)
SZC
S,D
ol00aaddddbbssss
2
14-30 (1)
JMP
DISP
000l0000cccccccc
2
10(15)
SZCB
5,0
0101aaddddbbssss
2
14-30(1)
JNC
DISP
00010111cccccccc
2
8/10(15)
TB
DISP
00011111cccccccc
2
12(8)
JNE
DISP
00010110cccccccc
2
8/10(15)
x
5
00000100 lObbssss
2
8-16 (7)
JNO
DISP
000ll00lcccccccc
2
8/10 (15)
XOP
S,R
00 10 11 rrrrbbssss
2
44-52 (4)
JOC
DISP
000ll000cccccccc
2
8/10 (15)
XOR
S,R
001010rmbbssss
2
14-22 (1)
INCT
• The number in brackets identifies the instruction's machine cycle sequence, as defined in the preceding text.
3-43
14-52 (16)
The minimum and maximum number of clock periods for the execution of each instruction are shown in the
CLOCK PERIODS column of Table 3-3. Remember that a machine cycle consists of two clock periods. The
bracketed number after the number of clock periods identifies the machine cycle sequence. Machine cycle sequences associated with each bracketed number are listed below. In the machine cycle list below, the follow:ing abbreviations are used:
R represents a memory read machine cycle as identified in Figure 3-4.
A represents an ALU machine cycle as illustrated in Figure 3-3.
W represents a memory write machine cycle as illustrated in Figure 3-5.
C represents a CRU machine cycle as illustrated in Figures 3-6 and 3-7.
A subscript associated with any machine cycle notation identifies that machine cycle repeated a number of times. Thus
A3 is equivalent to -A-A-A-.
M represents memory address computation machine cycles. Memory address computations were described earlier in
this chapter. In su mmary. here are the various possibilities for M:
Register addressing:
R
Implied memory addressing:
R-A-R
Implied memory addressing with auto-increment (for byte operand): R-A-W-R
Implied memory addressing with auto-increment (for word operand): R-A-A-W-R
Direct addressing:
A-A-R-A-R
Direct. indexed addressing:
R-A-R-A-R
(1)
(2)
(3)
(4)
(5)
(6)
(7)
~)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
R-A-M-A-M-A-W
R-A-M-A-R-A18-W-A-W
R-A-M-A-R-A-A-R-Ax-W-A-W (51 ~ x :::; 35)
R-A-M-A3-R-A-W-A-W-A-W-A-W-A-R-A
R-A-M-A-W
R-A-M-A3-W-A
R-A-M-A
R-A-A-R-R-R-A
R-A-M-A-A-W
R-A-M-A-A-W-A-W-A-W-A-R-A
R-A-M-A4-R-A-C x-A (16 ~ x ~ 1)
R-A-M-A-R-A-A-Cx-Ay-W (16 ~ x ~ 1. 11 ~ y ~ 5)
R-A-A-R-A-C
R-A-A-C-A-A
R-Ax (x=3 or 4)
R-A-R-A-A-R-Ax-W-A (18 ~ x ~ 3)
R-A-A-R-R-A-W
R-A-R-A-R-A-A
R-A-A-R-A-W
R-A-A-R-A
R-A-A-R-A3
R-A-A-W
R-A-M-A-R-A4-W
THE TMS 9980A AND THE TMS 9981 MICROPROCESSORS
The TMS 9980A and the TMS 9981 are low-cost variations of the TMS 9900. The principal differences between the TMS 9900 series and TMS 9980 series microprocessors are summarized in Table 3-4. Differences
between the TMS 9980A and the TMS 9981 are summarized in Table 3-5.
This discussion of the TMS 9980 series microprocessors covers only differences as compared to the TMS 9900.
The TMS 9980 series microprocessors are manufactured using N-channel silicon gate MOS technology. They are
packaged as 40-pin DIPs. The TMS 9980A uses three power supplies: -5V. +5V. and +12V. The TMS 9981 useS two
power supplies: +5V and +12V.
Typically. a c'lock cycle time of 400 nanoseconds will be used with TMS 9980 series microprocessors. This generates
instruction execution times ranging between 4 and 14 microseconds.
3-44
Figure 3-14 illustrates that part of general microcomputer system logic which is implemented by the TMS
9980 series microprocessors. This figure is identical to Figure 3~ 1. with the exception of clock logic. which is now
shown present.
Programmable registers are implemented and used in exactly the same way the TMS 9900 and TMS 9980
series microprocessors. Note. however. that the TMS 9980 series microprocessors address a 2048-bit CRU;
therefore. bits 1 through 11 of Register R12 identify the origin of any CRU bit field. The TMS 9900 uses bits 1 through
12 of Register R12 to identify the CRU origin within a 4096-bit CRU.
Table 3-4. A Summary of Differences Between the TMS 9900 and TMS 9980 Series Microprocessors
FUNCTION
Addressable external memory
DIP pins
Data Bus
Address Bus
External interrupt priorities
CRU field width
Clock logic
TMS 9900
TMS 9980AITMS 9981
32.768 x 16-bit words
64
16 bits
16.384 x 8-bit words
40
8 bits
13 bits
4
15 bits
15
4096 bits
2048 bits
Four external inputs
One external input
or internal (TMS 9981
only)
Table 3-5. A Summary of Differences Between the TMS 9980A and TMS 9981 Microprocessors
FUNCTION
Power supplies
Clock logic
Pin incompatibility ties
TMS 9980A
TMS 9981
-5V. +5V. +12V
One external input
+5V. +12V
One external input
or crystal only
DO - 07. INTO - INT2. cJ>3
The TMS 9980 series microprocessors have a 14-line Address. Bus, used to address up to 16,384 bytes of
memory. In contrast. the TMS 9900 addresses up to 32.768 16-bit words of external memory. Thus. TMS 9980 programs address memory as bytes. while externally generated addresses also select bytes. The TMS 9900. by way of contrast. addresses memory as bytes within the CPU. but as 16-bit words externally.
The TMS 9980 series microprocessors use exactly the same memory and CRU addressing techniques as the
TMS 9900. General-purpose registers are used in the same way. and instruction object codes are identical.
The Status register and Status flags used by the TMS 9980 series microprocessors are identical to those which
we have already described for the TMS 9900.
TMS 9980 SERIES MICROPROCESSOR PINS AND SIGNALS
Figure 3-15 illustrates pins and signals for the TMS 9980A. Figure 3-16 provides the same information for the
TMS 9981. In both of these illustrations. signal names conform to Texas Instruments nomenclature. For the Data and
Address Busses. our notation is given in brackets. Differences result from the fact that we number bits from right to left
(0 being the low-order bit). while Texas Instruments numbers bits from left to right (0 becomes the high-order bit). TMS
9980AITMS 9981 pin-out differences are shaded in Figures 3-15 and 3-16 so that you can identify them
quickly.
For descriptions of the individual signals, refer to the earlier TMS 9900 discussion.
3-45
Accumulator
Registens)
Data Countens)
Programmable
Timers
Figure 3-14.
I/O Ports
Memory
Logic of the TMS 9980A and TMS 9981 Microprocessors
3-46
HOLD
HLDA
IAQ
(LSB) (AO) CRUOUT/A13
(A1) A12
(A2) A11
(A3) A10
(A4) A9
(A5) A8
(A6) A7
(A7) A6
(A8) A5
(A9) A4
(A10) A3
(A11) A2
(A12) A1
(MSB) (A13) AO
DBIN
CRUIN
(+5V) VCC
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
TMS
9980A
17
18
19
20
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
(LSB)
(MSB)
Type
Description
Pin Name
AO-A13
Address Bus
Tristate, output
00-07
Data Bus
Tristate, bidirectional
CKIN
Clock signal in
Input
cJ>3
Synchronizing clock
Output
Memory Enable
Tristate, qutput
IAQ
I nstruction Fetch
Output
DBIN
Data Bus in
Tristate, output
WE
Write Enable
Tristate, output
READY
Memory Ready
Input
WAIT
Wait State indicator
Output
CRUCLK
I/O clock
Output
CRUOUT
Serial I/O out
Output
CRUIN
Serial I/O in
Input
INTO, INT1, INT2
I nterrupt request and priority
Input
HOLD
DMA request
Input
HOLDA
Hold acknowledge
Output
Power and Ground reference
Figure 3-15. TMS 9980A Signals and Pin Assignments
3-47
HOLD-~"
1
2
IAQ .....1----1 3
(LSB) (AO) CRUOUT/A13~~--I 4
(A1) A12 ~---I 5
(A2) A 11 ....1----1 6
(A3) A10 ....- - - 1 7
(A4) A9 .....1----I 8
(A5) A8 ....- - - 1 9
( A6) A 7 .....'""----1 10
(A7) A6 ~I----I 11
(A8) A5 ....-----1 12
(A9) A4 ....- - f 13
(A 1 0) A3 ....1----1 14
(A11) A2 ....1----4 15
(A12) A1 ....1----1 16
(MSB) (A 13) AO ...1----1 17
OBI N ....I----t 18
CRUIN
19
(+5V) VCC - - - - I 20
HLDA~---I
TMS
9981
33
32
31
30
29
28
27
26
25
24
23
22
21
(LSB)
(MSB)
Type
Description
Pin Name
AO-A13
Address Bus
Tristate, output
00-07
Data Bus
Tristate, bidirectional
CKIN
Clock or crystal connection
InptAt
OSCOUT
Crystal connection
Output
Synchronizing clock
Output
MEMEN
Memory Enable
Tristate, output
IAQ
I nstruction Fetch
Output
DBIN
Data Bus in
Tristate, output
WE
Write Enable
Tristate, output
READY
Memory Ready
Input
WAIT
Wait State indicator
Output
CRUCLK
I/O clock
Output
CRUOUT
Serial I/O out
Output
CRUIN
Serial I/O in
Input
INTO, INT1, INT2
I nterrupt request and priority
Input
HOLD
DMA request
Input
HOLDA
Hold acknowledge
Output
v CC'
Power and G round reference
V DO' V SS
Figure 3-16. TMS 9981 Signals and Pin Assignments
3-48
TMS 9980 SERIES MICROPROCESSOR TIMING AND INSTRUCTION EXECUTION
The TMS 9980A and TMS 9981 microprocessors have the same signal relationships and instruction execution
sequences as the TMS 9900. The few minor waveform differences are identified in the data sheets at the end of
this chapter.
The only significant difference between the TMS 9900 and TMS 9980 series is in clock logic. The TMS 9900 re e
quires four clock inputs. as identified in Figure 3-3.
The TMS 9980A requires a single clock signal. input via CKIN. The frequency of this clock input
must be four times the desired clock frequency. That is to say. CKIN will be divided by four in
order to create one clock period. The TMS 9981 can operate with the same CKIN input as the TMS
9980A; however. you can also connect a crystal across CKIN and OSCOUT. This may be illustrated
as follows:
TMS 9980
SERIES
CLOCK
LOGIC
CKIN ~------~------------____~
D
OSCOUT t-------....---------.
C1 and C2 must have values between 10pf and 25pf. typically 15pf.
-
The crystal must be of the fundamental frequency type. The frequency will be divided by four in order to create the internal clock frequency.
Both the TMS 9980A and the TMS 9981 output <1>3. a synchronizing clock Signal. <1>3 is the inverse of the <1>3 clock signal shown in Figure 3-3 and in subsequent timing diagrams for the TMS 9900.
Thus you can create the timing diagram for any TMS 9980 operation by looking at the equivalent timing diagram for
the TMS 9900 and replacing the four TMS 9900 clock signals by a single timing pulse which will be the complement of
<1>3.
The following operations are identical within TMS 9900 and TMS 9980 systems:
• Memory references. However. note that memory reference will consist of two memory access cycles. as a 16-bit word
is handled as two bytes.
·CRU I/O operations (remember that the TMS 9980 series CRU is only 2048 bits wide).
·CRU control operations
·The Wait state
•The Hold state and direct memory access operations
•The Halt state
•The interaction of Hold and Halt states
Refer to the TMS 9900 discussion for any of the above topics.
TMS 9980 SERIES INTERRUPT LOGIC
The TMS 9980A and TMS 9981 microprocessors support four levels of external interrupt. together with a Reset and a
Load. Reset and Load are non-maskable interrupts. In contrast. the TMS 9900 supports 15 levels of external interrupt.
along with Reset.
The TMS 9980 series microprocessors identify external interrupts via the INTO, INT1, and INT2 inputs as
shown in Table 3-6. Figure 3-17 shows the interrupt vector map.
3-49
Table 3-6. TMS 9980 Interrupts
I nterrupt Decoded
INTO
INT1
INT2
0
0
0
Reset
0
0
1
Reset
0
1
0
Load
0
1
1
Level 1 (Highest Priority)
1
0
0
Level 2
1
0
1
Level 3
1
1
0
Level 4 (Lowest Priority)
1
1
1
No Interrupts
Observe that the TMS 9980A and the TMS 9981 have no iNTRffi input. Also. the Reset and Load non-maskable interrupts are decoded from the INTO - INT2 inputs.
Figure 3-18 shows some pin connections for various levels of interrupt complexity in a TMS9980 series microcomputer system. The three illustrations shown are self-evident; they simply implement the INTO - INT2 codes defined
above.
The TMS 9980 series microprocessors provide all 16 XOP software interrupts available with a TMS 9900.
Observe that Figure 3-17 shows memory as 8-bit units in contrast to Figure 3-11, which shows memory as 16-bit
units. This reflects the fact that external memory is addressed as bytes by the TMS 9980A and the TMS 9981.
3-50
Memory
Address
Reset
~
~
('
External Interrupt.
<
Una..'oned Memo",
for Programs or Data
tt
Memory Byte
Content
0000
WP
HI
0001
WP
La
0002
PC
HI
-
0003
PC
La
0004
WP
HI
0005
WP
La
0006
PC
HI
...
t
WP
Reset Vector
). PC
-
<
). WP
,
0007
PC
- > ~C t
La
0008
WP
HI
0009
WP
La
OOOA
'PC
HI
OOOB
PC
La
OOOC
WP
HI
0000
WP
La
OOOE
PC
HI
OOOF
PC
La
0010
WP
HI
0011
WP
La
0012
PC
HI
0013
PC
La
Leve.! 1 Vector
~
>WP
}
Level 2 Vector
). PC
~
>WP
}
Level 3 Vector
; PC
1WP
1
} Leve' 4 Veeto'
J
PC
L
0040
WP
HI
0041
WP
La
0042
PC
HI
0043
PC
}
}
WP
La
}
PC
}wp }
xap 0 Vector
xap Vectors, Use
Same Memory Space
as the TMS 9900
Unassigned Memory ,
.
007C
WP
HI
0070
WP
La
007E
PC
HI
007F
PC
3FFC
WP
HI
3FFO
WP
LO
LO
}
xap 15 Vector
PC
t
Loed {
3FFE. PC
HI
PC
La
3FFF
~wp
PC
}
Figure 3-17. TMS 9980 Memory Map
3-51
Load Vector
LEVEl.. 1
RES'ET
r
INTO
I
r--f......}-
INTO
TMS 9980AI
TMS 9981
INT1
LEVEl.. 4
INT2
TMS 9980A/
TMS 9981
A) Using Reset and One
External Interrupt
RESET
INT1
['5Ai5
INT2
B) Using. Reset, I..oad and
One External Interrupt
Vee
)
!
~H
RESET
...
Lc5A5
I..EVEI..1
LEVEL 2
LEVEL 3
LEVEL 4
...
...
~
--
Power, Ground
Type
Output
Output
Input
Output
Input
Output
Figure 3-23. TIM 9904 Signals and Pin Assignments
OSCOUT provides a clock frequency four times that of the 1 _ _...-./
\~------------------~/
4>2 ______________~1
\~
\
________________
4>3 ______________________~/
\
4>4~~______________________________~/
3·68
\
~r_
When an external quartz crystal is used to drive the TIM 9904, the following connections are required:
TANK 1
0.47 J.LH
......_ .......~_--1 2
TANK 2
19
XTAL 1
D
·TIM
9904
17
20 ohm to. 75 ohm crystal,
2 mw power dissipation.
(May substitute a
0.1 J.LF capaciton)
OSCIN
OSCIN must be tied to a high logic level for the internal clock logic to work properly.
Required capacitor and inductance values are shown in the illustration above for a TMS 9900 microprocessor operating
with its standard 3 MHz frequency. The crystal must have a resonant frequency of 48 MHz. For 48 MHz operation. a
third overtone crystal is used.
For less precise timing. the quartz crystal may be replaced with a 0.1 JLf capacitor. The LC-tuned circuit now establishes the clock frequency according to the following equation:
fosc = 1/(2",$)
where L is the inductance. with units of Henries. and C is the capacitance with units of Farads. This includes the
capacitance of the circuit into which the components are mounted.
If an external clock signal is input, it must occur at OSCIN. The crystal connections XTAL 1 and XTAL2 should be
connected to VCC as follows:
+ 5V
NOT
{TANK 1
CONNECTED
TANK 2
20
2
19
TIM
9904 18
17
3-69
XTAL2
XTAL1
OSCIN
} TIED TO LOGIC '1'
CLOCK INPUT
The clock input OSCIN must have a frequency which is four times the clock period frequency and has a 25% duty cycle.
Thus. for a 3 MHz frequency, a 12 MHz signal must be input via OSCIN:
1..........- - OSCIN
"1
83.3ns
--Jr---\,__________--.',..--, ______
\~
~20.8ns~
I
In TMS 9900 microcomputer systems, the 0 input is used for an asynchronous reset; Q is output as a synchronous reset. This may be illustrated as follows:
Vee
)
.
10Kn:
100.0
.AAA
1
!
-
TIM
TMS
9904
9900
a
0
..-
RESET
1}LF ---
I-
The illustration above shows recommended resistor and capacitor values.
THE TMS 9901 PROGRAMMABLE SYSTEM INTERFACE (PSI)
The TMS 9901 Programmable System Interface (PSI) is a special support part designed for the TMS 9900 series
of microprocessors. This relatively primitive device uses 32 bits of the TMS 9900 CRU bit field to support
parallel I/O and interrupt request logic. Programmable timer logic is also available.
Figure 3-24 illustrates that part of general microcomputer system logic which has been implemented on the
TMS 9901 PSI.
The TMS 9901 PSI is packaged as a 40-pin DIP. It uses a single +5V power supply. All inputs and outputs are TTL-compatible. The device is implemented using N-channel silicon gate MOS technology.
3-70
Clock Logic
Accumulator
Registensl
Arithmetic and
Logic Unit
Data Countens)
Stack Pointer
Program Counter
Read Only
Memory
Figure 3-24.
Logic of the TMS 9901 Programmable System Interface
3-71
--
--....
RS'fi
CRUOUT
CRUCLK
CRUIN
iN'i'6
iN'f5
-- .-.
INT4
iNT3
Cii
INTREQ
(LSB) IC3
IC2
IC1
(MSB) ICO
Vss
wn
INT2
P6
P5
---
..
--
----
-
...
:
- ----
....
Pin Name
CRUIN
CRUOUT
CRUCLK
PO - P15
iNTT -INT15
INTREQ
ICO - IC3
CE
SO - S4
RSTf
4l
VCC' Vss
1
2
3
4
5
6
7
. 8
9
10
11
12
13
14
15
16
17
18
19
40
39
38
37
36
35
34
TMS
9901
-20
33
32
31
30
29
28
27
26
25
24
23
22
21
--- -..
-- ..
...
----- ---.-.
-- .....
-- ..-..
--- ---- ..
---- --.....
-.
-- -
Description
CRU data output
CRU data input
CRU data input strobe
I/O data
External interrupt requests
Interrupt request to CPU
Interrupt priority designation
Chip Enable
CRU bit select
Chip reset
Synchronizing clock signal
Power, Ground reference
VCC (+5V)
SO (MSB)
PO
P1
S1
S2
INT7/P15
INT8/P14
INT9/P13
INT10/P12
INT11/P11
iN'fi2/P10
INT13/PS
INT14/P8
P2
S3
S4 (LSB)
INT15/P7
P3
P4
Type
Output
Input
Input
Input or Output
Input
Output
Output
Input
Input
Input
Input
Figure 3-25. TMS 9901 Programmable System Interface Signals and Pin Assignments
In the illustration above, Address lines have been numbered using our standard notation, whereby A 14 is the highestorder address line and AO is the lowest-order address line. This is the opposite of Texas Instruments' notation. The CRU
select lines are numbered according to Texas Instruments' notation and Figure 3-25. Therefore, S4 is connected to
AO, and SO is connected to A4.
3-72
TMS 9901 PSI PINS AND SIGNALS
The TMS 9901 pins and signals are illustrated in Figure 3-25. The signals which connect the TMS 9901 to a
TMS 9900 series microprocessor are quite straightforward; they consist of the CRU and interrupt signals.
The CRU signals include CRUIN. CRUOUT. and CRUCLK.
The interrupt signals consist of INTREQ. ICO, IC1, IC2, and IC3.
For a description of CRU and interrupt signals, refer back to our TMS 9900 discussion.
Device select logic includes a chip enable input, CE, together with five CRU bit select pins, SO - S4.a and SO S4 will connect to the Address Bus as follows:
..--:.-
··
-
-
.6.3
.;:
AD
-.
1
A5
A4
-
.':
.r •••• r
A14
A2
Ai
S4
DEVICE
S3 _
SELECT
S2 ..
.~
S1 •
TMS
9901
SO ..
CE.
In the illustration above, Address lines have been numbered using our standard notation. whereby A 14 is the highestorder address line and AO is the lowest-order address line. This is the opposite of Texas Instruments' notation. The CRU
select lines are numbered according to Texas Instruments' notation and Figure 3-25. Therefore, S4 is connected to
AO. and SO is connected to A4.
3-73
Device select logic determines the CRU address space that will be reserved for the TMS 9901 PSI. This may be illustrated as follows:
=
..
;,
~
···
·•...
':,
...
:-=
-
A14
A13
A12
A11
A5
A4
A3
A2
A1
AO
r ......
l)
--..
CRU
ACCESS
I,
M"E'M'EN
DEVICE
SELECT
1,
I
000
n
n
n
n
EE
SO
II
n
n
n
x
,r
S1
x
S2 S3 S4
x
r ••
x
x
~ '---....- - -.....v-...---",?"~
These three bits zero
and MEMEN inactive
(high) indicate a
CRU address
These seven bits
identify the
TMS 9901 address
space.
These five bits select
a CRU bit in the
TMS 9901 PSI
The high-order three address lines. which we call A 14. A 13. and A 12. are all zero during a CRU access. at which time
MEMEN is inactive (high). Thus we decode address lines A 11 through A5 to select a particular TMS 9901 device. .
Since the TMS 9980 uses the Address Bus differently during a CRU operation. TMS 9901 device select logic would
connect to the Address Bus in a different way. The CRU bit select lines SO - S4 would be tied to lines A5 - A 1; device
select logic would decode lines A11 - A6; and lines A13 and A12. along with MEMEN. would indicate a CRU access.
We illustrate this as follows:
MSB
A13
o
LSB
A12
0
A11
A10
A9
AS
A7
A6
A5
n
n
n
n
n
n
x
A4
x
A3
x
A2
x
A1
x
AO
. - Address Bus
~ ~,------..--~v-~----------~~ ~'----"--~~~-------'~
These six bits identify
the TMS 9901
address space
These five bits
select a CRU bit
CRUOUT
These two bits zero,
along with MEi\ii"Ei\i
inactive, indicate a
CRU address
is the complement of <1>3. For
the TMS 9900. <1>3 is generated by the TMS 9904. The TMS 9980 outputs <1>3 directly.
The best way of understanding the interface between a TMS 9901 and external logic is to look at functions performed. as illustrated in Figure 3-26.
3-74
INTERRUPT
MASK BITS
o
SELECT BIT
ICO ~--.r---,
IC2 .....~-.......
INTERRUPT
PRIORITY
ENCODER
IC3
_ _......J
IC1.---I
~--L
INTREQ .....- - - (
CLOCK
LOAD
BUFFER
CRUOUT - -__...
CRU
CRUCLK
----t. . INTERFACE~----------'
CAUIN ....I---~
SO
S1
S2
'S3
~
PO-P6
CRU
BIT
SELECT
LOGIC
S4
CE
Figure 3-26. TMS 9901 PSI General Data Flows and CRU Bit Assignments
3-75
From the programmer's viewpoint. a TMS 9901 looks like 32 contiguous CRU bits. Thus. you will access any part of a
TMS 9901 device's logic using CRU input and output instructions.
As you read through the TMS 9901 description that follows, you should bear in mind the power of multi-bit CRU
load and store instructions as they apply to TMS 9901 architecture. A single instruction transferring an appropriate bit pattern can frequently perform mu Itiple control and data transfer operations.
The manner in which CRU bits are used by the TMS 9901 is not straightforward. This is because CRU bits share
functions and pins. Functions and pins are shared in different ways.
Let us first look at pin connections. CRU bits 1-6 connect to pins TNiT - INT6; thus. in interrupt mode each of these CRU
bits has its own dedicated input pin.
.-
CRU bits 7-15 share nine input or output pins with CRU bits 23-31. CRU bits share pins as follows:
31
30
29
7
8
28
27
26
25
24
10
11
12
13
14
23
15
t
9
4
4
,.
...,.
~
~
~
~
~
..
~
.
34
33
32
31
~
~O
~
29
28
27
4
~
~
--I
t
Device Pins
23
I
These CRU bits support interrupt logic
These CRU bits are dedicated to data I/O
Each of the CRU bits shown above shares a pin with another CRU bit. That is to say. within the illustrated CRU address
range. there are two CRU bits which will access the same pin. although each CRU bit performs a different operation.
Thus you use the same pin in one of two different ways. using a bit address to select one operation. This may be illustrated as follows:
If you select CRU bit 27.
Pin 30 supports data I/O
If you select CRU bit 11,
and interrupt mode,
Pin 30 serves as an
interrupt request input
27
')
I
30
.--11-....'
CRU bits 16-22 connect to parallel I/O pins. These bit addresses are not shared with any other TMS 9901 functions.
CRU bit 0 is a select bit that is not connected to any pin. A 1 written into this bit causes bits 1-15 to support realtime clock logic. A 0 written into CRU bit 0 selects interrupt logic. When CRU clock logic is selected. bits 1-14 function
as two 14-bit real-time Clock Buffer registers -: one a read-only register. the other write-only. Real-time clock logic is
separate from. and operates simultaneously with. and/or parallel 110 logic. That is to say. the process of selecting realtime clock logic does not disable any other logic. The select bit merely chooses which registers CRU addresses will access. rather than enabling or disabling any operations.
TMS 9901 PSI INTERRUPT LOGIC
The easiest place to start understanding the TMS 9901 is at its interrupt logic.
External logic can input data to CRU bits 1-15 via their connected pins. These input data signals will be interpreted as interrupt requests if interrupts are enabled. If interrupts are disabled, then these CRU bits act
simply as data input.
3-76
You access interrupt logic through the CRU when the select bit, CRU bit 0, contains a O.
CRU bit addresses 1-15 each access separate read-only and write-only locations. The read-only location stores the signal level input at the attached pin. The write-only location accesses an interrupt mask bit. This may be illustrated as
follows:
WriteOnly
Mask
Bit
ReadOnly
Data
Bit
--
N
NT 1
{
CRU Bit N + 1
{
+1
}
From Pins
I
I
I
I
I
CRU Bit N
-
f4I
I
-
N
N
I
--
Signals arriving at pins connected to CRU bits 1-15 are immediately reflected by CRU bit contents:
ReadOnly
Data
Bit
0
I
I
~
Low
High
A low level (that is. a 0 bit) is interpreted as an interrupt request. The interrupt request is passed on to the mask bit. If
the mask bit contains 1. the interrupt is enabled and the interrupt request is passed on:
WriteOnly
Mask
Bit
Interrupt
Request
ReadOnly
Data
Bit
I
I
I·
-4
0
I
I
I:
Low
High
If the mask bit is O. the interrupt request is disabled and therefore denied:
WriteOnly
Mask
Bit
0
ReadOnly
LJata
Bit
I
I
I·
0
3-77
I
I
I:
Low
High
Quite apart from interrupt logic. the CPU can at any time read the contents of one or more CRU bits in the address
range 1-15. Here are some instructions that may access CRU bits 1-15 in various ways:
LDCR
R12.PSI+1
R1.MASK
R1.15
LOAD CRU BASE ADDRESS INTO R12
LOAD INTERRUPT MASK BITS INTO R1
OUTPUT TO WRITE-ONLY MASK LOCATIONS
STCR
R2.15
INPUT CRU BITS 1 THROUGH 15 AS DATA TO R2
LI
LI
For some randomly selected data levels. CRU bits 1-15 may be illustrated as follows:
Interrupt
Mask Bits
CRU
Bits
1 Bits Pass on
Interrupt Reque sts
1/
-
2
1
2
3
0
3
1
4
1
4
0
5
1
5
1
6
0
6
0
7
0
7
0
8
0
8
1
9
1
1
10 0
'\-
t
*-
1
1
-
11
12
1
-
14 1
13 0
15 0
1
--
...
...
-
0
1
9
1
10
0
11
0
12
0
13 1
---
14 0
15 1
~Bit Number~CRU Data
to CPU
o Bits
Generate Interrupt
Requests
.
..-..
..
----··..
-----
--.
...
...
·
-
-
If one or more CRU bit's interrupt requests are low. and the corresponding mask bit is 1. then interrupt priority encoder
logic outputs INTREQ low. Simultaneously. the level of the active interrupt request which has highest priority is identified via ICO - IC3.
""iNTf.
input to CRU bit 1. has highest priority;
INT15. input to CRU bit 15. has lowest priority.
The levels at ICO - IC3 are maintained until the interrupt request signal is removed at the external pin. or the interrupt
mask bit for the level is reset to O.
TMS 9901 PSI DATA INPUT AND OUTPUT
You can use CRU 1/0 instructions to input, output, or test external data at CRU bits 16-31. Data is output from
the CPU to the TMS 9901 via CRUOUT; it is input from the TMS 9901 to the CPU via CRUIN. Bits are addressed via SOS4. as we have already described.
Following a reset, pins connected to CRU bits 16-31 are in input mode. In this mode. external logic can assert high
or low levels at connected pins. in which case one or two CRU bits will be affected: a signal input to PO - P6 will generate data in CRU bits 16-22; if interrupt mode is selected (by a 0 in CRU bit OJ. a signal input to INT?IP15-INT15/P? will
3-78
generate data in two CRU bits, one in the CRU bit range 7-15, the other in CRU bit range 31-23. In interrupt mode. if
the CPU inputs data from CRU bits 7-15 or 31-23. then it will input the same data. but in reverse order. This may
be illustrated as follows:
CRU
Bits
R12,PSI + 7
Rl,9
LI
STCR
0196
16
Loaded into R1
7
9
10
LI
STCR
0003
R12,PSI + 23
Rl,9
16
Loaded into R1
0
8
11
0
1
12
0
13
0
Pins
34
4--0
14
33
~1
15
32
~1
31
4--0
23
24
1
25
0
26
27
0
28
0
30
~1
29
28
..--0
.--0
27
23
4--1
.--1
29
30
31
0
Note that. as in all CRU transfers, the first CRU bit transferred goes to the least significant bit position of the destination
register.
As soon as the CPU outputs data to any bit capable of supporting data output. the I/O logic associated with this
bit is put into output mode. In this mode, a pin will output a voltage level reflecting data in the corresponding CRU bit.
External logic cannot input data to a CRU bit that is in output mode; in fact. driving input currents into an output pin
may damage the TMS 9901.
Once a CRU bit has been placed in output mode. it remains in output mode until the TMS 9901 is reset. That is to
say, you cannot selectively return CRU bits from output mode to input mode. However, you can always read output bits
back to the CPU; that is, although external logic must never attempt to input to a pin that is in output mode. the
CPU can always read the contents of any 1/0 bit. whether it is an input or an output.
You cannot output data via CRU bits 7-15, even though these bits are connected to the same pins as CRU bits 31-23.
When you output data to CRU bits 7-15, the data is routed to one of two write-only locations, depending on the contents of CRU bit 0: if the select bit is 0, the data goes to interrupt mask bits 7-15; if clock mode is selected (CRU bit
contains 1). the data goes to the Clock Load Buffer register (bits 7-14) and RST2 (bit 15).
°
In interrupt mode you can input external data from CRU bits 1-6. Once again. you cannot output data via these CRU bit
addresses, since any data output will be routed to corresponding interrupt mask bits or Clock Load Buffer bits.
3-79
TMS 9901 REAL-TIME CLOCK LOGIC
If you write a 1 into CRU bit 0 of a TMS 9901 device, then CRU bits 1-14 are used as two 14-bit Clock buffers,
which may be illustrated as follows:
234
CRUOUT
5678910111213
---.1........._........._........._........_. . . . . _. . .
.....11........._
MSB
CRUIN
14"-CRUBitNumber
Clock Load Buffer
....."'"
LSB
~ I.... . . ._.a..-............._
....................._ .................._ ......-...... Clock Read Buffer
Besides these two buffers. real-time Clock logic contains a decrementing r,gister which we call the Clock
Counter register. The CPU loads the Clock Counter register via the Clock Load Buffer. and reads the Counter contents
via the Clock Read Buffer. We illustrate this in the following way:
....
-
CRUOUT
....~
CLOCK
BUFFER
REGISTER
'<
~
CLOCK
COUNTER
REGISTE(R
CRU
INTERFACE
"< >'
CRUIN
-
A.
...
I
I
CLOCK
READ
BUFFER
The Clock Counter register decrements continuously as long as the TMS 9901 is powered up. This will cause no
problems as long as the clock interrupt is disabled.
When you write any non-zero value into the Clock Load Buffer (CRU bits 1-14), the Clock Counter register
starts decrementing from that value. A decrement occurs once every 64
clock pulses. Thus. with a 3 MHz clock. a
decrement occurs once every 21.3 microseconds. When the CRU Clock Counter register decrements to 0, an interrupt request is generated, the previously output starting value is reloaded, and the clock starts to decrement
again. Thus. with a 21.3-microsecond time interval between decrements. the maximum time interval between interrupt requests will be 249 milliseconds.
pulse which causes the Clock Counter to decrement.
• An exit from clock mode.
Thus, the Clock Read Buffer register is updated whenever the TMS 9901 leaves clock mode, and every time the
Clock Counter decrements outside of clock mode.
Beware - even if CRU bit 0 contains a 1, the TMS 9901 will exit clock mode for as long as it sees a 1 on select line SO;
this will happen whether or not CE is active. Thus the Clock Read Buffer will not hold the same value indefinitely
just because the TMS 9901 select bit is set. The PSI will leave clock mode whenever the CPU reads to or writes from
CRU bits 16-31, or if any device accesses a memory address with a 1 on the address line connected to SO (A4 in a TMS
9900 system).
The logic c'ontrolling clock mode and the Clock Read Buffer may be illustrated .as follows:
7$-----1
DECREMENT
CLOCK
COUNTER
+64
SELECT BIT
(CRU BIT 0)
UPDATE
CLOCK READ
BUFFER
SO
This logic summarizes our discussion above. There are two important things to note about clock mode and Clock Read
Buffer update. First. you cannot inadvertently exit clock mode while you are reading the Clock Read Buffer, since you
access it as CRU bits 1-14. Second, you cannot enter clock mode solely by accessing CRU bits 0-15; SO changes clock
mode only when the select bit is 1 (clock mode selected).
In order to read the most recent Clock Counter value, you must do two things:
• Exit clock mode so the Clock Read Buffer will receive the current Clock Counter contents.
• Enter clock mode so the Clock Read Buffer will be stable during the read itself.
Here is the appropriate instruction sequence:
LI
SBZ
SBO
STCR
R12,PSI+1
-1
-1
R1,14
LOAD PSI CRU BASE ADDRESS
EXIT CLOCK MODE TO UPDATE READ BUFFER
ENTER CLOCK MODE TO STABILIZE READ BUFFER
READ 14-BIT CLOCK READ BUFFER
TMS 9901 RESET LOGIC
You can reset a TMS 9901 in one of two ways:
1)
2)
By inputting a low signal at RSf"i.
By using a programmed reset via RST2, a CRU bit.
In order to use RST1, a low level must be input at 'this pin for at least two clock periods.
You can reset the TMS 9901 under program control only when clock mode is selected (CRU bit 0 is 0). At this time,
writing a 0 to CRU bit 15 (RST2) causes the device to be reset. Thus, the following instruction sequence causes a TMS
9901 device reset:
LI
SBO
SBZ
R12,PSI
o
15
LOAD PSI CRU BASE ADDRESS
ENTER CLOCK MODE
RESET PSI
When the TMS 9901 is reset. the INTREQ signal is output high, ICO through IC3 are output low, all interrupt requests
are disabled, and all I/O CRU bits are placed in input mode.
3-81
THE TMS9902 ASYNCHRONOUS COMMUNICATIONS CONTROLLER
The TMS9902 microprocessor family includes two serial I/O parts. The TMS9902 is a simple, asynchronous
communications device; the TMS9903 is a more powerful, recently introduced multifunction device. Both of
these parts are peculiar to the TMS9900 since they communicate with the CPU via its CRU logic. The two parts
are also pin-compatible; that is, the same 20-pin socket can hold either the TMS9902 (an 18-pin part) or the
TMS9903.
The TMS9902, which we are about to describe, offers asynchronous I/O capabilities comparable with those of
parts which we describe in Volume 3. The TMS9902 lacks some features which other parts offer:
1)
There are no external clocking signals for received or transmitted data. Receive and transmit rates are computed by
logic internal to the TMS9902.
2)
There is a single interrupt request which has no accompanying status output lines. Thus interrupt service routines
must interrogate status in order to correctly service the interrupt.
3)
The TMS9902 has only three Modem control lines and no other lines for handshaking with peripheral logic.
One advantage of the TMS9902 is that it occupies very little board space. It is an 18-pin part the smallest serial
I/O controller on the market. It requires less surrounding logic because it uses the system clock for its time base, and
because it provides almost no external status or handshake lines.
Another advantage of the TMS9902, when compared to other serial I/O parts, is the presence of real-time clock logic.
Anyone who has worked with serial I/O logic will appreciate the ability to generate interrupt requests at fixed time intervals.
The TMS9902 is fabricated using NMOS technology. It is packaged as an 18-pin DIP and requires a single +5V power
supply. All Signals are TTL-level compatible.
TMS9902 ACC PINS AND SIGNAL ASSIGNMENTS
TMS9902 pins and signal assignments are illustrated in Figure 3-27. These signal assignments are the same as
those of pins 1 through 9 and 12 through 20 of the TMS9903.
INT
TxD
RxD
CRUIN
RTS
CTS
DSR
CRUOUT
VSS (GND)
1
2
3
4
5
6
7
TMS9902
18
17
16
15
14
13
12
11
8
9
10
VCC (+5V)
CE
~
CRUCLK
SO (MSB)
S1
S2
S3
S4 (LSB)
Pin Name
Description
Type
CRUIN
CRUOUT
CRUCLK
CRU data output to CPU
CRU data input from CPU
CRU data strobe
Device select
CRU bit address
Synchronizing clock
Data set ready indicator
Request to send indicator
Clear to send indicator
Serial data in
Serial data out
Interrupt request to CPU
Power, Ground reference
Output
Input
Input
Input
Input
Input
Input
Output
Input
Input
Output
Output
CE
SO-S4
iii
DSR
m
CTS
RxD
TxD
iNT
VCC' VSS
Figure 3-27. TMS9902 Asynchronous Communications Controller Pins and Signal Assignments
3-82
Table 3-9, 'TMS9902 Control and Status Register Bit Interpretations
CRU!REGISTER
BIT NUMBER
CONTROL REGISTER (WRITE)
Device Reset (write 1 or 0)
STATUS REGISTER (READ)
31
Any interrupt pending'
30
One or more of control bits 17, 14, 13, 12, or 11 set to
l'
29
DSR or CTS input level change detected, Reset by writing 1 or 0 to CRU bit 21 ,
28
27
Complement of CTS input level
Complement of DSR input level
26
Complement of RTS output level
Timer time out', Reset by writing 1 or 0 to CRU bit 20,
25
24
Timer overrun error', Reset by writing 10r 0 to CRU bit
20,
23
Transmit Shift register empty', Automatic reset,
22
Transmit buffer empty', Reset by writing to high-order
Transmit buffer bit.
Enable interrupts on DSR or CTS input level change
(1 = enable, 0 = disable)
21
Receive buffer loaded', Reset by writing 1 or 0 to CRU
bit 18,
Enable timer interrupts
(1 = enable, 0 = disable)
20
DSR or CTS input level change interrupt pending', Reset
by writing 1 or 0 to CRU bit 21,
Enable transmitter interrupts
(1 = enable, 0 = disable)
Enable receive interrupts
(1 = enable, 0 = disable)
19
Timer interrupt pending', Reset by writing 1 or 0 to CRU
bit 20,
Transmit Break
(1 = enable, 0 = disable)
17
Transmit interrupt pending', Reset either by writing 0 to
CRU bit 19 or by writing to high-order Transmit buffer. bit,
Enable transmit I~
(Complement of RTS output)
16
Receive interrupt pending', Reset by writing 1 or 0 to
CRU bit 18.
Test mode select
(1 = Test mode, 0 = normal operation)
15
RxD input level
Write to Parameter register
14
Receive start bit detected', Reset automatically at end
of received character,
Write to Timer register
13
Receive first data bit detected'. Reset automatically at
end of received character,
Write to Receive Data Rate register
12
Receive framing error detected', Reset automatically by
error free received character.
Write to Transmit Data Rate register
11
Receive overrun error detected', Reset automatically by
error-free received character,
10
Receive parity error detected." Reset automatically by
error-free received character,
9
Any receive error detected." Reset automatically when
Status register bits 12, 11, and 10 are all 0,
18
/
Receive Data Rate register
or
Transmit Data Rate register
~
I
I
Parameter register,
Timer register,
or Transmit buffer
I
= "true" condition,
0
= "false"
1'1
I
I
4
t
,
'1
8
7
6
5
3
2
1
o
condition,
3-83
~
>
I
Signals that connect the TMS9902 to a TMS9900 series microprocessor include the three CRU signals CRUIN.
CRUOUT. and CRUCLK. together with device select logic signals CE and SO-S4. The TMS9902 uses these signals exactly as described for the TMS9901. CE must be low for the TMS9902 to be selected; if the TMS9902 is
selected. then data transfers occu r via the CRUIN or CRUOUT lines. SO-S4 identify the CRU bit within the selected
TMS9902. Table 3-9 summarizes the way in which the TMS9902 assigns its 32 CRU bit addresses for read and
write operations.
DSR. RTS. and CTS are standard handshaking control signals for communications devices.
DSR is a general purpose input signal; its level is reported in Status register bit 27. You can program DSR to generate
an interrupt request when it makes a high-to-Iow or low-to-high transition. However. DSR plays no part in enabling
either transmit or receive logic.
The TMS9902 outputs RTS low while tranSmit logic is enabled. But the transmitter will not actually start transmitting
data until CTS is input low.
In a standard asynchronous protocol system. TMS9902 transmit logic will output Ri'S low and sometime later receive a
low CTS input - at which time it will actually start transmitting data. But if TMS9902 transmit logic finds CTS low
when it outputs RTS low. it will start transmitting immediately.
For a discussion of Modem handshaking control signals. see Volume 1. Chapter 5.
Serial data is input via RxD and output via TxD. External logic does not provide signals that clock the serial input
or output data. Instead. the CP synchronizing clock input signal is used to derive data transmit or receive rates. Usually.
3 (the complement of CPU clock <1>3). However. you may use any clock signal that
satisfies the timing requirements given in the TMS9902 data sheet at the end of this chapter.
TMS9902 DATA TRANSFER AND CONTROL
The various addressable locations within the TMS9902 are summarized in Figure 3-28.
When you write to CRU bits 31 through 11 you will always access the Control register; when you read these
bits you will access the Status register. CRU bits 10 and 9 are also read-only status flags.
CRU bits 0 through 7. on a read. always access the Receive buffer; but via CRU bits 0 through 10 you can send
data to a variety of write-only locations.
The Control register contains four address bits. each of which corresponds to one of the write-only
TMS9902
locations. When an address bit is set to 1. the associated write-only register will receive
REGISTER
data output via CRU bits 0 through 10. If more than one write-only location is selected. then
ADDRESSING
the select priorities shown in Table 3-10 apply. The Transmit Buffer is selected when all
four address bits contain O. If any address bit is set to 1. Status register bit 30 will also contain 1.
When you write to the high-order (highest numbered) bit of the Parameter register. the Timer register. or the
Receive Rate register. you automatically reset that location's address bit in the Control register.
Table 3-10. TMS9902 Write-Only Register Select Scheme
CRU Output Bit
CRU Bits in
Location
Addressed Location
14
13
12
11
1
0
0
0
0
X
1
0
0
0
X
X
1
X·
0
X
X
X·
1
0
Parameter register
Timer register
Receive Rate register
Transmit Rate register
Transmit buffer
7-0
7-0
10-0
10-0
7-0
"X" means "does not matter"
• If both bits 11 and 1 2 are set to 1, data will be written to both Rate
registers at the same time.
----- -
Following a device reset. all write-only location address bits in the Control register are
set to 1. This allows you to write data to registers in the priority order shown in Table 3-10
during the device initialization process. without having to reset individual address bits. Thus
the initialization process will consist of these steps:
1)
Reset the TMS9902 by writing to Control register bit 31.
2)
Write to the Parameter register.
3-84
TMS9902 DEVICE
INITIALIZATION
3)
Write to the Interval Timer register.
4)
Write to the Receive Data Rate and Transmit Data Rate registers.
5)
Write to the Control register and Transmit buffer.
Transmit
Logic
CRUOUT
CRUIN
CRUCLK
INT
CE
SO
S1
S2
S3
S4
Transmit
Buffer
CPU
Interface
Logic
TxD
Transmit Rate
Register·
Status Register
Control Register
Parameter
Register·
Receive Rate
Register
Receive
Buffer
RxD
Receive
Logic
Timer
Logic
Figure 3-28. TMS9902 Functional Logic
Texas Instruments' literature suggests an initialization instruction sequence such as the following:
LI
SBO
LDCR
LDCR
LDCR
LDCR
R12.CRUBS
31
@CNTRL.8
@INTVL.8
@RDR.11
@XDR.12
INITIALIZE CRU BASE ADDRESS IN R12
RESET COMMAND
LOAD PARAMETER AND RESET BIT 14
LOAD INTERVAL AND RESET BIT 13
LOAD RECEIVE RATE AND RESET BIT 12
LOAD TRANSMIT RATE AND RESET BIT 11
3-85
In the sequence above, CRUBS represents the base address for the 32 CRU bits in the TMS9902. Fou r memory locations -labeled CNTRL. INTVL. RDR. and XDR - hold the values to be loaded into the write-only locations. Since CRU
bit 11 is not reset automatically. the instruction which writes to the Transmit Data Rate register writes 12 bits. the highorder bit being a 0 for CRU bit 11.
Let us now examine Control register bits in detail.
Control register bits may be divided into interrupt enable/disable bits. write-only location address
bits. the reset control. and the test mode control.
TMS9902
CONTROL
REGISTER
The test mode control (bit 15) is usually left at 0: this causes normal operations to occur. When
TMS9902
you set the test mode control bit to 1, RTS is internally connected to CTS and RxD is interTEST MODE
nally connected to TxD. Also. DSR is held low internally and the interval timer operates at 32
times its normal rate. You will operate the TMS9902 in this condition only when testing its logic.
You reset the TMS9902 by writing either a 0 or a 1 to Control register bit 31.
You will usually begin every event sequence with a Reset. The following instructions constitute
TMS9902 resets:
LI
SBO
R12.ACC
31
or
LI
SBZ
R12.ACC
31
ACC is a label identifying CRU bit 0 (the CRU base address) for the TMS9902.
When the TMS9902 is reset, the following events occur:
1)
All interrupts are disabled.
2)
RTS is output high: this is the inactive state for RTS.
3)
Control register bits 11. 12. 13. and 14 are set to 1. All other Control register bits are reset to O.
The TMS9902 should not be accessed for a minimum of eleven clock cycles following the reset command.
There are four interrupt enable control bits. They enable interrupts when set to 1 and disable
interrupts when reset to O.
TMS9902
INTERRUPT
Control bit 21 enables CTS and DSR input signal level change interrupt requests.
ENABLE
Control bit 20 enables timer time out interrupt requests.
Control bit 19 enables Transmit buffer empty interrupt requests.
Control bit 18 enables Receive buffer full interrupt requests.
In each case a Status register bit is set to identify the condition that can generate an interrupt request. But the interrupt
will not actually be requested unless the associated interrupt enable control bit has been set to 1.
r------.. . .
You acknowledge any interrupt other than a transmitter interrupt by writing to the interrupt's
TMS9902
enable control bit. To acknowledge an interrupt and leave it enabled. rewrite a 1 to the interINTERRUPT
rupt enable control bit. To acknowledge an interrupt and then disable it. write a 0 to the interACKNOWLEDGE
rupt enable control bit. But remember. you must write either a 0 or a 1 to the interrupt enable
control bit. since this is the mechanism used to reset the status flags that identify the interrupting condition.
You acknowledge a transmitter interrupt by writing to bit 7 of the Transmit buffer. If you write a 0 to CRU bit 19. you
will disable the interrupt. but you will not reset the status flag which was set by the emptying of the Transmit buffer.
Control register bits 16 and 17 directly control two TMS9902 operations.
Control register bit 16 is the complement of the RTS output. You must write a 1 to this bit in order to set RTS low. In
order to enable transmit logic. RfS must be output low while CTS is being input low. You must leave RTS low while the
transmitter is active. To disable the transmitter you raise RTS high again by writing 0 to Control register bit 16: if
transmit logic is part way through transmitting a character when you write a 0 to Control register bit 16, then it will
complete transmitting the character - and the character in the Transmit buffer. if the buffer is full- before outputting
RTS high.
Transmit break logic is controlled via Control register bit 17. When you set this bit to 1, a
TMS9902
break (continuous low output) will be transmitted following the next underrun (that is, when both
BREAK
the Transmit register and Transmit buffer are empty). You must end the break by writing a 0 to
LOGIC
Control register bit 17 before you can restart transmitting by writing new data to the Transmit
buffer. If you leave Control register bit 17 reset to 0, then following an underrun the transmitter will mark (output a continuously high signal). You can end the mark at any time. and start transmitting a new message. by writing fresh data
to the Transmit buffer.
3-86
When the break control bit is set to 1. Status register bit 30 will also contain a 1.
Let us now examine Status register bits; they may be grouped as follows:
1)
2)
Signal level indicators
Transmit operation status
3)
4)
5)
Receive operation status
Timer logic
Interrupt logic
TMS9902
STATUS
REGISTER
Status register bits 27 and 28 report the complement of the DSR and CTS input signal levels. Bit 26 reports the
complement of the RTS output signal level.
When the DSR or CTS input changes level, bit 29 is set. You reset bit 29 by writing to Control register bit 21.
There are three transmit logic status bits. Bit 22 is set when the Transmit buffer is empty. The
bit is reset when you next write data to the Transmit buffer. Bit 23 is set when the Transmit
Shift register is empty; this is an underrun condition. Following an underrun. a break or a mark
will be transmitted. depending on the level of Control register bit 17. Bit 30 of the Status register
contains a 1 if any of the following Control register bits are set to 1:
TMS9902
TRANSMITTER
STATUS
• Bit 17. the break control bit
• Bits 14. 13. 12. and 11. the write-only. location address bits
Thus Status register bit 30 will be set to 1 whenever Transmit buffer loading is disabled.
For receive logic. bit 21 is set when the Receive buffer is full. The CPU resets this bit by writing to
bit 18 of the Control register: usually the program will read the contents of the Receive buffer
before resetting the flag bit.
TMS9902
RECEIVER
STATUS
RxD, the serial data input line level. is reported at Status register bit 15.
The start of each received character is identified by Status register bits 14 and 13. When the start bit has been
detected, Status register bit 14 is set. One bit time later, when the first data bit is being detected, Status
register bit 13 is set. These two bits remain set until the end of the character. They are reset when the last stop bit has
been detected.
Framing, overrun, and parity errors are reported by Status register bits 12, 11, and 10, respectively. These error
status bits. once set. remain set until an error-free character is loaded into the Receive buffer.
If one or more of the three receive error conditions exist, then Status register bit 9 is set.
There are two timer status bits. Whenever the timer times out, Status register bit 25 is set to
1. This bit must be reset by writing 0 or 1 to Control register bit 20. If you do not do so before the
next time out. then Status register bit 24 will be set, indicating a timer error. The timer error is
also cleared by writing 0 or 1 to Control register bit 20.
TMS9902
TIMER
STATUS
The four interrupt generating conditions have associated status bits which are set following an interrupt request.
If the DSR or CTS input signal changes level, and the interrupt logic has been enabled, then
Status register bit 20 is set at the time that an interrupt request is generated.
If a time out occurs and timer interrupts have been enabled, then Status register bit 19 is
set at the time an interrupt request occurs.
TMS9902
INTERRUPT
STATUS
When the Transmit buffer becomes empty, if transmitter interrupts have been enabled, then Status register bit
17 is set at the time an interrupt request occurs.
When the Receive buffer is full, if receive interrupts have been enabled, then Status register bit 16 is set at the
time a receiver interrupt request is generated.
If one or more of these interrupt requests are active, then Status register bit 31 is set.
Interrupt status bits remain set until you reset either the status bit for the interrupting condition. or its interrupt enable
bit in the Control register. In most cases. writing to the enable bit resets the status bit.
For a Modem signal interrupt you must write to Control register bit 21 in order to acknowledge the interrupt. thus resetting the two Status register bits.
For a timer interrupt you must write to Control register bit 20 to reset the interrupt.
For a Transmit buffer empty interrupt you must write new data to the Transmit buffer in order to acknowledge the interrupt: specifically. you must write to bit 7 of the Transmit buffer.
3-87
For a Receive buffer full interrupt. you must write to Control register bit 18 in order to acknowledge the interrupt.
Let us now examine Parameter register contents.
After resetting the TMS9902, the next step is to identify subsequent operations by loading appropriate data into the Parameter register. Parameter register bits are interpreted as follows:
6
5
432
o
TMS9902
PARAMETER
REGISTER
~BitNo.
X
Parameter register
~--~--~~~--~~--~~
L{
OO
01
10
11
L-_ _ _ _ _ _ _ j 0
11
00
01
10
11
- 5-bit data words
- 6-bit data words
- 7 -bit data words
- 8-bit data words
- Divide! by 3 to generate ClK
- Divide by 4 to generate ClK
- No parity bit
- No parity bit
- Even parity bit
- Odd parity bit
00 - Select
01 - Select
10 - Select
11 - Select
11/2 stop bits
2 stop bits
1 stop bit
1 stop bit
The options presented by the Parameter register, as illustrated above, are self-evident. with
the exception of Parameter register bit 3. This bit is used to generate an internal clock sig!l81. CL~Depending on the setting of Parameter register bit 3, the ClK frequency will be
~/3 or <1>/4. ClK is then used to specify the time interval between bit sampling for serial
data input or output. as well as the interval timer rate. The frequency of ClK should not be
greater than 1.1 MHz; therefore if is faster than 3.3 MHz, Parameter register bit 3 should
be set to 1.
TMS9902 INTERNAL
CLOCK SIGNAL
TRANSMIT AND
RECEIVE DATA
RATE REGISTERS
After loading appropriate data into the Parameter register, you must load the Transmit and Receive Data Rate registers
in order to specify the time interval that will separate bit sampling. Data Rate register contents are interpreted as
follows:
10
9
8
6
5
4
3
o 4--BitNo.
Transmit or Receive Data Rate register
- - - - - - - - - - - Second scale factor (S).
Can have any value in the range 1 (000000001)
through 1023 (111111111)
L - - - - - - - - - - - - - - - - - - - F i r s t scale factor (F).
Can be 0 or 1. If F = 0, S should be
3-88
~
4
The time interval separating serial bits transmitted or received is given by the equation:
tClK x 2 x 8 F x S
For example. suppose the Receive Data Rate register contains 11000111000. S = 56810 and F = 1:
11000111000
~
L23816
~ ~
56810
S
'---------- 1 = F
If ClK
= $/3.
and
= 3 MHz.
+ (2
(1 x 106 )
If F
=
then the serial data transfer rate will be:
x 8 x 568) = 110.04 bits per second
O. then the serial data transfer rate becomes:
(1 x 106 )
+
(2 x 8 x 568) = 880.28 bits per second
Table 3-11 shows sample Data Rate register values for standard Baud rates. The assumed frequency produces very
precise Baud rates; it is also within the recommended operating range of TMS9900 series parts.
Table 3-11. Example of Data Rate Register Contents
for Standard Baud Rates
Frequency = 3. '!,!8 MHz
Frequency ClK = + 3 = 1.066 MHz
Data Rate Register
Contents
Data Rate in
Bits per Second
Decimal
F
S
0
0
0
0
0
1
1
1
1
55
110
220
440
880
220
440
600
880
Hexadecimal
9600
4800
2400
1200
600
300
150
110
75
037
06E
ODC
1B8
370
4DC
588
658
770
Date Rate = ClK
+
(2 x 8 F x S)
It is not strictly necessary to have data rates as precise as those we have shown in Table 3-11.
The devices which receive data from the TMS9902 will determine how precise the transmit rate must be.
TMS9902 Receive logic resynchronizes itself with the beginning of each incoming character. It does this by starting its
bit-time count at a high-to-Iow transition of RxD. When the TMS9902 has counted half a bit-time. it samples RxD; if the
line is still low. Receive logic assumes a valid start bit is present. It then samples the. line at single-bit-time intervals
after the first sample point. until a full character has been received:
Start
bit
RxD
Middle of start bit is
one-half bit time from
this falling edge
Character
~+ +
Stop
bit
'f
TMS9902 samples RxD at the middle of each bit
3-89
Because of this resynchronization. no skew errors will occur as long as the transmitted bit rate is within 4% of the
TMS9902 Receive data rate.
TMS9902 TRANSMIT OPERATIONS
Let us now examine a serial data transmit event sequence as illustrated in Figure 3-29.
~
.!!!
Q)
....
~
a:
.~
c:
!!!
g>:o
~~~ ~~
°~Q.·B~:::
c: 0.:
o3l
OE3l
U
~
~
.s~]!~
VI
~
1CJ)a:~
8
8 ~.~ 81 8 ~
~
IE
.0....
"l:I
~£~ ~~
Q)....
~
:!:
..Q:!:..Q
ClIO)
~~
0)
1.1 -b
I,{-__~----c::::::7+-~\~--L .LI .. .L..
:a;;data
.,-_ _. " A
r
Idle or Break
.t= 0 ..
..
Tran:::it Buf:: empty:Transmit Register empty,
_
Break control bit = 0
~
_______ ~
~ II ____ ~ 11-~~
.g ~
en
u5
One data
\
character
Move Transmit
Buffer contents to
Transmit register
\
Output data to
Transmit Buffer
o
1/
~
____ ____
~
~~
B .~
~_CJ).o~
______ ____ ____
~
~
~
START
BIT
........
0 ....
1/ II
C')N
NN
'V'
."
u
....
:.c
....c:
CII
.g
'2
'2
....
:.c
....c:
Data Character
CII
0+:
0)
0)
'iii
'iii
....
....VI
VI
CII
0
E
~
"l:I
CD
0+:
'0
CD
-....
C.
VI
:.c
~
.t;:
en ....
:Ell
.eC')
S~
en:.c
VI
.a
en
CII
CII
0....
Figure 3-29. TMS9902 Character Transmit Event Sequence
In this example. all operations will begin with a Reset. Remember. you reset the TMS9902 by writing a 0 or 1 to CRU bit 31.
Next. output appropriate codes to the Control and Parameter registers and enable appropriate interrupts.
Output Data Rate register settings.
Output the first character to the Transmit Buffer register.
3-90
TMS9902
SERIAL
TRANSMIT
EVENT
SEQUENCE
Transmit logic has now been initialized. You begin actual data transmission by setting RTS low. An appropriate initialization instruction sequence was given earlier.
Setting Ri'S low enables transmit logic within the TMS9902. but actual data transmission does not begin until external
logic inputs CTS low. If CTS is already low when RTS is reset low. then data transmission will begin as soon as RTS is
output low.
When a character is transmitted. the Transmit buffer contents are moved to the Transmit register. at which time Status
register bit 22 is set. If transmit interrupt logic has been enabled. an interrupt request will occur at this time and Status
register bit 17 will be set.
The character is transmitted as illustrated in Figure 3-29; options are specified in the Parameter register. As soon as
the character's stop bit (or bits) has been transmitted. transmission of the next data character begins. provided the CPU
has by this time loaded the next data character into the Transmit buffer. The CPU will normally have plenty of time to
reload the Transmit buffer. since it takes a long time. in terms of instruction execution times. to transmit a character.
Note that you must write to bit 7 of the Transmit buffer in order to reset the Transmit buffer empty flag. Thus even
though the character length is less than 8 bits. you will always write 8 bits to the Transmit buffer. You right-adjust
Transmit buffer characters; that is. bit 0 of the Transmit buffer is always the least Significant bit of the character.
If transmit interrupts have been enabled. an interrupt request will occur when Status register bit 22 is set. The CPU will
respond to the interrupt request by interrogating Status register bits to identify the nature of the interrupt. Upon
detecting a 1 in bit 17. the CPU will output another data character. If transmit interrupt logic has not been enabled.
then the CPU must periodically poll the Status register and output the next data character upon detecting bit 22 set to
1.
If the Transmit buffer is empty at the end of a data character transfer, then the TMS9902
may transmit a Break (if Control register bit 17 is 1), or it may terminate operations and go
into an idle state (if Control register bit 17 is 0).
The TMS9902 will transmit a Break if CTS is still low and Control register bit 17 is high. A Break is a continuous low
level output via TxD. External logic interprets a Break as a signal indicating temporary suspension of data transfer.
Break logic inhibits data transfers to the Transmit buffer. You must terminate a Break by resetting Control register
bit 17 to O. then loading the next data character into the Transmit buffer.
TMS9902 transmit logic will enter an idle state if CTS is input high by external logic or if CTS is input low. but no new
data is ready to transmit and break logic is off. During this idle state TxD will be held high (marking).
The level of the RTS output is not affected by a change in the
m
input level.
If CTS goes high during a transmit operation a~ou leave RTS output low. then as soon as CTS goes low again the
transmitter will be re-enabled; but if you output RTS high by writing 0 to Control register bit 16. then the CTS input will
be ignored. In order to re-enable transmit logic you must output 1 to Control register bit 16. again setting RTS low. If
CTS is low at this time. transmission will begin immediately; otherwise. transmission will begin as soon as CTS is input
low - after RTS has again been output low. This may be illustrated as follows:
Start
transmitting
Disable
transmit
logic
Disable
transmit
logic; stop
transmitting
at end of character.
--~-Enable
transmit
logic; start
transmitting
again
Stop transmitting at end of character
3-91
r-
!
e
UI
::I
~
0o
::-:'
Status register
bits 21. 14. and
13 all = 0
\
.."
\
cO·
t:
ro
w
W
9
Status register
bit 13 = 1
Least significant bit
-i
en
(0
\
(0
otv
w
~
po.)
()
:J"
Ql
(")
~
L
CD
00
'3"::;,
~ CD
CD
II)
a.
~
Q)
(")
<.
~~
Parity bit, if specified
m
<
CD
~
en
.0
t:
CD
:::J
(")
CD
/
~StoPbUISI
CD
Status register
_ 0
bits 14 and
. 13 both -
C/)
aJ --i
-»
--i:1]
--i
:l]
~ ~
CD
iii. 0
~~"
o;-~
~
I.-<
/
!!!.
•
Status bit
21 = 1
0:1]
CD 0 CD
~ ~!.
~~(i
/
/
RESET
CD
<"
CD
Output Control register and
Interrupt Enable register
settings
Ou~put
0?S"
CQ
."
c"
e
(;
W
fA
o
cUI
;....
CD
\
Most significant bit
:JJ
CD
(;
n
(Send "Receive logic
. . Ready" signal)
\
Q)
II)
both. Data Rate
. - register settmgs
\
~
r
:r
1____
....
til
....
~
~
--
:l]
CD
0
CD
(;
!!!.
<
CD
a.
Q)
2
i»
CD
~"
<
CD
::I
....
til
!e
a:
CD
CD
::I
n
!II
Q
!:tI
Q
r-::aooot
Om~
G)(")fI)
C:;!!!U)
t__
I .....
1 _ _ _ _ _ _.....
1....
11
---..
~.....
-,
Transmit Buffer
Parameter Register
Sync1 Register
Sync2 Register
Timer register
Receive CRC
Transmit CRC
3-95
.........- - TxC
Transmit
Logic
~-~
RTS
14o--m
Transmit CRC
CRUOUT
CRUIN
CRUCLK
INT
iP
CE
SO
51
52
53
54
Transmit Buffer
CPU
Interface
Logic
TxO
Status Register
Control Register
Parameter Register
Sync1 Register
Sync2 Register
RxO
Receive Buffer
Received SOLC
Check Character
Receive CRC
....1 - - - RxC
Receive
Logic
Timer Logic
Figure 3-31. TMS9903 Synchronous Communications Controller
-Functional Logic
As illustrated above, there are three cyclical redundancy check characters which can be read from the
TMS9903.
Transmit and receive logic each compute a cyclical redundancy character (under program con troll for transmitted and
received messages.
In SOLe mode only, the cyclical redundancy character for a received frame is isolated by receive logic and held in a
register out of which it can be read.
3-96
We will describe programming aspects of cyclical redundancy characters in more detail as the discussion of the
TMS9903 proceeds.
Transmit and receive logic are each buffered. Data is m9ved from the Transmit buffer to the Transmit Shift register,
whence it is output serially via TxD. You have one character transmit time within which to write another character to
the Transmit buffer, otherwise an underrun will occur.
Characters are assembled by receive logic in the Receive Shift register: when assembled, they are transferred to the
Receive buffer. You have one character receive time within which to read the contents of the Receive buffer, or else a
receive overrun will occur.
Data buffers within the TMS9903 are all nine bits wide; this gives you the option of appending a parity bit to
any 8-bit character. The Status register is 23 bits wide, the Control register is 20 bits wide, and the Parameter register
is 12 bits wide: these odd bit lengths cause no problems due to the nature of the CRU interface between the TMS9903
and the TMS9900 series microprocessor.
The Sync1 and Sync2 registers hold Sync characters; in certain protocols these two registers may hold special
control characters. Transmit logic may output the contents of one or both of these registers at the beginning of a
message and following an underrun. Receive logic uses the contents of the Syncl register to detect Sync characters in
a received data stream.
You specify the number of data bits per character for received data via Parameter register bit settings.
When receive logic is assembling characters in the Receive Shift register, it uses the bits-per-character specification
that was in effect when the current character started to be assembled. If you change the bits-per-character specification, the change will be recognized on the next receive character boundary.
The bits-per-character specification that you make in the Parameter register does not apply to transmit logic or
the Sync1 and Sync2 registers. For these three registers the number of data bits you write into the register
defines the number of data bits which will be transmitted. The most recently loaded Sync register determines
the character length for transmission of both Sync characters.
For example, if you output 6-bit characters to these three registers, then 6-bit characters are assumed by transmit logic.
Likewise, if you output 9-bit characters, then transmit logic will subsequently assume 9-bit Syncl and Sync2 characters.
Sync1 and Sync2 registers should have the same bits-per-character specifications. However, you could, for example,
output a 7-bit character to Syncl and then a 5-bit character to Sync2. If you did, the device would transmit just the
lower five bits of Syncl and Sync2. You could still specify 7-bit characters to receive logic: each received character
would be compared to all seven bits of Sync1. The Sync character bit length need not be the same as the bits-per'character specification in the Parameter register or even the number of bits specified by loading the Transmit buffer.
As with the Receiver, you can change the Transmit character length from character to character. As each character is
shifted from the Transmit buffer to the Transmit Shift register, transmit logic attaches the bits-per-character specification to the data in the Transmit Shift register. Therefore if you subsequently change the number of bits per transmit
character - namely, by loading a different-sized word into the Transmit buffer - it has no effect on the character
already in the Transmit Shift register.
Although Texas Instruments literature describes the TMS9903 as supporting six different
modes, in fact it supports three: Asynchronous, Synchronous, and SOLC/HOLC.
Asynchronous and Synchronous mode capabilities are quite standard.
In Synchronous mode you can approximate IBM standard Monosync or Bisync protocols.
Asynchronous mode is well suited to RS-232C and RS-449 EIA standard protocols.
The TMS9903 can be operated in a point-to-point SDLC or HDLC system: also, SDLC loop mode is supported.
The TMS9903, like the TMS9902, has on-chip timer logic.
TMS9903 PINS AND SIGNALS
TMS9903 pins and signals are illustrated in Figure 3-32. Pins 1 through 9 and 12 through 20 are functionally
equivalent to TMS9902 pins 1 through 18.
On its CPU interface the TMS9903 has the same standard TMS9900 Signals as the TMS9901 and
the TMS9902. These include:
1)
2)
The threestandard CRU signals: CRUIN, CRUOUT. and CRUCLK.
Five select lines (SO-S4) that address a 32-bit CRU field.
3-97
TMS9903 CPU
INTERFACE
SIGNALS
3)
4)
CEo an enable signal which must be low for the CPU interface to be enabled.
An input clock signal. normally connected to the TIM9904 ~3 clock.
Refer to our earlier discussion of the TMS9901 for a description of CPU interfacing logic.
INT
TxD
RxD
CRUIN
RTS
CTS
DSR
CRUOUT
(GND) VSS
TxC
1
2
3
4
5
6
7
8
9
10
TMS
9903
20
19
18
17
16
15
14
13
12
11
VCC (+5V)
et
iNf
Vce. VSS
Figure 3-32. TMS9903 Synchronous Communications Controller
Pins and Signal Assignments
Let us now examine transmit and receive logic signals.
TMS9903
Serial data is output by transmit logic via TxD, as clocked by TxC. Data is transmitted on
high-to-Iow transitions of"f'X'C.
SERIAL 1/0
SIGNALS
RTS and CTS are two Modem control signals associated with transmit logic. In order to
transmit data you must input CTS low while transmit logic is enabled. You have the option of connecting Ri'S to
transmit enable logic. If you do. RTS will be output low while transmit logic is enabled and it will be output high while
transmit logic is disabled. You also have the option of selecting the RTS output level under program control. in which
case RTS is disconnected from transmit enable logic.
Receive logic receives data via RxD as clocked by RxC. Data is sampled on low-to-high transitions of RxC.
DSR is shown in Figure 3-31 as a receive logic Modem input signal; in reality it is an unassigned input control signal. The DSR signal level is reported in a Status register bit. and can generate an interrupt whenever it changes state.
DSR does not contribute to receive enable logic.
TMS9903 PROGRAMMABLE REGISTERS
The two principal programmable registers of the TMS9903 are the Control and Status registers. We refer to
these as "principal" registers because they are automatically accessed by high numbered CRU bits on any CRU access.
Low numbered CRU bits transfer data to or from a variety of addressable locations. as specified by Control register bit
settings.
3-98
Table 3-12. TMS9903 Synchronous Communications Controller CRU Bit Assignments
When Writing to the TMS9903
MODE
... a::
iii~
=>~
a::=>
c.)z
31
30
29
28
C.)
z
>
U)
c(
X
X
X
C.)
Z
>
U)
X
X
X
X
X
27
26
C.)
..J
Q
U)
X
X
X
X
X
X
X
Cf
25
X
X
X
24
23
X
X
X
X
X
(C
(C
22
X
X
X
X
X
21
X
X
X
20
X
X
X
19
X
X
X
18
X
X
X
17
16
X
X
X
X
X
X
15
14
13
X
X
X
X
X
X
X
X
X
12
X
X
X
FUNCTION
1 or 0 = Reset device.
1 = Clear transmitter. 0 = Clear receiver. (In each case interrupts are disabled!'
1 = Clear transmit CRC register. 0 = Clear receive CRC register. (CRC register is reset to 0).
1 = Delete received Syncl characters (in Bisync mode only).
1 = Inhibit transmit logic's zero bit insertion.
1 = Load data at CRU bits 0 - 9 into Sync2 register.
1 = Load data at CRU bits 0 - 9 into Sync1 register (only for versions of Synchronous mode that use Sync1 register).
1 = Read received check character via CRU bits 0 - 1 5.
0= Reset Status register CRU bits 13 (Check Character Buffer full), 12 (Check Character overrun) and 10 (Zero insert detect error).
1 = Load data output to CRU bits 0 - 8 into Transmit buffer, and update the transmit CRC. Select the transmit eRC to be read via CRU bits 0 - 15.
0= Reset Status register bits 22 and 17 ..
1 = Update the transmit CRC with the next output to CRU bits 0 - 9. Read transmit CRC at CRU bits 0 - 15.
1 = Transmit break (low level output) during underrun. Reset this bit to 0 before loading new data into Transmit buffer to end underrun.
Specify synchronous modes' underrun options. (See text)(General and Bisync only).
1 = Transmit Sync2 register contents following an underrun. (Typically 7F16 for an HDlC abort).
0= Abort transmit following an underrun and set Status register bit 23. (General only).
1 = Enable abort interrupt and reset Status register bits 23 and 18. (General only!.
o = Disable abort interrupt and reset Status register bits 23 and 18. (General only).
1 = Enable data set change interrupts and reset Status register bits 29 and 20.
o = Disable data set change interrupts and reset Status register bits 29 and 20.
1 = Enable timer interrupts and reset Status register bits 25, 24 and 19.
o = Disable timer interrupts and reset Status register bits 25, 24 and 19.
1 = Enable Transmit buffer empty interrupts.
o = Disable Transmit buffer empty interrupts.
1 = Enable Receive buffer full interrupts and reset Status register bits 21 and 11.
o = Disable Receive buffer full interrupts and reset Status register bits 21 and 11.
1 = Enable Receive buffer full, Received Check Character buffer full and received abort interrupts. Reset Status register bits 21, 14, 11 and 9.
0= Disable interrupts listed abov!J:!.eset Status register bits22.: 14, 11 and 9.
o or 1 = Output complement via RTS and disable automatic RTS control logic.
1 = Enable transmitter logic.
o = Disable transmitter logic after transmitting available data.
1 = Test mode. 0 = Normal operation.
1 = load data at CRU bits 0 - 11 into Control register.
1 = Load data at CRU bits 0 - 7 into Timer register.
o = Move Timer register contents to timer and start timer.
1 = Update the Receive CRC with the next output to CRU bits 0 - 9. Read Receive CRC at CRU bits 0 - 15.
Let us begin by examining the Control register; bit interpretations are defined in Table
3-12.
When you write to a TMS9903, CRU bits 31 through 12 will always access the Control register.
TMS9903
CONTROL
REGISTER
Control register bits may be divided into the following groups:
Device reset
Register select
Variations within mode specifications Interrupt enable/disable
Direct device control
•
which are made in the Parameter register
In most cases. when you set or reset a TMS9903 Control register bit. this bit setting - and
its associated logic - remains in effect until you specifically change the bit setting. When
setting a bit to select a data register. be sure to reset any select bits that were previously
set. If two or more register select bits are set simultaneously. you will receive no error
message. but the device will probably malfunction.
TMS9903
REGISTER
SELECT
Let us now examine Control register bits by group.
There are three device reset CPU bits: 31. 30. and 29.
When you write a 0 or a 1 to CRU bit 31. the entire device is reset; all interrupts are disabled
and all flags and register select bits are reset to 0 (with the exception of Control register bit 14
and Status register bit 22. which are set to 1). This causes the first data to be loaded into the
Parameter register, while a transmit buffer empty condition is reported in the Status register.
TMS9903
DEVICE
RESET
After resetting the TMS9903 by writing a 1 or 0 to CRU bit 31 and loading the
Parameter register (CRU bits 0 to 11). you must next clear the transmitter and
receiver by writing a 1 and then a 0 to CRU bit 30. (It does not matter whether you clear
transmitter or receiver logic first so long as you do clear each set of logic before attempting
to use it.) You must also initialize CRC accumulation logic at the transmitter and the
receiver by writing 1 and then 0 to CRU bit 29.
TMS9903
INITIALIZE
TRANSMIT/
RECEIVE
INITIALIZE
CRC
In summary. the following steps are required to reset and initialize a TMS9903:
1)
2)
3)
4)
Write 1 or 0 to CRU bit 31. This resets the entire device and enables loading of the Parameter register.
Load the Parameter register (CRU bits 0-11l. establishing the operating mode and configuration.
Write 11 to CRU bits 30 and 29. This initializes the transmitter and transmitter CRC logic.
Write 00 to CRUbits 30 and 29. This resets the receiver and receive CRC logic.
(Note that when you write to CRU bits 31, 30, and 29, you will always access Control register bits 30, 31, and 29; only
CRU bits 0-11 have multiple destinations within the TMS9903.l
After resetting the TMS9903 and initializing transmit/receive logic, you will next select addressable locations to read
from or write to.
Selecting the data location from which you will read is straightforward. Normally, CRU bits 0-8
will contain the Receive buffer contents, while CRU bits 9-31 are taken from the Status register.
But you can also read one of three 16-bit CRC characters. We may illustrate TMS9903 register addressing during a CPU read as follows:
,._---------_-
TMS9903
READ
REGISTER
ADDRESSING
, . - - - - - - - - - - - - - - - - - - - - - - - - - - - From Status register
Computed CRC for received
message (Control register
bit 12 = 1)
/..-___-'A....._ ....- -....., " , . - - - - - _...~ .....-------~
Received CRC for received
·17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
SOLC frame (Control register
3130· •
!
IIIIIIIII IIIIIIIIII
'1
~--T
IIIII
..
. .T
,,/
~-------
3-100
~i:~:u~e~)CRC for transmitted
message (Control register
bit 24 or 25 = 1)
Otherwise:
From Receive buffer
From Status register
Note carefully that in SDLC mode you can read two receive cyclical redundancy check characters: the first is
computed under program control by receive logic for the received frame; the second is received at the end of
the frame.
The final 16 bits of the information field are the received 'cyclical redundancy character. To read the received cyclical
redundancy character. set Control register bit 26 to 1. To read the cyclical redundancy character computed by receive
logic for the received frame. set Control register bit 12 to 1. These two cyclical redundancy characters will be identical
if a valid message was received.
In Synchronous and Asynchronous modes there is no defined end-of-message. Rather. a control character in the
received data stream is interpreted as an end-of-message indicator. in which case two previously received data characters are interpreted as the received cyclical redundancy character. Your program logic must compare the two data
characters which are being interpreted as the received CRC character with the computed check character. read from
receive logic after setting Control register bit 12 to 1.
When the CPU reads from the TMS9903. if Control register bits 12. 24. 25. and 26 are all reset to O. then as the default
case CRU bits 0-8 are taken from the Receive buffer; higher numbered CRU bits are taken from the Status register. as
always.
When writing to the TMS9903. Control register address bits used to select a data location
for the low numbered CRU bits may be illustrated as follows:
31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9
Ixlxlxlxl
Ixlxlxl
8
7
6
5
4
TMS9903
WRITE
REGISTER
ADDRESSING
3
2
1
0
II
1
T
t
Control Register Bits
Control Register
Sync2 register
(up to 1 0 bits)
Sync 1 register
(up to 10 bits)
Transmit buffer (9 bits)
(also default location)
Transmit CRC register
(up to 10 bits)
Parameter register
(12 bits)
Timer register
(8 bits)
Receive CRC register
(up to 10 bits)
High numbered CRU bits always go to the Control register. Low numbered bits go to the write location whose
register select bit within the Control register is 1.
Following a reset. Control register bit 14 is set to 1. therefore data written to CRU bits 0-11 loads the Parameter register.
When you write into the high-order Parameter register bit (bit 11). Control register bit 14 is automatically reset.
But this is an exception. When you set any other register select bit in the Control register it remains set until you
specifically reset it.
If the Parameter register select bit (Control register bit 14) is set and you want to write to another addressable location.
then you must reset Control register bit 14 to 0 when setting another select bit to 1.
If all select bits in the Control register are O. then as a default case data will be written to the Transmit buffer.
3-101
You can only write into the Sync2 register in Synchronous or SOLC modes.
You can only write into the Sync1 register in Synchronous mode - and only in those variations of Synchronous mode
that use the Sync1 register. Variations of Synchronous mode are described later.
There are two Control register bits, 28 and 23, which you use to specify variations of mode specifications. We
wi" describe these two bits together with Parameter register bit settings, since Control register bits 23 and 28
are logica"y extensions of the Parameter register.
TMS9903
Five conditions capable of requesting interrupts have separate enable bits; these are Control register bits 22 through 18. When you write a 1 to any of these Control register bits. the
INTERRUPT
associated interrupt logic is enabled; when you write a 0 to that Control register bit. interrupt
ENABLE/DISABLE
logic is disabled. In most cases. when you write a 0 or a 1 to an enable/disable bit. you reset
any associated Status register bits. Exceptions are the Transmit buffer empty status and the Received CRC register full.
We will discuss individual interrupts in more detail later when looking at TMS9903 interrupt logic in general.
Direct device control bits consist of transmitter control and receiver controls.
Looking first at the transmitter. you must enable transmit logic, after clearing it, by setting
Control register bit 16 to 1; transmit logic remains enabled until you reset this bit to O. Transmit
logic will not disable itself in the middle of transmitting a character; if you write a 0 to Control
register bit 16 part way through a character's transmission. the character will be transmitted and
transmit logic will then be disabled.
TMS9903
TRANSMIT
CONTROLS
If you never write to Control register bit 17 following a reset, then the RTS output signal level is automatically
controlled by transmitter logic. As soon as you enable transmitter logic by writing a 1 to Control register bit 16. RTS is
output low; RTS remains low until you disable transmitter logic by writing a 0 to Contro~ister bit 16. But if you !!!!
write to Control register bit 17, you immediately disable the automatic control of the RTS output level. Now the RTS
output level becomes the reciprocal of Control register bit 17.
There are two ways in which you can include transmitted characters in any cyclical redundancy character computation.
If you select the Transmit buffer by setting Control register bit 25 to 1. then the character which you write to the
Transmit buffer is also included in the transmit cyclical redundancy character computation.
If you select the Transmit buffer as the default write location (j.e .. no address bits in the Control register are set to 1).
then the character which you write to the Transmit buffer will not be included in the transmit cyclical redundancy
character computation unless you set Control register bit 24 to 1 and then output the character to Transmit CRC logic.
That is. using bit 24 of the Control register you can write to either the Transmit buffer or to Transmit CRC logic. but not
to both at the same time.
When a large sequence of contiguous characters is to be included in the transmit cyclical redundancy character
computation, use Control register bit 25.
When characters are to be selectively included and excluded in the transmit cyclical redundancy character computation, use Control register bit 24.
There is no receiver enable control equivalent to the transmitter enable (Control register bit 16). As
TMS9903
soon as you clear receive logic. it is enabled and will begin to sample data arriving via RxO. As
RECEIVE CRC
each character is assembled. it is transferred to the Receive buffer. If a received character is to be
included in the computed receive cyclical redundancy character. program logic must output that character to Receive
CRC logic after reading it from the Receive buffer. When you set Control register bit 12. data output to CRU bits 0-9 will
go to Receive CRC logic.
Note that CRC logic is not necessarily connected to the transmitter or receiver. The cyclical redundancy calculation
registers may be used independently of transmit or receive logic.
The Test mode bit (Control register bit 15) is normally left reset to O. When you set this bit
to 1 the following connections occur:
TMS9903
TEST MODE
1)
2)
TxD is connected to RxO.
RTS is connected to CTS. and DSR is held low.
3)
TxC and RxC are both connected to the timer logic clock. which operates at 32 times its normal rate.
This is similar to TMS9902 Test mode. with the exception that. in the TMS9903. the timer determines Receive and
Transmit data rates in Test mode.
3-102
We will next describe the Parameter register. You will normally write into this register once
during any operation in order to define operating modes and options within these modes.
TMS9903
PARAMETER
After resetting the TMS9903 by writing to CRU bit 31. you simply output the parameter code to
REGISTER
CRU bits 0-11. Resetting the device automatically selects the Parameter register as the write location for the data at CRU bits 0-11. You could also select the Parameter register by writing Os to CRU bits 27. 26. 25. 24.
13. and 12. and writing a 1 to CRU bit 14. Parameter register contents are interpreted as follows:
11
10
9
8
65432
o
4 - - CRU Bit
~~~~~Io......._ ...I_. . . . . . .Io...""~""'''''''_'''''''''''1..-- Parameter register
~~~L{
o0 0 - 5
o0 1 - 6
o10 -7
o1 1-8
bits/character
bits/character
bits/character
bits/character
1 0 0 - 9 bits/character
.....- - - - - - - - - - {
0 - Divide (j) by 3 to generate timer clock
1 - Divide by 4 to generate timer clock
Non-SOLC
"--------------
{
0 0
o1
10
1 1
- No parity
- No parity
- Even parity
- Odd parity
o0 0
001
o10
o1 1
10 0
10 1
1 10
1 1 1
L-___________________ {
.....- - - - - - - - - - - - - - - - - - - - - - - {
SOLC
Point-to-point
Loop master
Loop slave - inactive
Loop slave - active
- Synchronous - general
- SOLC
- Monosync
- Bisync
- Unassigned
- Asynchronous with two stop bits
- Asynchronous with one stop bit
- Unassigned
0 0 - CRC-16 (X 16 + X 15 + X2 + 1)
01 -CRCC-12 (X12 + X11 + X3 + X2 + X + 1)
1 0 - Revised CRCC-16 (X16 + X14 + X + 1)
11 -CRC-CCIT (X16
+ x12 + X5 + 1)
0 - Transmit/receive at input clock rate
1 - Transmit/receive at input clock rate -7- 32. and
use zero-complementing NRZI encoding.
Parameter register bits 6, 7, and 8 determine the operating mode for transmit and receive logic, and some options within the selected mode.
When you select Asynchronous mode, you also select either one or two stop bits.
In Asynchronous mode, when you set Control register bit 23 to 1, then as soon as an underrun occurs transmit logic will output a continuous low level (break) on TxO. But note
carefully that setting Control register bit 23 to 1 does nothing until an underrun occurs. Once
an underrun does occur. you cannot load new data into the Transmit buffer until you reset
Control register bit 23 to 0 to end the break.
TMS9903
ASYNCHRONOUS
BREAK LOGIC
If Control register bit 23 is reset to O. then following an underrun a continuous high signal is output via TxD. You can at
any time r'estart transmission by loading data into the Transmit buffer - in which case the high level output at TxD
ends and the next character is transmitted according to the Asynchronous protocol options specified in the Parameter
register.
There are three Synchronous mode options and one SOLC mode option. These four options share Sync character
logic, as shown in Table 3-13. This table applies to transmit and receive logic.
Let us first consider SOLC transmit logic.
3-103
Table 3-13.
Parameter
Register
CRU Bit
TMS9903 Synchronous and SOLC Mode Sync
Character and Underrun Options
Underrun Fill Character
MODE
SYNC
Character
Control Register
CRU Bit 23 = 0
Control Register
CRU Bit 23 = 1
Abort
[SYNC2]
8
7
6
0
0
0
0
0
1
SDLC
7E16
Abort
[SYNC2]
0
1
0
SynchronousMonosync
[SYNC1]
[SYNC2]
[SYNC2]
0
1
1
SynchronousBisync
[SYNC 11 - [SYNC2]
[SYNC 1] - [SYNC 11
[SYNC21 - [SYNC 11
[]
SynchronousGeneral
None
Means: "contents of register named within brackets"
Every frame must begin with a flag character, therefore 7E 16 is always output as the leading Sync
TMS9903
character. You will subsequently reset Control register bit 23 to 0, since underruns are not allowed
SOLC
within an SOLC frame. Should an underrun occur, the transmitter will abort. outputting a conTRANSMIT
OPERATION
tinuous high signal and setting appropriate status bits. In order to transmit a valid end-ofmessage, you must read the computed transmit check character (selected via Control register bit
24 or 25). set Control register bit 23 to 1, load a flag (7E 16) character into the Sync2 register, and output the computed
transmit check character as two data bytes. Now allow an underrun to occur; the contents will be output when the underrun occurs. Since Sync2 contains a flag character, you will have terminated the frame by transmitting the message
check character and closing flag, as required by SOLC protocol.
There is another way of ending a frame's transmission.
Instead of allowing an underrun and outputting the frame's closing flag from the Sync2 register, you can suppress
SOLe 0 insertion by writing a 1 to Control register bit 28, then outputting the clOSing flag (or flags) as a simple sequence of 8-bit data characters.
If you are operating the TMS9903 using HOLC protocol. then you must output 7F16 as your
TMS9903
abort character. To obtain a valid HOLC abort following a transmit underrun you should write
HOLC ABORT
the HOLC abort character to the Sync2 register, then leave Control register bit 23 set to 1 while
the frame is being transmitted. Now if an underrun occurs, an HOLC abort character will be output from the Sync2
register.
When detecting a new frame, SDLC receive logic synchronizes itself on flag character 7E16,
which is also the specified Sync character. Consequently the setting of Control register bit 23
and the underrun fill character options shown in Table 3-13 do not apply. When receive logic
detects another flag character, it assumes it has received the frame's closing flag. SOLC receive
logic can a Iso detect an abort. SOLC receive logic sets appropriate status flags and generates an
interrupt request. if enabled.
TMS9903
SOLC
RECEIVE
LOGIC
The three Synchronous modes shown in Table· 3-13, together with their underrun fill character options, allow
you (under program control) to emulate any of the synchronous protocol options commonly encountered.
External synchronization uses no leading Sync characters at the head of a message. You can
emulate this protocol by choosing the general synchronous option.
ForTMS9903 transmit logic, make sure that CTS is low before you enable the transmitter; then as
soon as RTS goes low, message transmission begins. A station that receives this transmitted
message can use the low RTS output as its external Sync input.
TMS9103
EXTERNAL
SYNC LOGIC
TMS9903 receive logic will use the OSR Modem input as its external synchronization signal. The station which
transmits the signal to the TMS9903 must generate a low OSR input just before it starts transmitting a message. The
program controlling TMS9903 receive logic must detect the low '[5'S'R' input by interrogating the appropriate Status
register bit. and upon detecting OSR active should start receiving.
3-104
In Monosync mode a single Sync character occurs at the head of a new message. In Bisync mode two Sync
characters occur at the head of a new message. Both of these options are allowed.
The Monosync mode outputs the contents of the Sync1 register at the head of a transmitted
message and synchronizes on a received message by matching received characters against the
contents of the Sync1 register.
TMS9903
MONOSYNC
lOGIC
In Bisync mode Sync1 contents are output twice at the head of a transmitted message.
Receive logic assumes that a new message has been detected when two contiguous characters
match the contents of the Sync1 register.
BISYNC
lOGIC
By loading appropriate data into the Sync1 and Sync2 registers you can transmit and detect
ASCII. EBCDIC. or any other Sync characters.
When an underrun occurs in Monosync mode. a single Sync character is output. By loading the appropriate character
into the Sync2 register you can transmit and detect any underrun fill character.
In the Bisync option greater underrun flexibility is needed. In some cases. following any underrun two Sync characters
are transmitted; but in standard Bisync protocol DLE-SYN character combinations are output following an underrun.
When Control register bit 23 equals 0 the TMS9903 will output two Sync characters from the Sync1 register. To meet
the requirements of Bisync protocol you load the DLE character into the Sync2 register. load the SYN character into the
Sync1 register. and leave Control register bit 23 set to 1. Other bisync logic (in particular. the generation and detection
of special control character combinations) must be handled by a supervisory program.
Control register bit 28 adds some flexibility to the options shown in Table 3-13. However. this control bit applies
only to SOLC and Bisync modes. In SDLC mode, when Control register bit 28 is reset to O. TMS9903 transmitter logic
will insert a 0 after every five consecutive 1s transmitted. Setting Control bit 28 to 1 inhibits this zero bit insertion in
SOLC mode.
In Bisync mode. when Control register bit 28 is set to 1 any received character that matches
the contents of the Sync1 register is discarded. This allows you to strip received underrun
Sync characters.
TMS9903
SYNC STRIP
Parameter register bits 5 and 4 serve different functions in Synchronous and SOlC modes.
,..-----,
In Synchronous and Asynchronous modes Parameter register bits 5 and 4 are used to
TMS9903
specify odd parity, even parity, or no parity. When parity is specified. parity bits will automatPARITY
ically be generated for data characters that are transmitted and will be tested for data characters
OPTIONS
received. But parity does not apply to the contents of the Sync1 or Sync2 registers. You must
add your own parity bit to the contents of these registers if you want to transmit Sync characters with parity. The Sync
registers are each ten bits wide so that you can add one parity bit to the longest specifiable character (nine bits).
Receive logic will automatically check the parity of received Sync characters. since received logic treats all receive
characters as data.
InSOlC mode, Parameter register bits 5 and 4 specify Loop or Non-loop mode; in fact. they
specify the way in which an EOP character (7F 16) is handled.
TMS9903 SOlC
CONFIGURATIONS
In a point-to-point configuration the EOP character has no significance. and is ignored.
As a loop master. transmit logic pays no attention to the EOP character; however. receive logic treats the EOP character
as a frame's closing flag. This is necessary. since the polling EOP character which a loop master transmits around the
loop will eventually be received as the closing flag for the last frame transmitted by a loop secondary.
The loop slave inactive mode is selected for an SOlC loop secondary that is not transmitting data, but may be
receiving data. The loop slave active mode, in contrast, is selected for a secondary station in the SOlC loop that
wishes to transmit to the primary station.
In loop slave inactive mode, a TMS9903 will initially retransmit received data without delay. But. upon detecting an
EOP character in the received stream, the TMS9903 will introduce one bit delay before retransmitting received data. So
long as you never electrically disconnect a secondary station in an SOLC loop. the inactive slave mode will take care of
timing and protocol requirements of a secondary loop station coming on-line. But if you wish to electrically disconnect
3-105
a TMS9903 secondary station in an SOLC loop. you will require external logic which detours upstream data around the
electrically disconnected secondary. while breaking the detour and including the secondary when it is electrically connected. Here is the appropriate logic:
-------------------1
Up-loop data ----....
RxD
Down-loop data
IN LOOP - - - - - -.....-cJI
TxD------------------------~
You will normally leave a TMS9903 operating in loop slave active mode if it is functioning as a secondary station in an
SOLC loop. You will only switch to loop slave inactive mode when the secondary station has just entered the
loop and is not yet synchronized (has not received EOP). In the loop slave active mode. TMS9903 receive logic will
seek the next EOP character. Upon receiving an EOP character it will convert this character to a flag. which becomes
the opening flag for the frame which the station wishes to transmit to the primary. So long as a TMS9903 is left
operating in loop slave active mode, it will continue to trap receive EOP characters and transmit frames behind
them. When a TMS9903 has no further frames to transmit. you should leave it in loop slave active mode. but turn off
the transmitter by resetting Control register bit 16 to O.
For a discussion of SDLC loop secondary station logic see Chapter C1 in Volume 3.
Parameter register bits 0, 1, and 2 allow you to specify 5,6, 7, 8, or 9 data bits per received
character. Note that if parity is enabled. the parity bit is not counted in this specification.
TMS9903
RECEIVED
CHARACTER
SIZE
If Sync and control characters are eight bits wide, then you cannot specify less than 8-bit
characters in Synchronous mode. This is because receive logic does not automatically switch
from the specified bits per character to eight bits per character when receiving Sync or control
characters. Moreover. a program controlling receive logic cannot make this switch. since it does not know it has
received a Sync or special control character until the character is in the Receive buffer - by which time it is too late to
make a change.
Parameter register bit 11 allows you to transmit and receive data at the transmit and
receive clock rates, or at these clock rates divided by 32. This is normally a standard Synchronous mode option. With the TMS9903 it is available in all modes: Synchronous. SOLC.
and Asynchronous. This bit should be reset to 0 during operation as an SDLC loop slave.
TMS9903 CLOCK
RATE OPTION
NRZI SELECT
During synchronous or SDLC operation, if data is being sampled on every 32nd clock
pulse (Parameter register bit 11 is 1) then NRZI encoding and decoding of serial data is assumed; that is. the data
signal changes state to transmit a 0 or remains in the same state to represent a 1.
Parameter register bits 9 and 10 are used to specify the cyclical redundancy character
algorithm which will be used by transmit and receive logic.
TMS9903
eRC OPTIONS
CRC-16 is the normal algorithm used by synchronous and asynchronous protocols.
CRCC-12 is the algorithm used in synchronous and asynchronous protocols with 6-bit characters.
Revised CRCC-16 is the protocol frequently used in standard Bisync protocol.
CRC-CCIT is the standard SDLC algorithm.
Parameter register bit 3 is used by interval timer logic. This bit will be discussed later when we describe the interval
timer.
We will now examine TMS9903 Status register bit settings, which are summarized in Table
3-14. Status register bits may be divided into the following groups:
• Interru pt status
• Input signal levels
• Transmit logic status
• Receive logic status
• Timer logic status
3-106
TMS9903
STATUS
REGISTER
Table 3-14. TMS9903 Synchronous Communications Controller CRU Bit Assignments
when Reading from the TMS9903
MODE
!::ffi
mm
u
Z
u
U Z
(I)
...I
Q
c(
(I)
31
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
:::):E >
CZ::::)
30
29
28
27
26
25
24
23
22
21
20
19
Cf
o
-...I
18
17
16
X
X
X
X
X
X
X
X
X X X
X
X
x
15 IX IX I X
14 X
X
13 IX
X
12 I X
11 IX
10 X
I
X
X IX
X
X
9 IX IX
x
Function
1 = Any interrupt pending
1 = One or more Register Load Control flags set.
or automatic RTS signal level change occurred
1 = 55ft or
Complement of ffi input
Complement of DSR input
RTS level under automatic control. Transmitter active state if"RTS is under program control
1 = Timer decremented to 0
1 = Timer error. Bit 25 was already 1 when timer decremented to 0
1 = Abort followed an underrun (General only)
1 = Transmit buffer empty
1 = Receiver buffer full
1 = Interrupt request accompanying RTS, DSR, or CTS signal level change (Bit 29 = 1)
1 = Interrupt request accompanying timer time out (Bit 25 = 1)
1 = Interrupt request accompanying an abort (Bit 23 = 1)(General only)
1 = Interrupt request accompanying a Transmit buffer empty
1 = Interrupt request accompanying Receiver buffer full (Bit 21 = 1) only
1 = Interrupt request accompanying Receiver buffer full (Bit 21 = 1)
1 = Interrupt request accompanying Receiver buffer full (Bit 21 = 1) or abort received
(Bit 14 = 1) or Closing flag received and received check character ready to be read
(Bit 13 = 1)
RxD input level
1 = Start bit detected
1 = Abort received
1 = First character data bit received
1 = Closing flag has been received and check character may be read
1 = Receive framing error detected
1 = Overrun error detected - receive data overrunning previous frame's check character
1 = Receive overrun error detected
1 = Receive parity error detected
1 = Zero insert error detected
1 = Any receive error in most recently received character
1 = Flag detected
m
Reset Condition
No interrupt pending
No Control flag set
Output to CRU bit 21
Output to CRU bit 20
Output to CRU bit 20
Output to CRU bit 22
Output 0 to CRU bit 25
Output to CRU bit 18
Bit 29 = 0 or Output to bit 21
Bit 25 = 0 or Output to bit 20
Bit 23 = 0 or Output to bit 22
Output 0 to CRU bit 25
~ No active interrupting
~ condition
Stopbit(s) received
Output to CRU bit 18
Stop bit(s) received
Output 0 to bit 26
Error free character received
Output 0 to bit 26
Output to bit 18
Valid character received
Output 0 to bit 26
Valid character received
Output to bit 18
The interrupt status bits include CRU bit 31, which reports any active interrupt request. and CRU bits 20 through
16, which identify individual interrupts. These status bits are self-evident. In non-vectored interrupt configurations you
will test CRU bit 31 to find out if this particular TMS9903 has any active interrupt requests. In a vectored interrupt configuration you can ignore CRU bit 31, since the interrupt acknowledge process will identify the TMS9903 as the device
with the active interrupt request. I n each case, the interrupt service routine must examine CRU bits 20 through 16 in
order to determine which interrupt requests are active. The interrupt service routine must resolve its own interrupt
priorities.
Input Modem signals OSR and CTS modify Status register bits 27 and 28, respectively. The camp lement of the
input signal level is reported. Status bit 29 is set to 1 when either ~, CTS, or automatic RTS signal level
changes. This signal level change can cause an interrupt request, in which case Status register bit 20 is set. In
many serial I/O devices, CTS going high in the middle of a transmit operation forces a transmit abort. while DSR going
high in the middle of a receive operation disables receive logic. The TMS9903 does not make such critical decisions;
the supervisory program must respond appropriately.
When RTS output level is being controlled automatically, the complement of RT$ is reported in Status register
bit 26. But as soon as you start controlling RTS level by writing to Control register bit 17, Status register bit 26 reports
the active state of the transmitter.
The serial data input signal RxO has its level reported in Status register bit 15.
There are two status bits associated with transmitter logic: bit 22 reports Transmit buffer empty and bit 23 reports a transmitter abort (in those modes that can generate an abort). If interrupt logic for these conditions has been
enabled, then Status register bits 18 andlor 17 will also be set.
There are a number of Status register bit settings associated with receive logic, but there is only one interrupt status
bit associated with receive logic - bit 16. Therefore you must use the various receive status bits in order to identify
active error or non-error conditions within receive logic.
In all modes Status register bit 21 is set when the Receive buffer is full time.
and should be read within one character
In Synchronous mode, Status register bit 11 reports a receive overrun error, while Status register bit 10 reports
a receive parity error. Either of these errors causes Status register bit 9 to be set.
In Asynchronous mode, a receive framing error, overrun error, or parity error is reported in status bits 12, 11,
and 10, respectively. Status bit 9 reports one or more of these error conditions. In Asynchronous mode, two status
bits are also set at the beginning of each received character. Status bit 14 is set when a valid start bit has been
detected for the character, while status bit 13 is set when the first valid data bit has been detected.
In SOLC mode, a receive overrun is reported in status bit 11 and a receive zero insert error is reported in status
bit 10. The receive zero insert error means that five contiguous 1 bits were received, followed by a flag character. without the expected zero inserted between them. Thus, status bit 10 will be set when the sequence 01111110111112 is
received. While a frame is being received, Status register bit 14 is set when an abort is detected and Status
register bit 9 is set when any flag character is detected.
An unusual and interesting error reported in SOLC mode is the receive CRC overrun error. If a new frame's data
is received before you read the previous frame's cyclical redundancy check character, then status bit 12 is set.
There are two timer logic status bits; bit 25 is set to 1 whenever the timer decrements to zero. If timer interrupts
have been enabled, then status bit 19 is also set. You must acknowledge a time-out before another time-out occurs.
You acknowledge a time-out by outputting to Control register bit 20. If you do not do so, then on the next time-out
Status register bit 24 is set.
You can examine Status register bit 30 at any time to see if one or more write location select bits are set in the
Control register.
3-108
TMS9903 INTERRUPT LOGIC
There are seven conditions that can generate interrupt requests within the TMS9903. Three of the seven conditions combine to generate a single interrupt request status. Therefore, there are five interrupt request statuses for the
seven interrupt generating conditions. This may be illustrated as follows:
Status
Register
Condition
Bit No.
Receive buffer full
End of SOlC frame
Receive abort
Control
Register
Interrupt
Enable
Bit No.
Status
Register
Interrupt
Bit No.
29 - - 21 - - 20 25 - - 20 - - 19 22 - - 19 - - 17 23 - - 22 - - 18 21
13 7 1 8 - - 1 6 14
Interrupt
DSR. CTS, or automatic RTS level change
Timer time out
Transmit buffer empty
Transmit abort
Receive interrupt
The TMS9903 has no internal interrupt priority arbitration logic. When one or more conditions capable of requesting an interrupt occur, if the interrupt has been enabled, then INT is output low and Status register bit 31 is set to 1. An
interrupt service routine responding to the TMS9903 interrupt request must now interrogate Status register bits in
order to determine which interrupt requests are active. Program logic is responsible for all interrupt priority arbitration. These are the interrupt priorities which normally apply in serial I/O devices:
1)
HIGHEST PRIORITY. Receive buffer full (Status register bits 16 and 21 set)
2)
3)
4)
5)
6)
7)
Transmit buffer empty (Status register bits 17 and 22 set)
Modem Signal level change (Status register bits 20 and 29 set)
Receive abort detected (Status register bits 16 and 14 set)
Transmitter abort (Status register bits 18 and 23 set)
End of SDLC frame detected (Status register bits 16 and 13 set)
LOWEST PRIORITY. Timer interrupt (Status register bits 19 and 25 set)
TMS9903 INITIALIZATION PROGRAM LOGIC
The first step in any TMS9903 operation is usually to initialize the device. Here are the necessary steps:
1)
Reset the device by writing 0 or 1 to Control register bit 31.
2)
Now output appropriate Parameter register settings.
~-....;...-----
TMS9903 DEVICE
INITIALIZATION
3)
Output data to Control register bits 18 through 22 to enable appropriate interrupts.
4)
In Synchronous and SDLC modes, load appropriate codes into the Sync2 and/or Sync1 registers. These two
registers are not used in Asynchronous mode.
5)
To initialize receive logic, write 0 to Control register bit 30. If cyclical redundancy is being used, initialize receive
CRC logic by writing 0 to Control register bit 29. As soon as this step is complete. receive logic becomes active and
starts to assemble received data.
6)
To initialize transmit logic, write 1 to Control register bit 30. If cyclical redundancy is being used, initialize transmit
CRC logic by writing 1 to Control register bit 29. Transmit logic is now initialized, but it is not yet enabled.
7)
Transmit logic will not become active until you enable the transmitter by writing 1 to Control register bit 16. When
you enable the transmitter, you should also load data into the Transmit buffer. Refer to our earlier discussion of
Control register bits 25 and 24, where data output to the Transmit buffer is described, with or without associated
CRC accumulation.
TMS9903 ASYNCHRONOUS OPERATIONS
When you select Asynchronous mode, data will be transmitted with a parity bit if selected,
TMS9903
plus one or two stop bits, as specified by the Parameter register. Whenever the Transmit buffer
ASYNCHRONOUS
becomes empty. an interrupt request will be generated if the Transmit buffer empty interrupt
TRANSMIT
has been enabled, and appropriate status bits will be set - as described earlier. You have one
character time within which to respond by outputting another character, or else an underrun will occur. Following an
3-109
underrun, a continuous high (marking) signal or a continuous low (break) signal will be output. depending on the setting of Control register bit 23. See the break discussion given earlier for details.
When beginning a receive operation, sample the start bit detected status (Status register bit
TMS9903
14) to identify the beginning of a new received message. This status cannot generate an interASYNCHRONOUS
ru pt request. To process received characters, use Receive buffer fu II interrupt request logic. As
RECEIVE
characters are received, program logic must read characters out of the Receive buffer within - - - - - - - -.....
one character time, and check for any of the asynchronous receive error conditions by reading error Status register bits
at the same time. Received data and status can be read together by reading CRU bits 0 through 15 from the TMS9903.
There are no other special programming considerations associated with asynchronous operation of the TMS9903. Conversely, any other protocol requirements must be met by the supervisory program's logic.
TMS9903 SYNCHRONOUS OPERATIONS
Most of the logic associated with Monosync and Bisync protocols must be provided by the supervisory program
that controls TMS9903 transmit and receive operations. The only logic capabilities provided by the TMS9903 itself
are the various Sync register programmable options, the error and normal operation statuses reported, and the
character length definition.
For a discussion of the Sync character options, refer to our earlier description of the Parameter register.
For a description of the statuses reported, see the Status register discussion and interrupt logic summary.
TMS9903 SOLC OPERATIONS
When discussing the Parameter register we explained how you will use the Sync2 register in order to transmit
and receive frames; but there are additional SOLC protocol requirements and some common protocol variations
which need to be discussed.
SDLC and HOLC protocols are described in Chapter C1 of Volume 3. In SOLC protocol. the first byte of every frame is
the address field, while the second byte is a control field. In HOLC protocol the address field can have any length, while
the control field can be either one or two bytes long. Some variations of SOLC protocol insert a logical control field after
the control field; the logical control field can have any length. Address field. control field. and logical control field
characters are all eight bits wide. Information field characters can have any data bit width. The number of bytes
in a multibyte address or logical control field is determined by examining a specific character bit. For example, a protocol may specify that the last byte of an address field will have a 1 in the low-order bit. while all prior bytes have a 0 in
the low-order bit.
The TMS9903 has no on-chip logic designed to handle address, control, or logical control fields. Device programming can specify the number of bits per character - and this specification may change from character to characterand that is all. Moreover, the supervisory program must take into account primary or secondary station logic.
A supervisory program at a primary station must transmit secondary station addresses and must interpret received addresses as identifying a frame's source.
At a secondary station, the supervisory program must always transmit its own address at the head of a frame and must
examine the address at the head of a received frame to see if the rest of the frame should be read or ignored.
When the last byte of the control field (or logical control field) has been received, program logic must change the bitsper-character specification in the Parameter register before processing the first character of the information field should the information field use a different character length. Remember, the bits-per-character specification in the
Parameter register applies only to receiver logic; you specify transmit character size by the number of bits you output
to the Transmit buffer. Thus it is a simple matter to change character size from character to character as protocol may
require. It is also possible for a received character to have a different number of bits than a simultaneously transmitted
character.
TMS9903 receive logic in SOLC mode does have one very useful end-of-frame capability: the received check
character (which must be the 16 data bits preceding the frame's closing flag) is automatically loaded into a received
CRC register. The microprocessor can read this received check character and compare it with a computed check
character.
But there are some additional uses for this received CRC logic.
A valid SOLC frame must have at least 32 bits between the beginning and closing flags; these bits include an 8-bit control field, and a 16-bit cyclical redundancy check character. Frequently, 32-bit frames are transmitted and received to
pass a command or response with no associated data. An error occurring within such a short frame can cause complex
logic problems; it may be difficult to identify beginning and ending flags for subsequent frames, since the ending flag
3-110
for the short frame may go undetected. But you can use the TMS9903 receive CRC logic to identify short received
frames. If you do not get a valid status indicator telling you that the received check character is available. then you
know you have received a short frame.
SOLe protocols allow frames to be separated by a single flag character. which serves as the closing flag for one
frame and the opening flag for the next frame. Alternatively. a number of flag characters may separate two
frames. Either case can be handled by the TMS9903.
When describing the Parameter register. we explained how you can generate a frame's closing flag out of the Sync2
register. after allowing an underrun. or inhibit zero insertion and transmit flag characters as data. Since you can individually specify characters that will or will not be included in cyclical redundancy check accumulations. processing
non-data characters as though they are data characters presents no difficulties to a TMS9903. You can use either
method of ending a frame to separate frames with one or more flag characters.
If you have generated an end-of-frame using underrun logic. then loading the next frame's first address field byte while
a single flag character is being transmitted will cause a single flag to separate the two frames. If you let the underrun
last longer. then flag characters will continue to be output until you begin the next frame by writing the frame's first address field byte as data to the Transmit buffer. If you are transmitting flags as data without zero insertion. then the number of flags separating frames is strictly a function of program control - but you must make certain that an underrun
does not occur.
Let us now examine programming requirements within an SOLC loop.
There are no special programming requirements for the primary station in an SOLC loop. If you
generate an abort at the end of a transmitted frame's closing flag. then the flag's trailing 0 bit.
together with the first seven 1 bits of the abort. constitute an EOP character - which is transmitted around the loop in
order to poll secondaries. When this EOP character returns to the primary station's receive logic. it is treated as a closing flag. (Refer to our discussion of the TMS9903 Parameter register for details.)
Secondary stations within the SOLC loop should be run in the SOLC loop slave inactive mode until the secondary station has become synchronized with the loop - that is. has received the EOP character and begun retransmitting with a
one-bit delay. At this time. change the secondary station mode to SOLC loop slave active. In the active condition. the
secondary station will seek the next EOP character arriving at RxO. If the transmitter has been enabled. the TMS9903
will convert this received EOP character to the opening flag character for the frame which it wishes to transmit. The
program controlling the secondary SOLC can end the transmission with a closing flag and then an abort. or with an
EOP character. The closing flag and following abort generate an EOP character for the next downstream secondary and multiple flags between frames. A closing EOP character will be converted by the next downstream secondary to a
flag or will be passed on to the primary. which interprets EOP characters at receive logic as closing flags. A closing EOP
character. therefore. generates a single flag separating two transmitted frames.
For a discussion of normal status and error status that may occur during transmit and receive operations. refer to
our earlier description of the Status register. Also. refer to our earlier description of the Parameter register for logic
which you will use to abort a mistransmitted frame. or to detect an abort in a received frame. But remember. it is entirely up to the supervisory program to interpret status bits and to handle aborts as required by the local system logic.
TMS9903 INTERVAL TIMER LOGIC
The TMS9903 has an interval timer. You initialize the interval timer by loading a timer count into the Timer
register. Remember. you set Control register bit 13 to 1 in order to select the Timer register as the destination for data
output via CRU bits 0 through 7. As soon as you reset Control register bit 13 to 0 you enable the timer. which starts to
decrement. The rate at which the timer decrements depends on Parameter register bit 3 and Control register bit 15.
Parameter register bit 3 allows you to divide the 1.4>2.4>3.4>4
-100
All others
-400
IlA
IlA
4>1.4>2.4>3.4>4
4
mA
All others
8
rnA
54
MHz
48
Internal oscillator frequency. fosc
External oscillator pulse width. tw(osc)
25
ns
Setup time. FFO input (with respect to falling edge of 413). tsu
50
ns
Hold time. F F 0 input (with respect to falling edge of 4>3). th
-30
Operating free·air temperature. T A
0
3-07
ns
70
'c
TIM9904
Electrical Characteristics Over Recommended Operating Free-Air Temperature Range
(Unless Otherwise Noted)
TEST CONDITIONS
PARAMETER
MIN
H'gh·level input voltage
VIH
VIL
VT~
- VT_
0.5
Input voltage
OSCIN
0.8
HysteresIs
FFO
0.4
II
Other outputs
4
Low-level
VOL
output voltage
Other outputs
Input current at
FFO
maxImum input voltage OSCIN
IIH
IlL
lOS
=4.75 V, VOO = 11.4 V,
Vce =4.75 V,
V DO = 11 .4 V to 12.6 V
VCC
1, 2, 2L, cp3H
trJ>3L cp4H
Delay time, CP3 low' to 1/>4 h,gh
tcp4L,cplH
Delay time, 4 low to .pl h'gh
tcp1H,4>2H
Delay time, cpl high to ,p2 h,gh
t.p2H,cp3H
DeillY time, 1/>2 high to 'PJ high
tr/>3H, tP4H
Delay time, tP3 high to ,p4 hiqh
tcp4H,H,CPTL
MAX
12
330
UNIT
MHz
3
MHz
340
ns
20
ns
14
20
40
55
70
Oelay time, cp I low to <1>2 high
0
5
15
Oelay time, <1>2 low to <1>3 high
0
5
15
ns
ns
ns
ns
0
5
15
ns
0
5
15
ns
rJ>l,cp3,cp4: 100pF toGND
73
83
96
ns
tP2: 200 pF to GND
73
83
96
ns
Others: RL = 2 kH,
73
83
96
ns
CL "15 pF
73
83
96
ns
Delay time, 4>n high to n TTL low
-14
-4
6
ns
tC>L,<1>TH
Delay time, CPn low to r/>n TTl. high
-29
-19
-9
ns
tcp3L,OH
Delay time, tP3 low to FFQ output high
-18
-8
2
ns
ttP3L,QL
Delay time, cp310w to FFQ output low
-19
-9
1
ns
ttPL,OSOH
Delay time, cp low to OSCOUT high
-30
-20
-10
ns
tcpH,OSOL
Delay time, FFQ high to OSCOUT low
-27
-17
-7
ns
OutPllt loads:
0.7'll
.
I
~-t¢4l'¢IH
~2
f1
~--4-~t¢-3'H-.-O-4H--~~.----~:--------------~
a4V
I
I
I
I
~
tq,H, ¢TL ---I
I--
I
I
I
:~----------------------I
1
~
u
t-tol,¢TH:
I
I
U
I
1.3V
i
I
I
Lf-1
I
I
I
\JJ
--------------------~Wll
i3
u
I
OSCOUT
I
:
____________________
1._3Jvf
...-
-.I
~~N~F~~U~T
of; ; .~.; .~.;.;p;. ;U_T:. . .
tq,H,bsOl-l1'OL.OSOLh
I
I
-+f tsu
to3l, QH
_____________________
th
L..
\~_.::.0.~7..:.V_______________..;...--~----....f--.......
I
lr------~I------------------------~\
13V
0.7 V
0.7 V
I
t¢3l,4~RI\
I
I
-1
0.7 V
I
333
v
t--
II
\L'_3_V____________
I
~/
\~__~__~/
4+--+t_----------------------------------.....!I:::;;:;;;;;, I
----~l1.3
v
13
SWITCHING CHARACTERISTICS, VOLTAGE WAVEFORMS
3-09
t
o 3l. QL
v \"'_________
TIM9904
EQUIVALENT OF D INPUT
EQUIVALENT OF OSCIN INPUT
Vee ----4_--
EQUIVALENT QF XTAL 1 AND
XTAL 2 INPUTS
vee-......._vee---....- -
20 kU NOM
INPUT
INPUT
' - - - -..... GNDt
GNDt
EQUIVALENT OF
TANK INPUTS
TYPICAL OF
1/)1, 1/)2,1/)3 AND 414 OUTPUTS
GND1
TYPICAL OF OSCOUT, Q, AND
ALL ~TTL OUTPUTS
- - - -.......--Vee
vee---....-
-1~---,--,--VDD
- -.......,..,.........-OUTPUT
I NPUT-.....-
l . . - - t - - OUTPUT
....
GNDt
SCHEMATICS OF INPUTS AND OUTPUTS
TMS9901
TMS 9901 ELECTRICAL SPECIFICATIONS
Absolute Maximum Ratings Over Operating Free Air Temperature Range (Unless Otherwise Noted)·
Supply voltage, Vee ................. · ........................................... -0.3Vto 10V
All inputs and output voltages ..................................................... -0.3 V to 10 V
Continuous power dissipation ........................................................... 0.85 W
Operating free-air temperature range ................................................. ooe to 700 e
Storage temperature range .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. -65°C to 15Q"C
"Stress. beyond those listed under '"Absolute Maximum Ratings'" may cause permanent damage to the device. This is a strMS rating only and functional operation of
the device at IMM or any other conditions beyond those indicated in the '"ReoomrnencIed Operating Conditions'" section of this specification is not implied. Exposure to
absolute maximum rated conditions for extended periods may affect device reliability.
Recommended Operating Conditions·
PARAMETER
Supply voltage. Vee
Supply voltage. Vss
High-level input voltage. VIH
Low-level input voltage. VIL
Operating free-air temperature. TA
MIN
NOM
4.75
5.0
0
2.0
VSS-·3
0
3-D10
MAX UNIT
V
5.25
V
V
Vee
V
0.8
70
°e
TMS9901
Electrical Characteristics Over Full Range of Recommended Operating Conditions
(Un.... 0therwI.. Noted)·
PARAMETER
VOH
High level output voltage
Low level output voltage
Input current (any input)
II
ICC(avl Average supply current from Vcc
VOL
CI
Small signal input capacitance, any input
TEST CONDmONS
MIN
TVP
MAX UNIT
IOH- -100pA
2.4
Vcc
V
IOH - -200pA
2.2
Vcc
0.4
V
±100
pA
150
rnA
15
pF
IOL= 3.2mA
Vss
VI=OVtoVCC
te(4)1 = 330 ns.
TA=70·C
f= 1 MHz
V
Timing Requirements Over Full Range of Operating Conditions
PARAMETER
MIN
300
fe(4))
Clock cycle time
tr<4»
tMi
Clock rise time
tw~
Clock pulse width (high level)
225
tw(4)>L)
Clock pulse width (~ level)
tw(CC)
CRUCLK pulse width
tsu1
Setup time for CE, 50-S4, or CRUOUT before CRUCLK
Setup time for interrupt before if> low
Setup time for inputs before valid CRUIN
Hold time for CE, 50-54, or CRUOUT after CRUCLK
45
100
100
tsu2
tsu3
ttl
TVP
333
5
10
Clock fall time
MAX UNIT
2000
40
40
ns
ns
ns
ns
300
185
ns
ns
ns
60
200
60
ns
ns
ns
·NOTE: All voltage values are referenced to Vss.
Switching Characteristics Over Full Range of Recommended Operating Conditions
PARAMETER
tpd1
tod2
tod3
tpd
MAX
UNIT
CL = 100pF
300
ns
Propagation delay, 50-S4 to valid CRUIN
CL - 100pF
320
ns
Propagation delay, if; low to valid DiITREO, ICO-IC3
CL = 100pF
110
ns
Propagation delay, ~ to valid data out (PO-P15)
CL = 100pF
300
ns
TEST CONDITION
3-011
MIN
TVP
Propagation delay, ~ to valid CRUIN
TMS9901
twltPU
-l
'-
tr(t/»
-I !.--II- tf(rp)
~,..----':i'-f~
- - - - - - 1...
' - - - Ic(C
cou,. {XX)(X)(X)(XXt&soDO
~-"'-e---B-0"R-C
11,",1
.£oj
''"'
TIMING DIAGRAM
3-017
Chapter 4
SINGLE CHIP NOVA MINICOMPUTER
CENTRAL PROCESSING UNITS
In this chapter we are going to look at two microprocessors which are the world's first single chip reproductions
of established 16-bit minicomputers. We are going to describe two products which reproduce, on a single chip,
the logic of a Nova Central Processina Unit.
Nova minicomputers are built by Data General Corporation.
Data General Corporation offer a set of LSI chips centered on the MicroNova microprocessor. These chips are
described quite superficially in this chapter since Data General is not actively marketing them as LSI devices.
Rather, Data General favor the sale of MicroNova microcomputer systems.
Fairchild manufacture the 9440 microprocessor, which is sold primarily as an LSI device. The 9440 is therefore
described in some detail, together with standard Nova 1/0 bus and typical memory bus interface bus logic.
The Nova minicomputer was designed as a next generation enhancement of the PDP-8. The IM6100, which is described in Chapter 13 of the Osborne 4 & 8-Bit Microprocessor Handbook is a single chip implementation of the PDP-B
Central Processing Unit.
If you compare the Nova architectures, which we describe in this chapter, with the IM1600, the two products will
indeed look very different. But conceptually they are similar. Both the Nova and the PDP-8 Central Processing Units
have few addressable registers; for computing power they rely upon instructions which may perform complex
sequences of operations. Similarities between the Nova and the PDP-B will become more apparent if you compare
these two devices with the CP1600 and the TMS990 - which we have described in Chapters 2 and 3, respectively.
What is interesting about the Nova minicomputer is that it is one of the most popular in the world; and Data General
Corporation is the second largest minicomputer manufacturer in the world. despite the fact that many aspects of the
Nova Central Processing Unit may. on first inspection. appear to be very restricting.
.
The MicroNova is manufactured by:
OAT A GENERAL CORPORATION
Mail Stop 6-58
Southborough. MA 01772
The 9440 is manufactured by:
FAIRCHILD SEMICONDUCTOR
464 Ellis Street
Mountain View, CA 94040
The MicroNova and the 9440 are not the same; differences, however, are small.
The MicroNova is equivalent to the Nova 3 minicomputer. The Nova 3 is a low-end minicomputer recently introduced by Data General. Although it is a low-end product it includes a number of features not found in the basic
Nova arch itectu reo
The 9440 reproduces basic Nova architecture - that is, the lowest common denominator of architectural features
found in any Nova Central Processing Unit. As such, the 9440 lacks a number of logic features provided by the
MicroNova. The 9440, however. has higher instruction execution speeds.
Because the MicroNova and the 9440 are very similar. we are going to describe them together in this chapter.
The MicroNova is manufactured using NMOS LSI technology. The 9440 is manufactured using Isoplanar integrated injection logic (l3U technology.
Both products are packaged as 40-pin DIPs.
The MicroNova requires four power supplies: -4.25V, +5V, +10V and +14V. The 9440 requires two power supplies: +5V and +350 mA.
4-1
Using a 240 nanosecond clock. the MicroNova executes instructions in 2.4 to 10 microseconds. Using a 100 nanosecond clock. 9440 instructions will execute in 1 to 2.5 microseconds.
A PRODUCT OVERVIEW
Figure 4-1 illustrates that part of our general microcomputer system logic which has been implemented by the
MicroNova and the 9440.
Note that only the MicroNova has a Stack Pointer, and DMA logic.
Most Nova minicomputers do not have a Stack; the 9440 is a reproduction of.the basic Nova architecture. which is why
the 9440 lacks a Stack.
The MicroNova and Nova 3 do contain Stacks. because the addition of the Stack is technologically straightforward.
while the lack of a Stack had been one of the most distressing features of earlier Nova minicomputers.
Both the 9440 and the MicroNova have DMA request and DMA acknowledge signals; however. in response to a DMA
request. the 9440 does nothing except float the System Bus. It is up to you to provide any and all external logic needed
to actually perform a data transfer via direct memory access. The MicroNova. on the other hand. executes the required
sequence of I/O operations to actually perform the DMA transfer. That is why in Figure 4-1 DMA logic is shown as
being present on the MicroNova but not the 9440.
What about I/O ports? I/O ports interface logic is shown as absent in Figure 4-1 . The 1/0 port is a microcomputer concept.
In any microcomputer configuration. you will look upon I/O ports as the ultimate interface between the microcomputer
system and external logic. You need a conduit via which data bits or signals can be transferred to. or received from
logic beyond the microcomputer system. Each conduit becomes an I/O port and an I/O port becomes a set of pins.
which can be addressed as a unit on a support device. Minicomputers take a conceptually different approach to 110
operations. To begin with. data is generally transferred to or from the CPU - not signals. The data finishes up on a
System Bus. Therefore a minicomputer's interface with the outside world consists of an I/O System Bus and a memory
System Bus. In some cases the two busses are one; in other cases. such as the Nova minicomputers. these two are separate and distinct busses. Conceptually. what is important is the fact that the minicomputer anticipates transferring
data via its I/O System Bus to line printers. disk units. or other substantial devices each of which is capable of having a
Significant amount of local logic. Thus the System Bus is as far as the minicomputer attempts to go when defining its
interface to the outside world.
Figure 4-1 , including bus interface logic within the logic of the Central Processing U nit, needs some clarification. As we have just stated. the Nova minicomputer creates two separate System Busses: one for memory. the other
for I/O devices. All the signals of these two busses originate at card edge pins. There is nothing very expensive about
adding more pins to the edge of a card. as there is to adding more pins to a DIP. Therefore the Nova System Bus has 47
Signals. Since neither the MicroNova nor the 9440 can have 47 signals. neither of these two devices creates standard
Nova System Busses; but each device creates its own System Bus which could be used to drive external logic. That is
why interface logic is shown as being present in Figure 4-1.
4-2
Clock Logic
MicroNova anc:f~
r:':'::::::::::::::::::::)
.
::::::::.:.:.:.:.:.:.:.:::: MIt;:roNova only
:::~::::::::::::::::*.:.
'
Logic to Handle
Interrupt Requests
from Extemal Devices
~
w
Figure 4-1. Logic of the Data General MicroNova and the Fairchild 9440
There is one further major difference between the MicroNova and the 9440 which is not evident from Figure
4-1. The MicroNova provides transparent dynamic memory refresh logic. The 9440 has no dynamic memory
refresh logic.
The MicroNova, but not the 9440, contains an elementary interval timer capability. Providing interrupt timer logic
is enabled, the MicroNova will generate an interrupt request every 20,000 instruction cycles. Using a standard 8.333
MHz clock, this translates to an interrupt request occurring every 2.4 msec.
Note that the MicroNova and the Nova 3 interval timer logic differ. The Nova 3 provides four programmable interval
timer options; the MicroNova provides just one.
NOVA PROGRAMMABLE REGISTERS
These are the programmable registers of the MicroNova and the 9440:
o
lS
ACO } Primary Accumulator
AC1
AC2 Accumulator and Index register
AC3 Accumulator, Index register and
I Stack Pointer
J Frame Pointer
14
Subroutine Return Address register
}
MicroNova Only
o
Data General literature numbers registers and memory words from left to right. rather than as illustrated above, from
right to left. Also Data General is one of the few minicomputer manufacturers that uses octal numbering. In order to remain consistent with the rest of this book, we will use hexadecimal numbers, and we will number registers from right to
left; where confusions may arise, we will show both our standard numbers and Data General equivalents.
ACO and AC1 are typical primary Accumulators. AC2 and AC3 may be used as Accumulators or as Index
registers. The Jump-to-Subroutine instruction automatically stores the return address in AC3. If one subroutine
is going to call another (j.e., you are nesting subroutines). then the calling subroutine must save the contents of AC3
before itself calling another subroutine.
Only the MicroNova has a Stack Pointer. The only instructions that access the Stack Pointer are "Push" and "Pop"
instructions.
The MicroNova, but not the 9440, also contains a Frame Pointer register. The Frame Pointer register is an address
buffer used to access the Stack. This may be illustrated as follows:
MEMORY
Stack Pointer identifies
current top of Stack
Use Fram e Pointer
to hold important
Stack ad dresses
The Frame Pointer is a buffer register; it is not a Data Counter. There are no instructions that access the memory location addressed by the Frame Pointer.
Observe that we show no programmable registers identified as Data Counters, even though in Figure 4-1 we show
Data Counter logic as being present. This is because the Data Counter is another microcomputer concept - in effect. a
subset of the Index register. If a memory reference instruction specifies direct. indexed addressing with a zero displacement. then Index Registers AC2 and AC3 are equivalent to Data Counters.
4-4
NOVA MEMORY ADDRESSING MODES
Both the MicroNova and the 9440 offer the following standard Nova memory addressing modes:
1)
2)
3)
4)
5)
6)
7)
Base page, direct addressing
Program relative, paged, direct addressing
Indirect addressing
Indirect addressing with auto-increment
Indirect addressing with auto-decrement
Direct, indexed addressing
Pre-indexed, indirect addressing
These addressing modes have been described in Volume 1. Chapter 6.
Nova memory addressing modes are heavily influenced by the fact that every Nova instruction generates a single 16bit object code - just as the predecessor PDP-8 instructions each generated a single 12-bit object code. Even memory
reference instructions are confined to 16 bits of object code; therefore the memory reference instruction can only provide a short address displacement. Whereas PDP-8 memory reference instructions provide a 7-bit address displacement. the Nova provides an.8-bit address displacement. which is handled in a much more intelligent fashion.
Nova instructions that use simple, direct addressing treat the 8-bit displacements as a direct, page zero address, or as a signed binary, program relative displacement. Thus you can directly address the first 256 words of
memory. or you can address any location within +127 to -128 words of the memory reference instruction itself:
.
~
MEMORY
yy can directly address
base page
i
I
yy can be added, as a
signed binary number,
to xxxx, to address
program relative page
~"
I
OOFF
0100
I
(xxxx) + FFSO (FFSO = -SO)
(xxxx) + FF81 (FF81 = -7F)
(xxxx) + FF82 (FF82 = -7E)
Address displacement
equals vy
I
I
0001
0002
0003
1000E
§§
Memory reference instruction
0000
,, ,-"-~
IVY
)O(x-1
xxXx
xxxx+ 1
(xxxx) + 70
(xxxx) + 7E
(xxxx) + 7F
Remember. in microcomputer applications. program relative direct addressing is fine for Jump instructions. but is of
limited value when accessing data memory. When a microcomputer program is stored in read-only memory. program
relative. direct addressing can be used to read constant data only.
Nova instructions that specify direct, indexed addressing, compute the effective memory address as the contents of either AC2 or AC3, plus the 8-bit displacement provided by the instruction object code. The 8-bit dis-
4-5
placement is treated as a signed binary number. Since the Index registers are 16 bits ·wide, direct indexed addressing
allows you to address any memory word. This may be illustrated as follows:
Accumulator AC2 or AC3
15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
~Bit No.
Ixlxlxlxlxlxlx~xlxlxlxlxlxlxlxlxl
Instruction Code
9 8 7 6
e
x x x x x x x x x x x x x x x x
z z z z z z z z y y y y y y y y
Sum is the effective memory.address
' - - - - - - - -..... 0 selects AC2
1 selects AC3
Indirect addressing may be superimposed on any of the memory addressing options described thus far. Indirect addressing is identified by a "1" in bit 10 of the Memory Reference instruction's object code. When indirect addressing is specified, the effective memory address is the
contents of the directly addressed memory word.
NOVA
DIRECT
MEMORY
ADDRESSING
Let us examine the various indirect addressing options. First there is page zero indirect addressing:
NOVA
INDIRECT
PAGE ZERO
ADDRESSING
8
5
2
O~BitNo.
~~~~~~~~~~~~~~
~----~----------P~Z«o
'--------+-------- Indirect addressing
Arbitrary
Memory
I~I:0001
0002
0003
i
I
0024
1------1 0025
1236
0026 .....~----.....
0027
t - - - - - t 0030
1235
1 - - - - - 1 1236
1 - - - -.......
1237
1240
In the illustration above. arbitrary, real memory addresses have been selected to make the illustration easier to understand.
4-6
Program relative, indirect addressing may be illustrated as follows:
NOVA
INDIRECT
PROGRAM
RELATIVE
ADDRESSING
1216
~----------__~____
~·55
This instruction
126B
fetched from this
~------------- Program relative
memory location
----------------Indirect
Arbitrary
Memory
MEMORY
Address
1215
1216
1217
1220
1221
1222
126A
736~
126B
126C
1260
126E
12GF
1270
7362
7363
7364
7365
7366
7367
7370
7371
7372
4-7
NOVA
INDIRECT
INDEXED
ADDRESSING
Indirect. indexed addressing may be illustrated as follows:
Accumulator AC2
15 1<4 13 12 11 10 9 8 7 6 5 <4 3 2 1 0
10101010101011101110101011111111'
~
Bit No.
Instruction Code
6 5 <4 3 2
028F
- - - - - -.....-0020
02BC
.....- - - - - - - - - - - - Index via AC2
.....- - - - - - - - - - - - - - I n d i r e c t
MEMORY
Arbitrary
Memory
Address
02~
736"
0288
02BC
0280
02BE
02BF
7362
7363
7364
7365
7366
7370
The illustration above arbitrarily uses indexed addressing via Accumulator AC2. Also the computed effective memory
address is identical to that which was obtained in the indirect. program relative addressing illustration.
Observe that Nova indirect addressing logic results in pre-indexed indirect addressing. As described in Volume 1.
Chapter 6. this is less desirable than post-indexed indirect addressing.
4-8
If. and only if indirect addressing has been specified by a "1" in bit 10 of a Memory Reference instruction's object code. then the contents of the data fetched from memory are treated as a direct
address. providing the high-order bit of the direct address is O. If the high-order bit of the address
is 1. then the address is treated as another indirect address pointer. This may be illustrated as
follows:
Interpret
NOVA
MULTIPLE
INDIRECT
ADDRESSING
as. last memory adciIr.a
Note carefully that multilevel indirect addressing will occur only when indirect addressing is specified in the first place.
If you execute a direct memory reference instruction. data will never be interpreted as an address.
The Nova indirect addr~ssing logic means that. given a 16-bit indirect address. only 15 bits actually address memory;
therefore you are limited to a 32.768 word memory address space:
l' 14 13 12 11 10 9 8 7 6 .5 4 3 i
I I II
I I I I I I I
1 0 ~.No.
g--1ncInIc:t
MIiNIry
~
1
The Nova minicomputers and microcomputers also provide indirect addressing with auto-increment and autodecrement addressing. If you indirectly address one of the eight memory locations. 001016 through 001716. then the
contents of the addressed memory location are incremented at the beginning of the memory access. Thus you have indirect addressing with auto-increment.
If you indirectly address anyone of the locations. 001816 through 001 F16 then the contents of the addressed memory
location will be decremented at the beginning of the memory access. Thus you have indirect addressing with autodecrement.
Neither the MicroNova nor the 9440 provide memory mapping logic. Memory mapping is a technique whereby
more than 32.768 words of addressable memory may be accessed. The Nova 3 minicomputer is capable of supporting
memory mapping as an option.
Nova minicomputers have separate memory and I/O device spaces. I/O instructions include six
bits which identify one of 64 I/O devices. Because Nova minicomputers and microcomputers treat
I/O devices in a manner that differs significantly from the typical microcomputer. we will defer our
discussion of I/O addressing until we have looked at pins. Signals and System Busses.
4-9
NOVA I/O
DEVICE
ADDRESSING
NOVA STATUS FLAGS
Nova minicomputers contain just one status flag, as we would define it, and that is the Carry status. Instructions are able to test for a zero or nonzero condition occurring at the conclusion of an instruction's execution,
but no permanent zero status flag exists.
MicroNova also has these interrupt related status flags:
•
•
•
•
Interrupt Enable
Real Time Clock Enable
Real Time Clock Request
Stack Overflow Request
}
MicroNova Only
The interrupt related status flags do not occur as addressable locations in any Status register; rather they represent flipflops which are set or reset during the course of interrupt handling.
The interrupt enable bit is a master enable which is set to 1 in order to enable all interrupts. Specific instructions allow
all interrupts to be enabled or disabled.
The MicroNova has a Real Time Clock interrupt enable bit and a Real Time Clock request bit. The Real Time Clock enable bit must be set to 1 in order to enable Real Time Clock interrupts; as soon as a Real Time Clock interrupt occurs. the
Real Time Clock enable bit and the Real Time Clock request bit are reset to O.
The Stack Overflow request bit is only present in the MicroNova. since only the MicroNova has a Stack. A Stack overflow condition occurs if. following a push operation. the incremented contents of the Stack register have zeros in the
eight low-order bits. What this implies is that the Stack must reside within a 256-word memory page:
Arbitrary
Memory
I~RYI:~
0801
:
§· .
·
0802
0803
.
:08FD
08FE
08FF~
'0900
0901
•
•
=
.
~
09FD
Pushes that increment Stack Pointer
from XXFF to XYOO will cause a Stack
Overflow interrupt
09FE
/
09FF~
OAOO~
OA01
OA02
When a Stack overflow occurs. the Stack Overflow request bit is set to 1 and an interrupt is requested.
MICRONOVA AND 9440 CPU PINS AND SIGNALS
As we stated earlier in this chapter, minicomputer Central Processing Units are implemented on cards, not
DIPs; therefore they usually have System Busses containing more than 40 signals. The standard Nova System
1/0 Bus contains 47 signals; furthermore, the Nova System Bus is, in effect, two busses: one communicating
with memory, while a separate and distinct bus communicates with 1/0 devices:
NOVA
MEMORY BUS
I/OBUS
CPU
I/O DEVICES AND
EXTERNAL LOGIC
MEMORY
4-10
Table 4-1 briefly defines the functions of bus signals. The I/O Bus is standard for ali Nova line computers, while the
Memory Bus is different for each model. We give the Memory Bus signals of the Nova 2 in Table 4-1.
Table 4-1. Nova System Bus Signals
STANDARD NOVA SYSTEM I/O BUS
SIGNAL
DIRECTION
DsO-DS5
DATAo - i5Ai'Ai5
To Device
Bidirectional
To Device
To Device
To Device
DATOA
DAnA
DATOB
DATIB
DATOC
DATIC
STRT
CLR
10PLS
Sllii
SELi5
RciENB
iNTR
iNi'P
INTA
MSKO
50iR
OCHP
OCHA
DcHMo,DcHMi
To
To
To
To
Device
Device
Device
Device
To Device
To Device
To Processor
To Processor
To Device
To Processor
To Device
To Device
To Device
To Processor
To Device
To Device
To Processor
FUNCTION OR INDICATION
Device selection
Data and address lines
Data out to device's A buffer
Data in from device's A buffer
Data out to device's B buffer
Data in from device's B buffer
Data out to device's C buffer
Data in from device's t: buffer
Start device-clear Done flag, set Busy flag and clear devi.ce's INT REO
flip-flop
Oear device's Busy and Done flags and INT REO flip-flop
I/O Pulse - user-defined function
Selected device's Busy flag is set
Selected device's Done flag is set
Enable interrupt or DMA requests
Interrupt request
Interrupt priority
Interrupt acknowledge
Interrupt mask out
Data channel request (DMA request)
Data channel priority
Data channel acknowledge
Data channel mode:
l5CHMo
H
H
L
L
DCHI
OCHO
OVFLO
10RST
To
To
To
To
Device
Device
Device
Device
DCHMI
H
L
H
L
Data out
Increment memory
Data in
Add to memory
Data channel in
Data channel out
Overflow: result of memory increment or add exceeds FFFF II
Clear all I/O devices
THE NOVA 2 MEMORY BUS
SIGNAL
DIRECTION
AO-A14
DATAO - DATA15
INHIBIT SELECT
BMEMEN
WRITE
BRMW
WE
SYNC ENABLE
RELOAD DISABLE
WAIT
MEM CLOCK
EXTERNAL SELECT
EXTERNAL MBLD
To Memory
Bidirectional
To Memory
To Memory
To Memory
To Memory
To Memory
To Processor
To Memory
To CPU
To Memory
To Memory
To Memory
FUNCTION OR INDICATION
Memory address lines
Memory data lines
Inhibits selection of memory module
Starts memory cycle
Memory write
Causes pause between read and write
Enable write after pause in read-pause-write cycle
CPU hold control
Inhibits loading of memory buffer
Disables other memory modules during write portion of memory cycle
Memory Clock
Allows module to be selected despite contents of address lines
Allows data to be stored in memory buffer without starting a memory
cycle
If you are using the MicroNova or 9440 in a new product. then there is no reason why you should create the standard
Nova System Busses. Providing the signals generated by the MicroNova or the 9440 are adequate for your needs, you
can interface external logic directly to these two devices.
Let us first look at the MicroNova pins and signals, which are illustrated in Figure 4-2.
Two clock signals, <1>1 and <1>2, must be input to synchronize all MicroNova logic.
4-11
The Memory Bus consists of a 16-bit Address/Data Bus, plus three control signals: SAE, P
and WE.
MICRONOVA
MEMORY BUS
The Address/Data Bus connects to pins MBO - MB 15. P is a synchronization signal. SAE is a read
enable and WE is a write enable.
The I/O Bus consists of just four signals:
MICRONOVA
I/O BUS
I/O CLOCK synchronizes I/O transfers.
I/O DATA1 and I/O DATA2 are bidirectional data and control signals.
I/O INPUT identifies the direction of data transfers occurring via-:"I/~0~D-:"A~T~A~1 and I/O DATA2.
As compared to other microcomputers described in this book. the MicroNova I/O interface is very unusual. Only the
TMS 9900 I/O logic is at all similar. A 16-bit I/O data transfer occurs as two 8-bit serial units. This may be illustrated as
follows:
I/O CLOCK
~~~__~I_B_IT_O~I~~_T_1~I~B_IT__2~I_B_rr_3~I_B_rr_4~I_BI_T_5~I_B_rr_6~1_~_T_7JI
\
~ 0 I ~ 1 I ~T 2 I ~T 3 1 ~ 41 ~ 5 1 ~ 6 1~T 7 I
Eight serial bits are input in less than one microsecond: therefore this method of handling I/O is as fast as the parallel
data input operations described for other microcomputers.
Each data transfer is preceded by one of four_codes generated by levels output via I/O DATA 1 and I/O DATA2. These
are the four codes:
1/0 DATA1
I/O DATA2
1
o
o
o
o
1
INTERPRETATION
Accompanying I/O low pu Ise may be used to synchronize interrupt requests
and DMA requests.
DMA request acknowledge.
I/O data transfer. The transfer direction is specified by I/O INPUT.
I/O command out.
Thus every I/O operation will begin with I/O DATA 1 and I/O DATA2 being output during a low I/O CLOCK pulse. I/O INPUT will be low at this time since data is being output via I/O DATA1 and 110 DATA2. Providing 110 DATA1 andl70
DATA2 specify a data transfer to follow. the actual data transfer will occur via I/O DATA 1 and I/O DATA2 with I/O INPUT identifying the data transfer direction.
4-12
VBB
P
40
39
nc
WE
38
37
VDD
HALT
36
35
CLAMP
SAE
4
DCH INT
5
'OOiNT
6
7
VGG
VSS(GRO~
MOO
Mii1
Mii2
MB3
8
9
10
11
12
MICRONOVA
Vss (GROUND)
nc
34
nc
33
32
31
PAUSE
11>1
30
29
I/O DATA1
I/O DATA2
i70iNPUT
MB4
Ma5
MB6
MB7
13
14
28
27
15
16
26
25
VCC
MB8
17
18
MB9
MiiO
19
24
23
22
20
21
11>2
I/O CLOCK
Vss (GROUND)
MiT5
Mei4
Ma1'3
Me12
Mi1i'
PIN NAME
DESCRIPTION
TYPE
11>1. cJI2
Clock Signals
Input
MeO-'MBi5
Address/Data Bus
Memory Synchronization
Memory Read Enable
Output
Output
Memory Write Enable
Output
I/O Synchronization
Data and Control
Transfer Direction
P
SAE
WE
I/O CLOCK
I/O DATA1. I/O DATA2
Bidirectional
HALT
Power-On Reset
CPU Halted
Bidirectional
Bidirectional
Output
Input
Output
i5CHiNT
00iNf
DMA Request
External Interrupt Request
Input
Input
PAUsE
Memory Bus Grant
Output
VBB VDD. VGG. Vss
Power and Ground
i70iNfiijT
Ci:AMP
Figure 4-2. MicroNova CPU Signals and Pin Assignments
There are two CPU control signals which are not part of either the Memory Bus or the I/O Bus.
Following power-up, the MicroNova CPU will not perform any operation until a high input occurs at CLAMP.
When CLAMP goes high. interrupts are enabled. Real Time Clock and Stack Overflow interrupt requests are cleared.
and the CPU is halted. Once CLAMP has been input high. it is ignored until the MicroNova is powered down and then
powered up again.
The HALT signal is output by the MicroNova as a high pulse while the MicroNova CPU has been halted response to execution of a Halt instruction. or following CCAi\iiP going high.
either in
There are two MicroNova signals associated with interrupt logic. DMA requests are made via DCH INT while
any external interrupt is requested via EXT INT. Both the DMA request and the interrupt request must be. synchronized with instruction execution timing. This synchronization is provided by I/O DATA1 and I/O DATA2. as we
have already described. The DMA acknowledge occurs via 1/0 DATA1 and I/O DATA2. There is no external interrupt
acknowledge signal; however. such a signal can be derived from the Memory Bus. as we will describe later in this
chapter.
PAUSE is output low by the CPU when devices other than the CPU· are permitted to access memory.
Now look at 9440 pins and signals, which are illustrated in Figure 4-3.
These pins and signals create a single System Bus. No attempt is made to create separate Memory and I/O
Busses.
You may connect a crystal across CP and XTL in order to create a master clock signal, or you may input a clock
signal via CPO
4-13
C3
C2
C1
CO
DCH REO
00
40
39
M1
38
37
36
CLK OUT
6
35
34
01
8
INTON
GND
33
32
31
RUN
9
10
11
IINJ
CARRY
12
13
(high-order bit) 180
14
iB1
15
16
i'Nf'REa.
iB2
IB3
184
iSs
iB6
MO
1
2
3
4
9440
17
18
30
M2
CP
XTL
MA
SYN
MBUSY
VCC
GND
29
28
27
iBf5 (low-order bit)
26
25
24
IB12
IB11
19
23
22
20
21
iii14
iii13
IB10
iB9
iBii
iB7
PIN NAME
DESCRIPTION
TYPE
xn.cP
Cock Signals
Synchronization Signal
Output
SYN
iBO -iBiS
MO-Mi
System Cock
Data/ Address Bus
Memory Controls
MBUSY
Memory Busy
00.01
I/O Control
Interrupt Request
ClK OUT
iNiiiEQ
Input
Output
Bidirectional
Output! Input
Input
Output
OCHREQ
Interrupt Enable
DMA Request
Input
Output
Input
RUN
CARRY
CO-C3
CPU Running
Carry Status
Front Panel/Console Control Signals
Output
Input
MR
Master Reset
Input
IINJ. vcc. GND
Power and Ground
INTON
Output
Figure 4-3. 9440 CPU Signals and Pin Assignments
The 9440 generates a single synchronizing output (SYN). The CPU clock is output to the
system via ClK OUT.
IBO - IB15 provides the 9440 with a multiplexed 16-bit Data and Address Bus. This bus carries addresses to memory and I/O devices. and it carries bidirectional data between the CPU and
memory or I/O devices. IBO - IB 15 are low true; a low signal level represents a 1 bit.
9440
SYSTEM
BUS
IBO is the high-order bus line while IB15 is the low-order bus line. This agrees with Nova conventions. This
chapter, and this whole book describe the low-order bit as bit 0 - exactly the reverse of IBO - IB15.
There are three control signals on the 9440 CPU-memory interface.
MO is output low to identify a memory read.
M1 is output low to identify a memory write.
M2 is output low to identify a memory address being output.
MO - M2 have open-collector outputs; you can use these lines as inputs to make the timing of a non-memory machine
cycle conform to the timing of a memory cycle. We will discuss this further when we discuss 9440 timing and instruction execution.
External memory interface logic inputs MBUSY low while it is responding to any memory access. MBUSY is similar to
the WAIT signals that we have described for other microcomputers; it can be used to make the CPU wait for slow
memory to respond to a CPU access request.
4-14
The 9440 has two I/O control signals 00 and 01. These two control signals define I/O and memory accesses as follows:
o
o
01
01
01
01
00
00
1 00
1 00
o
Instruction Fetch
1 Data Channel Access
o Execute I/O Operation
1 No I/O
There are two signals associated with 9440 interrupt logic.
An external interrupt is requested by inputting INT REO low.
INT ON indicates whether or not interrupts are enabled. This signal is high when interrupts are enabled: if this signal is low. interrupts are disabled.
A DMA request is made by inputting DCH REQ low. The DMA request is acknowledged by 01 and 00 being output
low and high. respectively.
There are seven signals provided by the 9440 specifically to support a front panel or console.
Two of the front panel or console signals are outputs; these are the RUN and CARRY signals.
RUN is output high while the CPU is executing programs: it is output low while the CPU is halted. RUN is used to generate an appropriate front-panel display light: it is also equivalent to a Halt acknowledge. as described in this book for
many other microcomputers.
CARRY represents the condition of the Carry status. This signal is output specifically to drive a front-panel light.
Five input control signals are provided for switches on a front-panel. Four of these signals are CO. C1. C2 and
C3; they perform the following operations:
C3
C2
0
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
0
0
1
1
1
1
0
0
0
0
1
1
1
1
C1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
CO
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
FUNCTION
Display ACO contents at console
Display AC 1 contents at console
Display AC2 contents at console
Display AC3 contents at console
Increment Program Counter and then display contents of addressed memory word
Display contents of addressed memory word
Load memory from console switches
Halt
Deposit switches into ACO
Deposit switches into AC1
Deposit switches into AC2
Deposit switches into AC3
Load Program Counter from console switches
Continue/Run
Increment Program Counter and then load memory from console switches
No Operation
The first 9440 devices decoded the C lines in a slightly different manner. The following combinations were
different operations:
C3
o
o
1
1
C2
C1
0
1
0
1
CO FUNCTION
o
o
Load Program Counter from console switches
Not used
1 Load memory from console switches
o Continue/Run
MR is the Reset input to the 9440. When this line is pulled
low: the 9440 halts immediately and clears the Interrupt
Enable flip-flop. Once MR goes high. the CPU will remain in the Halt state until it receives the "Run" command from
lines C3 - CO. Reset has no further effect on the 9440. It is up to your hardware to load the Program Counter by
manipulating lines C3- CO and the Information Bus.
4-15
The following sequence is sufficient to start operation of a 9440 system with a "bootstrap"
program in non-volatile memory:
9440
INITIALIZATION
SYN
C3 - CO
iBO
-1815
NO OP
I
LOAD PC
I
(
STARTING
ADDRESS
)
NOOP
X
I
RUN
NO OP
(
I
RUN
The hardware must provide the program starting address while issuing the "Load Program Counter" command via the
Clines. C line codes other than "No Operation" are held for two machine cycles to ensure that the CPU reads them. The
"No Operation" code between "Load PC" and "Run" gives the CPU time to finish executing the C line command. See
the data sheets at the end of this chapter for more detailed timing information.
4-16
r-Da-ta-D-u-tP-ut-T-o-~-s-tin-a-tio-n---------------------------------------------------------------------, ~
Inhibit data output option
Second source
and destination
Jo. _ - - - -.....
ACO
Arithmetic
~
ACI
First source
....
~1~~A~C~2~~~---~~-J~
I
AC3
1
and
Boolean
Logic
Data output
•
.
I
I
Shifter
and
Byte Swap
Logic
r
Test
Data
for
Skip
.-...
,.....__
I
Program Counter
Test
Carry
I
:~
~
h~lc-~~--_J
1514 13 12 11 10
8 7 6 5 4 3 2 1 0 ___
Bit No.
'II I I I I I I I I I I I I I I I
It._________________________________ Arithmetic And Logic Instruction Code
Figure 4-4. The Nova Arithmetic and Logic Unit
1514 13 12 1110 9 8 7 6 5 " 3 2 1 0
IllS SID DITITITIH Hie elLlKIKIK
~Bit No.
Arithmetic/Logic instruction
LoooNo,"'
001 Always skip
010 Skip if Carry is Zero
011 Skip if Carry is One
100 Skip on Zero result
101 Skip on nonzero result
110 Skip on either Carry or result zero
111 Skip on Carry and result both nonzero
oStore result in destination Accumulator
1 Discard result
' - - - - - - - - - - 00 Preserve current Carry status
01 Zero Carry
10 Set Carry to 1
11 Complement current Carry
Thase operations are performed on Carry before
entering the ALU
' - - - - - - - - - - 00 No operation
01 Left rotate one bit position
10 Right rotate one bit position
11 Swap bytes
These operations are performed pn the ALU output
' - - - - - - - - - - - - - 000 Complement
001 Twos Complement (Negate)
010 Move
011 Increment
100 Add Complement
101 Subtract
110 Add
111 AND
OOACO
Destination ACCUmUlator} 01 AC1
' - - - - - - - - - - - - - - - - - - - - - - - - - Source Accumulator
10 AC2
11 AC3
Figure 4-5. Arithmetic/Logic Instruction Object Code Interpretation
4-17
~
r--:=:--l-..J·~D~.t~a~ou~tpu~t....~--.......L - -...
~T~~=l--------J
i
t
Carry
for
Skip
-ttll
Carry
Status
I
CPU LOGIC AND INSTRUCTION EXECUTION
The manner in which the Nova CPU executes instructions differs markedly from microcomputers described
earlier in this book. We will therefore begin our discussion of CPU operations by looking at overall CPU architecture.
Our discussion of Nova CPU logic is tied to instruction object code bit patterns; this happens to be the simplest way of
describing the Nova CPU. We will look at instructions from a programmer's perspective when we eX'amine the Nova instruction set.
Nova instructions may be divided into these three groups:
1)
2)
Arithmetic. Boolean and logical operations which are essentially internal to the CPU.
Memory reference instructions which offer a variety of memory addressing modes and very little else.
3)
I/O instructions which are designed to allow a considerable amount of intelligence in I/O devices.
Let us examine each group of instructions and associated CPU logic.
ARITHMETIC/LOGIC INSTRUCTIONS
The power of the Nova CPU lies in the fact that many logic functions are implemented sequentially along a
single data path through the CPU. This is illustrated in Figure 4-4. This figure shows how individual bits of
arithmetic and logic instruction object codes directly identify the many options available as data makes a single tour
through the CPU. Figure 4-5 provides specific arithmetic and logic instruction object code interpretations.
Data to be operated on is always fetched from the Accumulators. Results are always returned to an Accumulator. For
two-operand instructions. such as binary addition. the Destination Accumulator also serves as the second Source Accumulator. For one-operand instructions. such as a complement. there will be one Source Accumulator and one
Destination Accumulator; the same Accumulator may serve as source and destination.
As the source and destination definitions would imply. the Nova has no Secondary Memory Reference (or Memory
Operate) instructions as we define them; for example. you cannot directly add the contents of a memory word to the
contents of an Accumulator.
In addition to one or two 16-bit data words. the Carry status is input to the Arithmetic and Boolean logic unit. You may
input the Carry status as is. or you may complement it. reset it to 0 or set it to 1. If you modify the Carry status. then the
modified Carry status becomes the new input to the Arithmetic and Boolean logic.
You may specify one of eight Arithmetic and Logic operations. The Move operation serves both as a Move and a No
Operation. By specifying the same Accumulator as the source and destination for a Move. Arithmetic and Boolean logic
is bypassed. Notice that only one Boolean operation. the AND. is provided. This is an inconvenience rather than a problem. As discussed in Volume 1. Chapter 2. you can combine the AND and complement operations to generate an OR or
an Exclusive-OR. The following Nova instruction sequences substitute for the OR and Exclusive-OR:
;OR the contents of ACX with
COM
ACX.ACX
AND
ACX.ACY
ADC
ACX.ACY
;Exclusive-OR ACX with ACY.
;ACZ is needed for temporary
MOV
ACY,ACZ
ANDZL
ACX.ACZ
ADD
ACX.ACY
SUB
ACZ,ACY
ACY. Leave the result in ACY
Complement ACX
AND ACX with ACY. Result to ACY
Add original ACX. Result to ACY
Leave the result in ACY.
data storage
Save ACY in ACZ
Store twice ACX AND ACY in ACZ
Add ACX to ACY
Subtract twice ACX AND ACY
The 16-bit output from the Arithmetic and Boolean logic. together with the Carry status. passes to the Shifter and Byte
Swap logic; here the 17-bit data unit may be rotated left or right. high and low-order bytes of the 16-bit data unit may
be swapped. or this logic may be bypassed.
The Shifter and Byte Swap logic outputs 16 bits of data. plus the Carry status. The data and the Carry status may be
tested separately. and based on one of eight identifiable conditions. the Program Counter contents may be incremented; this provides conditional skip logic. Figure 4-5 defines the eight conditions that may cause a skip.
Finally you have the option of preventing results from being stored in the Destination register; this enables conditional
branch logic without modifying the contents of any Accumulator.
4-18
In summary, the five operations that can be specified by a single arithmetic/logic instruction may be illustrated
as follows:
[ ]
[[]]
A
A 3-bit skip-on-condition field which is used by the Register-Register Operate instructions.
Coded Character Result Bits Operation
option omitted
000
No operation
SKP
001
Always skip
SZC
010
Skip if Carry = 0
SNZ
011
Skip if Carry = 1
SZR
100
Skip if result = 0
SNR
101
Skip if result =1= 0
SEZ
110
Skip if either carry or resu It = 0
SBN
111
Skip if both carry and result =#= 0
Generates the address EA
@ is the indirect bit. If @=1 then indirection is specified.
DISP is an 8-bit address value.
(IX)
is a 2-bit field which indicates the addressing Mode:
Bits are Mode
00
Zero page addressing. DISP is an unsigned address
between 0 and 256.
EA = DISP
01
PC relative addressing. DISP is a signed two's
complement address displacement.
EA = DISP+ [ PC]
10
Indexed addressing via AC2. DISP is a signed
two's complement address displacement.
EA = DISP+ [ AC2]
11
Indexed addressing via AC3. DISP is a signed
two's complement address displacement.
EA = DISP+ [ AC3]
A 2-bit 1/0 test field whose meaning depends on whether the CPU or another device is referenced.
CPU
t
Device
Test for lr;te';TUpt On=1 00 'TeStTor Busy=1
Test for Interrupt On=O 01 Test for Busy=O
Never skip
10 Test for Done= 1
Always skip
11 Test for Done=O
Bits y through z of the quantity x. [AC] <5,0> is the low six bits of the specified Accumulator.
Contents of location enclosed within brackets. If a register designation is enclosed within the brackets,
then the designated register's contents are specified. If a memory address is enclosed within the
brackets, then the contents of the addressed memory location are specified.
Implied memory addressing; the contents of the memory location designated by the contents of a
register.
Logical AND
Data is transferred in the direction of the arrow.
Under the heading of STATUS in Table 4-2, an X indicates statuses which are modified in the course of the instruction's execution. If there is no X, it means that the status maintains the value it had before the instruction was executed.
4-37
Table 4-2. MicroNova and 9440 Instruction Set Summary
STATUS
TYPE
MNEMONIC
OPERAND(S)
OPERATION PERFORMED
9440
BYTES
C
NIOIfl
DEV
2
X
[DEVBD) - f
DIA(f)
AC,DEV
2
X
Set the device's Busy and Done flags according to I/O command.
[AC) - [DEVA)
[DEVBD) - f
DIB(f)
AC,DEV
2
X
Read device's A buffer into Accumulator, Set the device Busy arid Done fla9s,
[AC) - [DEVB]
[DEVBD] - f
DIC(f)
AC,DEV
2
X
DOA(f}
AC,DEV
2
X
DOB(f)
AC,DEV
2
X
DOC If)
AC,DEV
2
X
SKP(t)
DEV
2
X
~
w
g
CD
10RST
~
_L--
X
Read device's B buffer into Accumulator, Set the device Busy arid Done. flags,
[AC] - [DEVC]
[DEVBD) - f
Read device's C buffer into Accumulator, Set the device Busy and Done flags,
[DEVA) - [AC)
[DEVBD) - f
Write Accumulator into device's A buffer. Set the device Busy and Done flags,
[DEVB) - [AC)
[DEVBD) - f
Write Accumulator into device's B buffer. Set the device Busy and Done flags.
[DEVC) - [AC)
[DEVBD] - f
Write the Accumulator into device's C buff~r. Set the Busy and Done flags.
If T is true for DEV, [pc) - [pc) + 1
Skip if I/O test true,
[PM]-O
liON) -102
The Busy and Done flags in all I/O devices are set to
interrupts are turned on.
O.
The Priority Mask is set to
0 and
Table 4-2. MicroNova and 9440 Instruction Set Summary (Continued)
STATUS
tyPE
MNEMONIC
OPERAND(S)
9440
BYTES
OPERATION PERFORMED
I
C
w
»u
ex:ex:z
::!:::!:II:
LOA
AC,( ",) DISP (.IX)
2
X
[AC]- [EA]
STA
AC,( Ii) OISP (,IX)
2
X
Load contents of memory to Accumulator.
[EA]- [AC]
«Ow
f~~II:
Store contents of Accumulator into memory.
ADD(CS#)
SUB (CS#)
.J:o
W
CD
w
~
«
ex:
w
a.
NEG (CStJ)
S,D(SKCND)
S,OtSKCNO)
S,DtSKCND)
2
2
2
X
X
X
X
[0]- [0]+ [S]
X
Add contents of Source register to contents of Destination register. Perform the specified
options.
[0]-[0]- [5]
X
Subtract contents of Source register from contents of Destination register. Perform the
specified options.
[0] - [Si + 1 (twos complement)
X
Place twos complement of the Source register contents in the Destination register. Perform
the specified options.
[0]- [0]+ [5]
0
ex:
w
~
III
C;
ADC(CSti)
S,OI.SKCNO)
2
X
w
ex:
ci:
w
~
MOV(CS")
5,0 (.SKCND)
2
X
X
Add the ones complement of the Source tegister contents to contents of Destination register.
Perform the specified option.
[0]-[5]
INC (CS fI)
5,0 (.SKCND)
2
X
X
Move contents of Source register to Destination register. Perforj'Tl the specified options.
[0]-[5]+1
COM (CStt)
S,Oi.SKCNO)
2
X
X
Place incremented Source register contents into Destination register. Perform specified options.
[0]-[5]
X
Complement the Source register contents, then move to Destination register. Perform
specified options.
[0]- [0] A [5]
III
C;
w
II:
AND (CS fI)
S,OtSKCNO)
2
X
AND the Source register contents with the Destination register contents. Perform specified
options.
Table 4-2. MicroNova and 9440 Instruction Set Summary (Continued)
STATl,IS
TYPE
MNEMONIC
OPERANDIS)
BYTES
OPERATION PERFORMED
9440
C
a:
w
I-
MUL
2
DIV
2
[ACO] -(( [ACl]
[Acll-l( [ACl]
sw~
wI-;:)
a:
[AC2])+ [ACO}) < 15,0>
Multiply contents of AC 1 by contents of AC2 and add contents of ACO to result.
[ACll-( [ACOJ.[ACll)/[AC2] (quotient)
[ACO] - ( [ACOl,[ACl])/ [AC2] Iremainder)
Divide the 32-bit quantity contained in ACO (high order) and AC1 (low order) by the
contents of AC2.
[SP] - [SP] + 1; [[SP)) - [AC)
Push the Accumulator onto the Stack.
[AC] - [[SP]; [SP] - [SP] - 1
Pop the top of the Stack to the Accumulator.
[(SP] + 1]- [ACO]
[(SP] +2] - [ACll
[(SP]+3] - [AC2]
[(SP] +4] - [AC3]
([SP]+5] <14,0> - [PC]
[(SP] +5] <15> - [C)
[SP] - [Spj + 5
[FP] - [SP]
Save a return block in the Stack.
[SP] - [AC1<14,0>
Move the low 15 bits of the Accumulator to the Stack Pointer.
[FP] - [AC] <14,0>
Move the low 15 bits of the Accumulator to the Frame Pointer.
Table 4-2. MicroNova and 9440 Instruction Set Summary (Continued)
STATUS
TYPE
MNEMONIC
OPERAND(S)
9440
BYTES
OPERATION PERFORMED
C
Q
w
MFSP
AC
2
MFFP
AC
2
JMP
(u)DISP(,IX)
2
X
JSR
(u)DISP(,IX)
2
X
[AC] <14,0> - [SP]
[AC] <15>-0
MOve the Steck Pointer to low 15 bite of Accumulator.
[AC] <14,0> - [FP]
[AC] <15>-0
Move the Frame Pointer to the Accumulator.
~~
Uz
:!i=
cn Z
0
y
f"
~
Q,
[PC]-[EA]
Branch unconditIoneI.
[AC3]-[PC]+1
[PC] ...... [EA]
Branch to subroutine.
[SP] - [FP]
[C) - [[SP)) <15>
[PC] - [[SP)) <14,0>
[AC3] - [(SP] - 11
[AC2] - HSP] - 21
[ACl] - HSP] - 31
[AC2] - HSP] - 4]
[SP] - [SP] - 5
Return from subroutine end pop e retum block off the Stick.
RET
2
RTCEN(f)
2
X
[ION] - f
RTCDS(f)
2
X
Eneble Reel Time Clock then let ION vie I/O commend.
[ION] - f
DiAbIe R_I Time Clock then let ION vii I/O commend.
X
:E
~
"')
w
2~
i=g
,J,J
~u
a:
Table 4-2. MicroNova and 9440 Instruction Set Summary (Continued)
STATUS
TYPE
MNEMONIC
OPERAND(S)
9440
BYTES
OPERATION PERFORMED
C
c
~~z~
owO!-
ISZ
(II)DISPI.IX)
2
x
~!-ci.Q
wc(-z
~~~8
DSZ
(II)DISPI.IX)
2
X
1
N
INTEN
2
X
[lON]-1
INTDS
2
X
Enable interrupts. Same as NIOS CPU.
[lON]-O
2
X
INTA(f)
AC
!a.
:;)
I:£:
a:
w
!-
:!!:
[PC] + 1
Increment memory contents and skip if zero.
[EA]- [EA]- 1
If [EA] = 0 then [PC] - [PC] + 1
Decrement memory contents and skip if zero.
0
~
[EA] - [EA] + 1
If [EA] =0 then [PC] -
MSKO(f)
AC
X
Disable interrupts. Same as NIOC CPU.
[AC] <5,0> -DEV
[lON]-f
The 6-bit device code of the device closest to the CPU that is requesting an interrupt is loaded
into the low six bits of the Accumulator. Set ION via I/O command.
[PM]- [AC]
[ION]-f
Move contents of Accumulator to Priority Mask. Set ION via I/O command.
TRAP
SKPIt)
2
CPU
2
[26,.] - [PC]
X
[PC]- [27,.]
Performs a software interrupt.
If t is true. [PC]- [PC] + 1
If interrupt or power fail condition satisfied. skip next instruction.
I
HALT(fI
2
X
[lON]-f
Set ION via I/O command. then halt.
I
Table 4-3. MicroNova and 9440 Instruction Set Object Codes
INSTRUCTION
ADC(CS#)
ADDlCS#)
ANDlCS#)
COMICS#)
OlAf
DIBf
DICf
DIV
DOAf
DOBf
DOCf
DSZ
HALTf
INC(CS#)
INTAf
S,DI.SKCND)
S,DI.SKCND)
S,DI.SKCND)
S,DI.SKCND)
AC,DEV
AC,DEV
AC,DEV
AC,DEV
AC,DEV
AC,DEV
(f1:)DISPI,IX)
S,DI.SKCND)
AC
INTDS
INTEN
IORST
ISZ
JMP
JSR
LOA
MFFP
MFSP
MOV(CSIf)
MSKOf
MTFP
MTSP
MUL
NEG(CS11)
NIOf
POPA
PSHA
RET
RTCDSf
OBJECT CODE
1ssdd 1OOrrccnwww
1ssdd 110rrccnwww
1ssdd 111 rrccnwww
1ssddOOOrrccnwww
011 aaOO1 ffpppppp
011aaOllffpppppp
011aal0lffpppppp
7641
011aa01Offpppppp
011aa l00ffpppppp
011aallOffpppppp
000llixxbbbbbbbb
011aallOffllllll
lssddOllrrccnwww
011aaOllffllllll
60BF
607F
( f1i)DISPI,IX)
('f1;) DISP I,IX)
( f1i)DISPI,IX)
AC (· f1i),DISP (,IX)
AC
AC
S,DI.SKCND)
AC
AC
AC
S,DI.SKCND)
DEV
AC
AC
RTCENf
SAY
SKPt
SKPT
DEV
STA
SUB(CS#)
CPU
AC,( ,,) DISP (,IX)
TRAP
S,DI.SKCND)
011aa01Offllllll
000 10ixxbbbbbbbb
OOOOOixxbbbbbbbb
00001 ixxbbbbbbbb
011aaixxbbbbbbbb
011 aaOOO 100000o 1
011aaOl010000001
lssddOl0rrccnwww
011aal00ffll1111
011aaOOOOOOOOOOl
011aaOl000000001
76Cl
1ssddOO 1rrccnwww
01100000ffpppppp
011aaOl110000001
011aaOl100000001
6581
0110101Offl11111
011100lOffllllll
6501
01100111ttpppppp
01100111ttllllll
010aaixxbbbbbbbb
1ssdd 101 rrccnwww
1ssddqqqqqqq 1000
CLOCK
PERIODS
9440
2
2
2
2
2
2
5/7
5/7
5/7
X
X
X
5/7
15
15
2
15
123
10
X
X
X
X
BYTES
2
2
2
2
2
2
2
2
2
2
2
10
10
8/10"
10
5/7
15
10
10
10
2
2
2
8/10"
6/8·
2
2
2
6/8*
8
7
2
2
5/7
10
2
2
6
2
2
2
2
2
2
2
2
2
2
2
2
2
2
7/9*
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
6
86
5/7
10
7
7
15
10
10
16
15/17
15/17
6/8*
5/7
9
X
X
X
X
X
X
X
X
*Direct addressing. For indirect addressing, add two clock periods for each level of indirection. For autoincrement or auto- decrement locations, add three clock periods, plus two for each level of indirection.
The following symbols are used in Table 4-3:
aa
bbbbbbbb
cc
dd
ff
n
pppppp
rr
ss
Two bits selecting an Accumulator
8-bit signed two's complement address displacement
Two bits selecting the carry option
Two bits selecting the destination Accumulator
Two bits selecting the I/O command
One bit selecting indirect addressing
One bit choosing the no load option
Six-bit device number
Two bits determining the shift option
Two bits choosing the source Accumulator
4-43
tt
www
xx
Two bits choosing the I/O test
Three bits selecting the skip-an-condition option
Two bits selecting the index option
Execution times shown are for MicroNova. Where two execution times are shown (for example. 5/7). the second is the
instruction time if the skip or branch is taken. See Table 4-4 for 9440 execution times.
Table 4-4 shows the sequences of machine cycles by which the 9440 executes instructions, interrupt and
data channel requests, and commands received via lines C3 - CO.
4-44
Table 4-4.9440 Instruction Execution
NO.
INSTRUCTION
OR
OPERATION
CYCl E TYPE AND SEQUENCE··
EXECUTION
TIME"
FETCH READ WRITE LD MAR I/O OUT I/O IN WAIT DCH
1 Jump
1
2 Jump Indirect
3
3 Jump to Subroutine
1
4 JSR Indirect
3
(/JS)
1.5
1
2
4.5
1
2
4.5
1.5
5 Increment and Skip if Zero
3
1
2
4.5
6 ISZ Indirect
5
1,3
2,4
7.5
7 Decrement and Skip if Zero
3
1
2
4.5
8 DSZ Indirect
5
1,3
2,4
7.5
2
6.0
9 Load Accumulator
10 LOA Indirect
2
1
4
1.3
3.0
11 Store Accumulator
3
1
2
4.5
12 STA Indirect
5
1,3
2,4
7.5
13 Complement
1
14 Negate
1
1.5
.;A~!I{
1.5
15 Move
1
1.5
16 Increment
1
1.5
17 Add Complement
1
1.5
18 Subtract
1
1.5
19 Add
1
1.5
20 AND
1
1.5
1,2
3.0
21 ALU with Skip
22 I/O Data In
2
23 1.0 Data Out
2
24 Skip on Busy or Done
2
25 Interrupt
5
1
2.5
1
2.5
2.5
1
3
2.4
1
7.5
26 Data Channel
1
1
27 Wait
1.0
1.0
2.0
28 Examine Accumulator
2
1
.29 Deposit Accumulator
2
1
2.0
30 Load PC
2
1
2.5
31 Examine Memory
2
1
2.5
32 Examine Next
2
33 Deposit Memory
34 Deposit Next
35 Continue
1
2.5
3
2
1
4.0
3
2
1
4.0
1
2.5
2
2nd cycle - WRITE
··e.g., No.6, ISZ Indirect:
·For 9440 System using
3rd cycle - READ
1st cycle - READ
a 10 MHz oscillator.
Reprinted by permission of Fairchild Camera and Instrument Corporation.
4-45
4th cycle - WRITE
5th cycle - FETCH
!"""-
() +5V
I
.
iBi -187
~
G
.-
Ai
...
..>
-A7
)
...
I
PRE
IB8 - 'iB'i5
Q
'--- 0
LS74
LS377
G
.
AS - A15
CK
)
...
-- ~CK
"V
r
LS377
~
~
MBUSY
01
;::J
00
.
~
-
1
A
..
iBO - iB7
I('
A
A
B
r
'"
A
Mi50 - MD7
G
DIR
IB8 - IB15
G
A
,
r
LS245
~
MD4 -
I
..)
Mi59
G
I
L
MD8 - Mi51'5
...
'"
...
iR4 -IR9
CK
..
LS378
G
...
~
)
~
r
iR10 - iRT5 ...
)
..
LS378
,J~
.-
01
00
iR7
.
IBO -
iB7
,
.
Instruction
Register
(12 bits)
1
--..
DSO - 055 ".
G
DIR
A
A
B
r
i5ATAO - DAi'A7
.
LS245
A
~
IB8 - iBi5
..
..
,
iiffi -IB15
G
DIR
A
..L
DATA8 - DA'i'A15
Figure 4-18. 9440 Information Bus Demultiplexing Logic
4-46
..
...
)
"
I/O
Address
Bus
..
B
LS245
Memory
Data Bus
B1~ ~
r
iiii'i5'1O-MD15
..
~
,
LS245
DIR
I
Memory
Address
Bus
r
1/0 Data
Bus
1
9440 - NOVA BUS INTERFACE
We will now examine logic which expands the 9440 pins and signals to the standard Nova 1/0 bus and to a typical microcomputer memory bus. Table 4-1 identifies the Nova 1/0 bus that is created.
We will also illustrate that part of 1/0 device interface logic which is common to any 1/0 device associated with Busy, Done and Interrupt flags.
that is, logic
Our discussion of logic needed to create a memory bus is quite general. reflecting the fact that there is no standard
Nova memory bus. We will therefore limit ourselves to demonstrating. in general. how typical memory bus signals may
be created from 9440 signals. But we will be specific in describing logic that expands the 9440 interface to a standard
Nova I/O bus.
The 9440-Nova bus interface description is divided into three parts:
1)
ExpanSion of the Information Bus into various Address and Data Busses required by the I/O and memory
references.
2)
Creation of I/O interface control signals.
3)
Creation of memory interface control signals.
We will examine each of the three logic expansions in turn.
9440 INFORMATION BUS EXPANSION
These four busses must be created out of the bidirectional 16-bit Information Bus:
1)
A bidirectional. 16-bit Memory Data Bus.
2)
An output only. 15-bit Memory Address Bus.
3)
A bidirectional. 16-bit I/O Data Bus.
4)
An output only. 6-bit I/O Device Address Bus.
We must also latch I/O instruction object codes into a buffer out of which I/O instruction code bits can be read by I/O
control signal logic.
The 9440 Information Bus is low true; this means a low signal level represents a binary 1, while a high signal
level represents a binary O. Standard Nova 1/0 Data and Address Busses are also low true; we therefore do not
need to invert signals during multiplexing and demultiplexing.
There are many ways in which the 9440 Information Bus may be multiplexed to create the four required busses.
We illustrate one possibility in Figure 4-18. This logic uses LS245 8-bit bidirectional tristate buffers to generate the
two bidirectional Data Busses. while 8-bit and 6-bit gated. edge-triggered flip-flops create the Address Busses and the
Instruction Object Code register.
The Data Bus buffers each have a gate (output enable) input and a data direction input. The gate inputs are low true.
Logic shown in Figure 4-18 selects the LS245 buffers while valid memory data or valid I/O data can exist. Within
these select periods a data direction control signal is created to ensure that data flows in the correct direction.
For the Memory Data Bus. MDO - MD15. the LS245 buffers must be selected either during a read or a write operation.
as identified by MO or M1. But these two signals span addr~ and data occurring on the Information Bus. Valid data
exists on the Information Bus when MBUSY is high while SYN is low:
valid data from
memory or CPU
CPU drives address here
4-47
This timing is also illustrated in Figures 4-12 and 4-13. The logic of Figure'4-18 uses an LS74 flip-flop clocked by
the low-to-high transition of MBUSY. This ensures that data is not driven in the time shaded in the preceding illustration - between the high-to-Iow transition of SYN and the high-to-Iow transition of MBUSY.
MO is used as the Memory Data Bus data direction control.
The I/O Data Bus buffer logic is somewhat simpler. The Information Bus is dedicated to transferring I/O data for the entire duration of a data input or data output machine cycle, as defined by 01 high and 00 low: these two signals are
therefore used to create gate (output enable) logic. The direction of the I/O data transfer is taken from IR7: this bit of the
I/O instruction object code defines the direction of an I/O data transfer, as illustrated in Figure 4-18.
For the Address Busses we do not use buffers: rather. we use gated-clock, edge-triggered flip-flops. This allows the address being output to be held stable on the Memory Address Bus, or the I/O Address Bus, after it is no longer on the Information Bus.
In the case of the Memory Address Bus, the gate inputs are tied to M2, which will be low whenever a memory address
is being output on the Information Bus. The high-to-Iow transition of SYN is intended to act as a memory address
strobe: therefore it is inverted to clock the Memory Address Bus flip-flops when M2 is low. Observe that there are only
fifteen lines on the Memory Address Bus: the high-order bit of a 16-bit memory address is reserved to indicated an indirect address. Note also that the LS377 outputs are not tristate: therefore the Memory Address Bus will always hold
the address of the most recently accessed memory location.
Two LS378 6-bit gated-clock flip-flops are used to latch the lower 12 bits of instruction object codes off the Memory
Data Bus, c~ati~ the I/O Address Bus and the Instruction register. The six low-order output lines provide the I/O Address Bus, SO - S5. As you can see in Figure 19-8, only the low-order 11 bits of the I/O instruction need to be decoded
by I/O logic: therefore we use the 16-pin LS378 parts, rather than the 20-pin LS377s which we used for the Memory
Address Bus. Like the Memory Address Bus, the Instruction register and I/O Address Bus will always hold the most recently latched data. The Instruction register flip-flops are clocked by the low-to-high transition of M'imSY whenever an
instruction object code is on the Memory Data Bus. This condition is guaranteed by logic which enables the clock only
when 01 and 00 are both low, signifying an Instruction Fetch machine cycle. If we wished to latch only I/O instructions, we could change the gate logic as follows:
P
MDT
------~
~---.....
~ to G of each LS378
;
MOO
Mg~
Latching the instruction object code only when its upper three bits are 011 (MOO high, MD1 and MD2 low) means that
the I nstruction register will only hold I/O instructions. Latching all instructions is sufficient since an I/O execution
machine cycle (01 high and 00 low) follows the fetch of an I/O instruction. Our logic will use lines 01 and 00 to indicate execution of an I/O instruction.
Let us now examine I/O bus control signal logic.
9440-NOVA I/O BUS INTERRUPT SIGNALS
Three signals on the standard Nova I/O bus are used by interrupt logic: INTR. INTA and INTP.
INTR is the standard interrupt request signal. This signal can be tied directly to the9440 INT REO input.
The interrupt acknowledge signal INTA is created in response to execution of the interrupt acknowledge instruction.
We will describe logic which creates INTA along with other I/O bus control signals when we discuss Figure 4-19.
INTP is the initial input to the highest priority device in an interrupt daisy chain. This may be illustrated as follows:
INTA
I
T
T
Device
Device
Device
1
2
3
4
j
"PQij'f
PiN = Priority In
POUT = Priority Out
4-48
4
atc.
INTP may be connected to the complement of the 9440 output INT ON. in which case priorities within a daisy chain will
not be resolved while interrupts are disabled. Frequently the initial j5jj\j'input to a daisy chain will be tied to ground and
INTP will not be used. Now interrupt priorities will be arbitrated whether or not interrupts have been enabled.
As you will see. it takes very little logic to expand the 9440 interrupt signals to standard Nova I/O bus interrupt lines.
But a considerable amount of interrupt-related logic must be present at external device controllers -logic which we
will describe later in this chapter.
9440-NOV A DMA CONTROL SIGNALS
The only DMA logic provided by the 9440 consists of a DMA request signal. DCH REO. When input low. this signal
causes the 9440 to complete the instruction currently being executed. then to disable interrupts and wait. The DMA request is acknowledged by outputting 01 low and 00 high.
All logic which actually implements any DMA transfer must be implemented external to the 9440. We will discuss
briefly what logic would be required.
The Request Enable line. ROENB. goes true to permit both interrupt and DMA requests. Central DMA control logic
would contain an Enable flip-flop. analogous to the CPU's Interrupt Enable flip-flop. The output of this flip-flop. ANDed
with INT ON from the 9440. would I\)rovide ROENB as follows:
DMAenable
INT ON
D
---------1
- - - - - - RciENB
The DMA request line DCHR may be connected to the 9440 DCH REO input. Thus requests will be accepted and
granted by the 9440 CPU. DCHA. the acknowledgment signal. is simply decoded from lines 01 and 00:
g~--------9C::)o------~
DCHP is a priority line just like INTP. DMA daisy chain priorities would be implemented similarly to interrupt priorities.
The 9440 surrenders control of the System Bus when it acknowledges a DMA request; therefore external logic must
perform all signal manipulations and data transfers. DCHI and DCHO. which indicate the direction of data transfer. are
signals output by external DMA control logic.
The DMA control logic will input DeHMO and DCHM1 from the device requesting memory access. Of the four encoded
modes shown in Table 4-1. "Data Out" and "Data In" can be handled with relative ease. especially if you use an LSI
chip designed for DMA control. Implementing the other two functions. "Increment Memory" and" Add to Memory". requires much more logic since some arithmetic is required. Indeed. a one-chip microcomputer might supply this logic.
Since OVFLO is true when an "Increment Memory" or an "Add to Memory" operation produces a result greater than
FFFF16. this signal would be produced by the logic which performs those operations.
Figure 4-19 shows 3-to-8 and 2-to-4 decoders creating Ndva I/O Bus control signals. The signal logic directly
interprets I/O instruction object code bits illustrated in Figures 4-8, 4-9, and 4-10. Note that the Instruction
register bits from Figure 4-18 are low true. and that Instruction register lines are numbered according to Nova conven~
tion. where the low-order line is IR15.
4-49
iBO
SELB
iBi
SELD
SKJ5'
YO
C
iR'5
iR6
DATOC
5iC
Y2
B
iR7
DCre
Yi
DATIe
DOB
Y3
A
DATOB
LS13S
Y4
G1
01
00
om
ffi
Y5
(ffij
Va
Y7
OATIS
DOA
DATOA
i5IA
DATIA
REffi
iiITO
iim
1R12
m
10RST
CPU OP
iR14
iRT5
DoB
MSKo
DiS
INTA
YO
B
iR9
1/2
A
SKP
Vi
LS139
Y2
Y3
G
01
00
iOP
10PLS
cr.
CLR
ST
STRT
+5V
CPUOP
SYN
PRE
Q
0
CK
LS74
Ci
Prs
CIA
01
00
Figure 4-19. Creation of Nova I/O Bus Control Signals from 9440 Signals
4-50
an -
Instruction object code bits are continuously read out of the Instruction register
IR15), but I/O control signals are created only during an I/O Execute machine cycle (when 01 is high and 00 is low).
The logic of Figure 4-19 may be divided into these four sections:
1)
Creation of simple data transfer control signals
2)
Creation of I/O skip logic.
3)
Creation of interrupt control signals.
4)
Creation of control signals STRT, CLR and 10PLS.
Let us first consider simple data transfer control signals. There are six signals: DATIA, DATIB, DATIC. DATOA.
DATOB. and DATOC. These are created by the LS138 3-to-8 decode.r in Figure 4-19. and enabled when the I/O
device address is other than 3F 16.
If you look at Figure 4-8 you will see that instruction bits 10 and 9 (iR5 and iR'6) select one of the three registers that
may exist at an I/O device, while bit 8 (iR7) differentiates between I/O data input and I/O data output. These three bits
are input to the LS138 decoder so that the six data transfer signals and the Skip signal are decoded at the outputs. The
decoder is enabled only during an I/O Execute machine cycle - that is. when 01 is high and 00 is low. However. if the
I/O device address is 3F16. thef\ CPU OP will be true and no data transfer signal will go true. It is not strictly necessary
to disable the signals with CPU OP; sirrce none of the I/O devices will be assigned the address 3F16. none of them will
respond to I/O instructions with that address.
ma
The Skip control. SKP. output from the LS138 decoder. is used to enable
and SELD onto Information Bus
lines IBO and IB1. This is done using three-state buffers enabled by SKP low; the buffers in Figure 4-19 might be part
of an LS 125 or an LS367 chip. SELB and SELD are inputs to the buffers. while the outputs are connected to Information
Bus lines IBO and iBi'. We assume that as soon as any I/O deviCe is selected. it immediately connects its Busy and Done
statuses to the SELB and SELD control lines of the I/O bus. However. SELB and SELD will not appear on Information Bus
lines IBO and iBi unless a Skip I/O instruction has been executed.
When an I/O instruction is executed specifying device 3F16. a set of interrupt-related I/O instructions is executed.
as illustrated in Figure 4-10. Most of the instructions illl;lstrated in this figure specify events internal to the CPU. For
example. "enable interrupts" and "disable interrupts" apply to CPU interrupt logic; moreover. the Skip instructions interrogate interrupt request status and power fail status within the CPU. "Acknowledge Interrupt" (lNTA). "Output
Interrupt Mask" (MSKO) and "Clear All I/O Devices" (lORST) are the only instructions which require control
signals to be generated on the I/O bus. These control signals are generated by qualifying the decoder of the instruction bits with a device 3F16 select code. The device 3F 16 select code. CPU OP. is created by ANDing the loworder six instruction bits (lR10 through.IR15). Thus the gates producing INTA. MSKO. and 10RST are effectively
switched on and off by CPU OP Note that 10RST is generated either by execution ofa "Clear I/O Devices" instruction.
or by the master system Reset signal. RESET.
Let us next consider logic needed to create STRT. CLR. and 10PLS.
These control signals should be activated after the appropriate I/O transfer has taken place. Thus the logic in Figure
4-19 provides a gating signal. PLS. which goes low on the low-to-high transition of SYN. PLS is the IT output of the
LS74 flip-flop in Figure 4-19. The timing for STRT, CLR. or. 10PLS results as follows:
10PlS. STRT.
or ClR
01
00
I
Instruction Fetch
~~4"'----------1/0 Execute----~I-------_tl...
CPU or 1/0 Device reads
data here·
4-51
The LS 139 2-to-4 decoder decodes instruction object code bits 7 and 6 (iFrn and imil. providing that the I/O instruction
being executed is not an I/O Skip instruction. (The other half of the LS139 chip could be used to decode lines 01 and
00. instead of the gating logic shown in Figures 4-18 and 4-19.1
DSO
Address
Logic
DS5
STRT
rue
Q
BUSY
CK
LD74
Q
ffii
I/O COMPLETE
+5V
10RST
SELD
Q
CLR
DONE
SELECT
CK
LS74
Q
ffii
STRT
SELECT
10RST
PRE
i5ATAii
D
MsKo
Q
iN'fR
CK
§YN
LS74
Q
+5V
Figure 4-20. Busy. Done. and Interrupt Status Logic Required by I/O Device Controllers on the Nova I/O Bus
NOVA I/O DEVICE CONTROLLER LOGIC
Interface logic which an external device needs in order to connect to the standard Nova I/O bus depends on the
nature of the external device. A minicomputer device controller may be very complex, even costing more than
the minicomputer itself; that is because minicomputer devices that connect to the I/O bus are peripherals, such
as printers, disks. etc. When we reduce the Nova to microprocessor terms. however. external devices connected to the I/O bus reduce to such primitive elements as parallel I/O ports or serial data lines. Within this
4-52
reduced context we can synthesize the minimum necessary elements of an I/O interface as consisting of three
status flags: Busy, Done, and an Interrupt request. We can implement these three status flags using three LS74
flip-flops, as illustrated in Figure 4-20. Device select logic in this figure is limited to showing a select signal
which will be generated true when the appropriate device code appears on the I/O device Address Bus. We
have discussed I/O device select logic at various points earlier in this chapter.
Let us look at the BUSY and the DONE status logic. These are the operations which may affect the condition of the
BUSY and DONE statuses:
1)
At the start of an I/O operation BUSY must be set while DONE is clear. This condition is identified by 01 in bits 7
and 6 (lR8 and iR9) of the I/O instruction object code, which generates the STRT control signal of the I/O bus.
2)
At the completion of an operation BUSY is cleared and DONE is set. This change in status setting must be implemented automatically by I/O device interface logic: it alone knows when the I/O operation has been completed.
3)
BUSY and DONE may be cleared by the CPU. This is specified by 10 in bits 7 and 6 (lR8 andl"R9) of the I/O instruction, which generates the CLR control signal on the I/O bus.
4)
There is a "Clear All I/O Devices" instruction. This instruction generates 10RST on the I/O bus: it clears BUSY and
DONE statuses at all I/O devices.
5)
A Master Reset must also clear the BUSY and DONE statuses. This Master Reset signal can also create 10RST, as illustrated in Figure 4-19.
Two D-type flip-flops implement the BUSY and DONE status logic. These two D-type flip-flops are clocked by an
"I/O Complete" signal which local device logic must generate. The BUSY and DONE statuses are generated by the flipflop Q outputs which must connect to SELB and SELD, as required by I/O skip logic, which we have already described.
The BUSY flip-flop uses its Set and Clear logic to control the BUSY status. The BUSY status is set by STRT and SELECT
both true. This combination of STRT and SELECT sets the device BUSY status while it resets the DONE status.
Either CLR and SELECT both true or 10RST true will activate the Clear input of the BUSY flip-flop.
Neither of these conditions will be present when BUSY is set by the STRT pulse. Subsequently,
when STRT or SELECT goes false, BUSY will stay true until it is reset by "I/O Complete" or by an
active Clear input. which will occur when either 10RST or both CLR and SELECT are true.
10RST
CLR
STRT
The DONE status is set by the "I/O Complete" pulse after BUSY has been set. Once DONE is set. it will remain true until
the flip-flop is cleared. These conditions are provided by OR logic at the D input to the DONE flip-flop. The Clear input is
Clctivated by anyone of the following conditions being true:
1)
STRT and SELECT both true: thus the DONE status is reset at the same time as the BUSY status is set.
2)
The master Reset. 10RST.
3)
CLR and SELECT both true.
The device interrupt may be individually disabled by a Mask Out instruction's execution: this creates the MSKO control
signal used to permit the clocking of the interrupt mask flip-flops. Accompanying execution of the Mask Out instruction, a 16-bit data value is output on the I/O Data Bus. An I/O Device's interrupt logic is controlled by one bit of this
mask, the bit transmitted via I/O Data Bus line DATAn. Therefore DATAn becomes the D input to the interrupt mask
flip-flop. A 1 in the mask bit (DATAn low) disables interrupts from the I/O device. In Figure 4-20, the Q output of the
flip-flop becomes the interrupt enable signal. INT ENABLE. which gates the device's interrupt request onto INTR.
The bottom flip-flop in Figure 4-20 implements interrupt logic for the I/O interface. Let us summarize the conditions that can affect I/O interface interrupt logic. Providing interrupts are enabled at the I/O interface, an interrupt
will be requested whenever an I/O operation is completed, as identified by the DONE status going true. If INT ENABLE
is true, INTR will go low as soon as the DONE status is set.
Interrupt logic may be enabled by a master I/O reset therefore 10RST is connected to the flip-flop Preset input.
9440 MEMORY BUS
There being no standard Nova Memory Bus, we will look at the signals available to you when you interface
memory to the 9440.
First return to Figure 4-18. This figure shows how stable Data and Memory Busses may be demultiplexed off
the 9440 Information Bus. In order to create a Memory Bus of any type, all you need is control signals to accompany the Memory Data Bus and the Memory Address Bus.
Figure 4-21 presents an example of memory control signals derived from 9440 signals, and Figure 4-22
shows the timing for these signals. The four D-type flip-flops of an S175 chip, along with some combinatorial logic.
constitute a state machine to generate Signals required by memory and the 9440 CPU. The four flip-flops are triggered
4-53
by MEMORY CLOCK. In order that all worst-case delay times be satisfied. the frequency of MEMORY CLOCK should not
exceed 23.8 MHz; if LS parts are used. the maximum worst-case MEMORY CLOCK frequency is 10.8 MHz. The common clear of the four flip-flops will be activated if none of the lines MO. M1. or M2 is true.
SYN
MEM ENABLE
------e_---------------c.
MEMORY CLOCK -----+--------------e_~
CK
1/4 S175
J----------+---t D
Q
t - - - - - - ADDRESS VALID
CK
)--------+---ID 114 S175
Q
CK
a t-----1~- MBtiSY
114 S175
~~~D
Qt----~
CK
ENABLE
ot-------,
M2
MT
---------~~
-------4~_t
Mo--~.....-
.....WRITE
READ
Figure 4-21. Memory Controls Derived from 9440 Signals Using State Machine Logic
4-54
Figure 4-22. Timing for 9440-Based Memory Controls
Provided that either Mi. M1 or MO is low (signifying a memory access machine cycle) the MEM ENABLE signals
will go true on the first MEMORY CLOCK after SYN goes low. MEM ENABLE will stay on until the first MEMORY CLOCK after SYN goes high again.
On the rising edge of MEMORY CLOCK after MEM ENABLE goes on, MBUSY and ADDRESS VALID will go true.
Memory control logic must return MBUSY to the CPU. since the 9440 requires interaction of the MBUSY and SYN signals in order to complete memory access cycles. We discussed this earlier in this chapter. in the text accompanying
Figures 4-12 and 4-13. A more complex memory interface could use MBUSY to lock out CPU memory accesses while
the memory is busy -for example. while memory is responding to a direct memory access.
The memory logic itself may require some signal to be true as long as a valid address is on the Memory Address Bus.
Thus. our logic provides the signal ADDRESS VALID. which goes on after the contents of the Memory Address register
(shown in Figure 4-18) have had time to settle. and remains until the end of the memory cycle. In Figure 4-18. the
Memory Address register is clocked by the high-to-Iow transition of SYN. but a system might use the leading edge of
ADDRESS VALID to clock the Memory Address register.
On the MEMORY CLOCK after MBUSY goes on. the ENABLE flip-flop clocks on. If MO or Ml is low at this time,
then READ or WRITE will go on and stay on until the MEMORY CLOCK after MEM ENABLE goes off. The signals
READ and WRITE tell the memory chips the direction of the data transfer.
Of course. different system signal and timing specifications would require different implementations of memory Signals. A memory system might use one-shots or delay lines to create pulsed Signals. and simple combinatorial logic for
signal levels. A state machine implementation could use a Counter or Shift register. or perhaps a field-programmable
logic sequencer such as the Signetics 82S105.
4-55
DATA SHEETS
This section contains specific electrical and timing data for the following devices:
• MicroNova
·9440
4-01
MICRONOVA
ABSOLUTE MAXIMUM RATINGS·
Supply Voltage Range Vaa
-2
to
-7 Volts
Supply Voltage Range V CC
-0.3 to
+ 7 Volts
Supply Voltage Range VDD
-0.3 to
~Volts
Supply Voltage Range VGG
-0.3 to + 17 Volts
Input Voltage Range VI
-0.3 to
Input Current Range II
o
Operating Temperature Range T A
o to
Storage Temperature Range T stg
Average Power Dissipation
+ 7 Volts
to _6_mAmps
+70
°c
-55 to + 125°C
1 Watt
NOTES All voltage8 in this document are
referenced to V 88 (ground).
*Subjecting a circuit to conditions either
outside these limits or at these limits for an
extended period of time may cause irreparable
damage to the circuit. As such, these ratings
are not intended to be u-sed during the
operation
of
the
circuit.
Operating
specifications are given in the DC (STATIC)
CHARACTERISTICS ,TABLE.
Data sheets on pages 4-02 through 4-03 reprinted by permission of Data General Corporation.
4-D2
MICRONOVA
D. C. (STATIC) CHARACTERISTICS
mN601
OPERATING SPECIFICATIONS
TA
Vee
range_O_to~C
5,0.25 Volts
Voo = 10 • 1.0 Volts
VGG
=14.1.0Volts
ICC
~mAmps
Average
IBB
V BB
= -4.25
100
~mAmps
Average
ISS
Vss
= 0 - 0.0 Volts
~mAmps
Average
CHARACTE R1STIC
INPUT L(YN VOLTAGE
INPUT CURRENT FOR
L(YN STATE
INPUT HIGH VOLTAGE
INPUT CURRENT FOR
HIGH STATE
t
.25 Vol!s
SYMBOL
VIL
IlL
VIH
IIH
PINS
UNITS
Volts
mAOlps
Volts
mAmps
~mAmps
LIMITS
MIN.
MAX.
.0.5
121,3 and 122,4
-2.0
MB 0-1> , CLAMP
EXTINT. DCH INT
-1.0
+1.0
10 CLOCK. I 0 DATA I,
I o DATA 2
al,3 and 122,4
MB 0-1.
EXTINT. 0CIflN't, CLAMP
I 0 CLOCK. I 0 DATA I,
I o DATA 2
-1.0
.0.5
-2.0
-4.0
01,3 and '02,4
+13.0
+15.0
MB 0-1.
EXTINT,
, co;m;
IlClr'lNT
I 0 CLOCK. I 0 DATA I,
I o DATA 2
al,3 and 02,4
MB 0-1.
10 CLOCK. 10 DATA I, I 0 DATA 2
EXTIN' . OCR INC
+.01
-2.0
-2.0
-4.0
+4.25
+5.8
+2.5
+5.8
-.01
-.06
-1.0
-.02
+.001
co;m;
+3.0
OUTPUT LOW VOLTAGE
VOL
Volts
MB 0-1.
I 0 INPUT: PAUSE,
SAEG. WEG. PG
1/0 CLOCK JlO DATA 1 1/0 DATA 2
PG. I 0 INPUT
OUTPUT CURRENT FOR
LOW STATE
10L
mAntps
OUTPUT HIGH VOLTAGE
VOH
Volts
OUTPUT CURRENT FOR
HIGH STATE
10H
mAnl~
.0.4
+0.5
+4.0
MB 0-1. . 1 o CLOCK
I o DATA I,. o DATA 2
~. SAEG. PG. HALT
+2.0
MB 0-1.
I 0 CLOCK. I 0 DATA 1, 10 DATA 2
I 0 INPUT. PAUSE.
SAEG. WEG. PG
+4.25
HALT
C -0.
M80-1.
I o INPUT. PG
10 CLOCK. I 0 DATA 1, I 0 DATA 2,
PAUSE SAEG. WEG
01,3 and .2,4
-.01
-.06
-.02
-.01
100
CLAm>
INPUT CAPACIT ANCE
CI
pF
MB 0-1. ,I o CLOCK
I 0 DATA 1, I 0 DATA 2
Eiffiij'f.
0CIflIrr
NOTE
Logic "I" is defined as the more positive voltage as are the maximum
figures given under voltage limits. Logic "0" is defined as the more
negative voltage as are the minimum figures given under voltage limits.
Positive current, in the conventional sense, is defined as flo'wing into
t,he pin.
On power-up. VBB must be within its specified operating range (with
respect to Vss) before any of the other power supply voltages are applied to the circuit.
4-03
10
Average
9440
ABSOLUTE MAXIMUM RATINGS (beyond which the useful life of the device may be impaired)
Storage Temperature
Ambient Temperature Under Bias
Vee Pin Potential to Ground Pin
Input Voltage (de)
Input Current (de)
Output Voltage (Output HIGH)
Output Current (de) (Output LOW)
Injector Current (IINJ)
Injector Voltage (VINJ )
_65° to 15O"C
-55 to +125°C
-0.5 to +6.0 V
-0.5 to +5.5 V
-20 to +5 mA
-0.5 to +5.5 V
+20 rnA
+500 rnA
-0.5 to +1.5 V
DC CHARACTERISTICS OVER OPERATING TEMPERATURE RANGE (0 to 75°C)
hNJ(min) = 300 rnA. hNJ(max) = 400 rnA. Vee(min)
= 4.75 V.
Vee(max)
= 5.25 V
LIMITS
CHARACTERISTIC
SYMBOL
VIH
Input HIGH Voltage
MIN
TYP
MAX
2.0
VIL
Input lOW Voltage
Vco
Input Clamp Diode Voltage
VOH
Output HIGH Voltage
RUN. CARRY. INT ON. SYN. ClK OUT. 0 0 • 0 1
2.4
Q.utput.!'IGH Voltage
IBo - IB15
2.4
UNITS
TEST CONDITIONS
V
Guaranteed Input HIGH Voltage
0.8
V
Guaranteed Input lOW Voltage
-1.5
V
Vee = 4.75 V. liN = -18 rnA
IINJ = 300 mA
3.4
V
Vee = 4.75 V. IOH
IINJ = 300 mA
= -4OO,.,.A
3.4
V
Vee = 4.75 V.IOH
IINJ = 300mA
= -1.0 rnA
1.0
mA
Vee = 4.75 V. VO H
IINJ = 300mA
= 5.25 V
0.25
0.5
V
Vee = 4.75 V. 10L
IINJ = 300 mA
= 8.0 mA
-0.9
ICEX
Output leakage Mo. M1• M2
VOL
Output lOW Voltage
IIH
Input HIGH Current _ _ _ _ _ _ _
Co-C3. DCH REO, INT REO, MBSY, MR
1.0
20
,.,.A
Vee = 5.25 V. VIN
IINJ = 300 mA
= 2.7 V
Input HIGH Current
CP
2.0
40
,.,.A
Vee = 5.25 V. VIN
IINJ = 300mA
= 2.7 V
!!!put H..!.GH Current
IBo - IB15 (3-State)
5.0
100
,.,.A
Vee = 4.75 V. VIN
IINJ = 300 mA
= 2.7 V
1.0
mA
Vee = 4.75 V. VIN
IINJ = 300mA
= 5.5 V
Input HIGH Current
All Inputs
IlL
Input lOW Current
All inputs except CP
-0.21
-0.36
mA
Vee = 5.25V. VIN = 0.4 V
IINJ = 300 mA
Input lOW Current
CP
-0.42
-0.72
mA
Vee = 5.25 V. VIN
IINJ = 300 mA
100
,.,.A
Vee = 5.25 V. VO UT
IINJ = 300mA
= 2.4 V
-0.36
mA
Vee = 5.25 V. VOUT
IINJ = 300 mA
= 0.4 V
-100
mA
Vee = 5.25 V. VOUT
IINJ = 300 mA
= 0.0 V
200
mA
Vee
V
IINJ
10ZH
OFF State (Hig.!!.lmp~ance)
Output Current IBo - IB15
10ZL
OFF State (Hig~lmp~ance)
Output Current IBo - IB15
los
Output Short CircuiLCu~nt_
All Outputs Except Mo. M1• M2
Icc
Supply Current
150
VINJ
Injector Voltage
1.0
-0.21
-15
= 0.4 V
= 5.25 V
= 300 mA
Data sheets on pages 4-04 through 4-010 reprinted by permission of Fairchild Camera and Instrument Corporation.
4-04
9440
AC CHARACTERISTICS: T A = 0 to 75° C - Figures 8 & 9
LlMITS-ns
SYMBOL
CHARACTERISTIC
tCPSYL
Propagation Delay, CLOCK to SYN going LOW
150
tCPSYH
Propagation Delay, CLOCK to SYN going HIGH
160
MIN
TYP
MAX
NOTE
70
tMBSYL
Propagation Delay, MBSY going HIGH to SYN gOing LOW
tMBw
MBSY Min Pulse Width (HIGH)
tMBS
Set-up Time, MBSY HIGH to CLOCK
tMBHO
Hold Time, MBSY HIGH after CLOCK
tCPMH
Propagation Delay, CLOCK to M2, M1, Mo going HIGH
160
tCPML
Propagation Delay, CLOCK to M2, M1, Mo going LOW
170
tCPOH
Propagation Delay, CLOCK to 01, 00 going HIGH
160
Fig.90nly
tCPOL
Propagation Delay, CLOCK to 01, 00 going LOW
170
Fig. 8 Only
tCPAH
Propagation Delay, CLOCK to ADDRESS 180-15 going HIGH
170
30
-40
60
tCPAL
Propagation Delay, CLOCK to ADDRESS 180-15 going LOW
180
tMBAF
Propagation Delay, CLOCK toADDRESS 180-15 gOing 3-state
110
tos
Set-up Time, DATA 180-15 to CLOCK
tOHO
Hold Time, DATA 180-15 after CLOCK
130
tcs
Set-up Time, C3, C2, C1, Co to CLOCK
-110
tCHO
Hold Time, C3, C2, C1, Co after CLOCK
130
tCPRH
Propagation Delay, CLOCK to RUN HIGH
160
tCPRL
Propagation Delay, CLOCK to RUN LOW
tocs
Set-up Time, DCH REO to CLOCK
-110
170
-110
tOCHO
Hold Time, DCH REO after CLOCK
tiS
Set-up Time, INT REO to CLOCK
tlHO
Hold Time, INT REO after CLOCK
120
tCPCYH
Propagation Delay, CLOCK to CARRY HIGH
160
tCPCYL
Propagation Delay, CLOCK to CARRY LOW
150
tCPIOH
Propagation Delay, CLOCK to INT ON HIGH
200
tCPIOL
Propagation Delay, CLOCK to INT ON LOW
190
130
-100
NOTES:
1. The Information Bus is driven as a result of the previous cycle.
2. The Fetch and Read cycles will be stretched out for slower memories.
3. Applies to console operation using this cycle type.
4-05
Fig. 8 Only
9440
I ...
·-----------------FETCHCYCLE-----------------~.I
CLKOUT
~
"
tcPSYL~
~
-+l
"'BSYL
14-
....\\\\lil
MBSY
k-
I...w~
"
,.
I
-l
__ I...S
,
I4--tcPSYH---l
1-1-
"
"
"'.HO ~ .....
---
"
"I-
LrL'n..1-t.ru-t-
-~
~tcP"H~
-lk-
I--- tcPML----.J
"
'-
JL
"
i
00
"
I4--ICPOL -...j
0,
'-
iio-15
\.G)
_lcjAH_1
..J
_IC~AL---t
4-tDHD~
X
"
_ICPRH~
--
DATAIN
--ilcs
::
---+-l
,lit
RUN
-+I
IV/, DATAl/,(
JL
CO-3
i-ICPRL
los
tMBAF:i
ADDRESS OUT
""
X
i--ICHO-+l
------------ - - - - - - -4~- ----
'=-t~~
- - - - - - --
DCHREQ
..... 'ocs 14-
:",,""-ICPCYH----.j
-\::E
CARRY
....-ICPCYL
--J
-
~~
~
"
~
~IIHO_
"
INTON
MEMORY
BUSY
tiME
Fig. 8
<»
I4--ICPIOL
""'lit
--+-I
~tCP10H-----"
I
Fetch Cycle
' " " r . - - - - - - - - - - - - - - - - - - R E A D CYCLE
CLK OUT
liS
-----------------~.I
~
-+l
~
IMBSYL 1-
,.
unh.h..1l..JL
.j
"
"
~\\\\lil
IMBw-+l
4-1 ...
51
'4--tCPSVH
1
~I
-I-
1-4=
1_ IMBHO' """"
-f~lcPMH--+l
~
"
f4--ICPML~
MO
'.-
-f-
00
_lcPOH_1
0,
"
/
......-......tCPAH~
i'io-15
b.4 BAF
ADDRESS OUT
- 4 - - tCPAL--...
:i
~loHD_1
-+lIDS I--
J ~DATA~
-lIes
JL
DATA IN
1-.
®
Co-3'
MEMORY
BUSY
TIME
RUN, DCH REO, INT REO, CARRY, INT ON
unaffected durlnS! this cycle.
Fig. 9
Read Cycle
4-06
-
9440
AC CHARACTERISTICS: TA
= 0 to 75°C- Figures 12. 13. 14. 15
LlMITS-ns
SYMBOL
CHARACTERISTIC
tCPSYL
Propagation Delay, CLOCK to SYN going LOW
150
tCPSYH
Propagation Delay, CLOCK to SYN going HIGH
160
tCPMH
Propagation Delay. CLOCK to M2, M" Mo going HIGH
160
tCPML
Propagation Delay, CLOCK to M2. M" Mo going LOW
170
tCPOH
Propagation Delay, CLOCK to 0,,00 going HIGH
~O
tCPOL
Propagation Delay, CLOCK to 0,,00 going LOW
170
170
MIN
TYP
tCPOH
Propagation Delay, CLOCK to DATA IBo-15 going HIGH
tCPOL
Propagation Delay, CLOCK to DATA IBo-15 going LOW
180
tCPOF
Propagation Delay, CLOCK to DATA IBo-15 going 3-state
110
tos
Set-up Time, DATA IBo-15 to CLOCK
-110
tOHO
Hold Time, DATA IBo-15 after CLOCK
130
tcs
Set-up Time, C3, C2, C" Co to CLOCK
-110
tCHO
Hold Time, C3, C2, C" Co after CLOCK
130
MAX
NOTE
Fig. 12 Only
Fig. 13 Only
Fig. 14 Only
NOTES:
6. DUring DCH, the 94~ IS not driving the M lines An external device can control the memory when a LOW IS applied to the appropriate Milne.
7. The 9440 floats the 180-15. The Information Bus is available to the I/O devices and the memory as needed.
- - - - - - - - - - - - ItO IN
------------I.~I
PLA
CLKOUT
SYN _ _ _ _ _
of
M,
---------+------~
ij,-----+-----'
MO _____+-___-'
OO-----+___
--tc-P-ML-~-~---------------~-----O'-----+~--t-cP-M-H~~1
~o,~-----------------------~~~~~
co")
RUN _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ _
."i5CHiiEci
iNTiiEa
CARRY _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ____
INTON _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ _
Fig. 13
1/0 In Cycle
4-09
9440
t-.-----------WAIT----------_
CLK OUT
8YN
---------of
Ml __________+_-------'.
M' _________+_------'
~---------+--------'
Oo _________~------~
0, _ _ _ _ _ _ _ _ _ _ _ _ _- '
RUN __________________________________________________________
CARRy _________________________________________________________
INTON ______________________________________________________
Flg.14 Walt Cycle
I-
~-----------DCH----------~·~I
PLA
CLK OUT
8YN
iiiii
Ml
(i)
ii,
ilo
00
o
01 __________
~--------~----------------------------------------
iiio-15 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
~-3 _____________________________________________________________
RUN __________________________________________________________
CARRy _________________________________________________
INTON ______________________________________________________
Flg.15 Data Channel Request Cycle
4-010
Chapter 5
THE INTEL 8086
The 8086 is Intel's first 16-bit microprocessor. It is significantly more powerful than any prior microprocessor.
The 8086 assembly language instruction set is upward compatible with 8080A - but at the source program
level only. That is to say. every 8080A assembly language instruction can be converted into one or more 8086 assembly language instructions. There is no reason why anyone would try to convert 8086 assembly language instructions.
one at a time. into one or more 808~A assembly language instructions. but if you did. you would soon become
hopelessly tangled in conflicting memory allocations and special translation rules. That is why we say that the 8086
and 8080A assembly language instruction sets are "upward" compatible.
The 8086 and 8080A assembly language instruction sets are not compatible at the object code level. which means that
8080A programs stored in read-only memory are useless in an 8086 system.
The 8085 and 8080A assembly language instruction sets are identical. with the exception of the 8085 RIM and SIM instructions. The 8085 RIM and SIM instructions cannot be translated into 8086 instructions. This is because the RIM and
SIM instructions use the serial 110 logic of the 8085. which has no 8086 counterpart. Without the RIM and SIM instructions. the 8085 and 8080A assembly language instruction sets are identical; therefore the 8086 assembly language
instruction set must also be upward compatible with the 8086 assembly language instruction set - apart from
the RIM and 81M Instructions.
The 8085 and 8080A assembly language instruction sets are object code compatible - with the exception of the 8085
RIM and SIM instructions. That is to say. a program existing in read-only memory could be used with one
microprocessor or the other.
The 8080A assembly language instruction set is a subset of the l80 assembly language instruction set. That is to say.
the l80 will execute an 8080A object program - but the reverse is not true. The 8080A cannot execute l80 programs
when the full l80 instruction set is used. The 8086 assembly language instruction set is not upward compatible
with the Z80 assembly language instruction set.
As a historical note. it is worth mentioning that the 8008 microprocessor. which preceded the 8080A. was also compatible only at the source program level. That is to say. there is an 8080A assembly language instruction for every 8008
assembly language instruction. but the two microprocessor object code sets are not the same.
The various instruction set compatibilities that we have described may be illustrated as follows:
-..
(Excluding RIM and
SIM instructions)
8086
•
~
8085
+
IL
______ ..JI
_______
8080A
- - Source program of lower microprocessor
can be assembled to generate upper
microprocessor object program
+
- - - Lower microprocessor instruction set is a
subset of upper microprocessor instruction
set at the object program level
8008
5-1
~
These are the most interesting innovations to be found in 8086 hardware design:
1)
8086 Central Processing Unit logic has been divided into an Execution Unit (EU) and a Bus Interface Unit (BIU).
These two halves operate asynchronously. The Bus Interface Unit handles all interfaces with the external bus; it
generates external memory and I/O addresses and has a 6-byte instruction object code queue. Whenever the EU
needs to access memory or an I/O device, it makes a bus access request to the Bus Interface Unit. Providing the
Bus Interface Unit is not currently busy, it acknowledges the bus access request from the EU. When the Bus Interface Unit has no active pending bus access requests from the EU, it performs instruction fetch machine cycles to
fill the 6-byte instruction object code queue. The CPU takes its instruction object codes from the front of the queue.
Thus instruction fetch time is largely eliminated.
2)
The 8086 has been designed to work in a wide range of microcomputer system configurations, ranging from a simple one-CPU system to a multiple-CPU network. To support this wide flexibility, a number of 8086 pins output
alternate Signals. This may be illustrated as follows:
Maximum Configurations
Minimum Configurations
These signals
do not change
These Signals
do not change
+5 V
8086
8086
MN/MX
Simple control output
for use in one-CPU
system
MN/MX
Complex control signals
useful in multi-CPU
networks
The same pins output these two sets of signals, based on a level of MN/MX. This wholesale re-allocation of Signals
is a highly imaginative and innovative first for the microprocessor industry.
3)
The 8086 has built-in logic to handle bus access priorities in multi-CPU configurations. (This is not a new concept
National Semiconductor's SCIMP has had it for years.)
4)
In mUlti-CPU configurations, each 8086 CPU can have its own local memory, while simultaneously sharing common memory. The common memory may be shared by all CPUs, or by selected CPUs.
5)
The 8086 has been designed to compete effectively in program intensive applications that have been the domain
of the minicomputer. Up to a million bytes of external memory can be addressed directly. All memory addressing is
base relative; this memory addressing technique naturally generates relocatable object programs. (Relocatable object programs can be moved from one memory address space to another and re-executed without modification.)
Also, since the 8086 utilizes stack-relative addressing, re-entrant programs are easily written. (Re-entrant programs
can be interrupted in mid-execution and re-executed. For example, a subroutine which calls itself is re-entrant a
program which can be interrupted in mid-execution by an external interrupt, and then re-executed within the interrupt service routine, is also re-entrant.
6)
The 8086 uses prefix instructions that modify the interpretation of the next instruction's object code.
The 8086. like its predecessor. the 8080A. is really one component of a multiple-chip microprocessor configuration.
In addition to the 8086 microprocessor itself. you must have an 8284 Clock Generator/Driver. Yau cau Id create
the required clock Signal using alternative logic, but it would be neither practical nor economical to do so.
The third device necessary in some 8086 microprocessor configurations is the 8288 Bus Controller.
You will usually have an 8288 Bus Controller between an 8086 and its System Bus (or busses), just as you will usually
have an 8288 System Bus controller between an 8080A and its System Bus. In the case of the 8086, however, you can
dispense with the 8288 Bus Controller in Single-bus configurations - and pay no penalty for it.
5-2
The 8086 has a large family of support devices. In this chapter we describe the following support devices:
• The
• The
• The
• The
8284 Clock Generator/Driver
8288 Bus Controller
8282/8283 8-bit input/output latches
8286/8287 8-bit parallel bidirectional bus drivers
The 8088. an 8-bit version of the 8086. is also described.
The primary manufacturer of the 8086 is:
INTEL CORPORATION
3065 Bowers Avenue
Santa Clara. California 95051
Second sources are:
MOSTEK CORPORATION
1215 West Crosby Road
Carrollton. TX 75006
NEC MICROCOMPUTERS INC.
Five Militia Drive
Lexington. MA 02173
SIEMENS AG
Components Group
Balanstrasse 73. D8000
Munich-80. West Germany
The 8086 is manufactured using N-channel depletion load. silicon gate technology. It is packaged in a 40-pin DIP. A
single +5 V power supply is required. All signals. with the exception of the clock input. are TTL-level compatible. The
clock input must be an MOS level signal: it is generated by the 8284 Clock Generator/Driver device. which is described
later in this chapter.
Instruction execution times will vary depending on how effectively instruction queuing is used. Typically. between 2
and 30 clock cycles are required to execute an instruction. Multiplication and division instructions require more execution time. Clock cycles may be as short as 125 nanoseconds.
An 8 MHz version of the 8086 has been announced: it is identified as the 8086-2. The 4 MHz version is called the 8086-4. The standard 5 MHz 8086 is referred to without a suffix. There is no
difference between the three versions other than maximum allowed clock speeds.
5-3
8086
8086-2
8086-4
THE 8086 CPU
Functions implemented on the 8086 microprocessor chip are illustrated in Figure 5-1.
Interrupt priority arbitration logic is shown as only half present external logic. such as an 8259A. must provide a
device code identifying an interrupt. but all arbitration and vectoring logic is subsequently handled by logic within the
CPU.
It is worth noting that bus interface logic. which is shown as present in Figure 5-1. is much more extensive than other
microprocessors provide. One could rightfully demand that bus interface logic therefore be shown as absent in equivalent figures for other microprocessors.
Clock Logic
Interface Logic
Programmable
Timers
I/O Ports
Figure 5-1. Logic of the Intel 8086 CPU
5-4
Memory
8086 PROGRAMMABLE REGISTERS AND ADDRESSING MODES
We describe 8086 programmable registers in conjunction with 8086 addressing modes, since many 8086 programmable registers are there only to support memory addressing logic. 8086 programmable registers are illustrated in Figure 5-2.
Shaded registers are 8086 equivalents for 8080A registers. 8080A register names are shown in the left margin.
Let us first examine the general purpose registers, AX, BX, ex, and OX. These locations
are treated as four 16-bit registers or eight 8-bit registers; they also reproduce the 8080A
general purpose registers as follows:
AH has no 8080A equivalent. Do not confuse it with the 8080A PSW.
8086 AND
8080A
REGISTERS'
COMPATIBILITY
AL is equivalent to the 8080A A register.
BH is equivalent to the 8080A H register.
BL is equivalent to the 8080A L register.
CH is equivalent to the 8080A B register.
CL is equivalent to the 8080A C register.
DH is equivalent to the 8080A D register.
DL is equivalent to the 8080A E register.
Consistent with 8080A register utilization. register AX serves as a primary Accumulator. Input
and output instructions pass data through AX (or AU in preference to other general purpose
registers; also. selected instruction access AX (or AU contents only.
In addition to serving as a general purpose Accumulator, register BX can serve as a base
register when computing data memory addresses.
8086 BX
REGISTER
Register ex serves as an Accumulator; it is also used as a counter by multi-iteration instructions; these instructions terminate execution when register CX contents increment or decrement
to O.
8086 ex
REGISTER
Some 1/0 instructions move data between an identified 1/0 port and the memory location
addressed by Register OX. Register DX may also serve as an Accumulator.
When looking at general purpose registers AX. BX, Cx. and Ox. there is plenty of opportunity to
be confused by terminology.
Intel literature identifies the four 16-bit registers via the labels AX, BX, CX, and DX. Each of these 16-bit registers is
subdivided by Intel literature into two 8-bit registers, as follows:
o +- AX bit numbers
o +- AH, AL bit numbers
15
7
07
AH
I
AL
~
AX
15
7
I
I
o +- BX bit numbers
o +- BH, BL bit numbers
07
BH
1
B7
BX
15
7
I
CL
--.,CX
15
7
1
ox
I
o +- OX bit numbers
o +- OH, OL bit numbers
07
OH
I
o +- CX bit numbers
o +- CH, CL bit numbers
07
CH
I
OL
I
The 8080A Accumu lator must be reproduced by AL since selected 8080A and 8086 instructions access this register
and none other.
BH and BL must reproduce the 8080A Hand L registers, since only BX can contribute to an 8086 data memory address.
On the surface this would appear to present a problem, since the 8080A has a limited number of instructions that use
the BC and DE registers to provide 16-bit memory addresses. When 8080A source programs are reassembled to execute on an 8086 microprocessor, 8080A instructions that seek memory addresses out of the BC or OE registers
become 8086 instructions that use Index registers.
5-5
15
7
A
07
.------
o
o
+- One 16-bit register
+- Two 8-bit registers
AX (= AH, AL) Primary Accumulator(s)
H,L
BX (= BH, BL) Accumulator(s). and Base register
B,C
CX (= CH, eL) Accumulator(s) and Counter register
0, F
OX (= DH, DL) Accumulatorisl and I/O Data Counter
~ These names apply to 16-bit registers
' - - - - - - - 4 t - - - - - - - These names apply to 8-bit registers
15
Sf'
o
+- Bit number
Stack Pointer (SP)
Base Pointer (BP)
}
r-.::c._ _ _ _ _ _ _ _....,0 +- Bit number
I ndex registers
Source Index (Si)
:=================:IDestination Index
(01)
r-:~~~~-""~~~"" +- Bit number
PC '--'-'-'-'-'-'-'-'-'-'-'-'-'-'-'-'-""""1 Program Counter (PC)
15
8
15
0 +- Bit n·umber
Code Segment (CS)
}
Data Segment (OS)
Stack Segment (SS)
Segment registers
Extra Segment (ES)
0 +- Bit number
Ir ~------"'I Status
Shaded registers are 8086 equivalents for 8080A registers.
8080A register names are shown in the left margin.
Figure 5-2. 8086 Programmable Registers
5-6
All 8086 memory addresses are computed by summing the contents of a Segment register
and an effective memory address. The effective memory address is computed via a variety of
addressing modes, as it would be for any other microprocessor. The selected Segment register
contents are left-shifted four bits, then added to the effective memory address to generate the actual address output as follows:
Segment Register contents:
Effective memory address:
Actual address output:
8086
SEGMENT
REGISTERS
xxxxxxxxxxxxxxxxoooo
+ OOOOyyyyyyyyyyyyyyyy
XXXZZZZZZZZZZZZZyyyy
x, Y and Z represent any binary digits.
Thus a 20-bit memory address is computed dressed directly.
which allows 1.048.576 bytes of external memory to be ad-
The Segment registers of the 8086 are unlike any other microprocessor registers described in this book. They act as
base registers which can point to any memory location that lies on an address boundary that is an even multiple of 16
bytes. Using arbitrary memory addresses, this may be illustrated as follows:
Memory
Address
334DF16
CS Segment register - - - - .
contains 234E'6
234E016
ES Segment register --+contains OA3216
OA320;6
OS Segment register - + ~:":"":":"":":""'""-:":""'""-'""-.0....1 021 FO 16
contains 021F'6
As illustrated above, each Segment register identifies the beginning of a 65,536-byte memory segment. Since the
8086 has four Segment registers, there will at any time be four selected 65,536-byte memory segments. The actual address output will always select a memory location within one of these four segments. For example, if an actual address
output is the sum of the OS Segment register and an effective memory address, then the actual address output must
select a memory location within the OS segment: that is to say, within the address range 021 F016 through 121 EF16 in
the illustration above. Likewise, an actual address output. which is the sum of the CS Segment register and an effective
memory address, must select a memory location within the CS segment. which in the illustration above will lie in the
address range 234E016 through 3340F16.
No restrictions are placed on the contents of Segment registers. Therefore, 8086 memory is not divided into 65,536byte pages, nor do the four Segment registers have to specify non-overlapping memory spaces. Each Segment register
identifies the origin of a 65,536-byte memory segment that may lie anywhere within addressable memory and mayor
may not overlap with one or more other segments.
Even though Segment registers can create overlapping or non-overlapping segments, they do have dedicated addressing functions. That is to say, different types of memory accesses compute memory address within specific segments.
5-7
During an instruction fetch, the Program Counter contents are added to the Code Segment
register (CS) contents in order to compute the memory address for the instruction to be fetched.
This may be illustrated as follows:
i5
7
o
o
07
8086 CODe
SEGMENT
REGISTER
AND PROGRAM
COUNTER
AX = AH+AL
BX = BH+BL
CX = CH+CL
OX = DH+OL
o
15
5P
BP
51
01
M M M M
PC
o
15
C5
N N
N N
05
--
55
E5
M, N, and P represent any hexadecimal digits.
5-8
OMMMM
NNNNO
PPPPM
Actual program memory
address output.
Any Stack instruction such as a Push. Pop. Call. or Return adds the Stack Pointer contents to
the Stack Segment register (55) contents in order to compute the address of the Stack location
to be accessed. This may be illustrated as follows:
15
7
AX
0
0
07
= AH+Al
BX
= BH+Bl
CX
= CH+Cl
OX
= OH+Ol
15
SP
0
M
M
M
M
N
N
N
N
BP
51
DI
PC
15
CS
OS
S5
OMMMM
E5
NNNNO
PPPPM
Once again. M. N. and P represent any hexadecimal digits.
5-9
Actual Stack operation
address output.
8086 STACK
SEGMENT
AND STACK
POINTER
REGISTERS
Instructions that process data strings use the SI and 01 Index registers. together with the
Data Segment register (OS) and the Extra Segment register (ES), in order to identify string
source and destination addresses. This may be illustrated as follows:
15
7
AX
= AH+AL
BX
= BH+BL
CX
= CH+CL
DX
= DH+DL
o
o
07
8086 EXTRA
SEGMENT.
SOURCE INDEX
AND
DESTINATION
INDEX
REGISTERS
o
15
SP
BP
SI
J
J
J
J
DI
K
K
K
K
~
..-
PC
Destination string address.
OKKKK
NNNNO
RRRRK .......- -
15
0
CS
DS
j
M M M M
SS
~
ES
\
N
N N
N
l/
Actual data string
address output.
Source string address.
-
OJJJJ
MMMMO
SSSSJ .........- -
Actual data string
address output.
J. K. M. R. and S all represent any hexadecimal digits.
As the above illustration would imply. instructions that process strings require that the source and destination strings
reside within a single 65.536-byte address range but not necessarily the same 65.536-byte range.
5-10
Instructions that access data memory add an effective memory address to the Data Segment register (OS) or the Stack Segment register (55). This may be illustrated as follows:
Program
Memory.
as addressed
by PC and CS
15
7
8086 DATA
SEGMENT
AND STACK
SEGMENT
REGISTERS
o
07
o
AX =AH+Al
BX = BH+Bl
~
CX = CH+Cl
OX = OH+Ol
15
0
SP
i, -
BP
SI
01
PC
OXXXX
y y Y Y0
ZZZZ
15
.....- - - Effective address segment base.
X"
Actual data address output.
o
CS
OS
~------------~
SS
ES
x, y, and Z represent any hexadecimal digits.
When a data memory address is created, as illustrated above, the BX, BP, SI. and DI registers' contents, plus a displacement coming from the instruction object code, may contribute to the effective memory address. There are, however,
very specific register and displacement combinations that can create an active memory address, as summarized in Table 5-1. Each case specifies either the DS or SS register as the default source for the segment base address.
5-11
Table 5-1. A Summary of Intel 8086 Memory Addressing Options
Possible Displacements
Memory
Reference
Segment
Register
Base
Register
Index
Register
16-Bit
Unsigned
8-Bit.
High-Order Bit
Extended
None
51
X
X
X
None
OS
(Alternate":
CS, 55 or ES)
BX
Normal Data
Memory
Reference
01
X
X
X
51
X
X
X
01
X
X
X
None
X
X
X
None
X
51
X
X
X
01
X
X
X
None
X
X
OS
None
55
(Alternate" :
CS, OS, or ES)
BP
Stack
55
SP
None
String
Data
OS
None
51
ES
None
01
Instruction
Fetch
CS
PC
None
Branch
CS
PC
None
I/O Data
OS
OX
None
X
" The segment override allows OS or 55 to be replaced by one of the other segment registers
X These are displacements that can be used to compute memory addresses.
When creating any data memory address. you can add a prefix to an instruction to select a Segment register other than
the default Segment register. You can only select a Segment register other than the default Segment register when addressing data memory. You must live with the default Segment register when creating program memory addresses.
Stack addresses. or string instruction addresses.
It is very important to note that the 8086 has a whole set of data memory addressing options aimed at accessing the Stack as though it were a data area. That is to say. in addition to the normal "Push" and "Pop" type Stack instructions. the 8086 allows normal data memory access instructions to address the Stack. Many assembly language
programmers use the Stack to store addresses. and as a general depository for data that must be transmitted between
program modules. Anyone favoring this assembly language programming philosophy will be delighted with 8086 data
memory addressing options.
5-12
Let us now examine the various data memory addressing options in detail. Refer to Table 5-1.
In the simplest case. we have straightforward direct memory addressing. A 16":bit displacement provided by two instruction object code bytes is added to the Data Segment register in order
to create the actual memory address. This may be illustrated as follows:
15
7
AX
= AH+AL
r-----.
...
8086 DIRECT
MEMORY
ADDRESSING
o
o
07
1 - - - - - + - -........
BX = BH+BL
CX = CH+CL
OX = OH+OL
o
15
5P
BP
1----------1
51
,0 M M M M
01
PC
1---------1
M
M
M
,NNNNO
M
PPPPM
.\
15
C5
N
N
N
N
OS
R
R
R
R
55
E5
o
HHLL"'-{
RRRRO
Actual data memory address output for
direct memory addressing.
- -......
~.. 5 5 5 5 L
Program
Memory
PPPPM
L
L
PPPPM+1
H
H
PPPPM+2
H. L. M. N. P. R. and S all represent any hexadecimal digits.
Note that a 16-bit address displacement. when stored in program memory. has the low-order byte preceding the highorder byte. This is consistent with the way the 8080A stores addresses in program memory.
DS must provide the Segment base address when addressing data memory directly. as illustrated above.
5-13
Direct, indexed addressing is also provided. The SI or 01 register may be selected as the Index
register. You have the option of adding a displacement to the contents of the selected Index
register in order to generate the effective address. If you do not add a displacement. then you
have, in effect, implied memory addressing via the SI or DI register. This may be illustrated as
follows:
o
15
7
8086 IMPLIED
MEMORY
ADDRESSING
o
07
AX = AH+AL
BX = BH+BL
CX = CH+CL
OX = OH+OL
o
15
5P
BP
51
01
PC
15
C5
05
R
R
R
R
55
--
OXXXX
RRRRO
5 S S 5 X 4 - - Actual data memory address output for
implied memory addressing
E5
(You may substitute CS, S5 or E5 for 05 by executing a 1-byte instruction prefix.)
X, R. and S represent any
hexad~cimal
digits.
If a displacement is added to the contents of the selected Index register, then you may
8086 DIRECT.
specify an 8-bit displacement or a 16-bit displacement. A 16-bit displacement is stored in two
INDEXED
object code bytes; the low-order byte of the displacement precedes the high-order byte of the disADDRESSING
placement as illustrated for direct memory addressing. Wan a-bit displacement is specified. then
the high-order bit of the low-order byte is propagated into the high-order byte to create a 16-bit displacement. This
may be illustrated as follows:
Displacements:
Sign extended:
5-14
We may now illustrate direct, indexed addressing as follows:
o
o
15
7
AX = AH+AL
07
t----+----t
BX = BH+BL
CX = CH+CL
t----+---.........
OX = OH+OL
15
0
SP
BP
SI
01
PC
M
M
M
M
OMMMM
NNNNO
P P PPM
15
CS
N
N
N
N
OS
R
R
R
R
SS
ES
Program
MemorY
C
Oyyyy_{
OXXXX
PPPPM
PPPPM+1
PPPPM+2
PPPPM+3
Actual data memory address output for - - - . . Z Z Z Z Z
direct, indexed memory addressing.
(You may substitute CS, SS or ES for OS by executing a 1-byte instruction prefix.)
M. N, p, R, X, y, and Z all represent any hexadecimal digits.
YYYY is the 16-bit or 8-bit displacement taken from program memory.
XXXX is the index taken from either the 01 or the SI register.
The effective memory address can be computed using base relative addressing. You have
two sets of base relative addressing ~ptions:
1)
Data memory base relative addressing, which is within the DS segment (data memory).
2)
Stack base relative addressing, which is in the SS segment (Stack memory).
5-15
'8086 BASE
RELATIVE,
INDEXED
ADDRESSING
Data memory base relative addressing uses the BX register contents to provide the base
for the effective address. All of the data memory addressing options thus far described
are available with base relative data memory addressing. In effect, base relative data
memory addressing merely adds the contents of the BX register to the effective memory
address which could otherwise have been generated. Here. for example. is an illustration of
base relative direct addressing:
8086 DATA
MEMORY BASE
RELATIVE
ADDRESSING
o
15
7
o
07
AX = AH+AL
K
BX = BH+BL
K
K
K
~
CX = CH+CL
OX = OH+OL
0
15
SP
BP
SI
01
PC
M
M
M
M
--
OMMMM
~
PPPPM
. ~ NNNNO
0
15
CS
N
N
N
N
OS
R
R
R
R
SS
ES
Program
Memory
~I-------I
~OKKKK
V
~---I
OHHLL
R R R R 0 . . . -{
PPPPM
PPPPM+1
PPPPM+2
Actual data memory address output for
base relative. direct. indexed memory - - . S S S S S
addressing.
(You may substitute CS. ES or SS for OS by executing a 1-byte instruction prefix.)
Simple. direct addressing. which we described earlier. always generated a l6-bit displacement. Base relative. direct addressing allows the displacement. illustrated above as HHLL. to be a l6-bit displacement. an 8-bit displacement with
sign extended. or no displacement at all.
5-16
Base relative implied data memory addressing simply adds the contents of the BX register to the selected Index register
in order to compute the effective memory address. This may be illustrated as follows:
15
7
AX
= AH+AL
= BH+BL
CX = CH+CL
BX
OX
0
0
07
K
K
K
K
= OH+OL
15
SP
BP
SI
01
PC
15
OKKKK
CS
OS
SS
R R
R R
..
OXXXX
RRRRO
SSSSS .....
__- - Actual data memory address output for
base relative, implied memory addre'ssing.
ES
(You may substitute CS, SS or ES for OS by executing a 1-byte instruction prefix.)
5-17
Base relative. direct. indexed data memory addressing may appear to be complicated. but in fact it is not. We simply
add the contents of the BX register to the effective memory address. as computed for normal direct. indexed addressing. Thus. base relative. direct. indexed data memory addressing may be illustrated as follows:
15
7
BX
= BH+BL
CX
= CH+CL
OX
= OH+OL
0
0
07
K
K
K
K
15
SP
BP
SI
01
PC
M
M
M
M
OMMMM
NNNNO
PPPPM
15
CS
N
N
N
N
OS
R
R
R
R
SS
ES
OKKKK
\
Program
Memory
{
PPPPM
PPPPM+1
OYYYY~
OXXXX
Actual data memory address output for
base relative. direct. indexed memory ~ Z Z ZZ Z
addressing.
(You may substitute CS. S5 or ES for OS by executing a 1-byte instruction prefix.)
5-18
PPPPM+2
PPPPM+3
The 8086 also has Stack memory addressing variations of the base relative, data memory addressing options just described. Here, for example, is base relative, direct Stack memory addressing:
15
7
AX
= AH+AL
BX
= BH+BL
CX
= CH+CL
OX
= OH+OL
0
0
07
15
0
SP
BP
K
K
K
K
M
M
M
M
SI
01
PC
I---+-~
0 MM MM
NNNNO
P P PPM
~
15
CS
N
N
N
N
OS
SS
R
R
R
R
Program
Memory
1-------1
PPPPM
OKKKK
{
PPPPM+1
o H H L L . - - 1-----1
R R R R0
PPPPM+2
ES
Actual Stack memory address output for
~~SSSSS
base relative, direct memory addressing. - -..
(You may substitute CS, ES or SS for OS by executing a 1-byte instruction prefix.!
In the illustration above, the displacement HHLL must be present either as a 16-bit displacement or as an 8-bit displacement with sign extended. Remember, base relative, direct data memory addressing also allows no displacement.
However, base relative, direct Stack memory addressing requires a displacement. These options are summarized in Table 5-1.
5-19
Here is an illustration of base relative, implied Stack memory addressing:
0
0
15
07
7
AX = AH+AL
BX
= BH+BL
ex =
OX
CH+CL
= OH+OL
0
15
SP
BP
K
K
K
K
R
R
R
R
SI
01
PC
15
CS
OS
SS
RRRRO
SSSSS
ES
'4
Actual Stack memory address output for
base relative, implied memory addressing.
(You may substitute CS, OS or ES for SS by executing a 1-byte instruction prefix.)
X, R, and S represent any hexadecimal digits.
5-20
Here is an illustration of base relative. direct. indexed Stack memory addressing:
15
7
AX
= AH+AL
BX
= BH+BL
CX
= CH+CL
OX
= OH+OL
0
0
07
15
0
SP
BP
K
K
K
K
51
01
PC
M M M M 1---+---+-11"
-------~
NNNN0
PPPPM
15
CS
0 MMMM
N
N
N
N
Program
Memory
t-------t
OS
55
A
A
A
\
A
PPPPM
t----tPPPPM + 1
ES
~
{
PPPPM + 2
1------41
Actual Stack memory address output for
base relative, direct, indexed memory addressing.
PPPPM + 3
(You may substitute CS, OS or ES for 55 by executing a 1-byte instruction prefix.!
There is one anomolous 8086 addressing mode that can cause confusion. One variation of I/O in8086 1/0
structions addresses an 1/0 port via the OX register. The OX register contents are output on
PORT
the Address Bus, to be interpreted as an liD port address. This means you can have up to 65.536
ADDRESSING
I/O port addresses. Since the DX register contents are being output as an I/O port address. it is not
added to any Segment register contents. Thus. the DX register outputs an address in the range 000016 through
FFFF16. This is the only case in which a register's contents are output directly as an address on the Address Bus. without first passing through segmentation logic.
All 8086 Branch-on-Condition instructions use program relative addressing. This feature
allows dynamically relocatable code. The Branch-on-Condition instruction provides an 8-bit.
signed binary displacement that is added to the contents of the Program Counter. Thus. Branchon-Condition instructions have an addressing range of -128 through + 127 bytes from the location of the Branch-on-Condition. The queuing of instruction object codes has no impact on
Branch-an-Condition logic, or the branch addressing range.
5-21
8086
PROGRAM
RELATIVE
ADDRESSING
8086 Jump and Subroutine Call instructions offer these addressing options:
1)
Program relative addressing. An 8-bit or 16-bit displacement is added to the contents of the Program Counter.
2)
Direct addressing. New 16-bit addresses provided by the instruction are loaded into the Program Counter and the
CS Segment register.
3)
Indirect addressing. Any of the data memory addressing options may be used to read data
8086
from data memory. However. the data input is interpreted as a memory address. You have two
INDIRECT
indirect addressing options. A single 16-bit data word may be read. in which case it is loaded
ADDRESSING
into the Program Counter and the Jump or Call references a memory location within the current CS segment. You can also read two 16-bit data words; the first is loaded into the Program Counter and the
second is loaded into the CS Segment register. Thus you can Jump or Call indirectly any addressable memory location.
8086 STATUS
The 8086 has a 16-bit Stack register with the following satus bit assignments:
15141312 11 10 9
1-1-1-1-1
I I I I
8
7
0 I D II I Tis I
6
5
4
3
2
1 0 +- Bit No.
z I-I A I-I pi-I C I
I
I
I t
L
Status register
Reserved bits. normally 0
Carry
Parity
Auxiliary Carry
Zero
Sign
Trap
Interrupt enable/disable
Direction
Overflow
The Carry, Auxiliary Carry, Overflow, and Sign statuses are quite standard; see Volume 1 for a description of these
statuses. The Auxiliary Carry status is identical to the 8080A status with the same name. It represents carries out of bit
3 in an 8-bit data unit as described in Volume 1. Chapter 2.
Subtract instructions use twos complement arithmetic in order to subtract the minuend from the subtrahend.
However, the Carry status is inverted. That is to say, following a subtract operation, the Carry status is set to 1
if there was no carry out of the high-order bit, and the Carry status is reset to 0 if there was a carry out of the
high-order bit. The Carry Status therefore indicates a borrow.
The Parity status is set to 1 when there is an even number of 1 bits in the result of a data operation; an odd number of
1 bits causes the Parity status to be reset to O.
The Zero status is completely standard. It is set to 1 when the resu It of a data operation is zero; it is set to 0 when the
result of a data operation is not zero.
The Direction status determines whether string operations will auto-increment or auto-decrement the contents
of I ndex registers. If the Direction status is 1. then the SI and Dllndex registers' contents will be decremented; that is to
say. strings will be accessed from the highest memory address down to the lowest memory address. If the Direction
status is O. th.en the SI and Dllndex register contents will be incremented; that is to say. strings will be accessed beginning with the lowest memory address.
The Interrupt status is a master interrupt enable/disable. This status must be 1 in order to enable interrupts within
the 8086. If this status is O. then all interrupts except the NMI (Non-Maskable Interrupt) will be disabled.
The Trap status is a special debugging aid that puts the 8086 into a "single step" mode. The single step mode is described in detail together with 8086 interrupt logic. since it depends on this interrupt logic for its existence.
The Carry, Auxiliary Carry, Parity, Sign, and Zero statuses are also found in the 8080A. The Overflow, Direction,
Interrupt, and Trap statuses are new in the 8086.
5-22
8086 CPU PINS AND SIGNALS
8086 CPU pins and signals are illustrated in Figure 5-3.
GND
AD14
AD13
AD12
AD11
AD10
AD9
AD8
AD7
AD6
AD5
AD4
AD3
AD2
AD1
ADO
NMI
INTR
ClK
GND
·..
- -
-- ...
---
·..
--... ..·
-- ·..
-
.-..
8086
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
..
----.
·
-
---
VCC
AD15
A16/S3
A17/S4
A18/S5
A19/S6
B'FfE"/S7
MN!MX
RD
RO/GTO, HOLD
'FfCi/GTf, HLDA
LOCK, WR
52, M/iO
---- ---..
..
·
--·.. Sf, DT/R'
--·
-
----
SO, DEN
OSO, ALE
OS1,INTA
TEST
READY
RESET
Pin Name
Description
Type
Data/Address Bus
Address/Segment Identifier
Address/Interrupt Enable Status
Address/Status
High-order Byte/Status
Read Control
Wait State Request
Wait for Test Control
Interrupt Request
Non-maskable Interrupt Request
System Reset
System Clock
= GND for a Maximum System
Machine Cycle Status
Local Bus Priority Control
Instruction Oueue Status
Bus Hold Control
= VCC for a Minimum System
Memory or I/O Access
Write Control
Address latch Enable
Data Transmit/Receive
Data Enable
Interrupt Acknowledge
Hold Request
Holr;! Acknowledge
Power, Ground
Bidirectional, Tristate
Output, Tristate
Output, Tristate
Output, Tristate
Output. Tristate
Output, Tristate
Input
Input
Input
Input
Input
Input
READY
TEST
INTR
NMI
RESET
ClK
MN/MX
SO, 5152
Jm/G'rn, m/GTi
OSO, aS1
IOCK
Minimum
System
Signals
- -
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
ADO-AD15
A 16/S3, A17/S4
A18/S5
A19/S6
BHE/S7
Ri5
Maximum {
System
Signals
-- -..
-- ·.......
MN/MX
M/IO
WR
ALE
DTiR
om
INTA
HOLD
HlDA
VCC,GND
Output. Tristate
Bidirectional
Output, Tristate
Output, Tristate
Output,
.Output,
Output.
Output,
Output,
Output
Input
Output
Tristate
Tristate
Tristate
Tristate
Tristate
Figure 5-3. 8086 Pins and Signal Assignments
5-23
The 8086 outputs a 20-bit memory address. Data is accessed as 16-bit words. subdivided into a low-order byte and a
high-order byte. Therefore the 8086 needs a 20-line Address Bus and a 16-line Data Bus. In order to have a 40-pin
package. the low-order 16 Address Bus lines are multiplexed with the Data Bus.
i'H'E may be looked upon
as an additional Address Bus line, since it is used to identify the high-order byte of a
memory word. while ADO identifies the low-order byte of the memory word.
The four high-order Address Bus lines. together with 8HE. are multiplexed with five status I.ines. thus. we can illustrate
Address Bus line multiplexing as follows:
Data/Status
Address
Status
16-bit data word
S7
BRt High-order byte of selected word
DO
AO Low-order byte of selected word
{ D1 .... 015
Status
S3 - S6
A1 - A15 }
A16 - A19
Address of a 16-bit word
8086
EXTERNAL
MEMORY
ADDRESSING
It is easy to become confused when looking at how the Address Bus, together with BHE, is used
to access memory. As seen by external memory. Address Bus lines are interpreted as follows:
Select
Memory
select
logic
....
AO
ADO
AD1
AD7
AD8
j~
•
07
...
,
.=.
-.
_.
::: ·
-.
_.·
..
:. ADO
AD1
··-. ..
~
-._.·
SHE -
AD15
DO
A18
...
•
Low-order
byte
memory
bank
~
j
j
A19
~
AO
Select
..-
...
...
~
~
DO
A18
Memory
select
logic
..
-
5-24
·.-.
··:
~
~~
07
High-order
byte
memory
bank
AD7
AD8
AD15
A19
'8HE
In the previous illustration you will see that memory is indeed organized as bytes.
The data pins of the low-order byte memory bank connect to ADO-AD7. The high-order byte memory bank data pins
connect to AD8-AD15.
The low-order and high-order byte memory banks each have memory select logic that decodes AD1-A 19. These 19 address lines become inputs AO-A 18 at the illustrated memory select logic. Since each memory bank receives 19 address
lines. select logic can address up to 524.288 (512K) bytes of memory. These two memory banks. taken together. constitute the advertised one million bytes of directly add ressable memory.
Now. you may well ask why one should bother dividing memory into separate low-order and high-order byte banks. If a
sixteen-bit word lies on an even-byte address boundary. then we could ignore the memory select logic connections to
ADO and BHE. The address on AD 1-A 19 becomes an address identifying a 16-bit word. which just happens to be implemented as two separate 8-bit memory banks.
If an 8086 16-bit memory word does lie on an even-byte address boundary. then the low-order byte address is. in fact.
the only address output. BHE is pulsed low while the low-order byte address is being output. and both memory banks
consider themselves selected even though (in theory) the high-order memory bank's address has not been output.
To illustrate what happens. consider the memory addresses 02A4016 and 02A4116. One would normally expect the
two addresses to be output sequentially in order to access the low-order byte and then the high-order byte of the 16-bit
word. This may be illustrated as follows:
ADO-AD7
AD8-AD15
AD16-AD19
Input
data
from
byte
Output
address
02A40 16
02A4016
Output
address
02A41 16
Input
data
from
byte
02A41 16
But we could just as easily output the low-order byte address only. using BHE as an extra address line to substitute for
the odd-byte address - which is never output. This may be illustrated as follows:
ADO-AD7
AD8-AD15
AD16-AD19
Output
address
02A4016
Assume
address
02A4116
has been
output
5-25
If a word lies on an odd-byte address boundary, then two byte addresses must be output to access the two
halves of the 16-bit word. This may be illustrated as follows:
First memory
access is to
a byte in
the highorder byte
memory
bank. i.e.
an odd byte
address. with
ADO high
Return lowcrder byte
of 16-bit
word via
AD8-AD15
Second
memory
access is to a
byte in the
low-order
byte memory
bank. i.e.
an even byte
address. with
ADO low
Return highorder byte
of 16-bit
word via
ADO-AD7
When a 16-bit word lies on an odd-byte address boundary. as illustrated above. the low-order byte is input first via
AD8-AD15. then the high-order byte is input via ADO-AD7. Logic internal to the 8086 switches the data bytes into
their correction locations.
Intel could have elected to implement external memory as 16-bit words, which would eliminate BHE along with
the Address Bus complexities we have just described. But this would have forced all instruction object codes,
and data, to be accessed as 16-bit units. Why not do it?
One of the most interesting hindsight discoveries that 8080A users have made is the fact that the 8080A is extremely
efficient in its use of memory. By having a large number of 8-bit object codes. the 8080A generates object programs as
compact as the most powerful minicomputers on the market.
But if the 8086 is to keep 8-bit object codes. and therefore the efficient memory utilization of the 8080A. then it can no
longer guarantee that data will lie on even-byte address boundaries. The first 8-bit object code will force the next instruction or data entity to begin on an odd-byte boundary.
By including BHE and the extra logic needed to access 16-bit data units origined at odd-byte boundaries. the 8086 has
allowed instructions to generate 1-byte. 3-byte or other odd-byte object codes. rather than 2-byte. 4-byte. and evenbyte object codes only.
Simply stated, this is the trade-off: simplify memory addressing so that external memory is accessed only as
16-bit data units and you will use memory less efficiently. Intel elected to make memory addressing logic more
complex and memory utilization more efficient.
Moving on from the Data/Address Bus, 8086 signals may be grouped into those that do not change with system
complexity, and those that do. let us first look at the unchanging signals.
elK is the single clock signal output by the 8284 clock generator to synchronize all 8086 logic.
READY is the Wait state request which slow external logic inputs if it requires more time to respond to an access. A
high READY input occurring at the proper time early in a machine cycle causes the 8086 to extend the machine cycle
by inserting Wait state clock periods.
5-26
RD is a single
bus control signal that does not change with system configuration. This signal is output low when the
CPU is inputting data from any external source.
Even though RD is output by the same physical pin under all circumstances. this signal is functionally part of the group
that changes its nature depending on signal complexity. We will therefore refer again to RD when describing the signals that are a function of system complexity.
There are four interrupt and interrupt-related signals.
INTR is a normal interrupt request input.
NMI is a non-maskable interrupt request input.
RESET is a system reset signal; it must be input high to the 8284 clock generator for at least
four ClK clock periods. The 8284 transmits a synchronized RESET signal to the CPU. When
the 8086 is reset, the following events occur:
1)
The Status register is cleared. This disables external interrupts.
2)
The Program Counter and the three Segment registers. DS. SS. and ES. are cleared.
3)
The CS Segment register is set to FFFF16. Following a Reset. program execution therefore restarts with the instruction located at memory byte FFFF016·
These reset operations take approximately 10 clock periods to occur should occur.
during which time no other operations
Following power-up. at least 60 microseconds shou Id elapse before the 8086 is reset.
An interrupt request via INTR should not occur sooner than 9 clock periods after the end of the 8086 device reset. An
earlier interrupt request will cause one entire instruction to be executed before the interrupt request is recognized or
acknowledged.
A nonmaskable interrupt request should not be made during the first clock period following the end of a reset.
TEST is not really an interrupt input. but it is used by program logic that otherwise would rely upon an interrupt. The
8086 has a special "Wait-for-Test" instruction that puts the CPU into an Idle state; this Idle state ends when
the TEST input goes low.
An 8080A (and other microprocessors) will duplicate the logic of the 8086 "Wait-for-Test" instruction by executing a
"no operation" loop. which is terminated by an interrupt request:
SELF
ENI
Enable interrupts
JMP SELF
Only an interrupt will terminate loop execution
5-27
There are eight pins that can output one of two signals. depending on whether MN/Mx is
tied to power or ground. By having two sets of signals. the 8086 can be used in simple configurations. best served by elementary control signals. or in complex configurations. where control
signals must provide sufficient information to resolve the contentions and access conflicts that
complex microcomputer systems may encounter.
8086
DUAL BUS
COMPLEXITY
The two sets of signals may be illustrated as follows:
Minimum
Maximum
Systems
Systems
MN/MX = Vee
MN/MX
= GND
M/TIS
DT/R - - - DEN
S2
51
SO
H·
0 [.·.;.g.·.P
. . . . . .•. .•. :. . 0.·. •. . ;.:.:.·:•·. .• .1
O
' :0;:;::":)::1':::
1.. :il:>
0
1
I
I
I
N
0
0
T
R
IN
A
INTA
OS1
0
ALE
OSO
0
0
1
0
1
N
a
Q
Q
0
0
P
8
E
B
WR
HOLD
HLDA
1
S
LOCK
RO/GTO
RO/GTl
Let us first look at the simple set of control signals that are output when MN/MX is connected to +6 V. These are completely standard microprocessor control signals.
8086 SIMPLE
CONTROL
Since data and addresses are multiplexed on a single bus. ALE is output high to identify a valid
memory address.
SIGNALS
When data are being transmitted or received via the Data/Address Bus. WR is pulsed low to identify data output.
while RD is pulsed low as a request for external logic to place data on the Data/Address Bus. We have already
described RD. It is not one of the changing signals: nevertheless. it is used by both simple and complex system busses.
For a read or write operation. M/IO indicates whether memory (M/IO high) or an I/O port (M/IO low) is being accessed.
DTIR and DEN are two new control Signals not found in earlier Intel microprocessors. These two control signals have
been designed specifically to control 8286/8287-type bidirectional transceivers. DTIff identifies the data direction.
while DEN is the output enable. The 8286 and 8287 transceivers are described later in this chapter.
HOLD and HLDA are standard hold request!8cknowledge signals. When external logic inputs HOLD high. the 8086
CPU enters a Hold state upon completing the current instruction's execution: the 8086 acknowledges the Hold State
by outputting HLDA high. We will describe the Hold state in more detail later in the chapter.
Let us now look at the complex System Bus that is generated when MN/MX is tied to
ground. Control signals are output as a three-signal combination. decoded by a 3-to-8 decoder.
and a two-signal combination. decoded by a 2-to-4 decoder. Complex System Bus signals have
been designed to act as inputs to an 8288 Bus Controller.
8086 COMPLEX
CONTROL
SIGNALS
S2. ST. and SO are decoded to provide eight separate control signals. However. the simple system signals M/IO. DT/R
and DEN represent a subset of the eight S2. ST. and SO combinations. In our earlier illustration. we identify this simple
system subset by shading the applicable complex system S2. ST. and SO levels.
5-28
The eight combinations of 52. 51. and SO generate the following control signals:
S2
51
so
0
0
0
0
1
1
1
1
0
0
1
1
0
0
1
1
0
1
0
1
0
1
0
1
INTA
lOR
lOW
HALT
IFETCH
MEMR
MEMW
NONE
Interrupt acknowledge
I/O device read
I/O device write
CPU has executed a HALT instruction and is in the Halt state
The CPU is fetching an instruction object code byte
Memory read
Memory write
The System Bus is inactive
The control signal descriptions above use the words "read" and "write" as seen by the CPU. That is to say, a .. read"
operation moves data from a memory device or 1/0 port to the CPU, while a "write" operation moves data from the CPU
to a memory location or I/O port.
050 and 051 combine to identify conditions within the 8086 instruction object code queue soon. The QSO and QS1 combinations are interpreted as follows:
QSO
051
o
o
o
1
1
1
o
1
NOOP
OB1
OE
OBS
which we will describe
No operation. This is the default case
The first instruction object code in the queue is being executed
The queue is empty
An instruction object code other than the first one in the queue is being executed
Observe that the simple bus signals INTA and ALE do not correspond to any combination of OSO and 051. This is in
contrast to MilO, DT IR and DEN, which constitute a subset of 52, ST, and SO.
LOCK, RO/GTO, and RO/GT1 are not related to their simple system equivalent signals: WR. HOLD, and HLDA. LOCK.
iiO/GTO. and RQ/GT1 provide the 8086 with its System Bus priority and control logic in complex configurations.
LOCK is output high to prevent the 8086 from losing bus control while executing a sequence of machine cycles that
must not be interrupted. Typically these will be a memory access combination of read-modify-write machine cycles.
where an error could result if the CPU lost bus control after the read and before the write.
RO/GW and RQ/GT1 are two-bus priority, bidirectional type signals. They are used to determine which CPU in a multiCPU configuration will at any time have control of a shared bus. We will discuss these signals in more detail later in the
chapter when looking at the capabilities of the 8086 in mUlti-CPU shared bus configurations.
5-29
8086 TIMING AND INSTRUCTION EXECUTION
The most important concept to understand when looking at 8086 instruction execution tim8086
ing is the fact the 8086 bus control logic has been separated from the 8086 instruction exEXECUTION
ecution logic. That is to say, the 8086 has an Execution Unit (EU), and a Bus Interface Unit
UNIT (EU)
(BIU).
- - - - -..
The Execution Unit (EU) contains Data and Address registers, the Arithmetic and Logic Unit,
plus the Control Unit. The Bus Interface Unit (BIU) contains bus interface logic, Segment
registers, memory addressing logic, and a six-byte instruction object code queue. This may
be illustrated as follows:
8086 BUS
INTERFACE
UNIT (BIU)
---,
I
AH
AL
I
BH
BL
CS
0000
CH
CL
I
I
OS
0000
OH
OL
I
55
0000
I
I
I
I
I
ES
0000
SP
BP
51
01
~
n
I
:1
I
I
Arithmetic
and Logic
Unit (ALU)
--=>
I
I
Control Unit
(CU)
I
I
Status
--
I
Instruction Reg. ~
~
I
I
I
L __________ --L
1
PC
I
I
1
I
I
.oj
I
I
,
18086
).
.oj
Bus
Control
Logic
I Bus
~
II'
I
1
2
3
4
5
6
Instruction
object code
queue
I
I
I
I
I
I
I
1
__ ....JI
The Execution Unit (EU) and the Bus Interface Unit (BIU) operate asynchronously. Whenever
8086
the Execution Unit is ready to execute a new instruction. it fetches the instruction object code
INSTRUCTION
from the front of the Bus Interface Unit instruction queue. then it executes the instruction in some
QUEUE
number of clock periods that have nothing to do with machine cycles. If the instruction object
code queue is empty. then the Bus Interface Unit (BIU) executes an instruction fetch machine cycle - and the CPU
waits for the instruction object code to be fetched. But the queue will rarely be empty. for reasons that will soon
become apparent; therefore. the EU will usually not have to wait while an instruction fetch is executed.
5-30
If memory or an I/O device must be accessed in the course of executing an instruction, then the EU informs the BIU of
its needs. The BIU executes an appropriate external access machine cycle in response to the EU demand .
_-...
..--_.......
The Bus Interface Unit (BIU). for its part. is independent of the Execution Unit (EU). and attempts to
8086
keep the six-byte queue filled with instruction object codes. If two or more of these six bytes are
INSTRUCTION
empty, then the Bus Interface Unit (BIU) executes instruction fetch machine cycles - providing
QUEUE
the EU does not have an active request for bus access pending. If the EU issues a request for bus - - - - - -.....
access while the BIU is in the middle of an instruction fetch machine cycle, then the BIU will complete the instruction
fetch machine cycle before honoring the EU bus access request.
8086 BUS CYCLES
If we look at the way clock logic is used by the 8086, the term "machine cycle" no longer applies. The EU does
not use machine cycles; it executes instructions in some number of clock periods that are not subject to any type of
machine cycle grouping. The only time clock periods are grouped is when the bus control logic wishes to access
memory or I/O devices. Each access requires four clock periods. This is the minimum amount of time required to handle the normal bus protocol that accompanies any transfer of information between a microprocessor and logic beyond
the microprocessor. Since this is the only time the 8086 groups clock periods, it is more accurate to talk about
8086 bus cycles, rather than machine cycles.
Figure 5-4 illustrates two 8086 bus cycles executed back-to-back. In common with machine cycles, 8086 bus
cycles, as illustrated in Figure 5-4 assign individual clock periods to time specific events.
Memory and I/O device addresses are output on the Data/Address Bus during T1.
Data is transferred between the 8086 and memory or I/O devices during T3 and T4. If these two clock periods provide
external logic with insufficient time to respond to an access, then Wait state clock periods (TW) may be inserted between T3 and T 4·
T2 is a buffer clock period during which the Data/Address Bus stops outputting an address and starts outputting or inputting data.
During T4 the CPU identifies the status of the next bus cycle or clock period. In simple configurations when MN/MX is
tied to +5 V, DT /A is the only external signal that changes state during T4. When MN/MX is tied to ground, SO, S 1, and
S2 change state during T4. Thus, by examining these three status outputs, external logic knows whether to expect
another bus cycle. and, if so, what type of bus cycle.
Now if you look at Figure 5-4, there is very little about it that differentiates an 8086 bus cycle from any other
microprocessor's machine cycle. The characteristic of the bus cycle that differentiates it from standard machine
cycles is the fact that bus cycles occur only on demand.
:,...
...
. . - - - - - BUS CYCLE
:
T1
T2
---------i~
....: ..
4 J - - - - - - - BUS CYCLE ------I~
. .,:
T3
V//////A
V///////1
~
~
T4
i
T1
T2
T3
T4
i
r//////A
r///////1
~
VlllmllllllJ/j
~
t'lll/7lllllJll
V/I/I/III//IIII
VI//I/I///////A
Output address
duringT1
Turn Bus around
duringT2
Perform memory
accesses during T3
+L----------------~' W a i t
state clock
•..__-----periods
In complex systems,
status output in T4
identifies subsequent
operations.
Figure 5-4. Two 8086 Bus Cycles
5-31
8086 INSTRUCTION QUEUE
Consider what happens when an instruction is executed. Beginning with the simplest case, the instruction object code queue within the Bus Interface Unit will be empty. When the EU requests an object code byte there is
none, so the BIU executes a bus cycle that fetches the first byte of the instruction object code:
Tl
CLK
I
I
I
I
~
Bus cycle fetches
first object code
byte
Let us assume that this particular instruction requires two bytes of object code: keeping things simple. we will illustrate
another instruction cycle executed immediately to fetch the next instruction byte:
i T3 IT4
Tl
CLK
Tl : T2 : T3 : T4
I
I
I
I
I
I
I
I
~~
Bus cycle fetches
first object code
byte
Bus cycle fetches
second object
code byte
Let us suppose that this instruction reads a word of data from memory. then performs an arithmetic operation using
this data. The instruction is going to require some number of clock periods to compute the effective address for the
data memory location to be accessed (we will assume seven clock periods are needed). Some additional number of
clock periods will also be needed to perform the arithmetic operation (we will assume nine clock periods). In a normal
microprocessor. this instruction might be executed as the following sequence of machine cycles:
Machine Cycle 1
Tl : T2
I
i T3
: T4
I
I
Machine Cycle 2
Tl
I T2
I T3 I T4
I
I
Machine Cycle 3
Tl I T2
I
1
T3 : T4
Machine Cycle 4
Tl: T2 : T3 : T4
I:
I
Machine Cycle 5
Tl: T2
I T3
: T4
I
1
CLK
Fetch first
object code
byte
Machine Cycle 6
Tl : T2
I
: T4
1
I
Fetch second
object code
byte
Compute data
memory
address
Long Machine Cycle 7
Tl
Compute data
memory
address
Machine Cycle 8
i T3
i T41 T5
1
I
Tl
1
ClK
Execute arithmetic operation
in a standard machine cycle
and a long machine cycle
Start executing
next instruction by
fetching object
code byte
I
5-32
I
Fetch data
from memory
But the 8086. having asynchronous CPU and Bus Control Unit logic. will use clock periods to execute the instruction illustrated above as follows:
Bus Cycle 1
Tl
I T2 : T3
I
Bus Cycle 2
Bus Cycle 3
Tl I T2 I T3 I T4
I
I
i
Tl
I T2
I T3 I T4
I
:
Bus Cycle 4
Tl : T2 : T3 : T4
I
I
Bus Cycle 4
T1
I
ClK
EU
EU asks for an
object code byte.
There is none. so
the BIU fetches
one.
The EU needs a
second object
code byte.
The EU computes a data memory
address in 7 clock periods. At the
end of the 7th clock period the
CPU requests bus access.
The EU waits for the
requested data to be
fetched by the BIU
BIU
BIU fetches a
byte of object
code in one
bus cycle.
BIU fetches a
second byte of
object code in
one bus cycle.
Since the EU is not demanding
bus access. the BIU fetches the
next two object code bytes and
stores them in the queue. At the
end of bus cycle 4 the EU is
requesting bus access. so the BIU
services the EU.
BIU fetches data
from memory
location addressed
by the CPU.
Bus Cycle 5
Tl
Bus Cycle 6
Tl
I T2 i T3
Bus Cycle 7
: T4
Tl: T2
I
I
I T3
: T4
I
ClK
I
Ti
EU
Ti
Ti
Ti
Ti
Ti : Ti
,
I
etc.
Ti: Ti
The EU uses nine clock periods to execute : The EU takes the
etc.
I next object code
the required arithmetic operation.
, byte from the
BI U queue and
, starts executing the
! next instruction.
I
BIU
The BIU continues executing bus cycles to
etc.
I fill the instruction object code queue.
5-33
Now. the illustration above is not accurate because. you will recall. the 8086 fetches data in 16-bit increments. provided the data address lies on an even-byte boundary. Also. the BIU fetches instruction bytes and loads them into the
queue only when there are at least two free bytes in the queue. Let us assume that all data does lie on even-byte boundaries. This is how our timing will now look:
Bus Cycle 1
Tl
Bus Cycle 2
Tl : T2 : T3
: T4
I
I
I
Bus Cycle 3
I T4
Bus Cycle 4
Tl : T2 : T3 : T4
Tl: T2
I
I
I
: T4
I
Bus Cycle 4
T1: T2
: T3
I
i T4
Ti
Ti
Ti
I
ClK
EU
BIU
EU asks for an
object code byte.
There are none.
so the BI U executes a bus cycle.
The EU computes a data memory
address in 7 clock periods. At the
end of the 7th clock period the
EU requests bus access.
BI U fetches two
bytes of object
code in one bus
cycle. The CPU
takes both of
them. so the
queue is immediately emptied.
Ti
Ti
I
I
Ti
The EU waits for the
requested data to be
fetched by the BI U.
BI U fetches four bytes of object code
in two bus cycles and stores them in
the queue. which has two empty
bytes left.
I
BI U fetches data
from memory
location addressed
by the EU.
I
The EU uses 9 clock periods to execute
the arithmetic operation.
The BIU fetches
two more bytes
of object code and
stores them in the
queue which is
now full.
I The BIU is idle.
Ti
I
ClK
etc.
I
I
EU
BI U
Ti
t
The EU ends instruction execution
and fetches one byte of object code
from queue to execute next instruction.
The BI U remains idle since only one
byte of queue is empty.
There are some important points to note regarding 8086 bus cycle timing.
Bus cycles are a Bus Interface Unit (BIU) phenomenon.
So far as the EU logic is concerned. bus cycles do not exist. The EU experiences periods of activity while executing instructions. and periods of inactivity while waiting for instruction object codes or data that the BIU must process via bus
cycles. Periods of EU activity are timed by a sequence of clock periods. The EU makes no attempt to group clock periods
into machine cycles. nor do EU clock periods have to occur in any special numeric combinations.
The EU asks for memory operands before it needs them. so unless the BIU cannot get immediate bus access the maximum EU wait time is one clock cycle for bus access.
So far as the BIU is concerned. clock periods are grouped into bus cycles only when data must be transferred to or from
the 8086. First priority is given to a bus access request coming from the EU. If the EU is not requesting bus access. then
the BIU executes instruction fetch bus cycles until the queue is full. These are the prerequisites for the BIU to execute an instruction fetch bus cycle:
1)
The clock period that initiates the bus cycle would otherwise be an idle clock period.
2)
The EU does not have an active bus access request pending.
3)
There are at least two bytes empty in the queue.
If the queue is full. then the BIU ceases to execute bus cycles; as illustrated above. a sequence of idle clock periods occurs.
5-34
Note that the CPU may have to wait for bus access. In the illustrations above, the EU requires seven clock periods in
order to compute a data memory address. At the end of the seventh clock period, the EU issues a bus access request to
the BIU. But at this time the BIU is part way through executing an instruction fetch bus cycle. The BIU completes the instruction fetch bus cycle, then honors the EU bus access request.
In the final illustration above, no bus cycle accompanies the beginning of a new instruction's execution. We are assuming that the next instruction executed has one byte of object code. This object code byte is fetched from the front of the
queue - which then has just one empty byte. No bus cycle is executed to fetch the instruction object code, since it is
taken out of the queue. Subsequently, the BIU does not execute an instruction fetch bus cycle since there is only one
empty byte; there must be at least two empty bytes in the queue before the BIU will execute an instruction fetch bus
cycle.
Based on the foregoing discussion of 8086 instruction fetch queuing, we can see that the 8086 has essentially
eliminated instruction fetch time. The only time the EU will have to wait while the BIU fetches instruction object
codes is when a Branch-on-Condition instruction causes execution to branch out of the queue sequence, or when (for
any reason) the memory accesses accompanying an instruction's execution are so dense that the BIU has insufficient
idle clock periods within which to insert instruction fetch bus cycles.
8086 MEMORY AND 1/0 DEVICE READ BUS CYCLE FOR MINIMUM MODE
Figure 5-5 shows timing for an 8086 memory read bus cycle when MNiMX equals +5 V; that is to say, for the
minimum mode bus configuration.
I~
One Bus Cycle
T1
T2
I
T3
T4
ClK
ADO-AD15
A16-A19
BHE
ALE
MilO
RD
DT/R
DEN
Trailing edge of
ALE latches address
Figure 5-5. 8086 Memory Read Bus Cycle for a Minimum Mode System (MN/MX
5-35
=
+5 V)
·1
The memory or I/O device address is output via the Address Bus BHE during clock period T 1. ADO-AD15 starts floating
in T2 while turning around internal pin logic so that data can be input during T3 and T4. Address lines A 16 through
A 19 are all low when an I/O device address is being output. These address lines output status during T2. T3. and T4.
Close to the end of T4. A 16 through A 19 start to float.
BHE timing follows Address lines A 16-A 19; that is to say. BHE is output low for the time that A 16 through A 19 is outputting an address.
The trailing edge of the high ALE pulse should be used as the "valid address" strobe. If your 8086 configuration
demultiplexes the Data and Address Busses. then the Address Bus demultiplexing buffers should be the "pass
through" type and use the high-to-Iow transition of ALE as their latching strobe.
Remaining control signals consist of M/i5 and RD. which are directed at external memory or 1/0 devices. plus DTIAand
DEN. which are directed at bus buffers.
M/iO differentiates between a memory access and an I/O device access. M/IO will be high for a memory access bus cycle; it will be low for an 1/0 device access bus cycle. M/i5 will contribute to memory and I/O device select logic when
memory and I/O devices have similar addresses.
RD is pulsed low as a memory or I/O device read strobe. The addressed memory device must use this low signal to
place data on ADO - AD 15.
DT /R and DEN are control signals designed to control bidirectional latched buffers on the Data Bus. DT IR is output low
for the entire memory or I/O device read bus cycle; it should be used to turn the latched buffers around so that they will
transmit data to the CPU. DEN subsequently acts as a latching strobe. These two signals have been designed
specifically to work with the 8286 and 8287 Data Bus transceivers; however. their logic is quite general.
There is no difference between external timing for an instruction fetch or memory read bus cycle. Given the
pipelining instruction fetch logic of the 8086. this makes sense.
The only timing difference between a memory read bus cycle and an I/O device input bus cycle occurs at the
M/ffi signal. This signal will be low for the duration of an I/O input bus cycle. whereas in Figure 5-5 it is shown high
for the duration of a memory read bus cycle.
Except for this difference, Figure 5-5 also illustrates I/O input bus cycle timing for a simple 8086 configuration.
During any simple configuration memory access operation, the following status is output on address lines A 16
through A 19:
A 19/56 A 18/S5 -
Always 0
Interrupt enable status
A17/S4- 0
A 16/S3 - 0
0
1
1
0
1I
t
1
L- Data segment access
cOde segment access or no access
Stack segment access
~------- Extra segment access
_____
L...
The interrupt enable status appearing on A 18 may be used to illuminate an indicator on a control panel. should there
be one. This indicator will show whether interrupts are enabled or disabled at any time. This status has no other value.
S3 and 54 together identify the memory segment which is being accessed. This is not very useful information.
Even a code segment access cannot be interpreted as an instruction fetch. since data can be addressed out of the program segment.
8086 MEMORY OR I/O DEVICE WRITE BUS CYCLE FOR MINIMUM MODE
Figure 5-6 illustrates timing for an 8086 memory or I/O device write bus cycle when the 8086 is operating in a
minimum mode with MN/MX tied to +5 V.
Address output logic is identical in read and write bus cycles. As was the case for a read bus cycle. the address is output on the Address Bus. together with BHE. during T 1. External logic should use the high-to-Iow transition of the ALE
pulse in order to latch a valid address. During T2. ADO - AD15 switches to outputting data. while A16 - A19 outputs
status. The same status is output in read and write bus cycles.
M/iO is output high for the duration of a memory write bus cycle; it is output low for the dUration of an I/O device write
bus cycle.
WR is output low beginning early in T2 and ending shortly after T3. Note that RD does not go low for a read bus cycle
until halfway through T2.
5-36
For an 8286 or 8287 Bus Transceiver, or any similar device, DT/A" is output high for the entire duration of the write bus
cycle. This conditions the device to transmit data from the CPU to external logic. DEN is the chip enable signal provided
for the bus transceiver. DEN is output high from the end of T1 until the end of T4. Note that this high pulse is longer
than the DEN pulse accompanying a read bus cycle.
\-
.\
One Bus Cycle
T2
T1
I
I
T3
T4
ClK
ADO-AD15
Data Out
Status Out
A16-A19
BHE
ALE
MIlO
WR
DT/R
DEN
Trailing edge of
ALE latches address
Figure 5-6. 8086 Memory Write Bus Cycle for a Minimum Mode System (MN/MX = +5 V)
An I/O write bus cycle has timing identical to Figure 5-6, except that the M1i5 Signal will be low for the duration of the
bus cycle, rather than high as shown in Figure 5-6. Wherever a memory word and an I/O port may have the same address, M1i5 must contribute to device select logic in order to discriminate between memory and I/O devices.
The status output on A 16-A 19 is no more usefu I in a write bus cycle than it is in a read bus cycle.
8086 READ AND WRITE BUS CYCLES FOR MAXIMUM MODE
It is not very rewarding looking at maximum mode memory or I/O access bus cycle timing, if we look at timing for an
8086 device on its own. This is because in maximum mode, with MN/MX tied to ground, the 8086 has been designed
to operate with the 8288 Bus Controller.
Figure 5-7 and 5-8 provide maximum mode timing for the 8086 on its own when executing read or write bus cycles.
Only the status signal levels differentiate memory or I/O access bus cycles.
Timing for the Address/Data Bus is identical in minimum and maximum modes. The read strobe RD does not change.
However, remaining control Signals become control inputs to the 8288 Bus Controller.
5-37
Observe that OSO and OS1 change levels on a clock period by clock period basis in order to identify events for individual clock periods. SO, Sf, and S2 hold their levels from shortly before T 1 until shortly after the end of T2·
li-oI•. . . . - - - - - - - - - - - - o n e
T1
Bus c y c l e - - - - - - - - - - - - - o . , j l
T3
T2
T4
CLK
ADO-AD15 _ _......J....._ _...
A 16-A 19 ---...o+---C
QSO,QS1
SO,S1,S2
Figure 5-7. 8086 Memory or I/O Read Bus Cycle for a Maximum Mode System (MN/MX = 0 V)
li-oI•. . . . . - - - - - - - - - - - - - o n e
T1
T2
Bus C y c l e - - - - - - - - - - - - . J
I
T3
T4
,.---",,1
CLK
-"""---4
Address Out
Data Out
A16-A19 - - t - - - 1
Address Out
Status Out
ADO-AD 15
QSO,QSl
Figure 5-8. 8086 Memory or liD Write Bus Cycle for a Maximum Mode System (MN/MX = 0 V)
5-38
The 8288 Bus Controller. described later in this chapter. decodes SO. 51. and S2 in order to generate control signals
that are comparable to those illustrated in Figures 5-5 and 5-6. For a complete discussion of bus cycle timing in complex 8086 microcomputer configurations. see the discussion of 8288 Bus Controller.
THE 8086 WAIT STATE
8086 Wait state logic is independent of the MN/MX pin connection and the external access bus cycle being executed.ln any bus cycle it is possible to insert one or more Wait clock periods (Twl between T3 and T4. In order to extend a bus cycle with Wait clock periods. external logic must input a low READY signal during T2 of the bus cycle
which is to be extended. The READY input to the 8086 must be synchronized with the falling edge of ClK at the end of
T2; this synchronized READY input is created by the 8284 clock generator. External logic will normally input an
asynchronous READY to the 8284 clock device. which outputs a synchronous READY for the 8086. Wait clock periods
will continue to be inserted to the bus cycle until READY goes high again. Timing is illustrated in Figure 5-9. All output signal levels are maintained for the duration of the Wait state.
THE 8086 HOLD STATE
The 8086 can be forced into a Hold state, at which time all three-state signals are floated. The 8086 Hold state
is used to enable direct memory access logic, and in addition to disable inactive 8086 devices when more than
one CPU accesses the same System Bus in a mUlti-CPU configuration.
In a minimum mode configuration, when MN/MX is tied to +5 V, the 8086 has a traditional
Hold request input (HOLD) and a Hold Acknowledge output (HLDA). Upon receiving a high
HOLD input. the 8086 will complete execution of its current instruction bus cycle before entering
the Hold sate and outputting HlDA high. Timing may be illustrated as follows:
8086 HOLD
IN MINIMUM
MODE SYSTEM
T40rTi
elK
HOLD
HlDA
The 8086 samples the HOLD input on the low-to-high transition of ClK. Therefore. HOLD must make its transitions
away from this sampling point that is to say. HOLD must be stable when ClK is making its low-to-high transition.
The 8086 will acknowledge the Hold request by making HlDA high during any idle clock period. or at the end of a bus
cycle. If a bus cycle is being executed when a Hold request occurs. the Hold request will not be acknowledged until the
end of T4 for the currently executing bus cycle.
The Hold state will last until the HOLD input goes low again. The 8086 continues to sample the HOLD input on all lowto-high transitions of ClK; therefore. HOLD must make its high-to-Iow transition away from the rising edge of ClK.
When HOLD goes low. the Hold state will immediately end and HlDA will be forced low again.
In 8086 maximum mode configurations where MN/MX is tied to ground. the HOLD and HlDA
pins convert to bidirectional type control signals. There are two bidirectional signals; RO/GTO and
RO/GT1. RO/GTO has higher priority than RO/GT1.
5-39
8086 HOLD
IN MAXIMUM
MODE SYSTEM
T1
T2
TW
T3
T4
elK
RDY
READY
ROY comes from external logic to the 8284 clock.
READY comes from the 8284 clock, goes to the 8086.
Figure 5-9. The 8086 READY Input and Wait States
Any external logic that wishes to put an 8086 CPU into the Hold state transmits a low pulse to RO/GTO or RO/GT1, The
8086 CPU will acknowledge this Hold request immediately. if a bus cycle is not being executed. or at the conclusion of
a currently executing bus cycle. The 8086 acknowledges the Hold request transmitting by a low pulse via the same ROI
GT line; simultaneously the 8086 floats its three-state bus lines. External logic must allow atl~~t~~ clock period to
elpase following the Hold Acknowledge pulse. before attempting to input via the same pin. External logic terminates
the Hold state by inputting another low pulse. Timing may be illustrated as follows:
: T4 or Ti
I-- Hold State---j
I
I
I
eLK
RQ/GT
0
@
®
In the illustration above. ® identifies the instant at which external logic requests a Hold state by transmitting a low
pulse via either RO/GT line. The 8086 samples RO/GT on the rising edge of ClK; therefore. all signal transitions on
RO/GT must occur away from the ClK low-to-high transitions.
The 8086 will now acknowledge a Hold request during a bus cycle, If a bus cycle is in progress. then the Hold
acknowledge will occur at the end of the bus cycle - that is to say. at the end of T4. If a bus cycle is not in progress.
then the Hold request will be acknowledged immediately. In the illustration above. @ identifies the low pulse the
8086 will output as its Hold acknowledge, The Hold state will last until external logic again transmits a low pulse via
RO/GT. This is identified above as
Once again the 8086 samples RO/GT on the rising edge of ClK; therefore.
RO/GT should be stable at this time,
©
When the 8086 enters the Hold state. it continues executing instructions it takes out of the pipeline. until a bus
access is required. When the EU requires a bus access. it stops operating until the end of the Hold state - at which
time its bus access request will be honored by the Bus Interface Unit
In the event that Hold requests occur simultaneously on RQ/GTO and RQ/GT1. the acknowledge pulse will be
output on RQ/GTO. RO/GT1 will not be acknowledged until the Hold state initiated via RO/GTO has ended.
When one Hold state ends. another Hold state can begin immediately for either of these reasons:
1)
RO/GT1 was active when RO/GTO was acknowledged; the RO/GT1 Hold request. being of lower priority. was
denied and is pending,
2)
While the 8086 was in a Hold state. a new hold request occurs on the other RO/GT line.
If a new hold request occurs while the 8086 is in Hold state. priorities no longer apply. For example. if the CPU has
acknowledged a Hold request occurring at RO/GT1 and is in a Hold state. then it will deny a new Hold request arriving
via RO/GfO until the current Hold state has ended,
5-40
If there is an active Hold request when the CPU ends a Hold state. then the CPU will immediately acknowledge the
pending Hold request. This may be illustrated as follows:
Hold for RQ/GT1 in progress
Hold for RQ/GTO in progress
Low pulse
input at
RQ/GTO
and immediately
stand RQ/GTO Hold
When a Hold state ends. if the CPU has a bus access request pending. then the CPU bus access request will be denied
until all active Hold requests have been acknowledged.
Note that there are no 8086 instructions that specifically affect the level of RO/GTO or RO/GT1. That is to say. external
logic is entirely responsible for the interfaces to these two signals.
We will discuss RO/GTO and RO/GT1 in more detail later in this chapter when we look at some multiple CPU 8086 configurations.
THE 8086 HALT STATE
The 8086 enters a Halt state after a HALT instruction is executed. In the Halt state no signals are floated, and
undefined data is output on the Data/Address Bus. No bus cycles can be executed while the 8086 is in the Halt
state.
.
When a Halt instruction is executed. a bus cycle initiates the Halt state. This Halt state initializing bus cycle has
nothing to do with instruction fetch logic. If the Halt instruction object code is fetched by the CPU from the queue. then
there will be no preceding instruction fetch bus cycle. If the Halt instruction must be fetched from memory because the
queue is empty. or is at the conditional end of a Branch-on-Condition. then the Halt initializing bus cycle will be preceded by an instruction fetch bus cycle.
For a simple system, the HALT initialization bus cycle is given by Figure 5-5, except that RD, M/IO, DT/R and
DEN are not active. ALE is active, although the address output has no meaning.
For a complex system, the HALT initializing bus cycle is illustrated in Figure 5-10. The Halt state combination
occurring at SO. ST. and 52 causes the 8288 Bus Controller to issue an ALE pulse before entering the Halt state:
however. the occurrence of ALE could not be deduced simply by looking at 8086 timing.
5-41
CLK-\
so,51
52
I
,
,
I
I
I
I
I
T3
~
'\
I
J
\
I
I
\
T4
I
I
I
L
Undefined Data
\
\
\
\
""
\
\~
T2
?-
I
)
OSO, OS1
I
I
T1
\
A16-A19
-I
One Bus Cycle
I
-.,
ADO-AD15
SHE
I·
\
t
I
Undefined Data
/
"
\
J
Figure 5-10. 8086 HALT Instruction and Bus Cycle Timing for a Complex Bus Configuration
The Halt state is terminated by an interrupt request or a Reset.
You can freely enter and leave a Hold state within an 8086 Halt state via any of the means that we have just described. The fact that the 8086 is in the Halt state in no way modifies Hold logic.
THE 8086 LOCK
A potential for serious error exists in the Hold request/acknowledge logic of the 8086.
The 8086 will acknowledge a Hold request occurring on the RQ/GTO or RQ/GT1 lines at the end of the current bus cycle, if one is being executed. or at the next idle clock period. if a bus cycle is not being executed. The 8086 does not
wait until the conclusion of the current instruction's execution before acknowledging the Hold request. Therefore, if an
instruction reads the contents of a memory location (or I/O port), modifies these contents, then writes it back, a Hold
state may separate the read bus cycle from the write bus cycle:
I Read from memory location X
Modify data
I
Write back to location X
I
HOLD STATE
This can cause unexpected errors. If the 8086 enters a Hold state after reading memory location X contents and before
writing these contents back. then it is possible for external logic - either direct memory access logic or another
Central Processing Unit - to modify the contents of memory location X while the 8086 is in the Hold state. Now when
the 8086 writes back the modified word. it may destroy data that should have been preserved.
5-42
If a 16-bit data word lies on an odd-byte boundary. it will require two bus cycles to access the data word. Under normal
circumstances. a Hold request could be acknowledged between the first and second memory access bus cycles. But
what if the word being accessed gets modified during the Hold state? If the Hold state splits two memory read bus cycles. this is what the CPU is going to read:
o 7
7
....
..
'~,--
o
----~~~--------~~,--
High-order byte
was read after Hold
and is modified
--,~
Low-order byte
was read before Hold
and is not modified
If a Hold state splits two memory write bus cycles. this is what ultimately gets written:
o
7
7
~L,
I
'-,----v
High-order byte
is written after Hold
and is not modified
o
___-.~___. . ._"J
Low-order byte
was written before Hold
and gets modified
You use the 8086 LOCK instruction in order to prevent the types of errors described above.
When a LOCK instruction is executed, the LOCK signal is low for the duration of the next sequential instruction's execution. Also, while the next sequential instruction is being executed, a Hold request will not be
acknowledged.
You cannot extend protection against a Hold acknowledge beyond a single instruction's execution. For example. suppose you have two instructions. each of which is preceded by a Lock:
LOCK
AND
LOCK
OR
MEMX.AX
MEMX. BX
In the instruction sequence above. MEMX is a label which represents the address of a memory location. The contents of
this memory location are ANDed with a mask stored in AX. then ORed with a mask stored in BX. The contents of MEMX
are read. modified. and written back at each step.
Now. you may wish to inhibit Hold logic for both the AND and the OR operation. You cannot do so using the LOCK instruction. The first LOCK instruction will protect the following AND instruction from being interrupted by a Hold state:
however. any pending Hold state will be acknowledged before the second LOCK instruction is executed.
Each LOCK instruction extends protection against a Hold Acknowledge for the duration of the next sequential instruction only. The fact that the following instruction is also a LOCK is irrelevant. The second LOCK instruction will be the
first instruction executed following the Hold state. and it will guarantee that no new Hold state begins until it. and the
OR instruction. have both been executed.
You can use the LOCK instruction and signal to identify individual instruction execution
times. If for any reason external logic needs to know the execution time for certain instructions.
then by preceding these instructions with a LOCK instruction you will generate a high pulse on
the LOCK output. The width of this high pulse exactly equals the execution time of the instruction
which follows the LOCK.
8086 SINGLE
INSTRUCTION
TIME
IDENTIFIED
THE 8086 PROCESSOR WAIT FOR TEST STATE
The 8086 has a program-initiated Wait state that external logic must terminate via the TEST input signal. The
WAIT instruction initiates this Wait state. After the WAIT instruction is executed. the 8086 generates an endless sequence of idle clock periods. This sequence lasts until external logic inputs a low signal at the TEST input. TEST must
be high for at least four clock periods.
While the endless sequence of idle clock pulses is being executed. the System Bus is not floated and the Bus Interface
Unit may execute memory read bus cycles in order to fill up the instruction object code queue.
5-43
The processor Wait state can be used to synchronize an 8086 with any external time sequence. For example, you
could start two programs, executing in two separate 8086 systems, at exactly the same time, by preceding each program with a Wait instruction. If both 8086's receive low TEST inputs simultaneously, then both microprocessors will
start executing their programs at the same instant.
THE 8086 PROCESSOR ESCAPE
The 8086 has a special escape instruction (ESC) intended for use in mUlti-CPU configurations. When the ESC instruction is executed. the contents of an addressed memory location are input to the CPU. but the input data is not
stored anywhere. The purpose of the ESC instruction is to place the addressed data on the Data/Address Bus so that
any other microprocessor (or external logic) connected to the Data/Address Bus can receive the data.
We will examine the value of the ESC instruction later in the chapter when looking at the 8086 in multiple CPU
configurations.
THE 8086 RESET OPERATION
The 8086 has an asynchronous RESET input. This signal can be forced high at any time in order to reset the
8086. The high RESET must be at .least four clock cycles long.
The 8086 terminates all current operations as soon as the RESET input makes a low-to-high transition. Nothing
more happens until the RESET signal subsequently makes a high-to-Iow transition. It then takes approximately
ten clock periods in order to execute the following operations:
1)
The Status register is cleared. Among other things. this resets the interrupt enable flag to 0, thus disabling interrupts.
2)
The CS Segment register is set to FFFF16.
3)
The OS, SS, and ES Segment registers and the Program Counter are all reset to O.
4)
Program execution begins. Since the CS Segment register contains FFFF16 and the Program Counter contains 0,
the first instruction executed is taken from memory location FFFF016.
8086 INTERRUPT PROCESSING
The 8086 allows interrupts to originate in one of three ways:
1)
From software or within program logic.
2)
From external logic as a nonmaskable interrupt.
3)
From external logic as a maskable interrupt.
There is, in addition, a special "single step" condition that makes use of interrupt logic. We will describe single stepping after our discussion of interrupt logic.
In the event that two or more of the three interrupt types occur simultaneously, software generated interrupts have the
highest priority and maskable interrupts have the lowest priority.
These are the ways in which a software interrupt request may occur:
1)
Following an attempt to divide by O. A special divide by 0 interrupt request will occur any
time the divide instruction is executed with a 0 dividend.
2)
Following execution of an Interrupt instruction (lNT).
3)
Following execution of an Interrupt-on-Overflow instruction (INTO) -
8086
SOFTWARE
INTERRUPTS
if the Overflow status is set.
A nonmaskable interrupt request is initiated when external logic transmits a low-to-high
transition to the NMI pin. This is an edge-triggered Signal. A nonmaskable interrupt has lower
priority than a software interrupt. but higher priority than a maskable interrupt.
8086 NONMASKABLE
INTERRUPT
A maskableinterrupt request will be generated when external logic transmits a high level to
the INTR pin. This input is level sensitive; it is the high level at INTR that causes the interrupt requests to occur.
8086
MASKABLE
INTERRUPT
Central to all 8086 interrupt processing is a Vector table that can be up to 1024 bytes in
length, occupying absolute memory addresses 00000 through 003FF 16. This Vector table consists of up to 256 four-byte entries. Each entry contains two 16-bit addresses that get loaded into
the CS Segment register and the Program Counter.
8086
INTERRUPT
VECTOR
TABLE
Figure 5-11 illustrates the 8086 Interrupt Vector table.
5-44
A number of the Vector table entries serve specific interrupts. Other entries are reserved by Intel and should be
avoided if compatibility with Intel software is desired. These entries are identified in Figure 5-11. As illustrated in Figure 5-11, 32 of the 256 interrupt vectors are not available to external logic; that leaves 224 vectors available to maskable external interrupts - which is plenty.
Taking each of the three interrupt types in turn, let us examine the interrupt acknowledge p...r_o_c_e_s_s._ _ __
When any of the software interrupts are acknowledged, the following steps occur:
8086
SOFTWARE
INTERRUPT
1)
The Status register contents are pushed onto the Stack; Stack Pointer contents. in consequence. are decremented by two.
2)
The Interrupt and Test status flags are cleared; this disables maskable interrupts and single
step logic (which we describe after our discussion of interrupt logic).
3)
The CS Segment register contents are pushed onto the Stack; Stack Pointer contents. in consequence. are decremented by two.
4)
The new CS Segment register contents are taken from the appropriate interrupt vector location. With the exception of the INT instruction. software-generated interrupts have dedicated vector locations as illustrated in Figure
5-11. The INT instruction allows anyone of the 256 vector locations to be selected; a default option selects Vec•
tor 3.
5)
The Program Counter contents are pushed onto the Stack; Stack Pointer contents are decremented by two.
6)
The new Program Counter contents are taken from the interrupt vector.
When a nonmaskable interrupt is acknowledged, the following events occur:
8086
NONMASKABLE
INTERRUPT
1)
The Status register contents are pushed onto the Stack. The Stack Pointer contents are
decremented by two.
2)
The Interrupt and Test statuses are reset to 0; this disables nonmaskable interrupts and
single stepping mode.
3)
The CS Segment register and Program Counter are reloaded from Interrupt Vector 2. See Figure 5-11.
------....
When a maskable interrupt is acknowledged, the following steps occur:
1)
8086
Two interrupt acknowledge bus cycles are executed by the Bus Interface Unit of the 8086. An
MASKABLE
interrupt acknowledge bus cycle is identical to the memory read bus cycles. as illustrated in
INTERRUPT
Figures 5-5 and 5-7. with the exception that an interrupt acknowledge low pulse replaces
the memory read low pulse. For a minimum mode system. INTA will provide the low RD pulse shown in Figure
5-5. Figure 5-7 accura.tely illustrates timing for an interrupt acknowledge bus cycle in a maximum mode system;
however. SO. ST. and S2 will all be low. identifying an interrupt acknowledge. whereas a read I/O port or read
memory status combination would be output otherwise. LOCK is low beginning at T2 of the first interrupt
acknowledge bus cycle and ending at T2 of the second interrupt acknowledge bus cycle. This may be illustrated as
follows:
I.
I
ClK
I
Bus Cycle 1
T1
,
I
T2
,
I
T3
..
,
I
T4
I..
I
I
·1
Bus Cycle 2
Tl
I
I
T2
I
I
T3
I
I
T4
I
I
LOCK
2)
The acknowledged external device must send back a byte of data on lines ADO-AD7 in response to the second interrupt acknowledge bus cycle. This data byte is interpreted as a pointer into the interrupt vector. Multiplying this
8-bit value by 4 creates the correct beginning address for the interrupt vector.
3)
The Status register contents are pushed onto the Stack.
4)
The Interrupt and Test flags in the Status register are cleared. This disables further maskable interrupts and single
step logic.
5)
The CS Segment register contents are pushed onto the Stack.
6)
The next CS Segment register contents are taken from the interrupt vector location identified in Step 2.
7)
The Program Counter contents are pushed onto the Stack.
5-45
Memory
Interrupt
Addresses Vector Table
00000
CSO
00002
PCO
00004
CSl
00006
PCl
00008
CS2
OOOOA
PC2
OOOOC
CS3
OOOOE
PC3
00010
CS4
00012
PC4
00014
CS5
00016
PC5
I
00078
CS30
0007A
PC30
0007C
CS31
0007E
PC31
00080
CS32
00082
PC32
00084
CS33
00086
PC33
Vector 1 } Vector 2 -
Single step
mode
~onmaskable
Interrupt
} Vector 3 -INT software
interrupt (default option)
} Vector 4 -
~NTO software
Interrupt
I
} Vector 5
I
Reserved by Intel
} Vector 3010
f
Veeto, 3 t 10
Vector 3210
} Vector 3310
I
I
I
User vectors
I
003F8
CS254
003FA
PC254
003FC
CS255
003FE
PC255
I
-
-
> Vector 25410
Vector 25510
I
I
I
@
VeeW 0 - D;v;de by 0
I
I
•
®
t
,
Interrupt acknowledge sequence of
events is (D-@-@-@
CSN
PCN
~
Shaded row applies to EA and LABEL.
X These are displacements that can be used to compute
memory addresses.
5-56
The following abbreviations are used in Tables 5-4 and 5-5:
AH
AL
AL7
AX
AX15
BD
BH
BL
BRANCH
BS
BX
C
CH
CL
CS
CX
DADDR
DATA8
DATA 16
DH
DI
DISP
DISP8
DL
DS
DX
EA
ES
I
I/D
LABEL
N
o
OEA
PC
PDX
PORT
RB
RBD
RBS
RW
RWD
RWS
SEGM
SFR
SI
SP
SR
SS
Accumulator. high-order byte
Accumulator. low-order byte
The value of register AL high-order bit (0 or 1) extended to a byte (0016 or FF16)
Accumulator. both bytes
The value of register AH high-order bit (0 or 1) extended to a 16-bit word (000016 or FFFF16)
The destination is a byte operand (used only by the Assembler)
B register. high-order byte
B register. low-order byte
Program memory direct address. used in Branch addressing option shown in Tables 5-1 and 5-2
The source is a byte operand (used only by the Assembler)
B register. both bytes
Carry status
C register. high-order byte
C register. low-order byte
Code Segment register
C register. both bytes
Da\a memory address operands identified in Table 5-3
Eight bits of immediate data
16 bits of immediate data
D register. high-order byte
Destination Index register
An 8-bit or 16-bit signed displacement
An 8-bit signed displacement
D register. low-order byte
Data Segment register
D register. both bytes
Effective data memory address using any of the memory addressing options identified in Table 5-2
Extra Segment register
Status flag set to 1
Increment/decrement selector for string operations; increment if D is O. decrement if D is 1
Direct data memory address. as identified in Table 5-2
A number between 0 and 7
Status flag reset to 0
Offset data memory address used to compute EA:
EA =OEA + [DS] * 16
Program Counter
I/O port addressed by OX register contents; port number can range from 0 through 65.536
A label identifying an I/O port number in the range 0 through 25510
Anyone of the eight byte registers: AH. AL. BH. BL. CH. CL. DH. or OL
Any RB register as a destination
Any RB register as a source
Anyone of the eight 16-bit registers: AX. BX. CX. DX. SP. SP. SI. or DI
Any RW register as a destination
Any RW register as a source
Label identifying a 16-bit value loaded into the CS Segment register to execute a segment jump
Status Flags register
Source Index register
Stack Pointer
Anyone of the Segment registers CS. DS. ES. or SS
Stack Segment register
5-57
U
V
X
WD
WS
[[ ]]
[]
1=
Status flag modified. but undefined
Any number in the range a through 25510
Status flag modified to reflect resu It
The destination is a word operand (used only by the Assembled
The source is a word operand (used only by the Assembler)
Contents of the memory location addressed by the contents of the location enclosed in the double
brackets
The contents of the location enclosed in the brackets
Data on the right-hand side of the arrow is moved to the location on the left-hand side of the arrow
Contents of locations on each side of < - - are exchanged
The twos complement of the value under the Not equal to
INSTRUCTION EXECUTION TIMES AND CODES
Table 5-5 lists instructions in alphabetical order. showing object codes and execution times. for the 8086 and the
8088. expressed in whole clock cycles. Execution time is the time required from beginning execution of an instruction
that is in the queue to beginning execution of the next instruction in the queue. The time required to place an instruction from memory into the queue (instruction fetch time) is not shown in the table; because of queuing. instruction
fetch time occurs concurrently with instruction execution time and thus has no effect on overall timing. except as
specifically noted in the table.
Instruction object codes are represented as two hexadecimal digits for instruction bytes without variations.
Instruction object codes are represented as eight binary digits for instruction bytes with variations for the instruction.
The following notation is used in Tables 5-4 and 5-5:
[]
a
aa
bbb
DISP
ddd
rr
reg
indicate an optional object code byte
one bit choosing length:
in bit position a=O specifies 1 data byte; a= 1 specifies 2 data bytes
in bit position 1 a=O specifies 2 data bytes: a=1 specifies 1 data byte
two bits choosing address length:
no DISP = 00
one DISP byte = 01
two DISP bytes = 10. or 00 with bbb = 110
11 causes bbb to select a register. using the 3-bit code given below for reg.
three bits choosing addressing mode:
000 EA = (BX) + (SI) + DISP
001 EA = (BX) + (DI) + DISP
010 EA = (BP) + (SI) + DISP
011 EA = (BP) + (DI) + DISP
100 EA = (SI) + DISP
101 EA = (DI) + DISP
110 EA = (BP) + DISP
111 EA = (BX) + DISP
represents two hexadecimal digit memory displacement
represents three binary digits identifying a destination register (see reg.)
two binary digits identifying a segment register:
00 = ES
01 = CS
10 = SS
11 = DS
three binary digits identifying a register:
a
000
001
010
011
100
101
110
111
=
=
=
=
=
=
=
=
16-bit
AX
CX
DX
BX
SP
BP
SI
DI
8-bit
AL
CL
DL
BL
AH
CH
DH
BH
5-58
sss
PPOO
v
x
yy
yyyy
represents three binary digits identifying a source register (see reg)
represents four hexadecimal digit memory address
one bit choosing shift length:
o count = 1
1 count = (eU
"don't care" bit
represents two hexadecimal data digits
represents four hexadecimal data digits
one bit where z XOR (ZF) = 1 terminates loop
Execution time is less than or equal to instruction fetch time.
Includes up to eight clock cycles of overhead on each transfer due to queue maintenance. For conditional
jumps, the lesser figure is when the test fails (no jump taken).
Effective Address calculation and extra clock cycles:
Extra Clock Periods
bbb
8086(1)
EA
000
000
000
001
001
001
010
010
010
011
011
011
100
101
110
111
(BX) + (SI)
(BX) + (SI) + 0lSP8
(BX) + (SI) + OISP 16
(BX) + (01)
(BX) + (01) + 0lSP8
(BX) + (01) + OISP 16
(BP) + (SI)
(BP) + (SI) + olspa
(BP) + (SI) + OISP 1 6
(BP) + (01)
(BP) + (01) + olspa
(BP) + (01) + OISP 16
(SI) ir (01) or (BO)
or (BX)
+ olspa
+ 0lSP16
a-bit immediate
1 6-bit immediate
(1 )
Add another 4 clock cycles for each
1 6-bit operand or an odd address boundary.
(2)
Add anoter 4 clock cycles for each
1 6-bit operand.
8088(2)
7
7
11
11
a
12
12
a
12
12
11
15
a
12
16
8
12
16
7
7
11
11
5
11
15
5
9
9
6
6
9
13
6
10
Substitute the clock cycles shown above wherever EA appears in Tables 5-4 and 5-5.
5-59
Table 5-4. A Summary of 8086 and 8088 Instructions
Statuses
GI
a.
> Mnemonid Operand Is)
Object Code
Clock Cycles
t-
g
U1
m
IN
AL,PORT
E4 YY
10
IN
AL.[OX)
EC 1
8
IN
AX,PORT
E5 YY
10
IN
AX,[OX)
ED
8
OUT
AL,PORT
E6 YY
10
OUT
AL,[OX)
EE 1
8
OUT
AX,PORT
E7 YY
10
OUT
AX,[OX)
EF
8
LOS
RW,OAOOR
C5 aasssbbb
[OISPJ[OISP)
16+EA
LEA
RW,OAOOR
80 aasssbbb
[OISP)[OISP)
2+EA
LES
RW,OAOOR
C4 aasssbbb
[DISP][DISP)
16+EA
MOV
RB,OAOOR
8A aadddbbb
[OISP)[OISP)
8+EA
o
GI
Col
c
f
.;
II:
~
0
E
GI
~
~
III
.5
A:
Operation Performed
O D I T S Z APe
[AL) - [PORT)
Load one byte of data from 1/0 port PORT into AL
[AL)- [POX)
Load into AL one byte of data from 1/0 port whose address is held in the OX
register
[ALI - [PORT). [AH) - [PORT + 1)
Load 16 bits of data into AX, AL receives data from 1/0 port PORT, AH
receives data from 1/0 port PORT + 1
[AL) - [POX), [AH) - [POX+ 1)
Load 16 bits of data into AX, AL receives data from 1/0 port whose address is
held in the OX register. AH receives data from the 1/0 port whose address is
one higher
[PORT) - [AL)
Output one byte of data from register AL to 1/0 port PORT
[POX)- [ALI
Output one byte of data from register AL to the 1/0 port whose address is held
in the OX register
[PORT) - [ALI. [PORT + 1) - [AH)
Output 16 bits of data. The AL register contents are output to 1/0 port PORT.
The AH register contents are output to 1/0 port PORT + 1
[PORT) - [POXI. [PORT + 1) - [POX+ 1)
Output 16 bits of data. The AL register contents are output to the 1/0 port
whose address is held in the OX register. The AH register contents is output
to the 1/0 port whose address is one higher
[RW)- [EAI. [OS)- [EA+2)
Load 16 bits of data from the memory word addressed by OAOOR into
register RW. Load 16 bits of data from the next sequential memory word into
the OS register
[RW)-OEA
Load into RW the 1 6-bit address displacement which, when added to the
segment register contents, creates the effective data memory address
[RW)- [EAI. [ES)- [EA+2)
Load 16 bits of data from the memory word addressed by DADDR into
register RW. Load 16 bits of data from the next sequential memory word into
the ES register
[RBI- [EAI
Load one byte of data from the data memory location addressed by OAOOR to
register RB
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
•
D.
~
:a.,
Statuses
~nemonic
Operandls)
Object Code
Clock Cycles
MOV
RW,DADDR
8B aadddbbb
[DISP][DISP)
8+EA
MOV
DADDR,RB
9+EA
MOV
DADDR,RW
88 aasssbbb
[DISP][DISP]
89 aasssbbb
[DISP][DISP]
MOV
AL,LABEL
AO PPQQ
10
MOV
AX,LABEL
A1 PPQQ
10
MOV
LABEL,AL
A2 PPQQ
10
MOV
LABEL,AX
A3 PPQQ
10
MOV
SR,DADDR
8E aaOrrbbb
[DISP][DISP]
8+EA
MOV
DADDR,SR
8C aaOrrbbb
[DISP][DISP]
9+EA
XCHG
RB,DADDR
86 aaregbbb
[DISP][DISP)
17+EA
XCHG
RW,DADDR
87 aaregbbb
[DISP][DISP]
17+EA
D7
11
9+EA
~
c::
U1
~
~0
2
.,
Co)
c::
!
.,
of
II:
>
0
.,E
:E
>
l&
'E
A.
XLAT
o
Operation Performed
D
I
T, S; Z
A
P C
[RW] -[EA)
Load 16 bits of data from the data memory word addressed by DADDR to
register RW
[EA] - [RBJ
Store the data byte from register RB in the memory byte addressed by DADDR
.[EA] -[RW]
Store the 16-bit data word from register RW in the memory word addressed
by DAD DR
[AL] - [EA]
Load the data memory byte directly addressed by LABEL into register AL
[AX] - [EA]
Load the 16-bit data memory word directly addressed by LABEL into register
AX
[EA] - [AL]
Store the 8-bit contents of register AL into the data memory byte directly addressed by LABEL
[EA] - [AX]
Store the 16-bit contents of register AX into the data memory word directly
addressed by LABEL
[SR] - [EA]
Load into Segment register. SR the contents of the 1 6-bit memory word addressed by DADDR
[EA]- [SR]
Store the contents of Segment register SR in the 16-bit memory location addresed by DADDR
[RB] - - [EA)
Exchange a byte of data between register RB and the data memory location
addressed by DADDR
[RW]-[EA]
Exchange 16 bits of data between register RW and the data memory location
addressed by DADDR
[AL] - [[AL] + [BX))
Load into AL the data byte stored in the memory location addressed by summing initial AL contents with ex contents
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
GI
Q.
>
~
Mnemonic
Operand(s)
Object Code
Clock Cycles
Operation Performed
0
D
I
T 5 Z A P C
ADC
RB,DADDR
12 aadddbbb
[DISPJ[DISP]
9+EA
X
X X X X
ADC
RW,DADDR
13 aadddbbb
[DISPJ[DISP]
9+EA
X
X X
ADC
DADDR,RB
10 aasssbbb
[DISP][DISP]
16+EA
X
X
ADC
DADDR,RW
11 aasssbbb
[DISP][DISP]
16+EA
X
X X
ADD
RB,DADDR
9+EA
X
X
ADD
RW,DADDR
9+EA
X
X X
ADD
DADDR,RB
02 aadddbbb
[DISP)[DISP]
03 aadddbbb
[DISPJ[DISP]
00 aasssbbb
[DISP)[DISP]
16+EA
X
X X
ADD
DADDR,RW
01 aasssbbb
[DISP][DISP]
16+EA
X
X X
AND
RB,DADDR
22 aadddbbb
[DISP][DISP]
9+EA
0
X X
AND
RW,DADDR
23 aadddbbb
[DISP)[DISP]
9+EA
0
X X
AND
DADDR,RB
20 aasssbbb
[DISP)[DISP]
16+EA
0
X X
AND
DADDR,RW
21 aasssbbb
[DISP)[DISP]
16+EA
0
X X
CMP
RB,DADDR
3A aadddbbb
[DISP][DISP]
9+EA
X
X X
CMP
RW,DADDR
38 aadddbbb
[DISP][DISP]
9+EA
X
X X
X
'$
I!GI
Q.
0
~
0
E
GI
U1
m
~
~
GI
X
u
c
!!!
.!GI
a:
~
E
GI
~
~
IV
"uc
0
GI
(I)
[RB] - [EA] + [RB] + [C]
Add the contents of the data byte addressed by DAD DR, plus the Carry status,
to register RB
X X X [RW] - [EA] + [RW] + [C]
Add the contents of the 16-bit data word addressed by DADDR, plus the Carry status, to register RW
X X X [EA] - [EA] + [RB] + [C]
Add the 8-bit contents of register RB, plus the Carry status, to the data
memory byte addressed by DADDR
X X X [EA] - [EA] + [RW] + [C]
Add the 16-bit contents of register RW, plus the Carry status, to the data
word addressed by DADDR
X X X [RB] - [EA] + [RB]
Add the contents of the data byte addressed by DADDR to register RB
X X X [RW] - [EA] + [RW]
Add the contents of the 16-bit word addressed by DADDR to register RW
X X X [EA] - [EA] + [RB]
Add the a-bit contents of register RB to the data memory byte addressed by
DADDR
X X X [EA] - [EA] + [RW]
Add the 16-bit contents of register RW to the data memory word addressed
by DADDR
U X 0 [RB] - [EA] AND [RB]
AND the 8-bit contents of register RB with the data memory byte addressed
by DADDR. Store the result in RB
U X 0 [RW] - [EA) AND [RW]
AND the 1 6-bit contents of register RW with the data memory word addressed by DADDR. Store the result in RW
U X 0 [EA] - [EA) AND [RB)
AND the 8-bit contents of register RB with the data memory byte addressed
by DADDR. Store the result in the addressed data memory byte
U X 0 [EA] - [EA] AND [RW]
AND the 1 6-bit contents of register RW with the data memory word addressed by DADDR. Store the result in the addressed data memory word
X X' X [RB] - [EA]
Subtract the contents of the data memory byte addressed by DADDR from the
contents of register RB. Discard the result, but adjust status flags
X X X [RW] - [EA]
Subtract the 1 6-bit contents of the data memory word addressed by DADDR
from the contents of register RW. Discard the result, but adjust status flags
X
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
Q)
a.
> Mnemonic Operand(s)
~
Object Code
Operation Performed
Clock Cycles
0
D
I
T
S Z A P C
CMP
DADDR.RB
38 aasssbbb
[DISP][DISP]
9+EA
X
X X
X X X
CMP
DADDR.RW
39 aasssbbb
[DISP][DISP]
9+EA
X
X
DEC
DADDR
1111111a
aa001bbb
[DISP][DISP]
15+EA
X
X X X X
DIV
AX.DADDR
F6 aa110bbb
[DISP][DISP]
(S6-96 )+EA
U
U U U U U
DIV
OX. DAD DR
F7 aa110bbb
[DISP][DISP]
(150-16S)+EA U
U U U U U
IDIV
AX.DADDR
F6 aa111bbb
[DISP][DISP]
(107-11S)8+EA U
U U U U U
IDIV
DX.DADDR
F7 aa111bbb
[DISP][DISP]
(171)-190)+EA U
U U U U U
IMUL
ALDADDR
F6 aa101bbb
[DISP][DISP]
IMUL
AX.DADDR
F7 aa101bbb
[DISP][DISP]
X X X X
;;
CD
·Sc=
0
~
'!
f
CD
a.
C11
0,
W
C
>
~
E
CD
~
CD
u
c
G)
.!CD
II:
~
0
E
G)
~
~
III
"a
c
0
u
G)
CI)
X
U U U U )(
(134-160)+EA X
U U U U X
(S6-104)+EA
[EA] - [RB]
Subtract the 8-bit contents of register RB from the data memory byte addressed by DADDR. Discard the result, but adjust status flags
[EA] - [RW]
Subtract the 16-bit contents of register RW from the data memory word addressed by DADDR. Discard the result. but adjust status flags
rEA] - [EA] - 1
Decrement the contents of the memory location addressed by DADDR. Depending on the prior definition of DADDR. an 8-bit or a 16-bit memory location may be decremented
[AX] - [AX]/[EA]
Divide the 16-bit contents of register AX by the S-bit contents of the memory
byte addressed by DADDR. Store the integer quotient in AL and the remainder
in AH. If the quotient is greater than FF16. execute a "divide by 0" interrupt
[OX] [AX] - [OX] [AX]/[EA]
Divide the 32-bit contents of registers OX (high-order) and AX !low-order) by
the 16-bit contents of the memory word addressed by DADDR. Store the integer quotient in AX and the remainder in OX. If the quotient is greater than
FFFF 16. execute a "divide by 0" interrupt
[AX] - [AX]/[EA]
Divide the 16-bit contents of register AX by the 8-bit contents of the memory
byte addressed by DAD DR. treating both contents as signed binary numbers.
Store the quotient, as a signed binary number. in AL. Store the remainder. as
an unsigned binary number. in AH. If the quotient is greater than 7F 16. or less
than -S016. execute a "divide by 0" interrupt
[OX] [AX] - [OX] [AXl![EA]
Divide the 32-bit contents of register OX (high-order) and AX !low-order) by
the 16-bit contents of the memory word addressed by DADDR. Treat both
contents as signed binary numbers. Store the quotient. as a signed binary
number. in AX. Store the remainder. as an unsigned binary number. in AH. If
the quotient is greater than 7FFF16. or less than -S00016. execute a "divide
by 0" interrupt
[AX] - [AL] • [EA]
Multiply the S-bit contents of register AL by the contents of the memory byte
addressed by DADDR. Treat both numbers as signed binary numbers. Store
the 16-bit product in AX
[OX] [AX] - [AX] • [EA]
Multiply the 16-bit contents of register AX by the 16-bit contents of the
memory word addressed by DADDR. Treat both numbers as signed binary
numbers. Store the 32-bit product in DX (high-order word) and AX !low-order
word)
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
&
>-
Mnemonic Operand(.)
Statuses
Object Code
Clock Cycle.
~
o D
I
T S· Z
Operation Performed
A P C:
INC
DADDR
1111111a
a8000bb
[DISP}(DISP]
15+EA
X
X X X X
MUL
AL.DADDR
FS a8100bbb
[DISP][DISP]
(7S-831+EA
'X
U U U U X
MUL
F7
F7 aa100bbb
[DISP][DISP]
(124-1391+EA X
U U U U X
NEG
DADDR
1111011a
aa011bb
[DISP][DISP]
16+EA
NOT
DADDR
1111011a
aa010bbb
[DISP][DISP]
16+EA
OR
RB.DADDR
OA aadddbbb
[DISP)[DISP]
OR
RW.DADDR
OR
OR
~
Gl
~
c
c
+:
..;
0
~
X
X X X X X
9+EA
X
X X U X X
OB aadddbbb
[DISP][DISP]
9+EA
X
X X U X X
DADDR.RB
08 aasssbbb
[DISP][DISP)
1S+EA
X
X
DADDR.RW
09 aasssbbb
[DISP][DISP)
16+EA
X
X X U X X
Q.
(J1
I
m
-1=0
0
>~
E
Gl
!
Gl
u
c
~
II:
~
0
E
Gl
~
~
III
"uc
X U X X
0
Gl
en
[EA] +- [EA] + 1
Increment the contents of the memory location addressed by DADDR. Depending on the prior definition of DAD DR. an 8-bit or a 16-bit memory location may be incremented
[AX] +- [All • [EA]
Multiply the 8-bit contents of register AL by the contents of the m~mory byte
addressed by DADDR. Treat both numbers as unsigned binary numbers. Store
the 1S-bit product in AX
[OX) [AX) +- [AX) • [EA)
Multiply the 16-bit contents of register AX by the 16-bit contents of the
memory word addressed by DADDR. Treat both numbers as unsigned binary
numbers. Store the 32-bit product in DX (high-order word) and AX (low-order
word)
[EA) +- rEAl
Twos complement the contents of the addressed memory location. Depending on the prior definition of DADDR. an 8-bit or 16-bit memory location may
be twos complemented
[EA] +- NOT [EA]
Ones complement the contents of the addressed memory location. Depending
on the prior definition of DADDR. an 8-bit or 16-bit memory location may be
ones complemented
[RB] +- [EA] OR [RB]
OR the 8-bit contents of register RB with the data memory byte addressed by
DADDR. Store the result in RB
[RW) +- [EA) OR [RW)
OR the 1S-bit contents of register RW with the data memory word addressed
by DADDR. Store the result in RW
[EA) +- [EA) OR [RB)
OR the 8-bit contents of register RB with the data memory byte addressed by
DADDR. Store the result in the data memory byte
[EA] +- [EA) OR [RW)
OR the 16-bit contents of register RW with the data memory word addressed
by DADDR.
Store the result in the data memory word
----
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
t>-
I-
~nemonic Operand(s)
Statu.e.
Object Code
RCL
DADDR.N
110100va
aa011bbb
[DISP](DISP]
ROL
DADDR.N
110100va
aaOOObbb
Clock Cycles
N=1
15+EA;
N>1
4N+20+EA
o
X
Operation Performed
D
I
T
S
Z A
P C
X
I IT~ I
~
•
rg
0
9
i
;a.
0
m
(J1
~
E
•
Jj
[EA]
or DADDR may address a word:
~
.!c
(J1
Rotate the contents of the data memory location addressed by DADDR left
through the Carry status. If N = 1. then rotate one bit position. If N =CL. then
register CL contents provide the number of bit positions. Depending on prior
definition. DADDR may address a byte:
6
[EA)
[EA+1]
RCR
DADDR.N
110100va
aaOO1bbb
[DISP](DISP]
N=1 15+EA
X
X
As RCL. but rotate right
ROL
DADDR.N
110100va
aaOOObbb
[DISP](DISP]
N>1
4N+20+EA
X
X
Rotate the contents of the data memory location addressed by DAD DR left.
Move the left most bit.into the Carry status. If N = 1. then rotate one bit position. If N = CL. then register CL contents provides the number of bit positions.
Depending on prior definition. DAD DR may address a byte:
!
•c
Co)
i
~
~
0
E
:e•
~
all
'a
C
EJ...
0
Co)
•
(I)
+
I
[EA]
b
or DADDR may address a word:
EJ4
!~
[EAI
[EA+1]
6
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
G
~ Mnemonic Operand(s)
Object Code
Clock Cycles
t-
DADDR.N
Operation Performed
O D I T S Z A P C
110100va
aaOO1bbb
[DISP][DISP]
N=1 15+EA
X
X
SAL
As ROl. but rotate right
Shift the contents of the data memory location addressed by DADDR left.
Move the left most bit into the Carry status. If N = 1. then shift one bit position. If N = Cl. then register Cl contents provides the number of bit positions.
Depending on prior definition. DADDR may address a byte:
&-I
;;
G
-!=
[EA]
1--
0
or DADDR may address a word:.
~
0
g
~
i
;
Q.
0
Cf
m
m
~
E
G
~
G
SAR
DADDR.N
u
~
!
.!G
110100va
aa111bbb
[DISPJ[DISP]
N=1 15+EA;
N>1
.4N+20+EA
X
X X U X X
[EA]
[EA+1]
As SAL. but shift right and propagate sign:
-.J
CC
~
2
[EA]
~
or
q
E
G
:E
~
III
'tI
~
0
u
G
(/)
SBB
SBB
RB,DADDR
RW,DADDR
1A aaddd bbb
[DISPJ[DISP]
9+EA
1B aadddbbb
[DISPJ[DISP]
9+EA
X
X
[EA]
[EA+1]
~
X X X X X [RB] - [RB] - [EA] - [C]
X X X X
Subtract the contents of the data byte addressed by DAD DR from the contents of a-bit register RB, using twos complement arithmetic. Decrement the
result in RB if the Carry status was initially set
X [AW) - [RW) - [EA) - [e)
Subtract the contents of the 16-bit data word addressed by DAD DR from the
contents of the 16-bit register RW, using twos complement arithmetic.
Decrement the result in RW if the Carry status was initially set
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
G)
Q.
~
Mnemonic Operand(s)
SBB
~
CD
DADDA,AB
SBB
DADDA,AW
SHL
DADDA,N
SHA
DADDA,N
~
·Sc
0
Object Code
Clock Cycles
18 aasssbbb
[DISP][DISP]
16+EA
19 aasssbbb
[DISP][DISP]
16+EA
110100va
aa1p1bb
[DISP][DISP]
N=1 15+EA;
N>1
4N+20+EA
o D
X
X
I
Operation Performed
I
T
S
Z
A P C
X X X X
X X
X X
X
X
X
X X U X X
X
X
X U X
X
1
q
Q.
0
-.,j
~
E
G)
~
G)
u
c
...!
CD
SUB
AB,DADDA
2A aadddbbb
[DISP][DISP]
9+EA
X
X
SUB
AW,DADDA
2B aadddbbb
[DISP][DISP]
9+EA
X
X X X X X
SUB
DADDR,AB
28 aasssbbb
[DISP][DISP]
l6+EA
X
X X X X X
SUB
DADDR,RW
29 aasssbbb
[DISP][DISP)
l6+EA
X
X X X X X
TEST
DADDR,RB
84 aaregbbb
[DISP][DISP]
9+EA
0
X X U X 0
G)
a:
X X X X
~
0
E
CD
~
~
III
"a
[EA]
f---.E]
or
1!!
G)
a,
As SAL, but shift right:
O~
g
U'1
[EA] - [EA] - [AB] - [C]
Subtract the contents of 8-bit register AB from the data byte addressed by
DADDA, using twos complement arithmetic. Decrement the result in data
memory if the Carry status was initially set
[EA]- [EA] - [AW] - [C]
Subtract the contents of 16-bit register AW from the 16-bit data word addressed by DADDA, using twos complement arithmetic. Decrement the result
in data memory if the Carry status was initially set
This is an alternate mnemonic for SAL
c
o
u
G)
(/)
[EA]
[EA+1]
~
[RB] - [RB] - [EA]
Subtract the contents of the data memory byte addressed by DADDR from the
contents of a-bit register AB, using twos complement arithmetic
[AW] - [AW] - [EA]
Subtract the contents of the 1 6-bit data memory word addressed by DADDA
from the contents of l6-bit register AW, using twos complement arithmetic
[EA] - [EA] - [AB]
Subtract the contents of 8-bit register AB from the data memory byte addressed by DADDA, using twos complement arithmetic
[EA) - [EA) - [AW]
Subtract the contents of l6-bit register AW from the l6-bit data memory
word addressed by DADDA, using twos complement arithmetic
[EA) AND [RB)
AND the 8-bit contents of the data memory location addressed by DADDR
with the contents of 8-bit register RB. Discard the result, but adjust status
flags appropriately
----
I
!
I
I
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
G
Q.
> Mnemonic Operand(s)
Object Code
Clock Cycles
~
o
Operation Performed
0
I
T
S Z
A P C
TEST
DADDR,RW
85 aareg bbb
[DISP][DISP]
9+EA
0
X
X U X 0
CDu'a
c G
CD=
..
C
CD'CDC
XOR
RB,DADDR
32 aadddbbb
[DISP][DISP]
9+EA
0
X
X U X
o
~"i
e
....
XOR
RW,DADDR
33 aadddbbb
[DISP][DISP]
9+EA
0
X X U X
o
XOR
DADDR.RB
30 aasssbbb
[DISP][DISP]
16+EA
0
X X U X
0
XOR
DADDR,RW
31 aasssbbb
[DISP][DISP]
16+EA
0
X X U X 0
MOV
DADDR,
DATA8
C6 aaOOObbb
[DISP][DISP] yy
10+EA
MOV
DADDR,
DATA16
C7 aaOOObbb
[DISP][DISP] YYYY
10+EA
MOV
RB,DATA8
10110ddd YY
4*
MOV
RW,DATA16
10111ddd YYYY
4*
JMP
BRANCH
111010a1
DISP [DISP]
15**
JMP
BRANCH,
SEGM
EA PPOO PPOO
15*·
JMP
DADDR
FF aa100bbb
[DISP][DISP]
18+EA**
_....
a: o
>y
GI!
~~
~O
~~
5e
U'CD
CD'~
0
_
C1I
c'n
00
!
II
:sCD
E
oS
Q.
E
:s
..,
[EA] AND [RW]
AND the 16-bit contents of the data memory word addressed by DAD DR with
the contents of 16-bit register RW. Discard the result, but adjust status flags
appropriately
[RB] ....... [RB] XOR [EA]
Exclusive OR the 8-bit contents of register RB with the data memory byte addressed by DADDR. Store the result in RB
[RW] ....... [RW] XOR [EA]
Exclusive OR the 16-bit contents of register RW with the 16-bit data memory
word addressed by DADDR. Store the result in RW
[EA] ....... [RB] XOR [EA]
Exclusive OR the 8-bit contents of register RB with the data memory byte addressed by DADDR. Store the result in the addressed data memory byte
[EA] ....... [RW] XOR [EA]
Exclusive OR the 16-bit contents of register RW with the data memory word
addressed by DADDR. Store the result in the addressed data memory word
[EA] ....... DATA8
Load the immediate data byte OAT A8 into the data memory byte addressed
by DADDR
[EA] ....... DATA16
Load the immediate 1 6-bit data word OAT A 16 into the data memory word
addressed by DADDR
[RB] ....... DATA8
Load the immediate data byte DAT A8 into 8-bit register RB
[RW] ....... DATA16
Load the immediate 16-bit data word DATA 16 into 16-bit register RW
[PC] ....... [PC] + DISP
Jump direct to program memory location identified by label BRANCH. The
displacement DISP which must be added to the Program Counter will be computed as an 8-bit or 16-bit signed binary number, as needed, by the assembler
[PC] - DATA16, [CS] - DATA16
Jump direct into a new segment. BRANCH is a label which becomes a 16-bit
unsigned data value which is loaded into PC. SEGM is a label which becomes
another 16-bit unsigned data value that is loaded into the CS segment
register
[PC] ....... [EA]
Jump indirect in current segment. The 16~bit contents of the data memory
word addressed by DAD DR is loaded into PC
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
"D.
~
iMnemonic Operandls)
Statuses
Object Code
Clock Cycles
JMP
DADDR.CS
FF aa101bbb
[DISP] [DISP]
24+EA"
JMP
RW
FF 11100reg
11
CALL
BRANCH
E8 DISP DISP
1S""
CALL
BRANCH.
SEGM
SA PPQQ PPQQ
28""
CALL
DADDR
FF aa010bbb
[DISP][DISP]
21 +EA""
CALL
DADDR.CS
FF aa011bbb
[DISP][DISP]
37+EA""
'"S:;,
CALL
RW
FF 11010reg
16""
.&:I
RET
C3
8""
...c
0
g
a.
..,E:;,
c.n
m
co
~
II:
"
'a
c
ca
ii
Co)
e
:;,
en
RET
CS
CB
12""
RET
DATA16
C2 YYYY
17""
RET
CS.DATA16
CA YYYY
18""
o
D
I
TS
Operation Performed
Z A P C
[PC) - [EAI. [CS) - [EA+2)
Jump indirect into a new segment. The 16-bit contents of the data memory
word addressed by DADDRis loaded into PC. The next sequential 16-bit data
memory word's contents is loaded into the CS segment register
[PC]- [RW]
Jump to memory location whose address is contained in register RW.
liSP]] - [PCI. [SP] - [SP] - 2. [PC] - [PC] + DISP
Call a subroutine in the current program segment using direct addressing
liSP]] - [CSI. [SP] - [SP] -2. liSP]] - [PC]. [SP] - [SP] -2. [PC] DATA16.
[CS] - DATA 16
Call a subroutine in another program segment using direct addressing.
BRANCH and SEGM are labels that become different 16-bit data words; they
are loaded into PC and CS. respectively
liSP)) ..... [PC). [SP) - [SP) -2. [PC)- [EA)
Call a subroutine in the current program segment using indirect addressing.
The address of the subroutine called is stored in the 16-bit data memory
word addressed by DADDR
IISP]]- [CSI. [SP] - [S2] -2. liSP]] - [PC]. [SP]- [SP] -2. [PC] - [EAI.
[CS]- [EA+2]
Call a subroutine in a different program segment using indirect addressing.
The addre~s of the subroutine called is stored in the 16-bit data memory
word addressed by DADDR. The new CS register contents is stored in the
next sequential program memory word
[SP] - [PC].[SP] - [SP-21. [PC] ..... [RW]
Call a subroutine whose address is contained in register RW.
[PC] - [[SP]]. [SP] ..... [SP] + 2
Return from a subroutine in the current segment
[PC]- [[SP]1. [SP] ..... [SP] +2. [CS]- liSP]]. [SP] ..... [SP] +2
Return from a subroutine in another segment
[PC]"'" IISP]]. [SP]- [SP] +2 +DATA16
Return from a subroutine in the current segment and add an immediate displacement to SP
[PC]- [[SP)1. [SP) - [SP) +2. [CS)- liSP]]. [SP) ..... [SP) +2 +DATA16
Return from a subroutine in another segment and add an immediate displacement to SP
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
II
a. Mnemonic Operand(s)
>
Object Code
Clock Cycles
t-
o
I
T
S
Z A P C
X X X X
ADD
AL,DATAS
04 yy
4·
X
X
ADD
AX,DATA16
05 YYYY
4·
X
X
ADD
RB,DATAS
SO 11 OOOddd YY
4·
X
X
ADD
RW,DATA16
4·
X
X
ADD
17+EA
X
X
17+EA
X
X
ADe
DADDR,
DATAS
DADDR,
DATA16
AL,DATAS
S1 11000ddd
YYYY
SO aaOOObbb
[DISP][DISP] YY
S1 aaOOObbb
[DISP][DISP) YYYY
14 YY
4·
X
X
ADe
AX,DATA16
15 YYYY
4·
X
X
ADD
?'
....,
Operation Performed
O D
ADe
B,DATAS
SO 11010ddd YY
4·
X
X
f
G
a.
ADe
RW,DATA16
X
X
ADe
DADDR,
DATA8
S1 11010ddd
YYYY
SO aa010bbb
[DISP][DISP] YY
4·
!l1li
17+EA
X
X
ADe
DADDR,
DATA16
S1 aa010bbb
[DISP][DISP] YYYY
17+EA
X
X
AND
AL,DATAS
24 YY
4·
0
X
AND
AX,DATA16
25 YYYY
4·
0
X
AND
RB,DATAS
SO 11100ddd YY
4·
0
X
AND
RW,DATA16
4·
0
X
AND
DADDR,8
S1 11100ddd
YYYY
80 aa100bbb
[DISP][DISP] YY
17+EA
0
X
AND
DADDR,
DATA16
17+EA
0
X
!
0
:sG
E
-E
81 aa100bbb
[DISP][DISP] YYYY
[AL] - [All + DAT AS
Add S-bit immediate data to the AL register
X X X X [AX)- [AX) + DATA16
Add 16-bit immediate data to the AX register
X X X X [RB) - [RB] + DAT AS
Add S-bit immediate data to the RB register
X X X X [RW]- [RW] + DATA16
Add 16-bit immediate data to the RW register
X X X X [EA] - [EA] + DAT AS
Add S-bit immediate data to the data memory byte addressed by DADDR
X X X X [EA] - [EA] + DATA16
Add 16-bit immediate data to the data memory word addressed by DADDR
X X X X [AL] - [AL] + DAT AS + [e]
Add S-bit immediate data, plus carry, to the AL register
X X X X [AX]- [AX] + DATA16 + [e]
Add 16-bit immediate data, plus carry, to the AX register
X X X X [RB] - [RB] + DAT AS + [e]
Add S-bit immediate data, plus carry, to the RB register
X X X X [RW]- [RW] + DATA16 + [e]
Add 16-bit immediate data, plus carry, to the RW register
X X X X [EA] - [EA] + DATA8 + [e]
Add 8-bit immediate data, plus carry, to the data memory byte addressed by
DADDR
X X X [EA] - [EA] + DATA16 + [e]
Add 16-bit immediate data, plus carry, to the data memory word addressed
by DADDR
X U X 0 [ALI - [ALI AND DAT AS
AND S-bit immediate data with AL register contents
X U X 0 [AX]- [AX] AND DATA16
AND 16-bit immediate data with AX register contents
X U X 0 [RB] - [RB] AND DAT AS
AND 8-bit immediate data with RB register contents
X U X 0 [RW]- [RW] AND DATA16
AND 16-bit immediate data with RW register contents
X U X 0 [EA] - [EA] AND DAT A8
AND 8-bit immediate data with contents of data memory byte addressed by
DADDR
X U X 0 [EA]- [EA] AND DATA16
AND 1 6-bit immediate data with contents of 1 6-bit data memory word addressed by DADDR
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
G)
CL
>- Mnemonic Operand(s).
Object Code
Clock Cycles
~
;:;
G)
:I
cr
"""
~
I
T S
Z
Al,DATA8
3C YY
4·
X
X X
CMP
AX,DATA16.
3D YYYY
4·
X
X X
CMP
RB,DATA8
80 11111 ddd YY
4·
X
X
CMP
RW,DATA16t
4·
X
X X
CMP
DADDR,
DATA8
100000a1
1111ddd
YY [YY)
80 aa111bbb
[DISP)[DISP] YY
10+EA
X
X X
CMP
DADDR,
DATA16
10+EA
X
X X
OR
Al,DATA8
100000a1
aa111bbb
[DISP)[DISPI'YY\[YY]
OC YY
4·
0
X
OR
AX,DATA16·
00 YYYY
4·
0
X X
OR
RB,DATA8
80 11001ddd YY
4·
0
X
OR
RW,DATA16
81 11001ddd
YYYY
4·
0
X X
OR
DADDR,
DATAS
80 aa001bbb
[DISP)[DISP] YY
17+EA
0
X
OR
DADDR,
DATA16
17+EA
0
X X
SBB
Al.DATAS
81 aa001bbb
[DISP)[DISP)
YYYY
1C YY
4·
X
X
X
SBB
AX,DATA16:
10 YYYY
4·
X.
X
X,
0
CL
D
CMP
'Sc
g
!
I!
G)
0
X
X
0
!III
X
:s
G)
e
.5
X
Operation Performed
A
P C
X X X [All - DATA8
Subtract 8-bit immediate data from Al register contents. Discard result, but
adjust status flags
X X X [AX] - DATA16
Subtract 16-bit immediate data from AX register contents. Discard result, but
adjust status flags
X X X [RB] - DATA8
Subtract S-bit immediate data from RB register contents. Discard result, but
adjust status flags
X X X [RW] - DATA16
Subtract 16-bit immediate data from RW register contents. Discard result,
but adjust status flags
X X X [EA] - DATA8
Subtract 8-bit immediate data from contents of data memory byte addressed
by DADDR. Discard result, but adjust status flags
X X x. [EA] - DATA16
Subtract 16-bit immediate data from contents of 16-bit data memory word
addressed by DADDR. Discard result, but adjust status flags
U X o [All - [All OR OAT A8
OR 8-bit immediate data with Al register contents
U X 0 [AX] - [AX] OR DATA16
OR 16-bit immediate data with AX register contents
U X 0 [RB] - [RB] OR OAT A8
OR 8-bit immediate data with RB register contents
U X 0 [RW]- [RW] OR DATA 16
OR 16-bit immediate data with RW register contents
U' X 0 [EA] - [EA] OR DATA S
OR S-bit immediate ata with contents of data memory byte addressed by
DADDR
U X 0 [EA] - [EA] OR DATA16
OR 16-bit immediate data with contents of 16-bit data memory word addressed by DADDR
X X X [Al] - [Al] - OAT A8 - [e]
Subtract S-bit immediate signed binary data from Al register contents using
twos complement arithmetic. If the Carry status was originally 1 decrement
the result
X X X [AX) - [AX) - OATA16 - [C)
Subtract 16-bit immediate signed binary data from AX register contents
using twos complement arithmetic. If the Carry status was originally 1 decrement the result
I
I
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
II
Q.
> Mnemonic Operand(s)
...
Object Code
Clock Cycles
Operation Performed
0
I
D
T
S Z A
P C
X X
SBB
RB,DATA8
80 11011ddd YY
4'
X
X X X
SBB
RW,DATA16
100000al
11011ddd
yy [YY]
4'
X
X X X X
SBB
DAD DR,
DATA8
80 aaOllbbb
[DISP)[DISP] YY
17+EA
X
X
SBB
DAD DR,
DATA16
100000al
aaOl1bbb
[DISP)[DISP)YY [YY)
17+EA
X
X X X X
SUB
AL,DATA8
2C YY
4'
X
X
X X X
SUB
AX,DATA16
2D YYYY
4'
X
X
X X X
SUB
RB,DATA8
80 lll01ddd YY
4"
X
X X X X
SUB
RW,DATA16
81 11101ddd
YYYY
4'
X
X X X X
SUB
DADDR,
DATA8
80 aal0lbbb
[DISP)[DISP] YY
17+EA
X
X X X X
SUB
DADDR,
DATA16
100000al
aal 01 bbb
[OISP][DISP]YY [YY]
17+EA
X
X
A8 YY
4'
~
X X X
G)
:I
'Sc
Cf'
-..J
y
'"
$
fG)
0
Q.
0
$
III
:s
G)
E
.5
TEST
AL,DATA8
X X X
0
X X U X
[RB) +- [RB) - DAT A8 - [C)
Subtract 8-bit immediate signed binary data from RB register contents using
twos complement arithmetic. If the Carry status was originally 1 decrement
the result
X [RW] +- [RW] - DATA 16 - [C]
Subtract 16-bit immediate signed binary data from RW register contents
using twos complement arithmetic. If the Carry status was originally 1 decrement the result
X [EA] +- [EA] - DATA8 - [e]
Subtract 8-bit immediate signed binary data from contents of data memory
byte addressed by DAD DR using twos complement arithmetic. If the Carry
status was originally 1 decrement the result
X [EA] +- [EA] - DATA 16 - [e]
Subtract 16-bit immediate signed binary data from contents of l6-bit data
memory word addressed by DADDR using twos complement arithmetic. If the
Carry status was originally 1 decrement the result
X [ALI +- [ALI - DA T A8
Subtract the 8-bit immediate signed binary data from AL register contents
using twos complement arithmetic
X [AX] +- [AX] - DATA16
Subtract the 16-bit immediate signed binary data from AX register contents
using twos complement arithmetic
X [RB] +- [RB] - DA T A8
Subtract the 8-bit immediate signed binary data from RB register contents
using twos complement arithmetic
X [RW] +- [RW] - DATA16
Subtract the 16-bit immediate signed binary data from RW register contents
using twos complement arithmetic
X [EA] +- [EA] - DA T A8
Subtract the 8-bit immediate signed binary data from the contents of the data
memory byte addressed by DADDR using twos complement arithmetic
X [EA] +- [EA] - DATA 16
Subtract the 16-bit immediate Signed binary data from the contents of the
16-bit data memory word addressed by DADDR using twos complement
arithmetic
0 [All AND DAT A8
AND the 8-bit immediate data and AL register contents. Discard the result but
adjust status s
--
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
!
Statuse,
Mnemonic Operand(s) .
Object Code,
Clock Cycles
~
;:;
Q)
~
0
Operation Performed
D I
T S
Z A
P C·
TEST
AX,DATA16
AS YYYY
4-
0
X X U X
0
TEST
RB,DATAS
F6 11 OOOddd YY
5-
0
X
0
TEST
RW,DATA16
F7 11000ddd
YYYY
5-
0
X X U X 0
TEST
DADDR,
DATAS
F6 aaOOObbb
[DISP][DISP] YY
11+EA
0
X X U X 0
TEST
DADDR,
DATA16
F7 aaOOObbb
[DISP][DISP] YYYY
11+EA
0
X X U X 0
XOR
AL,DATA8
34 YY
4-
0
X X U X 0
XOR
AX,DATA16
35 YYYY
4-
0
X. X U X 0
XOR
RB,DATAS
SO 1111 Oddd YY
4-
0
X X U X 0
XOR
RW,DATA16
4-
0
X X U X
XOR
DADDR,
DATAS
S1 11110ddd
YYYY
SO aa010bbb
[DlsP][DISP] YY
17+EA
0
X X U X 0
XOR
DADDR,
DATA16
81 aa010bbb
[DISP][DISP]
YYYY
17+EA
0
X X U X 0
LOOP
DISPS
E2 DISP
5 or 17--
LOOPE
DISP8
E1 DISP
6 or 1S--
LOOPNE
DlsPS
EO DlsP
5 or 1S--
LOOPNZ
LOOPZ
JA
DIsP8
DlsPS
DISP8
77 DISP
4 or 16--
..;c
g
o
$
X U X
IV
Ii~
c.n
..:...
w
0
$
IV
:s
Q)
E
.E
c
~
:sc
0
(.)
c
0
~
u
c
I!
ID
0
[AX] AND DATA16
AND the 16-bit immediate data and AX register contents. Discard the result
but adjust status flags
[RB] AND OAT AS
AND the S-bit immediate data and RB register contents. Discard the result but
adjust status flags
[RW] AND DATA16
AND the 16-bit immediate data and RW register contents. Discard the result
but adjust status flags
[EA] AND OAT AS
AND the 8-bit immediate data and the contents of the data memory location
addressed by DADDR. Discard the result but adjust status flags
[EA] AND DATA16
AND the 16-bit immediate data and the contents of the 16-bit data memory
word addressed by DADDR. Discard the result but adjust status flags
[All - [All XOR OAT A8
Exclusive OR 8-bit immediate data with AL register contents
[AX] - [AX] XOR DATA16
Exclusive OR 16-bit immediate data with AX register contents
[REt] - [RB] XOR OAT AS
Exclusive OR 8-bit immediate data with RB register contents
{RW]- [RW] XOR DATA16
Exclusive OR 16-bit immediate data with RW register contents
[EA] - [EA] XOR OAT AS
Exclusive OR S-bit immediate data with contents of the data memory byte addressed by DADDR
[EA] - [EA] XOR DATA16
Exclusive OR 16-bit immediate data with contents of the 16-bit data memory
word addressed by DADDR
[eX] - [CX] -1 If [CX] i= 0 then [PC] - [PC] + DISPS
Decrement CX register and branch if CX contents are not 0
[CX] - [CX] -1 If [CX] i= 0 and [Z] = 1 then [PC] + DlsPS
Decrement eX register and branch if CX contents is not 0 and Z status is 1
[CX] - [CX] -1 If [CX] i= 0 and [Z] = 0 then [PC] - [PC] + DISP8
Decrement CX register and branch if CX contents is not 0 and Z status is 0
See LOOPNE
See LOOPE
[PC] - [PC] + DISPS
Branch if C or Z is 0
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
CD
a.
> Mnemonic Operand(s)
Object Code
Clock Cycles
~
-:aCD
..
:::s
c
c
C11
I
-.,j
~
JAE
DISPB
73 DISP
4 or 16"
JB
DISPB
72 DISP
4 or 16"
JBE
DISPB
76 DISP
4 or 16"
JCXZ
DISPB
E3 DISP
6 or lS"
JE
DISP8
74 DISP
4 or 16"
JG
DISPS
7F DISP
4 or 16"
JGE
DISPB
7D DISP
4 or 16"
JL
DISPB
7C DISP
4 or 16"
JLE
DISPB
7E DISP
4 or 16"
JNA
JNAE
JNB
JNBE
JNE
DISPB
DISP8
DISP8
DISP8
DISP8
75 DISP
4 or 16"
JNG
JNGE
JNL
jnle
JNO
DISP8
DISP8
DISPS
disp8
DISP8
71 DISP
4 or 16"
JNP
DISP8
7B DISP
4 or 16"
JNS
DISP8
79 DISP
4 or 16"
JNZ
JO
DISPS
DISP8
70 DISP
4 or 16""
JP
DISP8
7A DISP
4 or 16"
JPE
DISP8
0
g
c
~
:ac
0
u
c
0
s:u
c
f
l1li
Operation Performed
0
0
I
T
S Z
A P C
[PC] +- [PC] + DISPS
Branch if C is 0
[PC] + - [PC] + DISP8
Branch if C is 1
[PC] +- [PC] + DISPS
Branch if C or Z is 1
[PC] + - [PC] + DISP8
Branch if the CX register contents is 0
[PC] + - [PC] + DISPS
Branch if Z is 1
[PC] + - [PC] + DISPS
Branch if Z is 0 or the Sand 0 statuses are the same
[PC] + - [PC] + DISPS
Branch if the Sand 0 statuses are the same
[PC] + - [PC] + DISPB
Branch if the Sand 0 statuses differ
[PC] + - [PC] + DISPS
Branch if Z is 1 or the Sand 0 statuses differ
See JBE
See JB
See JAE
See JA
[PC] + - [PC] + DISP8
Branch if Z is 0
See JLE
See JL
See JGE
See JG
[PC] + - [PC] + DISP8
Branch if 0 is 0
[PC] + - [PC] + DISP8
Branch if P is 0
[PC] +- [PC] + DISP8
Branch if S is 0
See JNE
[PC] +- [PC] + DISPS
Branch if 0 is 1
[PC] + - [PC] + DISP8
Branch if P is 1
See JP
--
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
CD
a.
> Mnemonic Operand Is)
Object Code
Clock Cycles
~
..,l
JPO
0lSP8
0
JS
0lSP8
JZ
DlSP8
MOV
CD
..
c
Operation Performed
0
D
I
T S
Z A
P C
See JNP
[PC] ...... IPC] + 0lSP8
Branch if S is 1
See JE
780lSP
4 or 16"
RBO.RBS
8A11dddsss
2'
MOV
RWO.RWS
8B 11dddsss
2'
2
!
MOV
SR.RW
8E 11 Orrsss
2'
'61
MOV
RW.SR
8C 110rrddd
2'
XCHG
AX.RW
10010reg
3'
XCHG
RB.RB
86 11 regreg
4'
XCHG
RW.RW
87 11regreg
4'
CMPS
BO.BS
A6
22
X I/O
X
CMPS
WO.WS
A7
22
X /0
X X X X
LOOS
BO.BS
AC
12
I/O
.
LOOS
WO.WS
AO
12
I/O
I!
MOVS
BO.BS
A4
18
I/O
MOVS
WO.WS
A5
18
I/O
9
(,)
0
ID
>
0
II)
CD
[RBO] ...... [RBS]
Move the contents of any RB register to any RB register
[RWO] ...... [RWS]
Move the contents of any RW register to any RW register
[SR] ...... [RWS]
Move the contents of any RW register to any Segment register
[RWO] ...... [SR]
Move the contents of any Segment register to any RW register
[AX] ...... - [RW]
Exchange the contents of AX and any RW register
[RB] ...... - [RB]
Exchange the contents of any two RB registers
[RW] ...... - [RW]
Exchange the contents of any two RW registers
a::
.
I
!
II)
'61
CD
a::
CTI
~
CTI
~
~
II
II
CI)
"tI
c
II
-!c
~
~
u
0
iii
[[SI11 - [[01]]. [SIl ...... [SIl ± 1. [Oil ...... [Oil ± 1
Compare the data bytes addressed by the SI and 01 Index registers using
string data addressing'
X [[5111 - [[01]1. [SIl ...... [SIl ± 2. [01] ...... [Oil ± 2
Compare the 16-bit data words addressed by the 51 and 01 Index registers
using string data addressing'
[All ...... [[5111. [SIl ...... [SIl ± 1
Move a data byte from the location addressed by the 51 Index register to the
AL register using string data addressing
[AX] ...... [[5111. [SIl ...... [SIl ± 1
Move a data word from the 16-bit location addressed by the 51 Index register
to the AX register using string data addressing
[[0111 ...... [[51]]. [SIl ...... [SIl ± 1. [01]""" [Oil ± 1
Move a data byte from the location addressed by the 51 Index register to the
extra segment location addressed by the 01 register using string data addressing'
[[Olll ...... [[5111. [SIl ...... [51] ± 2. [01] ...... [Oil ± 2
Move a 1 6-bit data word from the location addressed by the 51 Index register
to the extra segment location addressed by the 01 Index register using string
data addressing'
• For these instructions. the default destination segment register cannot be
overri,den.
X X X X
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
CD
...~
Statuses
Mnemonic Operand(s)
Object Code
Clock Cycles
Operation Performed
0
0
I
T
S Z A
P C
REP
N
1111001z
+2 per loop
SCAS
BO,BS
AE
15
X I/O
X
SCAS
WO,WS
AF
15
X I/O
X X
X X
X
STOS
BO,BS
AA
11
X I/O
X X
X X
X
STOS
WO,WS
AB
11
X I/O
X X X X
X
AOC
RBO,RBS
12 11dddsss
3'
X
X X X X X
AOC
RWO,RWS
13 11 dddsss
. 3'
X
X X X X X
.
ADD
RBO.RBS
02 11dddsss
3'
X
X
X X X
X
ADD
RWO.RWS
03 11dddsss
3'
X
X X X X
X
&
0
...
AND
RBO.RBS
22 11dddsss
3'
0
X X U
o
'51
AND
RWO.RWS
23 11dddsss
3'
0
X X
U X 0
98
2'
X X
I/O
;;
CD
~
~c
0
.g
.c
X
X X X
2
a!
CD
en
'a
C
a!
•
';
c
!
~
~
u
0
iii
cr
en
"""
CD
!
!1ft
CD
X
a:
I
!
CBW
1ft
'51
CD
X
CMP
RBO.RBS
3A 11dddsss
3'
X
X X
eMP
RWO.RWS
3B 11dddsss
3'
X
X X X X X
99
5
a:
CWO
Repeat the next sequential instruction (which must be a Block Transfer and
Search instruction) until CX contents decrements to O. Decrement CX contents on each repeat. If the next instruction is CMPB, CMPW, SCAB, or
SCAW then repeat until CX contents decrements to 0 or Z status does not
equal N
[AL] - [[Dill, [Oil - [Oil ± 1
Compare AL register contents with the extra segment data byte addressed by
the 01 Index register using string data addressing
[AX] - [[Dill, [Oil - [Oil ± 2
Compare AX register contents with the extra segment 1 6-bit data ord addressed by the 01 Index register using string data addressing
[[0111 - [AL), [Oil - [Oil ± 1
Store the AL register contents in the extra segment data memory byte addressed by the 01 Index register using string data addressing
[[0111 - [AXI. [Oil - [Oil ± 2
Store the AX register contents in the extra segment 16-bit data memory word
addressed by the 01 Index register using string data addressing
[RBO) - [RBO] + [RBS) + [C]
Add the 8-bit contents of register RBS, plus the Carry status, to register RBO
[RWO) - [RWO] + [RWS] + [C]
Add the 16-bit contents of register RWS. plus the Carry status. to register
RWO
[RBO) - [RBO] + [RBS)
Add the 8-bit contents of register RBS to register RBO
[RWO) - [RWO) + [RWS)
Add the 16-bit contents of register RWS to register RWO
[RBO] - [RBO] AND [RBS]
AND the 8-bit contents of register RBS with register RBO
[RWO] - [RWO] AND [RWS)
AND the 1 6-bit contents of register RWS with register RWO
[AH]- [AL7)
Extend AL sign bit into AH
[RBO] - {RBS]
Subtract the contents of register RBO from register RBS. Discard the result.
but adjust status flags
[RWO] - [RWS]
Subtract the contents of register RWO from register RWS. Discard the result,
but adjust status flags
[OX) - [AX15)
Extend AX sign bit into OX
- -
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
•
...>
~
Statu.e.
!Mnemonic Operand(.)
Object Code
Clock Cycle.
o
Operation Performed
D I
T S Z
A P
C
OIV
RBS
F6 11110555
80-90
U
U U U U
u
DIV
RWS
F7 1111 Osss
144-162
U
U U U U
u
IDIV
RBS
F6 11111 sss
101-112
U
U U U U U
IOIV
RWS
F7 11111 sss
165-184
U
U U U U U
IMUL
RBS
F6 11101 sss
80-98
X
U U U U X
IMUL
RWS
F7 11101 sss
128-154
X
U U U U X,
MUL
RBS
F6 11100sss
70-17
X
U U U U
MUL
RWS
F7 11100sss
118-133
X
U. U UU X
OR
RBD.RBS
OA 11dddsss
3"
0
X X U X 0
OR
RWD.RWS
OB l1dddsss
3"
0
X X U X 0\
~
•
~
'Sc
0
cr
""
g
!
\!!
•
CL
..
0
!
II)
'81
•
II:
..
!
I
II)
'81
•
II:
X
[AX] +- [AX]/[RBS]
Divide the 16-bit contents of AX by the 8-bit contents of RBS. Store the integer quotient in AL and the remainder in AH. If the quotient is greater than
FF16. execute a "divide by 0" interrupt
[DX) [AX) +- [DX) [AX)/[RWS]
Divide the 32-bit contents of registers DX (high-order) and AX (low-order) by
the 16-bit contents of RWS. Store the integer quotient in AX and the remainder in DX. If the quotient is greater than FFFF16. execute a "divide by 0"
interrupt
[AX] +- [AX]/[RBS]
Divide the 16-bit contents of register AX by the 8-bit contents of RBS. treating both contents as signed binary numbers. Store the quotient. as a signed
binary number. in AL. Store the remainder. as an unsigned binary number. in
AX. Store the remainder. as an unsigned binary number. in AH. If the quotient
is greater than 7F16. or less than -8016. execute a "divide by 0" interrupt
[OX] [AX] +- [OX] [AXlI[RWS]
Divide the 32-bit contents of registerl>X (high-order) and AX (low-order) by
the 16-bit contents of RWS. Treat both contents as signed binary numbers.
Store the quotient. as a signed binary number. in AX. Store the remainder. as
an unsigned binary number. in AH. If the quotient is greater than 7FFF16. or
less than -800016. execute a "divide by 0" interrupt
[AX) +- [AL) " [RBS)
Multiply the 8-bit contents of register AL by the contents of RBS. Treat both
numbers as signed binary numbers. Store the 16-bit product in AX
[DX) [AX) +- [AX)" [RWS)
Multiply the 16-bit contents of register AX by the 16-bit contents of RWS.
Treat both numbers as signed binary numbers. Store the 32-bit product in DX
(high-order word) and AX (low-order word)
[AX] - [AL] " [RBS]
Multiply the 8-bit contents of register AL by the contents of RBS. Treat both
numbers as unsigned binary numbers. Store the 16-bit product in AX
[DX) [AX) - [AX] .. [RWS)
Multiply the l6-bit contents of register AX by the 16-bit contents of RWS.
Treat both numbers as unsigned binary numbers. Store th,e 32-bit product in
DX (high-order word) and AX (low-order word)
[RBD] - [RBD] OR [RBS]
OR the 8-bit contents of register RBS with register RBD
[RWD]- [RWD] OR [RWS]
OR the l6-bit contents of register RWS with register RWD
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
t>
t-
~
Statuses
Mnemonic Operand(s)
Object Code
Clock Cycles
Operation Performed
O 0
I
T
S
Z A
P C
SBB
RBD.RBS
lA lldddsss
3'
X
X X X X X
SBB
RWD.RWS
lB lldddsss
3'
X
X X X X X
SUB
RBD.RBS
2A lldddsss
3'
X
X X X X X
SUB
RWD.RWS
2B 11 dddsss
3'
X
X X X X X
TEST
RBD.RBS
8411regreg
3'
0
X
TEST
RWD.RWS
85 11 regreg
3'
0
X X U X
XOR
RBD.RBS
30 lldddsss
3'
0
X X U X 0
XOR
RWD.RWS
31 lldddsss
3'
0
X X U X
AAA
37
4'
U
U U X U X
AAD
D50A
SO
U
X X U X
AAM
D40A
83
U
X X U X U
AAS
3F
4·
U
U U X
DAA
27
4'
U
X X X X X
DAS
2F
4'
U
X X X X X
CD
:::I
~c
0
g
!
f
CD
a.
0
...
!III
"61
X U X 0
CD
II:
...I
$
III
0
"61
cr.....
Q)
II:
co
!
f
Q)
a.
0
U
U X
0
!
III
"61
CD
II:
DEC
RB
FE 11001ddd
3'
X
X
X X X
DEC
RW
01001ddd
2'
X
X
X X X
[RBD) - [RBD) - [RBS) ~ [C)
Subtract the 8-bit contents of register RBS from RBD using twos complement
arithmetic. If the Carry status was originally 1 decrement the result
[RWD) - [RWD) - [RWS) - [C)
Subtract the lS-bit contents of register RWS from RWD using twos complement arithmetic. If the Carry status was originally 1 decrement the result
[RBD) - [RBD) - [RBS)
Subtract the 8-bit contents of register RBS from RBD using twos complement
arithmetic
[RWD) - [RWD) - [RWS)
Subtract the lS-bit contents of register RWS from RWD using twos complement arithmetic
[RBD) AND [RBS)
AND the a-bit contents of register d and register RBS. Discard the result. but
adjust status flags
[RWD) AND [RWS)
AND the lS-bit contents of register RWD and register RWS. Discard the
result. but adjust status flags
[RBD) - [RBD) XOR [RBS)
E~clusi~e OR the 8-bit contents of register RBS with register RBD
[RWD) - [RWD) XOR [RWS)
~xclusive OR the 16-bit contents of register RWS with register RWD
ASC" adjust AI register contents for addition (as described in accompanying
text)
Decimal adjust dividend in AL prior to dividing an unpacked decimal divisor.
to generate an unpacked decimal quotient. (See accompanying text for
details)
After multiplying 0 unpacked decimal operands. adjust product in AX to
become an unpacked decimal result. (See accompanying text for details)
After subtracting two unpacked decimal numbers. adjust the difference in AL
so that it too is an unpacked decimal number. (See accompanying text for
details)
After adding two packed decimal numbers. adjust the sum in AL so that it too
is a packed decimal number. (See accompanying text for details)
After subtracting two packed decimal numbers. adjust the difference in AL so
that it too is a packed decimal number. (See accompanying text for details)
[RB] - [RB] -1
Decrement the 8-bit contents of register RB
[RW) - [RW) -1
Decrement the 1S-bit contents of register RW
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
Statuses
GJ
~ Mnemonic Operand(s)
t-
-:aGJ
:::s
c
"+l
1!
II
a.
..
!
0
cr
.....
co
1/1
"61
II
a:
Clock Cycles
Operation Performed
O 0
INC
RB
FE 11000ddd
3-
X,
INC
RW
01000ddd
2-
X
NEG
RB
F611011ddd
3-
X
NEG
RW
F7 11011ddd
3-
X
NOT
RB
F6 11010ddd
3-
RW
F7 11010ddd
RCL
RCL
RCR
RCR
ROL
ROL
ROR
ROR
SAL
SAL
SAR
SAR
SHL
SHL
SHR
SHR
RB,N
RW,N
RBN
RW,N
RB,N
RW,N
RB,N
RW,N
RB,N
RW,N
RB,N
RW,N
RB,N
RW,N
RB,N
RW,N
~ 10100vO 11010ddd
~10100v1 11010ddd
POP
DADDR
SF aaOOObbb
[DISP][DISP]
17+EA
POP
POP
RW
SR
01011ddd
000rr111
8
a
90
8.
FF aa1l0bbb
[DISP][DISP]
16+EA
~ 10100vO 11011ddd
~ 10100vl 110l1ddd
110100vO
110100v1
110100vO
~ 10100v1
110100vO
~ 10l00v1
110100vO
~ 10100v1
11 OOOddd
11000ddd
11001ddd
11001ddd
11100ddd
11100ddd
f1111ddd
1llllddd
110100vO 11101ddd
~ 10100vl 1ll0lddd
I
T
S
Z A
P C
X X X X
[RB] - [RB] + 1
Increment the a-bit contents of register RB
[RW] - [RW] + 1
X X X X
Increm..!!!! the 16-bit contents of register RW
X X X X X [RB] - [RB] + 1
Twos ~Iement the a-bit contents of register RB
X X X X X [RW] - [RW] + 1
Twos complement the 16-bit contents of register RW
[RB] -
3-
NOT
C
0
g
S
Object Code
N=1 2N>1 4N+8
N=l 2N>1 4N+8
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X X
X X
U
U
U
U
U
U
U
U
X
X
X
X
X
X·
X
X
~
u
S
en
POPF
PUSH
DADDR
X
X X
X X_ .X XX
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
[RBi
Ones c Mnemonic Operand(s)
Object Code
Clock Cycles
I-
....c
Operation Performed
0
PUSH
PUSH
RW
SR
01010rrr
000rr110
PUSHF
9C
10
INT
INT
INTO
3
V
CC
CD YY
CE
52
51
4 or 53
IRET
CF
24
CLC
Fa
2·
CLD
FC
2·
CLI
FA
2·
CMC
F5
2·
LAHF
9F
4·
0
I
T
S
Z
A
P C
[SP) +- [SP) -2, [(SPII +- [RW or SR)
Store the contents of the specified 1 6-bit register in the 16-bit Stack word
addressed using Stack addressing. Decrement SP by 2
[SP) +- [SP) +2, [[SPII - [SFR)
Store the Status flags register contents in the 16-bit Stack word addressed
using Stack addressing. Decrement SP by 2
11
10
0
g
~
u
ca
U;
J!!
a.
E
!
.E
U1
cO
o
0
0
0
Execute a software interrupt and vector through table entry 3
Execute a software interrupt and vector through table entry V
If the 0 status is 1, execute a software interrupt and vector through table entry 1016
Return from interrupt service routine
0
0
0
[C)-O
Clear Carry status
[0)-0
Clear Decrement/Increment select
(1)-0
Clea!:..!r'terrupt enable status, disabling all interrupts
X [C) -[C)
Complement Carry status
Transfer flags to AH register as follows:
0
0
0
7
6
5
4
3
2
1
0
S
Z
0
A
0
P
I
C
I I I II I III
Bit no.
AH register
1/1
af!
U)
SAHF
9E
4·
X
X X X X
Transfer AH register contents to status flags as follows:
7
6
5
4
3
2
1
I IIIII II
S
STC
F9
2·
STD
FD
2·
STI
FB
2·
1
1
Z
A
P
0
Bit no.
IAH register
C
1 [C)-1
Set Carry status to 1
(0)-1
Set Decrement/Increment status to 1
(1)-1
Set interrupt enable status to 1, enabling all interrupts
Table 5-4. A Summary of 8086 and 8088 Instructions (Continued)
II
a. IMnemonlc! Operand(s)
>-
Statuses
Object Code
ESC
~
-S
DADDA
HI-T
LOCK
0
11011xxx
aaxxxbbb
[DISP][DISP)
F4
FO
8+EA
001reg110
+2
98
90
3+5n
3"
2"
2"
0
SEG
WAIT
NOP
SA
Operation Performed
Clock Cycles
I-
D I
T
S Z A P C
7 - [EA)
The contents of the data memory location addressed by DADO A is read out of
memory and placed on the data bus; however, it is not input to the CPU
CPU Halt
Guarantee the CPU bus control during execution of the next sequential instruction
The next sequential allowed memory reference instruction accesses the segment identified by Segment register SA. See Table 20-1 for allowed memory
reference instructions
CPU enters the WAIT state until TEST pin receives a high input signal
No operation (This is the same object code as XCHG, AX, AX.)
C11
~
i
I
I
Table 5-5. 8086 and 8088 Instruction Mnemonics
Object Code
Bytes
Clock
Periods
37
D50A
D40A
3F
14 YY
15 YYYY
SO aa010bbb
[DISP) [DISP) YY
1 OOOOOa 1 aaO 1Obbb
[DISP) [DISP) YY[YY)
10 aasssbbb
[DISP) [DISP)
11 aasssbbb
[DISP) [DISP)
12 aadddbbb
[DISP) [DISP)
SO 11010ddd YY
12 11dddsss
13 aadddbbb
[DISP) [DISP)
100000a111010ddd
YY[YY)
13 11dddsss
04 YY
05 YYYY
SO aaOOObbb
[DISP) [DlSP] YY
1OOOOOa 1 aaOOObbb
[DISP) [DISP) YY[YY)
00 aasssbbb
[DISP) [DISP)
01 aasssbbb
[DISP) [DISP)
02 aadddbbb
[DISP) [DlSP]
SO 11 OOOddd YY
02 11dddsss
03 aadddbbb
[DISP) [DISP)
100000a1 11000ddd
YY[YY)
03 11dddsss
24 YY
25 YYYY
SO aa100bbb
[DISP) [DISP) YY
81 aa100bbb
[DISP) [DISP] YYYY)
20 aasssbbb
[DISP] [DISP]
21 aasssbbb
[DISP) [DISP]
22 aadddbbb
[DISP) [DISP]
SO 111 OOOdddYY
22 11dddsss
1
2
2
1
2
3
3,4 or 5
460
S3
44417+EA
3,4,5 or 6
17+EA
2,3, or 4
16+EA
2,3 or 4
16+EA
2,3 or 4
9+EA
3
2
2,3 or 4
439+EA
3 or 4
4-
2
2
3
3,4 or 5
34417+EA
3,4,5 or 6
17+EA
2,3 or 4
16+EA
2,3 or 4
16+EA
2,3 or 4
9+EA
3
2
2,3 or 4
4"
3"
9+EA
3 or 4
4"
2
2
3
3,4 or 5
3"
4"
4"
17+EA
4,5 or 6
17+EA
Instruction
AAA
AAD
AAM
AAS
ADC
ADC
ADC
AL,DATAS
AX,DATA16
DADDR,DATAS
ADC
DADDR,DAT A 16
ADC
DADDR,RB
ADC
DADDR,RW
ADC
RB,DADDR
ADC
ADC
ADC
RB,DATAS
RBD,RBS
RW,DADDR
ADC
RW,DATA16
ADC
ADD
ADD
ADD
RWD,RWS
AL,DATAS
AX,DATA16
DADDR,DAT AS
ADD
DADDR,DAT A 16
ADD
DADDR,RB
ADD
DADDR,RW
ADD
RB,DADDR
ADD
ADD
ADD
RB,DATAS
RBD,RBS
RW,DADDR
ADD
RW,DATA16
ADD
AND
AND
AND
RWD,RWS
AL,DATAS
AX,DATA16
DADDR,DATAS
AND
DADDR,DAT A 1"6
AND
DADDR,RB
AND
DADDR,RW
AND
RB,DADDR
AND
AND
RB,DATAS
RBD,RBS
5-82
2; 3 or 4
16+EA
2,3 or 4
16+EA
2,3 or 4
9+EA
3
23-
4"
Table 5-5. 8086 and 8088 Instruction Mnemonics (Continued)
Instruction
AND
RW,DADDR
AND
AND
CALL
CALL
CALL
RW,DATA16
RWD,RWS
SRANCH
BRANCH,SEGM
DADDR
CALL
DADDR,CS
CALL
CBW
CLC
CLD
CLI
CMC
CMP
CMP
CMP
RW
AL,DATAS
AX,PATA16
DADDR,DAT A8
CMP
DADDR,DAT A 16
CMP
DADDR,RB
CMP
DADDR,RW
CMP
RB,DADDR
CMP
CMP
CMP
RB,DATAS
RBD,RBS
RW,DADDR
CMP
RW,DATA16
CMP
CMPS
CMPS
CWO
DAA
DAS
DEC
RWD,RWS
BD,BS
WD,WS
DEC
DEC
DIV (S-bit)
RB
RW
AX,DADDR
DIV (16-bit)
DX,DADDR
DIV
DIV
ESC
RBS
RWS
DADDR
ESC
HLT
RW
DADDR
Clock
Periods
Object Code
Bytes
23 aadddbbb
[DISP) [DISP)
S1 11100sss YYYY
23 11dddsss
ES DISP DISP
9A PPOOPPOO
FF aa010bbb
[DISP) [DISP)
FF aa011 bbb
[DISP) [DISP)
FF 11010reg
9S
FS
FC
FA
F5
3C YY
3D YYYY
SO aalll bbb
[DISP) [DISP) YY
1 OOOOOa 1 aa 11 bbb
[DISP) [DISP) YY[YY)
3S aasssbbb
[DISP) [DISP)
39 aasssbbb
[DISP] [DISP]
3A aadddbbb
[DISP] [DISP)
SO 11 111 ddd YY
3A l1dddsss
3B aadddbbb
[DISP] [DISP]
1OOOOOa 1 11111 ddd
YY[YY]
3B l1dddsss
A6
A7
99
27
2F
1111 lllaa aaOOlbbb
[DISP) [DISP]
FE 11001ddd
01001ddd
F6 aa110bbb
[DISP] [DISP]
F7 aal10bbb
2,3 or 4
9+EA
4
2
3
5
2,3 or 4
4319-2S-21+EA
2,3 or 4
37+EA--
2
1
1
1
1
1
2
3
3,4 or 5
16-2
22224410+EA
3,4,5 or 6
10+EA
2,3 or 4
9+EA
2,3 or 4
9+EA
2,3 or 4
9+EA
3
2
2,3 or 4
439+EA
3 or 4
4-
2
1
1
1
1
1
2,3 or 4
322
22
5
4415+EA
2
1
2,3 or 4
32(S6-96)+EA
2,3 or 4
(150-16S)+EA
2
2
2,3 or 4
80-90
144-162
8+EA
2
1
2
2-
[DISP] [DISP]
F6 11110sss
F7 11110sss
11 011 xxx aaxxxbbb
[DISP) [DISP]
11011 xxx 11xxxreg
F4
5-83
Table 5-5. 8086 and 8088 Instruction Mnemonics (Continued)
Instruction
IDIV
AX.DADDR
IDIV
DX.DADDR
IDIV
IDIV
IMUL
RBS
RWS
AL.DADDR
IMUL
AX.DADDR
IMUL
IMUL
IN
IN
IN
IN
INC
RBS
RWS
AL[DX]
AL.PORT
AX.[DX)
AX.PORT
DAD DR
INC
INC
INT
INT
INTO
IRET
JA/JNBE
JAE/JNB
JB/JNAE
JBE/JNA
JCXZ
JE/JZ
JG/JNLE
JGE/JNL
JL/JNGE
JLE/JNG
JMP
JMP
JMP
RB
RW
3
V
DISP8
DISP8
DISP8
DISP8
DISPS
DISPS
DISPS
DISPS
DISPS
DISPS
BRANCH
BRANCH.SEGM
DADDR
JMP
DADDR.CS
JMP
JNE/JNZ
JNO
JNP/JPO
JNS
JO
JP/JPE
JS
LAHF
LOS
RW
DISPS
DISPS
DISPS
DISP8
DlSPS
DISPS
DISPS
LEA
RW.DADDR
LES
RW.DADDR
LOCK
RW.DADDR
Object Code
FS aa111bbb
[DISP) [DISP)
F7 aa111 bbb
[DISP) [DISP)
FS 11111sss
F7 11111 SSS
FS aa101bbb
[DISP) [DISP]
F7 aa101bbb
[DISP) [DISP)
FS 11101sss
F711101sS5
EC
E4 YY
ED
E5 YY
1111111 a aaOOObbb
[DISP) [DISP]
FE 11000ddd
01000ddd
CC
CD YY
CE
CF
17 DISP
73 DISP
7i DISP
7 DISP
63DISP
74DISP
7F DISP
7DDISP
7C DISP
7E DISP
111010a 1 DISP [DISP)
EA PPOO PPOO
FF aa100bbb
[DISP) [DISP]
FF aa101bbb
[DISP] [DISP)
FF 11100reg
75DISP
71DISP
6S DISP
79DISP
70DISP
7A DISP
7S DISP
9F
C5 aadddbbb
[DISP] [DISP]
SO aadddbbb
[DISP) [DISP)
C4 aadddbbb
[DISP] [DISP]
FO
5-84
Bytes
Clock
Periods
2.3 or 4
(107-118)+EA
2.3 or 4
(171-190)+EA
2
2
2.3 or 4
101-112
165-184
(8S-104)+EA
2.3 or 4
(134-1S0)+EA
2
2
1
2
1
2
2.3 or 4
80-98
128-154
8
10
8
10
15+EA
2
1
1
2
1
1
2
2
2
2
2
2
2
2
2
2
2 or 3
5
2.3 or 4
3
2
52
51
4 or 53
24
4 or 1s
4 or 16.
4 or 1s
4 or 1S·
S or 18.
4 or 1s
4 or 1S·
4 or 1S·
4 or 1s
4 or 1S·
15"
15
1S+EA
23 or 4
24+EA
2
2
2
2
2
2
2
2
1
2.3 or 4
0
0
00
0
00
0
0
00
0
0
00
0
00
4
4
4
4
4
4
4
11
or 1S·
or 16
or 1S·
or 1s
or 1s
or 1s
or 1s
4
16+EA
0
00
0
00
00
00
00
0
2.3 or 4
2+EA
2.3 or 4
4 or 16
1
2
0
00
Table 5-5. 8086 and 8088 Instruction Mnemonics (Continued)
Instruction
LODS
LODS
LOOP
LOOPE/LOOPZ
LOOPNE/LOOPNZ
MOV
MOV
MOV
BS
WS
DISP8
DISP8
DISP8
AL.LABEL
AX. LABEL
DADDR.DAT A8
MOV
DADDR.DAT A 16
MOV
DADDR.RB
MOV
DADDR.RW
MOV
DADDR.SR
MOV
MOV
MOV
LABEL.AL
LABEL.AX
RB.DADDR
MOV
MOV
MOV
RB.DATA8
RBD.RBS
RW.DADDR
MOV
MOV
MOV
MOV
RW.DATA16
RW.SR
RWD.RWS
SR.DADDR
MOV
MOVS
MOVS
MUL (8-bit)
SR.RW
BD.BS
WD.WS
AL.DADDR
MUL (16-bit)
AX.DADDR
MUL
MUL
NEG
RBS
RWS
DADDR
NEG
NEG
NOP
NOT
RB
RW
NOT
NOT
OR
OR
OR
RB
RW
AL.DATA8
AX.DATA16
DADDR.DAT A8
OR
DADDR,DAT A 16
Object Code
AC
AD
E2DISP
E1DISP
EO DISP
AO PPQQ
Al PPQQ
C6 aaOOObbb
[DISP] [DISP] YY
C7 aaOOObbb
[DISP] [DISP] YYYY
88 aasssbbb
[DISP] (DlSP]
89 aasssbbb
[DISP] [DISP]
8C aaOrrbbb
[DISP] [DISP]
A2 PPQQ
A 3 PPQQ
8A aadddbbb
[DISP] [DISP]
10110ddd YY
8A lldddsss
8B aadddbbb
[DISP] [DISP]
10111 ddd YYYY
8C 110rrsss
8B 11dddsss
8E aaOrrbbb
[DISP] [DISP]
8E 11 Orrsss
A4
A5
F6 aa100bbb
[DISP] [DISP]
F7 aa100bbb
[DISP] [DISP]
F6 11100sss
F7 11100
1111 011 a aaO 11 bbb
[DISP] [DISP]
F5 11011ddd
F711011ddd
90
1111 011 a aaO 1Obbb
[DISP] [DISP]
F611010sss
F711010sss
OC YY
00 YYYY
80 aaOOlbbb
[DISP] [DISP] YY
81 aaOOlbbb
[DISP] [DISP] YYYY
DADDR
5-85
Bytes
Clock
Periods
1
1
2
2
2
3
3
3.4 or 5
12
12
5 or 11**
6 or 18**
5 or 19**
10
10
10+EA
4.5 or 6
10+EA
2.3 or 4
9+EA
2.3 or 4
9+EA
2.3 or 4
9+EA
3
3
2.3 or 4
10
10
8+EA
2
2
2.3 or 4
4*
2*
8+EA
3
2
2
2.3 or 4
4*
*
*
8+EA
2
1
1
2.3 or 4
*
18
18
(76-831+EA
2.3 or 4
(124-139)+EA
2
2
2.3 or 4
70-71
118-133
16+EA
2
2
1
2.3 or 4
3*
3*
3*
16+EA
2
2
2
3
3.4 or 5
3*
3*
4·
4*
17+EA
4.5 or 6
17+EA
Table 5-5. 8086 and 8088 Instruction Mnemonics (Continued)
Object Code
Instruction
OR
DADDR,RB
OR
OR
DADDR,RW
RB,DADDR
OR
OR
OR
RB,DATAB
RBD,RBS
RW,DADDR
OR
OR
OUT
OUT
OUT
OUT
POP
RW,DATA16
RWD,RWS
Al,[DX]
Al,PORT
AX,[DX]
AX,PORT
DADDR
POP
POP
POPF
PUSH
RW
SR
PUSH
PUSH
PUSHF
RCl
RW
SR
RCl
RB,N
RCl
RCR
RW,N
DADDR,N
RCR
RB,N
REP
RET
RET
RET
RET
ROL
N
CS
CS,DATA16
DATA16
DADDR,N
ROl
RB,N
ROl
ROR
RW,N
DADDR,N
ROR
RB,N
ROR
SAHF
SALISHL
RW,N
SALISHL
SALlSHl
DADDR
DADDR,N
OB aasssbbb
[DISP] [DISP]
09 aasssbbb
OA aadddbbb
[DISP] [DISP]
BO 11001ddd YY
OA lldddsss
OB aadddbbb
[DISP] [DISP]
81 11001ddd YYYY
OB lldddsss
EE
E6 YY
EF
E7 YY
8F aaOOObbb
[DISP] [DISP]
01011ddd
OOOrrlll
9D
FF aall0bbb
[DISP] [DISP]
01010sss
000rrll0
9C
110100va aaOl Obbb
[DISP] [DISP]
11 01 OOvO 1 101 Osss
Bytes
Clock
Periods
2,3 or 4
16+EA
2,3 or 4
2,3 or 4
16+EA
9+EA
3
2
2,3 or 4
4"
3"
9+EA
4
2
1
2
1
2
2,3 or 4
4"
3"
8
10
8
10
17+EA
1
1
1
2,3 or 4
B
B
B
16+EA
1
1
1
2,3 or 4
11
10
10
N=l 15+EA
N> 4N+20+EA
N=l 2"
N>14N+B
2
11 01 OOv 1 1 10 1Osss
110100va aaOllbbb
[DISP] [DISP]
11 01 OOvO 1 10 11 sss
2
2,3 or 4
1111001z
C3
CB
CA YYYY
C2 YYYY
1 101 OOva aaOOObbb
[DISP] [DISP]
110100vO 11000ddd
1
1
1
3
3
2,3 or 4
1 101 OOv 1 11000ddd
110100va aaOOlbbb
[DISP] [DISP]
1 101 OOvO 11 001 ddd
2
2,3 or 4
2
1
2,3 or 4
RB,N
1 101 OOv 1 11001 ddd
9E
110100va aal00bbb
[DISP] [DISP]
1 101 OOvO 111 OOddd
RW,N
110100vl 11100ddd
2
DADDR,N
5-86
2
2
2
2
N=l 15+EA
N> 4N+20+EA
N=l 2"
N> 1 4N+B
+2
8""
12""
lB""
17""
N=l 15+EA
N>l 4N+20+EA
N=l 2"
N>4N + B
N=l 15+EA
N>l 4N+20+EA
N=12"
N>14N+B
4"
N=l 15+EA
N> 1 4N+20+EA
N=l 2"
N>14N+B
Table 5-5. 8086 and 8088 Instruction Mnemonics (Continued)
Instruction
SAR
DADDR,N
SAR
RS,N
SAR
SSS
SSS
SSS
RW,N
Al,DATAB
AX,DATA16
DADDR,DAT A8
SSS
DADDR,DATA16
SSB
DADDR,RB
SSB
DADDR,RW
SSB
RB,DADDR
SSB
SBB
SSB
RB,DATA8
RBD,RBS
RW,DADDR
SBB
RW,DATA16
SSB
SCAS
SCAS
SEG Prefix
SHR
RWD,RWS
BD
WD
SR
DADDR,N
SHR
RS,N
SHR
STC
STD
STI
STOS
STOS
SUB
SUB
SUB
RW,N
BD
WD
Al,DATAB
AX,DATA16
DADDR,DATA8
SUS
DADDR,DATA16
SUB
DADDR,RB
SUB
DADDR,RW
SUB
RB,DADDR
SUS
SUB
SUB
RB,DATAS
RBD,RBS
RW,DADDR
SUB
RW,DATA16
SUB
RWD,RWS
Clock
Periods
Object Code
Bytes
11 01 OOva aa 111 bbb
[DISP] [DISP]
11 0 1OOvO 11111 ddd
2,3 or 4
11 0 1OOv 1 111 11 ddd
1C yy
1D YYYY
80 aa011bbb
[DISP] [DISP] YY
1OOOOOa 1 aaO 11 bbb
[DISP] [DISP] YY[YY]
18 aasssbbb
[DISP] [DISP]
19 aasssbbb
[DISP] [DISP]
1A aadddbbb
[DISP] [DISP]
80 11 011 ddd YY
1A 11dddsss
lS aadddbbb
[DISP] [DISP]
100000a111011ddd
YY[YY]
1S 11dddsss
AE
AF
001 rrl 01
110100va aal 01 bbb
[DISP] [DISP]
110100vO 11101 ddd
2
2
3
3,4 or 5
4"
4"
17+EA
3,4,5 or 6
17+EA
2,3 or 4
16+EA
2,3 or 4
16+EA
2,3 or 4
9+EA
3
2
2,3 or 4
4"
3"
9+EA
3 or 4
4"
2
1
1
1
2,3 or 4
3"
15
15
+2
N=l 15+EA
N> 1 4N+20+EA
N=l 2"
N>14N+B
1 101 OOv 1 111 01 ddd
F9
FD
FS
AA
AB
2C YY
2D YYYY
80 aal01bbb
[DISP] [DlSP] YY
100000a1 aa1 01 bbb
[DISP] [DISP] YY[YY]
28 aasssbbb
[DISP] [DISP]
29 aasssbbb
[DISP] [DISP]
2A aadddbbb
[DISP] [DISP]
SO 111 01 ddd YY
2A 11dddsss
2B aadddbbb
[DISP] [DlSP]
100000alll101ddd
YY[YY]
2S 11dddsss
5-87
2
2
N=1 15+EA
N>l 4N+20+EA
N=12"
N> 1 4N+8
2
1
1
1
1
1
2
3
3,4 or 5
2"
2"
2"
11
11
4"
4"
17+EA
3,4,5 or 6
17+EA
2,3 or 4
16+EA
2,3 or 4
16+EA
2,3 or 4
9+EA
3
2
2,3 or 4
4"
3"
9+EA
3 or 4
4"
2
3"
Table 5-5. 8086 and 8088 Instruction Mnemonics (Continued)
Object Code
Instruction
TEST
TEST
TEST
AL,DATA8
AX,DATA16
DADDR,DATA8
TEST
DADDR,DAT A 16
TEST
DADDR,RB
TEST
DADDR,RW
TEST
TEST
TEST
TEST
WAIT
XCHG
XCHG
RB,DATA8
RBD,RBS
RW,DATA16
RWD,RWS
XCHG
XCHG
RB,RB
RW,DADDR
XCHG
XLAT
XOR
XOR
XOR
RW,RW
AL,DATA8
AX,DATA16
DADDR,DATA8
XOR
DADDR,DATA 16
XOR
DADDR,RB
XOR
DADDR,RW
XOR
RB,DADDR
XOR
XOR
XOR
RB,DATA8
RBD,RBS
RW,DADDR
XOR
XOR
RW,DATA16
RWD,RWS
AX,RW
RB,DADDR
A8 yy
A9 YYYY
F6 aaOOObbb
[DISP] [DISP] YY
F7 aaOOObbb
[DISP] [DISP] YYYY
84 aaregbbb
[DISP] [DISP]
85 aaregbbb
[DISP] [DISP]
F6 1000reg YY
84 11 regreg
F7 11 OOOreg YYYY
85 11 regreg
9B
10010reg
86 aaregbbb
[DISP] [DISP]
86 11 regreg
87 aaregbbb
[DISP] [DISP]
87 11 regreg
07
34 YY
36 YYYY
80 aa010bbb
[DISP] [DISP] YY
81 aa010bbb
[DISP] [DISP] YYYY
30 aasssbbb
[DISP] [DISP]
31 aasssbbb
[DISP] [DISP]
32 aadddbbb
[DISP] [DISP]
80 1111 Oddd YY
32 11dddsss
33 aadddbbb
[DISP] [DISP]
81 11110ddd YYYY
33 11dddsss
5-88
Bytes
Clock
Periods
2
3
3,4 or 5
4411+EA
4,5 or 6
11 +EA
2,3 or 4
9+EA
2,3 or 4
9+EA
3
2
4
2
1
1
2,3 or 4
5*
3*
53*
3+5n
3*
17+EA
2
2,3 or 4
4*
17+EA
2
1
2
3
3,4 or 5
4*
11
4*
4*
17+EA
4,5 or 6
17+EA
2,3 or 4
16+EA
2,3 or 4
16+EA
2,3 or 4
9+EA
3
2
2,3 or 4
43*
16+EA
4
2
4*
3*
Table 5-6. 8086 and 8088 Instruction Object Codes
Object Code
Mnemonic
Byte 1
Byte 2
Other Bytes
00
01
02
03
04
05
06
07
OS
09
OA
OB
OC
OD
OE
OF
10
11
12
13
14
15
16
17
1S
19
1A
1B
1C
1D
1E
1F
20
21
22
23
24
25
26
27
2S
29
2A
2B
2C
2D
2E
2F
30
31
32
33
34
35
36
aasssbbb
aasssbbb
aadddbbb
aadddbbb
[DISP)[DISP]
[DISP][DISP]
[DISP][DISP]
[DISP)[DISP]
yy
yy
yy
aasssbbb
aasssbbb
aadddbbb
aadddbbb
[DISP][DISP]
[DISP)[DISP]
[DISP)[DISP]
[DISP][DISP]
yy
yy
yy
aasssbbb
aasssbbb
aadddbbb
aadddbbb
[DISP)[DISP]
[DISP][DISP]
[DISP)[DISP]
[DISP][DISP]
yy
yy
yy
aasssbbb
aasssbbb
aadddbbb
aadddbbb
[DISP)[DISP]
[DISP)[DISP]
[DISP)[DISP]
[DISP)[DISPI
yy
yy
yy
aasssbbb
aasssbbb
aadddbbb
aadddbbb
[DISP)[DISP]
[DISP)[DISP]
[DISP)[DISP]
[DISP)[DISP]
yy
yy
yy
aasssbbb
aasssbbb
aadddbbb
aadddbbb
[DISP][DISP]
[DISP)[DISP]
[DISP][DISP]
[DISP)[DISP]
yy
yy
yy
aasssbbb
aasssbbb
aadddbbb
aadddbbb
[DISP][DISP]
[DISP)[DISP]
[DISP)[DISP]
[DISP)[DISP]
yy
yy
yy
ADD
ADD
ADD
ADD
ADD
ADD
PUSH
POP
OR
OR
OR
OR
OR
OR
PUSH
Not used
ADC
ADC
ADC
ADC
ADC
ADC
PUSH
POP
SBB
SBB
SBB
SBB
SBB
SBB
PUSH
POP
AND
AND
AND
AND
AND
AND
SEG
DAA
SUB
SUB
SUB
SUB
SUB
SUB
SEG
DAS
XOR
XOR
XOR
XOR
XOR
XOR
SEG
5-89
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD.DADDR/RBS
RWD.DADDR/RWS
AL.DATAS
AX.DATA16
ES
ES
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD.DADDR/RBS
RWD.DADDR/RWS
Al.DATAS
AX.DATA16
CS
(POP CS)
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD.DADDR/RBS
RWD.DADDR/RWS
Al.DATAS
AL.DATA16
SS
SS
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD.DADDR/RBS
RWD.DADDR/RWS
AL.DATAS
AX.DATA16
DS
DS
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD .DADDR/RBS
RWD.DADDR/RWS
AL.DATAS
AX.DATA16
ES
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD.DADDR/RBS
RWD.DADDR/RWS
AL.DATAS
AX.DATA16
CS
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD.DADDR/RBS
RWD.DADDR/RWS
Al.DATAS
AX.DATA16
SS
Table 5-6. 8086 and 8088 Instruction Object Codes (Continued)
Object Code
Mnemonic
Byte 1
37
38
39
3A
3B
3C
3D
3E
3F
40
41
42
43
44
45
46
47
48
49
4A
4B
4C
40
4E
4F
50
51
52
53
54
55
56
57
58
59
5A
5B
5C
50
5E
5F
60-6F
70
71
72
73
74
75
76
77
78
79
7A
7B
7C
Byte 2
Other Bytes
aasssbbb
aasssbbb
aadddbbb
aadddbtlb
[DISP)[DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
yy
yy
yy
DISP
DISP
DISP
DISP
DISP
DISP
DISP
DISP
DISP
DISP
DISP
DISP
DISP
AAA
CMP
CMP
CMP
CMP
CMP
CMP
SEG
AAS
INC
INC
INC
INC
INC
INC
INC
INC
DEC
DEC
DEC
DEC
DEC
DEC
DEC
DEC
PUSH
PUSH
PUSH
PUSH
PUSH
PUSH
PUSH
PUSH
POP
POP
POP
POP
POP
POP
POP
POP
Not used
JO
JNO
JB or JNAE or JC
JNB or JAE or JNC
JE or JZ
JNE or JNZ
JBE or JNA
JNBE or JA
JS
JNS
JP or JPE
JNP or JPO
JL or JNGE
5-90
RBD/DADDR.RBS
RWD/DADDR.RWS
RBD.DADDR/RBS
RWD.DADDR/RWS
AL.DATA8
AX.DATA16
OS
AX
CX
OX
BX
SP
BP
SI
01
AX
CX
OX
BX
SP
BP
SI
01
AX
CX
OX
BX
SP
BP
SI
01
AX
CX
OX
BX
SP
BP
SI
01
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
DISP8
Table 5-6. 8086 and 8088 Instruction Object Codes (Continued)
Object Code
Mnemonic
Byte 1
Byte 2
70
7E
7F
80
80
80
80
80
80
80
80
81
81
81
81
81
81
81
81
82
82
82
82
82
82
82
82
83
83
83
83
83
83
83
83
84
85
86
87
88
89
8A
8B
8C
8C
80
8E
8E
8F
8F
to 8F
90
91
92
93
94
DISP
DISP
DISP
aaOOObbb
aa001bbb
aa010bbb
aa011 bbb
aa100bbb
aa101 bbb
aa110bbb
aa111bbb
aaOOObbb
aa001bbb
aa010bbb
aa011bbb
aa100bbb
aa101bbb
aa110bbb
aa111 bbb
aaOOObbb
xxOO1xxx
aa010bbb
aa011 bbb
xx100xxx
aa101 bbb
xx110xxx
aa111 bbb
aaOOObbb
xx001xxx
aa010bbb
aa011bbb
xx100xxx
aa101bbb
xx110xxx
aa111bbb
aasssbbb
aasssbbb
aadddbbb
aadddbbb
aasssbbb
aasssbbb
aadddbbb
aadddbbb
aaOrrbbb
xx1 xxxxx
aadddbbb
aaOrrbbb
xx1 xxxxx
aaOOObbb
xx001xxx
xx111xxx
Other Bytes
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
yy
YY
yy
YY
YY
YY
YY
YY
YYYY
YYYY
YYYY
YYYY
YYYY
YYYY
YYYY
YYYY
YY
[DISP][DISP] YY
[DISP][DISP] YY
[DISP][DISP] YY
[DISP][DISP] YY
[DISP][DISP] YYYY
[DISP][DISP] YYYY
[DISP][DISP] YYYY
[DISP][DISP] YYYY
[DISP][DISP] YYYY
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP] [DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
JLE or JGE
JLE or JNG
JNLE or JG
ADD
OR
ADC
SBB
AND
SUB
XOR
CMP
ADD
OR
ADC
SBB
AND
SUB
XOR
CMP
ADD
Not used
ADC
SBB
Not used
SUB
Not used
CMP
ADD
Not used
ADC
SBB
Not used
SUB
Not used
CMP
TEST
TEST
XCHG
XCHG
MOV
MOV
MOV
MOV
MOV
Not used
LEA
MOV
Not used
POP
Not used
Not used
XCHG
XCHG
XCHG
XCHG
XCHG
5-91
DISP8
DISP8
DISP8
RBD/DADDR,DA T A8
RBD/DADDR,DAT A8
RBD/DADDR,DAT A8
RBD/DADDR,DAT A8
RBD/DADDR,DA T A8
RBD/DADDR,DAT A8
RBD/DADDR,DAT A8
RBD/DADDR,DATA8
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RBD/DADDR,DAT A8
RBD/DADDR,DAT A8
RBD/DADDR,DAT A8
RBD/DADDR,DAT A8
RBD/DADDR,DAT A8
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RWD/DADDR,DATA16
RWD/DADDR,DATA 16
RWD/DADDR,DATA 16
RBD/DADDR,RBS
RWD/DADDR,RWS
RBD/DADDR,RBS
RWD/DADDR,RWS
RBD/DADDR,RBS
RWD/DADDR,RWS
RBD,DADDR/RBS
RWD,DADDR/RWS
RWD/DADDR,SR
RWD,DADDR
SR,RWD/DADDR
RWD/DADDR
AX,AX (NOP)
AX,CX
AX,DX
AX,BX
AX,SP
Table 5-6. 8086 and 8088 Instruction Object Codes (Continued)
Object Code
Mnemonic
Byte 1
95
96
97
9S
99
9A
9B
9C
90
9E
9F
AO
A1
A2
A3
A4
A5
A6
A7
AS
A9
AA
AB
AC
AD
AE
AF
BO
B1
B2
B3
B4
B5
B6
B7
BS
B9
BA
BB
BC
BD
BE
BF
CO-C1
C2
C3
C4
C5
C6
C6
to C6
C7
C7
to C7
C8-C9
Byte 2
Other Bytes
PP
OOPPOO
PP
PP
PP
PP
00
00
00
00
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
yy
aadddbbb
aadddbbb
aaOOObbb
xxOO1xxx
xx111xxx
aaOOObbb
xx001xxx
xx111xxx
[DISP][DISP]
[DISP][DISP]
[DISP][DISP] YY
[DISP][DISP] YYYY
XCGH
XCHG
XCHG
CBW
CWO
CALL
WAIT
PUSHF
POPF
SAHF
LAHF
MOV
MOV
MOV
MOV
MOVS
MOVS
CMPS
CMPS
TEST
TEST
STOS
STOS
LODS
LODS
SCAS
SCAS
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
MOV
Not used
RET
RET
LES
LEA
MOV
Not used
Not used
MOV
Not used
Not used
Not used
5-92
AX,BP
AX,SI
AX,DI
BRANCH,SEGM
AL,LABEl
AX,LABEl
LABEL,AL
LABEL,AX
BD,BS
WD,WS
BD,BS
WD,WS
AL,DATAS
AX,DATA16
BD
WD
BS
WS
BD
WD
AL,DATAS
CL,DATAS
DL,DATAS
BL,DATAS
AH,DATAS
CH,DATAS
DH,DATAS
BH,DATAS
AX,DATA16
CX,DATA16
DX,DATA16
BX,DATA16
SP,DATA16
BP,DATA16
SI,DATA16
DI,DATA16
CS,DATA16
RWD,DADDR
RWD,DADDR
DADDR,DAT A8
DADDR,DAT A 1 6
Table 5-6. 8086 and 8088 Instruction Object Codes (Continued)
Object Code
Mnemonic
Byte 1
CA
CB
CC
CD
CE
CF
DO
DO
DO
DO
DO
DO
DO
DO
D1
D1
D1
D1
D1
D1
D1
D1
02
02
02
02
02
02
02
02
D3
03
D3
D3
D3
D3
03
D3
04
D5
06
D7
D8-DF
EO
E1
E2
E3
E4
E5
E6
E7
E8
E9
EA
EB
Byte 2
yy
Other Bytes
yy
yy
aaOOObbb
aa001bbb
aa010bbb
aa011bbb
aa100bbb
aa101bb1
xx110xxx
aa111 bbb
aaOOObbb
aa001bbb
aa010bbb
aa011bbb
aa100bbb
aa101bbb
xx110xxx
aa111bbb
[DISPJ[DISP)
[DISP)[DISP)
[DISPJ[DlSP)
[DISP)[DISP)
[DISPJ[DISP)
[DISP)[DlSP)
aaOOObbb
aa001bbb
aa010bbb
aa011bbb
aa100bbb
aa101bbb
xx110xxx
aa111 bbb
aaOOObbb
aa001bbb
aa010bbb
aa011bbb
aa100bbb
aa1 01 bbb
xx110xxx
aa111bbb
OA
OA
[OISP)[OISP)
[OISP)[DISP)
[DISP)[DlSP)
[DISP)[DISP)
[DISP) [DISP)
[DISP)[DISP)
aaxxxbbb
OISP
DlSP
DISP
DISP
[DISP)[DISP)
[DISPJ[DISP)
[DISP) [DISP)
[DISP)[DlSP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
[OISP) [DISP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
[DISP)[DISP)
yy
yy
yy
yy
DISP
DISP
PP
DISP
DISP
DISP
QQ PPQQ
RET
RET
INT
INT
INTO
IRET
ROL
ROR
Ret
RCR
SAL or SHL
SHR
Not used
SAR
ROL
ROR
RCL
RCR
SAL or SHL
SHR
Not used
SAR
ROL
ROR
RCL
RCR
SAL or SHL
SHR
SAR
ROL
ROR
RCL
RCR
SAL or SHL
SHR
Not used
SAR
AAM
AAD
Not used
XLAT
ESC
LOOPNE or LOOPNZ
LOOPE or LOOPZ
LOOP
JCXZ
IN
IN
OUT
OUT
CALL
JMP
JMP
JMP
5-93
CS.DATA16
3
V
RBD/DADDR.1
RBD/DADDR.1
RBD/DAODR.1
RBD/DADDR, 1
RBD/DADDR.1
RBD/DADDR.1
RBD/DADDR.1
RWD/DADDR.1
RWD/DADOR.1
RWD/DADDR.1
RWD/DADOR.1
RWD/DADOR,1
RWD/DADDR.1
RWD/DADDR.1
RBD/OAODR.N
RBD/OADDR.N
RBD/DADDR.N
RBD/OAODR,N
RBD/DAODR,N
RBD/DADDR,N
RBD/DAODR,N
RWD/DADDR,N
RWD/DADDR,N
RWD/DADDR,N
RWD/DADDR,N
RWD/DADDR,N
RWD/DADOR.N
RWD/DADDR.N
DADDR
DISP8
DISP8
DISP8
DISP8
AL.PORT
AX,PORT
AL,PORT
AX,PORT
BRANCH
BRANCH
BRANCH,SEGM
BRANCH
Table 5-6. 8086 and 8088 Instruction Object Codes (Continued)
Object Code
Byte 1
EC
ED
EE
EF
FO
F1
F2
F3
F4
F5
F6
F6
F6
F6
F6
F6
F6
F6
F7
F7
F7
F7
F7
F7
F7
F7
F8
F9
FA
FB
FC
FD
FE
FE
FE
to FE
FF
FF
FF
FF
FF
FF
FF
FF
Byte 2
Mnemonic
Other Bytes
IN
IN
OUT
OUT
lOCK
Not used
REPNE or REPNZ
REP or REPE or REPZ
aaOOObbb
xx001xxx
aa010bbb
aa011 bbb
aa100bbb
aa1 01 bbb
aa110bbb
aa111bbb
aaOOObbb
xx001 xxx
aa010bbb
aa011bbb
aa100bbb
aa101bbb
aa110bbb
aa111 bbb
[DISP][DISP] YY
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP] YYYY
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
HlT
CMC
TEST
Not used
NOT
NEG
MUl
IMUl
DIV
IDIV
TEST
Not used
NOT
NEG
MUl
IMUl
DIV
IDIV
ClC
STC
Cli
Al,DX
AX,DX
Al,DX
AX,DX
RBD/DADDR,DAT A8
RBD/DADDR
RBD/DADDR
RBD/DADDR
RBD/DADDR
RBD/DADDR
RBD/DADDR
RWD/DADDR,DAT A 1 6
RWD/DADDR
RWD/DADDR
RWD/DADDR
RWD/DADDR
RWD/DADDR
RWD/DADDR
511
aaOOObbb
aa001bbb
xx001 xxx
xx111xxx
aaOOObbb
aa001bbb
aa010bbb
aa011 bbb
aa100bbb
aa1 01 bbb
aa110bbb
xx111 xxx
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
[DISP][DISP]
CLD
STD
INC
DEC
Not used
Not used
INC
DEC
CAll
CAll
JMP
JMP
PUSH
Not used
5-94
RBD/DADDR
RBD/DADDR
DADDR
DADDR
RW/DADDR
DADDR,CS
RW/DADDR
DADDR,CS
DADDR
Table 5-7. 8080A to 8086 Instruction Mapping
8080A
Instruction
Equivalent 8086
Instructionls)
IN
OUT
DEV
DEV
IN
OUT
PORT
PORT
LDAX
B·
SI,CX
LDAX
0
STAX
B
STAX
0
MOV
MOV
LOA
STA
LHLD
SHLD
REG,M
M,REG
ADDR
AD DR
AD DR
ADDR
MOV
LODB
MOV
LODB
MOV
STOB
MOV
STOB
MOV
MOV
MOV
MOV
MOV
MOV
ADD
ADC
SUB
SBB
ANA
XRA
ORA
CMP
INR
OCR
M
M
M
M
M
M
M
M
M
M
ADD
ADC
SUB
SBB
AND
XOR
OR
CMP
INC
DEC
AL,DADDR
AL,DADDR
AL,DADDR
AL,DADDR
AL,DADDR
AL,DADDR
AL,DADDR
AL,DADDR
DADDR
DADDR
LXI
RP,DATA16
MOV
RW,DATA16
MVI
MVI
JMP
PCHL
M,DATA
REG,DATA
AD DR
MOV
MOV
JMP
JMP
DADDR,DAT AS
RB,DATAS
BRANCW·
BX
CALL
CC
AD DR
ADDR
CNC
AD DR
CZ
ADDR
CNZ
ADDR
CP
ADDR
CM
ADDR
CPE
ADDR
CPO
ADDR
CALL
JNB
CALL
JB
CALL
JNZ
CALL
JZ
CALL
JS
CALL
JNS
CALL
JPO
CALL
JPE
CALL
RET
BRANCH
next-inst
BRANCH
next-inst
BRANCH
next-inst
BRANCH
next-inst
BRANCH
next-inst
BRANCH
next-inst
BRANCH
next-inst
BRANCH
next-inst
BRANCH
RC
RNC
RET
Equivalent 808~
I nstructi on Is)
8080A
Instruction
RZ
SI,DX
RNZ
DI,CX
RM
DI,DX
RP
RB,DADDR
DADDR,RB
AL,LABEL
LABEL,AL
BX,DADDR
DADDR,BX
RPE
RPO
JNB
RET
JB
RET
JNZ
RET
JZ
RET
JNS
RET
JS
RET
JPO
RET
JPE
RET
next-inst
next-inst
next-inst
next-inst
next-inst
next-inst
next-inst
next-inst
ADI
ACI
SUI
SBI
ANI
XRI
ORI
CPI
DATA
DATA
DATA
DATA
DATA
DATA
DATA
DATA
ADD
ADC
SUB
SBB
AND
XOR
OR
CMP
AL,DATAS
AL,DATAS
AL,DATAS
AL,DATAS
AL,DATAS
AL,DATAS
AL,DATAS
AL,DATAS
JC
JNC
JZ
JNZ
JP
JM
JPE
JPO
ADDR
ADDR
ADDR
ADDR
AD DR
ADDR
ADDR
ADDR
JB
JNB
JZ
JNZ
JNS
JS
JPE
JPO
DISPS···
DISPS
DISPS
DISPS
DISPS
DISPS
DISPS
DISPS
MOV
XCHG
SPHL
d,s
MOV
XCHG
MOV
RBD,RBS
DX,BX
SP,BX
ADD
ADC
SUB
SSB
ANA
XRA
ORA
CMP
DAD
REG
REG
REG
REG
REG
REG
REG
REG
RP
ADD
ADC
SUB
SBB
AND
XOR
OR
CMP
LAHF
ADD
RCR
SAHF
RCL
AL,RBS
AL,RBS
AL,RBS
AL,RBS
AL,RBS
AL,RBS
AL,RBS
AL,RBS
BX,RW
AL
AL
or ADD BX,RW lunlike .DAD
will affect AF,PF,SF, and ZF)
5-95
Table 5-7. 8080A to 8086 Instruction Mapping (Continued)
INR
DCR
CMA
DAA
RLC
RRC
RAL
RAR
INX
DCX
8080A
Instruction
Equivalent 8086
Instruction(s)
REG
REG
RB
INC
RB
DEC
AL
NOT
DAA
ROL
AL
ROR
AL
RCL
AL
RCR
AL
LAHF
SAHF
or INC RW (unlike INX - will
affect AF, PF, SF, and ZF)
LAHF
RW
DEC
SAHF
or DEC RW (unlike DCX - will
affect AF, PF, SF, and ZF)
RP
RP
8080A
Instruction
PUSH
PUSH
RP
PSW
POP
POP
RP
PSW
Equivalent 8086
Instruction(s)
PUSH
LAHF
PUSH
POP
POP
SAHF
POP
XCHG
PUSH
STI
CLI
CALL
XTHL
EI
DI
RST
N
STC
CMC
STC
CMC
NOP
HLT
XCHG
HLT
RW
AX
RW
AX
SI
BX,SI
SI
S'N
AX,AX
'SOSOA registers map into SOS6 registers as follows:
8080A
8086
8080A
8086
A
B
C
D
E
H
AL
CH
CL
DH
DL
BH
L
BC
DE
HL
SP
PC
BL
CX
DX
BX
SP
IP
"Addresses on SOS6 jumps and calls are adjusted tb be self-relative.
"'Conditional jumps to a location out of the short self-relative range must be implemented by using a reversed-sense
conditional jump around a normal jump to the location, e.g.:
JC
ADDR
becomes
JNB
JMP
next-inst
BRANCH
Refer to Table 4-4 for a complete description of SOSA mnemonics shown above.
Refer to Table 20-4 for a complete description of SOS6 mnemonics shown above.
5-96
THE 8088 CPU
The 8088 is an 8086 microprocessor with an 8-bit Data Bus. The two parts are otherwise identical. Therefore
we will describe differences between the 8088 and the 8086 in the text which follows.
If you are going to use the 8088. first read the description of the 8086 given at the beginning of this chapter. then note
differences as described below.
8088 PROGRAMMABLE REGISTERS AND ADDRESSING MODES
8088 programmable registers and addressing modes are identical to the 8086 in every way.
8088 CPU PINS AND SIGNALS
8088 CPU pins and signals are illustrated in Figure 5-12. As compared to the 8086 pins and signals illustrated
in Figure 5-3, only pin 34 differs.
For the 8086. pin 34 outputs BHE. This signal discriminates between the high-order byte and the low-order byte on the
16-bit 8086 Data Bus. Since the 8088 has an 8-bit Data Bus. BHE and associated logic is irrelevant. The 8088 outputs
maximum mode SSO status at pin 34.
The 101M signal has opposite polarity for the 8088, as compared to the 8086. This makes the 8088 compatible
with the 8086.
Combining 101M, DT/R, and SSO, 8088 bus cycles can be decoded as follows:
10/M
Dr/A"
sso
0
0
0
0
0
0
0
1
0
1
0
0
0
1
0
Code segment access
Memory read
Memory write
No operations
Interrupt acknowledge
I/O read
I/O write
Halt
Since the 8088 has no BHE signal. nor need for any such signal. the discussion of external memory addressing and BHE
given for the 8086 will not apply to the 8088.
5-97
GND
A14
A13
A12
A11
A10
A9
A8
AD7
AD6
AD5
AD4
AD3
AD2
AD1
ADO
NMI
INTR
CLK
GND
Maximum {
System
Signals
-..
.....
..
--..
~
-- -.....
----..- --..---- -..
--- -.--...
~-
---..
~
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
8088
CPU
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
---..
..
--
VCC
A15
A16/S3
A17/S4
A18/S5
A19/S6
SSO
MN/MX
RD
-- .....-.
F«l/GTO.
lm/GTT:
---..
~.ImJ
--
~-
--.....
HOLD
HLDA
LOCK. WR
52.IO/M
Sl.DTlR
OSO, ALE
OS1, iN'fA
TEST
READY
RESET
Pin Name
Description
Type
ADO-AD7
A8-A15
A 16/S3. A 17 /S4
A18/S5
A19/S6
SSO
RD
READY
TEST
INTR
NMI
RESET
CLK
MN!MX
Address/Data Bus
Address Bus
Address/Segment Identifier
Address/Interrupt Enable Status
Address/Status
Status Output
Read Control
Wait State Request
Wait for Test Control
Interrupt Request
Non-maskable Interrupt Request
System Reset
System Clock
GND for a Maximum System
Machine Cycle Status
Local Bus Priority Control
Instruction Oueue Status
Bus Hold Control
= VCC for a Minimum System
Memory or I/O Access
Write Control
Address Latch Enable
Data Transmit/Receive
Data Enable
Interrupt Acknowledge
Hold Request
Hold Acknowledge
Power. Ground
Bidirectional. Tristate
Output. Tristate
Output. Tristate
Output. Tristate
Output. Tristate
Output. Tristate
Output. Tristate
Input
Input
Input
Input
Input
Input
'SO. ST. 52
RO/GTO. RO/GT1
OSO.OS1
LOCK
MN/MX
101M"
Minimum
System
Signals
---- -..
-
WR
ALE
DTIR
DEN
INTA
HOLD
HLDA
VCC.GND
=
Output. Tristate
Bidirectional
Output
Output. Tristate
Output.
Output.
Output
Output.
Output.
Output
Input
Output
Tristate
Tristate
Tristate
Tristate
Figure 5-12. 8088 Pins and Signal Assignments
5-98
8088 TIMING AND INSTRUCTION EXECUTION
The 8088 has a 4 byte instruction object code queue; the 8086. in contrast. has a 6 byte in8088
struction object code queue. The 8088 will start executing instruction fetch bus cycles to
INSTRUCTION
fill its 4 byte queue as soon as one or more queue bytes are empty. The 8086. in contrast.
QUEUE
will not start pre-fetching instruction object code bytes until two or more of its 6 queue bytes .....- - - - - -......
are empty. The description of bus cycles and queue logic given for the 8086 otherwise applies directly to the 8088.
8088 MEMORY AND I/O DEVICE ACCESS BUS CYCLES
Bus cycle timing for the 8088 and the 8086 differ only at the multiplexed Data/Address Bus lines. Timing
differences are confined to the eight Address Bus lines A8-A 15 and may be illustrated as follows:
:
T2
I
8~~:8A:~~~~; ~I; Add~s O~1 I
I
elK-
80: 8:D:~ : :
T1
I
I
'Of:
8088 AOO-AD7.
A~~s Om
I
3)
\
Add,.ss 0"(:
. •
T3
I
Dat:O~
I
D.~ In
\
T4
:
1
~ 1!
~ 1~ j j
!E:
CD
~
- Apart from the fact that the 8088 has no BHE signal, all timing for signals other than the Data/Address Bus is
identical for the 8086 and the 8088.
5-99
THE 8088 HALT STATE
When operating in minimum mode. the 8088 delays the ALE pulse by one clock period as compared to 8086 timing. This may be illustrated as follows:
T1
T2
T3
TH
TH
elK
101M
sso
OllA
ALE
HALT
Halt state logic and timing is otherwise identical for the 8086 and the 8088.
OTHER 8086 COMPATIBLE 8088 LOGIC
8086 and 8088 logic is absolutely identical for the following states and logic:
• The Wait state
• The Hold state
• RQ/GT logic
• Lock logic
• Wait for test state
• Processor escape
• Device reset
• Interrupt processing
• Single stepping mode
THE 8088 INSTRUCTION SET
The 8086 and 8088 instruction sets, listed in Table 5-4, are identical with the exception of execution times. Since the
8088 has an 8-bit bus, two bus cycles will have to be executed wherever the 8086 would have executed a single bus
cycle to fetch 16 bits of data. Table 5-5 provides execution times for the 8086 and the 8088.
5-100
THE INTEL 8284 CLOCK GENERATOR/DRIVER
The 8284 Clock Generator/Driver is a standard component that will be present in every 8086 microcomputer
system. In a mtiltimicroprocessor system. each 8086 microprocessor will have its own 8284 Clock Generator/Driver. While one could conceivably have a single 8284 servicing more than one 8086 microprocessor, it will rarely
make any economic sense to design a system in this fashion.
Logic implemented on the 8284 Clock Generator/Driver corresponds generally to the block labeled clock logic in
Figure 5-1. To be completely accurate, however. a small portion of the bus interface logic should also be illustrated as
provided by the 8284 device.
Figure 5-13 illustrates 8284 device internal logic.
The 8284 is manufactured using bipolar technology. It is packaged as an 18-pin DIP. All signals are TTL-level compatible.
8284 CLOCK GENERATOR/DRIVER PINS AND SIGNALS
8284 device pins and signals are illustrated in Figure 5-14. Figure 5-20 illustrates the 8284 device in a single
8086 microprocessor configuration.
Signals may be divided between timing and control logic.
Clock frequency is controlled by a crystal connected across the X1 and X2 pins. Clock frequency must be exactly
three times the required clock period. Since the standard 8086 clock period is 200 nanoseconds, a 15 MHz crystal frequency is required.
If an overtone mode crystal is employed, then it must be supported by an external LC network connected to
TANK to insure oscillation of the overtone frequency. This is standard clock logic practice; for the 8284 it is illustrated along with other normal connections in Figure 5-15.
You have the option of connecting a crystal across Xl and X2 in order to generate a fundamental frequency, or you can
inputthe fundamental frequency via EFI. The level of F/C determines whether an external crystal or a signal input will provide the fundamental frequency. If FIE is high, then the fundamental frequency is taken from the EFI input. If Fie is low, then the crystal connected across Xl and X2 provides the fundamental frequency.
Three clock outputs are generated:
1)
ClK is an MOS level signal designed to meet the requirements of the 8086.
2)
PClK Is a TTL level clock signal. output for support circuits. PClK runs at half the frequency of ClK.
3)
OSC is an oscillator output running at the crystal or EFI input frequency.
These timing signals may be illustrated as follows:
2
3
4
5
EFlor
Crystal
OSC
ClK
PClK
5-101
6
7
8
9
10
etc.
RES
Q
0
RESET
.ff
CK
X1
X2
XTAl
OSCILLATOR
OSC
TANK
ClK
+2
SYNC
+3
SYNC
Fie
PClK
EFI
CSYNC
RDY1
READY
SYNC
lOGIC
AEN1
AEN2
RDY2
Figure 5-13.
Logic of the 8284 Clock Generator and Driver
5-102
READY
CSYNC
PCLK
AEN1
RDY1
READY
RDY2
AEN2
ClK
GND
-.
-- -..
- ..
--- -
1
2
3
4
5
6
7
8
8284
9
18
17
16
15
14
13
12
11
10
------- ---- ..
VCC (+5 V)
X1
X2
TANK
EFI
Fie
OSC
RES
RESET
Pin Name
Description
Type
RESET
RES
RDY1.RDY2
AEN1. AEN2
READY
X1. X2
TANK
EFI
Control Signal Output to 8086
Reset Logic Input
Wait State Ready Inputs
Address Enable Qualifiers for RDY1 and RDY2
Control Signal Output to 8086
External Crystal Connections
Overtone Crystal Tank Circuit Connection
Alternate Clock Input
Clock Source Select
MOS level Clock Signal to 8086
TTL Clock for Peripherals
Crystal Oscillator Output
Clock Synchronizer
Power. Ground
Output
Input
Input
Input
Output
Input
Input
Input
Input
Output
Output
Output
Input
Fie
ClK
PClK
OSC
CSYNC
VCC.GND
Figure 5-14.
8284 Clock Generator and Driver Pins and Signal Assignments
5-103
XTAl
..dl§
Cx
R~DYand
Xl
RDYl
{
READY ~nable
Inputs
X2
asc
asc
AENl
ClK
RDY2
PClK
AEN2
READY
READY
RESET
RESET
ClK
PClK
8284
...-_..-------tRES
RESET)
I
_---tF/C
CR _
..,:_:_IY_NC...-_--..
r
--,
I
I
I
I
I
I
L..
CT
I
I
I
Tank circuit used
with overtone
crystals only.
I
I
I
--J
Notes:
1.
2.
Cx should be 3 to 10 pF
Cc (when used) should be 1 to 10 nF
3.
CR and RR determine Reset time constant
4.
CT and l T determine tank frequency: fo = ---V-==lT=C·T=-
1
Figure 5-15. Normal 8284 Clock Generator Circuit
5-104
In multi-CPU configurations you will probably need to synchronize all 8086 clock signals.
You use the CSYNC signal for this purpose. When CSYNC is input high. logic internal to the
8284 Clock Generator/Driver is stopped. When CSYNC subsequently goes low. clock outputs
restart. If the same CSYNC signal is input to a number of 8284 devices that receive the same
EFI input, then all microprocessors in a multi-CPU configuration will be exactly synchronized.
Appropriate logic is illustrated in Figure 5-16.
SYNCHRONIZING
MULTI-8086
CLOCK
SIGNALS
Note that you cannot use individual crystals for 8284 Clock Generator/Drivers that are supposed to be synchronized
with each; minor variations in crystal frequency. which must occur. will quickly distort clock Signal synchronization.
You can use a crystal to generate the fundamental frequency for one 8284 Clock Generator/Driver. then use the OSC
output of this Clock Generator/Driver as the EFI input to other 8284 Clock Generator/Drivers.
The 8086 requires its RESET input to be synchronized with clock logic. The 8284 will 18086
receive an asynchronous Reset input at RES and will generate synchronized RESET output,
RESET
which the 8086 requires. Appropriate logic is illustrated in Figure 5-15. Timing is illustrated in ''__ _ _ _ ___
the data sheets at the end of the chapter.
The 8284 RES input need not make a sharp transition. The 8284 inputs
RESET output. RES can make a slow low-to-high transition.
RES
to a Schmit trigger that generates the
We have described earlier in this chapter how external logic can extend a bus cycle by inserting
8284
Waitclock periods between T3 and T4. Figure 5-9 illustrates the READY input that controls Wait
WAIT STATE
states within the 8086 bus controller. As illustrated in Figure 5-9. the 8086 READY input must be
LOGIC
synchronized with the clock signal. The 8284 Clock Generator/Driver outputs an appropriately. - - - - - - synchronized READY signal to the 8086. The 8284 creates its READY output from one of two inputs: RDY1 or
RDY2. The 8284 has two READY inputs to support MULTIBUS configurations. A single 8086 may connect to two separate System Busses. Memory or I/O devices attached to either bus may wish to create a Wait state within a bus cycle.
Each System Bus may therefore have its own READY line. In order to arbitrate bus priorities. RDY1 and RDY2 have
companion enable signals AEN1 and AEN2. respectively. The 8284 will respond to RDY1 only when AEN1 is low.
Similarly. the 8284 will respond to RDY2 only when AEN2 is low.
AEN1 and AEN2 are general bus priority signals you must generate through your own bus priority arbitration logic. We
will describe these two Signals. and methods of generating them. later in this chapter.
5-105
1~tL
Master
Synchronizer
-
Q
D
7474
~ CLK
..-
--
Q
0
T
Xl
CSYNC
X2
7474
~ CLK
8284
.-
OSC
.......
r
CSYNC Input to
other 8284 devices
r
EFI input to
other 8284 devices
Figure 5-16. Clock Synchronization Logic in a Multi-CPU 8086 Configuration
5-106
THE INTEL 8288 BUS CONTROLLER
In configurations where the MN/MX signal is low, you must use an 8288 Bus Controller in order to decode the
SO, ST, and S2 status lines, and thus create System Bus control signals. You can also use the 8288 Bus Controller in order to connect more than one processor to a single System Bus, or in order to create more than one
System Bus for a single 8086.
Although the primary purpose of the 8288 Bus Controller is to decode the three status signals SO, S1, and S2, a
simple 1-of-8 decoder could accomplish this limited task. The 8288 has these additional capabilities:
1)
The 8288 can generate control signals for a System Bus or an liD device only bus.
2)
You can float a System Bus's control signals to enable direct memory access, or to arbitrate bus priorities.
3)
The two Write control lines have alternate advanced outputs designed for slow memories or liD devices.
4)
You can suppress control signals as a means of implementing memory protect logic in multi-bus or
multimicroprocessor configurations.
5)
The 8288 generates control signals needed by line drivers.
6)
The 8288 generates control signals needed by simple or complex interrupt logic.
The 8288 Bus Controller is manufactured using bipolar technology. It is packaged as a 20-pin DIP. All signals are TTLlevel compatible.
8288 BUS CONTROLLER SIGNALS AND PIN ASSIGNMENTS
Figure 5-17 illustrates 8288 Bus Controller signals and pin assignments. Figure 5-21 illustrates an 8288 within
an 8086 microcomputer system.
Control signals are generated from SO, S1, and S2 as follows:
80
81
82
0
0
0
Interrupt acknowledge
INTA and MCE
0
0
1
I/O read
10RC
0
1
0
I/O write
10WC, AIOWC
0
1
1
Halt
None
1
0
0
Instruction fetch
MRDC
MRDC
8086
8288 Control Output
1
0
1
Memory read
1
1
0
Memory write
MWTC,AMWC
1
1
1
No operation
None
5-107
51
DT/R
ALE
AEN
MROC"
lJI.Wr:
MWTC
GND
Pin Name
"SO.'ST.~
ClK
A£F1
CEN
lOB
JXRU"C"
~
1JAWC
mAC
mwc
AiOWC
TFJ"fA
MCE/PDEN
ALE
DT/R
DEN
VCC,GND
---...
...
lOB
ClK
---
~
---
-
...
1
2
3
4
5
6
7
8
9
10
8288
Bus
Controller
20
19
18
17
16
15
14
13
12
11
Description
Bus Cycle State Signals
TTL Clock Signal
Bus Priority Control/Enable
Command Enable
Mode Control
Memory Read Strobe
Memory Write Strobe
Early Memory Write Strobe
I/O Read Strobe
I/O Write Strobe
Early I/O Write Strobe
Interrupt Acknowledge
Cascade/Peripheral Data Enable
Address latch Enable
Data Direction Control
Data Buffer Enable
Power, Ground
--
-
-
VCC (+5 V)
SO
~
..
...
-..-
MCE/PDEN
DEN
CEN
INTA
mAC
AIOWC
mwc
Type
Input
Input
Input
Input
Input
Output.
Output.
Output.
Output.
Output,
Output,
Output,
Output
Output
Output
Output
Tristate
Tristate
Tristate
Tristate
Tristate
Tristate
Tristate
Figure 5-17. 8288 Bus Controller Pins and Signal Assignments
5-108
8288 and 8086 control signal timing is essentially the same. For details. see the data sheets given at the end of this
chapter.
If you look again at the Read and Write bus cycle timing descriptions given earlier in this chapter
for the 8086 you will see that Read control signals pulse low approximately one clock period
earlier than Write control signals. The 8288 creates two alternate Write control signals whose
timing is the same as the Read control signals. These alternative Write control signals are referred to as advanced Write control signals. because they go low one clock pulse in advance of the
standard Write control signals.
8288
ADVANCED
WRITE
CONTROL
SIGNALS
We can thus summarize 8288 System Bus control signals as follows:
MRDC is the memory read control.
MWTC is the memory write control.
AMWC is a memory write control whose timing conforms to MRDC.
INTA is a memory read control signal that is output during the two interrupt acknowledge bus cycles.
10RC is an I/O device read control signal.
10WC is an I/O device write control signal.
AIOWC is an alternative I/O device write control signal with timing that conforms to 10RC.
Devices connected to a bus are likely to use 10WC and MWTC or AIOWC and AMWC. but not all four signals. That is.
you will use either the normal write control signals or you will use the advanced write control signals.
All 8288 control signals are tristate. They can be disabled and thus disconnected from the System Bus.
You have two control options that modify the control signal logic of the 8288 Bus Controller.
Using the lOB pin, you can operate the 8288 device in I/O bus mode or in System Bus mode.
Using the CEN pin, you can suppress control signals.
Let us examine each of these capabilities in turn.
8288 I/O
When the lOB pin receives a high input. the 8288 Bus Controller generates an I/O bus. lOB high
floats MflDC. MWTC. and 7f.MWC all of the time but outputs INTA. 10RC. 10WC. and AIOWC. In
BUS MODE
I/O bus mode. these four I/O control signals cannot be floated. Since the four I/O control lines will
always be active. it is assumed that the I/O bus generated by an 8288 is a logic bus. You cannot share this local I/O bus
with another microprocessor.
The 8288 I/O bus has two control signals. PDEN and DTIR. which drive I/O ports and line drivers. DTIA. which we have
described for the 8086. is used to control a bidirectional bus driver. When high. DT/R puts the bus driver in output
mode. while when low. DTIff puts the bus driver in input mode. PDEN pulses low as a data enable signal. PDEN is
equivalent to DEN. the standard bus data enable signal output by the 8086.
When lOB is low. a normal System Bus is generated. All seven control signals are active: however. AEN is a bus enable
control (much as the BUSEN input is used by the 8228 Bus Controller in an 8080A system).
AEN is inactive when lOB is high and an I/O bus is being generated. AEN is active only when lOB is low and a System
Bus is generated.
When lOB is low and AEN is high. all contro~nals are floated. When lOB is low and AEN is low. control signals are
connected to the System Bus. You will use AEN to implement bus priority arbitration logic. or direct memory access
logic. as described later in this chapter.
CEN is used to disable, but not float, control signals. CEN can be used when an 8288 is
generating a System Bus or an I/O bus. CEN will normally be high. When CEN is low, control
signals are inactive. CEN does not float signals; it just disables the logic that might otherwise have made a control signal pulse low.
Table 5-8 summarizes the effect of lOB and CEN on control signals generated by the 8288
Bus Controller.
5-109
8288 BUS
CONTROLLER
MEMORY
PROTECT
Table 5-8.
Effect of lOB. CEN. and AEN on Control Signals Output by the 8288 Bus Controller
Control Unit
Effect on Control Output
lOB
CEN
AEN
0
0
0
0
0
0
0
1
0
1
0
0
0
1
1
0
1
1
1
1
1
1
1
1
i"NTA. iOR'C. iOWC. AffiWc
MiffiC.
MWTC. AMWC
Mode
Floated?
Active?
Mode
Floated?
Active?
System
System
System
System
1/0
1/0
1/0
1/0
Floated
Floated
Connected
Connected
Floated
Floated
Connected
Connected
Active
Inactive
Active
Inactive
Active
Active
Active
Active
System
System
System
System
Not Used
Not Used
Not Used
Not Used
Floated
Floated
Connected
Connected
Floated
Floated
Floated
Floated
Active
Inactive
Active
Inactive
Inactive
Inactive
Inactive
Inactive
The CEN control enables memory mapping. Here are some possibilities:
1)
In multi-bus configurations. one block of memory addresses may access memory on two or more busses. In order to
avoid contentions, you can use the CEN signal to selectively disable busses so that only one bus will actually respond when the 8086 accesses duplicated memory addresses.
2)
Privileged memory is frequently present in large microcomputer systems. Privileged memory is likely to become
more common in microcomputer systems as they grow larger. Privileged memory is memory that can be accessed
only under special circumstances. Frequently, system programs are run out of privileged memory, while application programs are run out of non-privileged memory. This prevents errors in application programs from destroying
system programs; it also prevents unauthorized access of reserved memory spaces.
DTIR and DEN, the two standard buffer control signals, are generated by the 8288 when it is creating a normal
System Bus. These two control signals. when generated by the 8288 Bus Controller. are identical in form and purpose
to the signals that the 8086 creates. DT/R determines the data direction for bidirectional buffers, while DEN is a latching strobe.
The 8288 generates two interrupt control signals: INTA and MCE. INTA is active on a
System Bus or an I/O Bus. MCE shares a pin with PDEN and is active only on a System Bus.
8288 BUS
CONTROLLER
INTERRUPT
SIGNALS
As we discussed earlier in this chapter. the 8086 executes two bus cycles when acknowledging
an interrupt. During each bus cycle. INTA is output as a low read pulse. On the second low INTA
pulse. the acknowledged device must return an 8-bit code. which the 8086 uses as an interrupt
vector. The INT A control signal generated by the 8288 Bus Controller is identical to the 8086 INT A control signal and
serves the same purpose. on a System Bus or an I/O Bus. The MCE control signal has been added for use in large 8086
microcomputer systems that use a variation of the 8259A Priority Interrupt Control Unit. When you have a master
8259A Priority Interrupt Control Unit and slave 8259A Priority Interrupt Control Units. you will use MCE as a control to
the master. while INT A becomes a control to slaves. The 8086 version of the 8259A Priority Interrupt Control Unit is not
described in this chapter.
5-110
THE 8282/8283 8-BIT INPUT/OUTPUT LATCH
These are simple unidirectional 8-bit latch buffers. The 8283 inverts inputs in order to create outputs; the 8282
does not. That is the only difference between these two devices.
Both devices have three-state outputs. When a device is not selected, its outputs are floated.
These devices are manufactured using bipolar technology. All signals are TTL-level compatible. Outputs have a high
drive capability. as defined in the data sheets at the end of this chapter. The devices are packaged as 20-pin DIPs.
THE 8282/8283 INPUT/OUTPUT LATCH PINS AND SIGNAL ASSIGNMENTS
Figure 5-18 illustrates the pins and signal assignments for the 8282 and 8283 8-bit input/output latches.
Data must be input at 010-017.
When STS is high. the internal latches appear transparent and data on the output pins track data on the input pins. The
transition from high to low of STS latches the data. The outputs remain stable while STS is low.
Data that is latched internally is output when OE is low. The 8282 outputs data unaltered. while the 8283 inverts
the data.
Were you to simply ground DE and tie STS to +5 V. the 8282 or 8283 I/O ports will function as simple bus drivers. The
outputs will continuously track the inputs. but will support heavier signal loads.
If you tie STS high. but use the low DE pulse. then input data is constantly available but outputs only become valid
while DE is low. Timing may be illustrated as follows:
DIO-DI7
5-111
-:-.
010
011
012
013
014
015
016
017
--..
-----
or
GNO
-.....
20
19
18
16
15
14
13
12
11
or
6
7
-
17
8282
8283
8
9
10
-..
~-
:.
-
----
Vee (+5 VI
000
001
002
003
004
005
006
007
STB
'in Name
Description
Type
010-017
000-007
Data Input
Data Output
Output Enable
Input Data Strobe
Power. Ground
Input
Output. Tristate
Input
Input
~
STB
Vee· GNO
Figure 5-18.
1
2
3
4
5
8282 and 8283 Input/Output Latch Pins and Signal Assignments
When the Strobe and Output Enable signal are both active. I/O port logic may be illustrated as follows:
DtO-Dt7
STB
Latches
A
B
A
5-112
e
AO
A1
A2
A3
A4
A5
A6
A7
OE
GND
Figure 5-19.
----
-
~
:.
- --
---
--
-- .~
--
1
2
3
4
5
6
7
8
9
10
8286
or
8287
20
19
18
17
16
15
14
13
12
11
-- ---- --- ------ ---- --..
--
Vee
BO
B1
B2
B3
B4
B5
B6
B7
T
Pin Name
Description
Type
AO-A7
BO-B7
OE
T
Vee,GND
Local Bus
System Bus
Output Enable
Direction Select
Power, Ground
Bidirectional, Tristate
Bidirectional, Tristate
Input
Input
8286 and 8287 Bidirectional Bus Transceiver Pins and Signal Assignments
THE 8286/8287 8-BIT BIDIRECTIONAL BUS TRANSCEIVERS
These two devices are used to buffer bidirectional lines on a System Bus. The 8286 transmits data unaltered,
while the 8287 inverts the data. The two devices are otherwise the same.
The 8286 and 8287 bidirectional bus drivers are manufactured using bipolar technology. All pins are TTL-level compatible. The devices are packaged as 20-pin DIPs.
8286 AND 8287 BIDIRECTIONAL BUS TRANSCEIVER PINS AND SIGNAL
ASSIGNMENTS
Figure 5-19 illustrates pins and signal assignments for the 8286 and 8287 bidirectional bus drivers.
AO-A7 constitute eight parallel data lines that connect with the microprocessor Data/Address Bus. BO-B7 constitute eight equivalent lines that connect with the System Bus. System Bus outputs have a higher line drive
capability (as defined in the data sheets at the end of this chapter); otherwise. there is no difference between the two
busses.
When the T input is low, data arriving at the B pins is output via the A pins. When T is high, data arriving at the A
pins are output via the B pins. The actual data transfer occurs only while OE is low. When used as an 8086 Data
Bus transceiver. T should be connected to DTfR and OE connected to DEN.
5-113
SOME 8086 MICROPROCESSOR BUS CONFIGURATIONS
We are now going to look at some 8086 microprocessor bus configurations.
The flexibility of the 8086 gives rise to such a bewildering array of system configuration possibilities that a
whole book could be written on the subject. We are going to fulfill the more limited objective of identifying
possibilities.
Figure 5-20 illustrates the simplest case. Here we are using the 8086 to generate a simple microcomputer system.
Addresses taken off the bidirectional 8086 Data/Address Bus are unidirectional. We therefore use 8282 I/O ports to
latch addresses of the 8086 Data/Address Bus.·ln Figure 5-20 we show just two 8282 I/O ports generating a 16-line
Address Bus. Address lines A 16 through A 19 are wasted. By adding one' more 8282 I/O port to the logic in Figure
5-37, you could include the four missing Address Bus lines.
In Figure 5-20, we ground the Output Enable inputs of the 8282 I/O ports; the Address Bus will therefore never be
floated. We use the 8086 ALE pulse to strobe addresses into the 8282 I/O ports.
Since the Data Bus is bidirectional. we use 8286 bidirectional Bus Transceivers in order to create a separate Data Bus
from the 8086 Address/Data Bus. Two 8286 bidirectional Bus Transceivers are required to create the 16-line Data Bus.
We can use the DT!R and DEN outputs of the 8086 as the 8286 T and CS inputs.
We can now illustrate timing for creation of the Address Bus and Data Bus during a read bus cycle, as follows:
T1
T4
T3
T2
eLK
M~--""+--(= 8282 01
AOO-AO 15--4-.....,1---1
~------"
' - -_ _.....~-....J
and 8286A)
ALE
M/iO
OT/R
8282 0 1 - - - - 4
828200
8286 B
(=
8286 T)
(= AOO....------,0.-015)
Address Out
~--4_--....,
AO-A15
Address Out
Oata In
00-015
(= AOO8286A-------------~~~_ _ _D_a_ta_l_n_ _ _~----------AD15)
5-114
M/TO
INTA
~
-
WR
~
ADO
AO
\
-
r
AD7
ALE
RD
+DE
DE
8282
A7
A8
ST~
8282
t.:
A15
SHE
V
~DO
8086
Vee
J\
\
/
KAD7
~T/R
MN/MX
DEN
TV
DE
AAD8
V
1\ AD15
--l
u
>0
-L<
r~
READY
PCLK
8284
~2
AEN1
~
ES
X1
X2
~Dl
2S.
=-:-=
Figure 5-20. Generating a System Bus for a Simple 8086 Configuration
The simple system illustrated in Figure 5-20 will not make use of the dual READY clock logic. A single READY input is
connected to RDY 1, and both of the READY enables are grou nded. Thus, the 8086 READY input will be created directly
from the 8284 RDY1 input.
Figure 5-21 illustrates a slightly more complex 8086 microcomputer configuration. Figure 5-21 uses an 8288
Bus Controller to generate System Bus control signals. The DEN, DTlA, and ALE control outputs, which in Figure
5-20 were generated by the 8086 microprocessor, are now generated by the 8288 Bus Controller.
As a stand-alone microcomputer configuration, Figure 5-21 offers little or no a,dvantage over Figure 5-20. In a Single
bus, single 8086 microcomputer configuration, there is no compelling reason to use the 8288 Bus Controller. All it does
is add an extra component to the system without offering any significant ,logic enhancement.
5-115
so
S1
52
GND
&
lOB
DEN
DT/R
8288
elK
~
c
ADO
AO
-
AD7
~STB •
A7
8282
llE
A8
~
STB
AD8
A15
8282
SHE
AD15.
J3HE
.ADO
ADO
-"'D7
AD7
T
J*
GND MN/MX
OE
>-
Cl
w
....J
~:,~~~
""\'
!--TClR1X
-
]
~-----t-"
1,---+",,\
-TCHRYX
1
TRYHCH~I
I
-TClAZ
-
1\
TDVCl-- '-TCLDX~1
1
~
A07-ADo
I\.
RO
READ CYCLE
.=L r-""",,
(NOTE I}
('NIl. INn c VOH}
OT/R
'
,;'' '- r:{ """- -
AD AD
"J--I" /'
'--------------r--+---~---
""'''- {/
5-D12
FLOAT
__
I'--
-TRHAV---l
f---..;,",
TRlRH
'\
Figure 11: 8088 Bus'Tlmlng -
\I
DATA IN
'
r--~-J
TCHCTV
""'''- )Qlr-+L--_____
Minimum Mode System
8088
ClK
(828~
Output)
[.0,-." --+--+------
WRITE
~~~~~
I
DE ..
-+----1-----.----1-.. 1---11---
ViR
~
__~--------------~-JI
AD7-ADo
DT/R
INTA CYCLE
NOTES 1.3
(RD. W1i -
VOH)
SOFTWARE HALT DEN.iiD,Wii.INTA = VOH
INVALID ADDRESS
AD7-ADo
SOFTWARE HALT
TClAV
NOTES:
I. All SIGNALS SWITCH BETWEEN VOH AND VOl UNLESS OTHERWISE
SPECIFIED.
2. RDY IS SAMPLED NEAR THE END OF T2. T3. Tw TO DETERMINE IF Tw
MACHINES STATES ARE TO BE INSERTED.
3. TWO INTA CYCLES RUN BACK·TO-BACK. THE 11088 LOCAL ADDAIDATA
BUS IS FLOATING DURING BOTH INTA CYCLES. CONTROL SIGNALS
ARE SHOWN FOR THE SECOND INTA CYCLE.
~. SIGNALS AT 8214 ARE SHOWN FOR REFERENCE ONLY.
5. All TIMING MEASUREMENTS ARE MADE AT I.SV UNLESS OTHERWISE
NOTED.
Figure 12. 8088 Bus Timing -
Minimum Mode System (cont.)
5-D13
8088
8088 MAX MODE SYSTEM (USING 8288 BUS CONTROLLER)
TIMING REQUIREMENTS
P.r. meter
Symbol
TCLCL
CLK Cycle Period
Min.
Max.
Unlls
200
500
ns
TCLCH
CLK Low Time
('hTCLCL)-15
ns
TCHCL
CLK High Time
('IJTCLCL) + 2
ns
TCH1CH2
CLK Rise Time
10
TCL2CLI
CLK Fall Time
10
ns
TOVCL
Data In Setup Time
30
Data In Hold Time
10
ns
TR1VCL
ROY Setup Time into 8284 (See Notes 1. 2)
35
ns
0
ns
ns
TCLRIX
ROY Hold Time into 8284 (See Notes 1,2)
TRYHCH
READY Setup Time InlO 8088
(,hTCLCL)-15
ns
TCHRYX
READY Hold Time IOto 8088
30
ns
TRYLCL
READY Inactive to CLK (See Note 4)
-8
ns
TlNVCH
Setup Time lor Recognition (INTR. NMI. TESn (See Note 2)
30
ns
TGVCH
RQ/GT Setup Time
30
RQ Hold Time into 8086
40
From 1.0V to 3.5V
-- ----_. - - - - - - - - ns
From 35V to 1.0V
--
TCLDX
TCHGX
Te.t Conditions
ns
-- f - - ns
TIMING RESPONSES
P.r.meter
Min.
MIX.
Unll.
TCLML
Command Active Delay (See Note 1)
10
35
ns
10
Symbol
TCLMH
Command Inactive Delay (See Note 1)
TRYHSH
READY Active to Status Passive (See Note 3)
TCHSV
Status Active Delay
TCLSH
Status Inactive Delay
TCLAV
35
ns
110
ns
10
110
ns
10
130
ns
Adaress Valid Delay
15
110
ns
TCLAX
Address Hold Time
10
TCLAZ
Address Float Delay
TCLAX
80
ns
TSVLH
Status Valid to ALE High (See Note 1)
15
ns
TSVMCH
Status Valid to MCE High (See Note 1)
15
ns
TCLLH
CLK Low to ALE Valid (See Note 1)
15
ns
TCLMCH
CLK Low to MCE High (See Note 1)
15
ns
TCHLL
ALE Inactive Delay (See Note 1)
15
ns
TCLMCL
MCE Inactive Delay (See Note 1)
15
ns
TCLDV
Data Valid Delay
15
110
ns
TCHDX
Data Hold Time
10
TCVNV
Control Active Delay (See Note 1)
5
45
ns
TCVNX
Control Inactive Delay (See Note 1)
10
45
ns
TAZRL
Address Float to Read Active
0
TCLRL
RD Active Delay
10
TCLRH
RD Inactive Delay
10
TRHAV
RD Inactive to Next Address Active
TCHDTL
Direction Control Active Delay (See Note 1)
50
TCHDTH
Direction Control Inactive Delay (See Note 1)
30
TCLGL
GT Active Delay
TCLGH
GT Inactive Delay
TRLRH
RD Width
NOTES: 1.
2.
3.
4.
--
ns
---
- - ------
-- f--- - - - . - - - -
--.---------
CL = 2().100 pF lor
ali 8088 Outputs
in addition to
internal loads
--ns
ns
165
- - - - - -1 - - - - 150
110
ns
---ns
.. _._----ns
--f-----.---
ns
-----ns
- - - - - - - 1-------ns
85
----_._._-----.- -
2TCLCL-75
Signal at 8284 or 8288 shown lor relerence only.
Setup requirement lor asynchronous signal only to guarantee recognition at next CLK.
A~lies only to T3 and wait states.
Applies only to T2 state (8 ns into T3 state).
5-014
---
ns
TCLCL-45
--
Tilt Conditions
ns
i
8088
T,
T,
ClK
VCl
OSo.OS,
S"S"So (EXCEPT HAL T)
I
ALE (8288 OUTPUT)
SEE NO!F 5
tROY (8284 INPUT)
I
READY (8088 INPUT)
I
READ CYCLE
TClAV--i
RD
I
TCHDTL
-I
DTiA
TClMH __ I
8288 OUTPUTS
SEE NOTES 5,8
It
MRDC OR IOhc
DEN
TCVNX-
I
Figure 13.8088 Bus Timing -
Maximum Mode System (Using 8288)
5-D15
8088
C1.K
Ii. s,. iii (EXCEPT HALTJ
WAITE CYCLE
AO,-AOO
DEN
.2811 OUIPUI'S
SEE NOTES 5,8
AiiWC 011 AiOWC
INTA CYCLE
A,.-Aa
(SEE NOTES 3.4)
FLOAT
AO,·AOo
TSVMCH
I
MCEI
PlmI
TCLMCH-
DT/Ii
82.
OUIPUI'S
SEE NOTES 5.6 \ INTA
DEN
TCVNX-
:W-~::N _ vOL;RI).~.IOJIe.iIW'fe.AM\vC.~.AroWC.iffn.. =VOH
AD, - ADo. A" -
Aa
INVALID ADDRESS
TCLAV
~
/------~\-------
\1...._ _ _ _ _.....J.
s..s..~
NOTES:
1. ALL SIGNALS SWITCH BETWEEN VO H AND VOL UNLESS OTHERWISE
SPECIFIED
2. ROY IS SAMPLED NEAR THE END OF T2. T,. Tw TO DETERMINE IF Tw
MACHINES STATES ARE TO BE INSERTED.
3. CASCADE ADDRESS IS VALID BETWEEN FIRST AND SECOND INTA
CYCLES
4. TWO INTA CYCLES RUN BACK·TO·BACK. THE 808a LOCAL ADORIDATA
BUS IS flOATING DURING BOTIi INTA CYCLES. CONTROL FOR
POINTER ADDRESS IS SHOWN FOR SECOND INTA CYCLE.
5. SIGNALS AT 828< OR 8288 ARE SHOWN FOR REFERENCE ONLY .
•. TliE ISSUANCE OF THE 8288 COMMAND AND CONTROL SIGNALS
(MIf!le. MWTl: . .{'f;lWt. IOIIC. roWC.lIOWl:. TRn AND DEN)' . GS THE
ACTIVE HIGH 1288 CEN
7. All TIMING MEASUREMENTS ARE MADE AT , .5V UNLESS OTHERWISE
NOTED.
a. STATUS INACTIVE IN STATE JUST PRIOR TO T•.
5-016
\. _ _ _ _ _ _
r--
8088
NOTE: 1. SETUP REQUIREMENTS FOA ASYNCHRONOUS
SIGNALS ONLY TO GUARANTEE RECOGNITION AT NEXT elK
Figure 15. Asynchronous Signal Recognition
~- Any CLK Cr~
__
I~
CLK
---1 TeLAV
1--
~_Any CLK cye;=!
}-----I
--tTClAVr:
Figure 16. Bus Lock Signal Timing (Maximum Mode Only)
,,,~""D-L ~
'CLGH
~_ +-- "CLCL-- _.-=\
V
i:---,'c"H ' ; ; . ,
- - -_.'
PULSE'
COPROCESSOR
:
.
:_TCLGH
;
PULSE :I
\
IOI1G1
P'.YIOUS gran.
Au-Ai
A"'-ADo
PULSE'
.
COPROCESSOR
I
Ct-
--,
A,oIIo· A,,,.,,
,---+------\
RELEASE
TCLAZ
~~ ~----------------------~
NOTE: 1. THE COf'JIOCUIOIII; IllAY NOT DRIVE THE
1
~'
:
.
RO.
AOIGT;
l
_I'CLG',__
~
'
~---(S-EE~N~OT~E-'I---------J
.UsaE' OUTSIOE THE KEQtOH
SHOWN WITHOUT ftllKING CONTENTION.
Figure 17. Request/Grant Sequence Timing (Maximum Mode Only)
r-"A
"CLKCYCU-
Clk
"\J'C
I
I
-..
-THVeH
(SEE NOTE
't
HOLO~
I
-~ I·~
~I._H:LH.V \i~j,J,
,
\
~------------~~----~
~
_____________~~____~
,~------~
1'~--~T~CL~Al~
_ _ _ _ _~,~__________~
COPROCESSOR
Figure 18. Hold/Hold Acknowledge Timing (Minimum Mode Only)
5·017
8282/8283
ABSOLUTE MAXIMUM RATINGS·
"NOTICE: Stresses above those listed under "Absolute Maximum Ratings"
may cause permanent damage to the device. This is a stress rating on:,. and
functional operattoh of the device at these or any other conditions above
those ;ndicated in the operational sections of this specification is hot
implied. Exposure to absolute maximum rating conditions for extended
periods may affect device reliability.
Temperature Under Bias .•.•.....•.....•. O·C to 70'C
Storage Temperature .•.... , ...... - 6SoC to + 1S0·C
All Output and Supply Voltages ........ - O.SV to + 7V
All Input Voltages .............. " •• - 1.0V to + S.SV
Power Dissipation .......................... 1 Watt
D.C. CHARACTERISTICS
Conditions: Vee = 5V ± 10%, TA =
Symbol
ooe to 70 e
0
Min
Parameter
Max
Units
-1
V
Power Supply Current
160
mA
Forward Input Current
-0.2
mA
V F = 0.45V
IR
Reverse Input Current
50
JAA
VA = S.2SV
VOL
Output Low Voltage
.45
V
IOL = 32 mA
VOH
Output High Voltage
V
IOH = -5 mA
± 50
JAA
V OFF = 0.45 to 5.25V
0.8
V
Vee= 5.0V
See Note
1
V
Vee= 5.0V
See Note
1
Ve
Input Clamp Voltage
Icc
IF
10FF
Output Off Current
Vil
Input Low Voltage
V IH
Input High Voltage
2.0
Input Capacitance
C IN
Not••: 1. Output
2.4
12
Test Conditions
Ie = -S mA
F= 1 MHz
VBIAS= 2.5V, Vce= 5V
TA=2SoC
pF
loading 10l =32 mAo IOH = - 5 mA, Cl = 300 pF
A.C. CHARACTERISTICS
Conditions: Vee
= 5V ± 10%, TA = ooe to 70 e
0
Loading: Outputs - 10l = 32 mA, 10H = - 5 mA, C l = 300 pF
Symbol
TIVOV
TSHOV
Parameter
Input to Output Delay
-Inverting
- Non-Inverting
STB to Output Delay
-Inverting
- Non-Inverting
Min
Max
Units
22
30
ns
ns
40
45
ns
ns
18
ns
30
ns
Test Conditions
(See Note 1)
TEHOZ
Output Disable Time
TELOV
Output Enable Time
10
TIVSL
Input to STB Setup Time
0
ns
TSLIX
Input to STB Hold Time
25
ns
TSHSL
STB High Time
15
ns
NOTE: 1. See waveforms and test load circuit on following page.
5-018
8282/8283
INPUTS
'" '
1=r~'==tm"-l)K
3~~~+",,----1,---_______
I
I
o,,~:--I-ii----1l-T-IV·-o'-1-l-~---------i'"o,;=,~ _,::~{'---------t------''t..l....__-=--------------'
f..--- TSHOV _
NOTE: 1.8283 ONLY -
VOL ,-.lV
SEE NOTE 1
OUTPUT MAY BE MOMENTARILY INVALID FOLLOWING THE HIGH GOING STB TRANSITION.
2. ALL TIMING MEASUREMENTS ARE MADE AT 1.5V UNLESS OTHERWISE NOTED,
1.SV
1.SV
180Q
33Q
OUT ~
.111
Q
20
pF LOAD
50
8282
U
III
/I)
Z
>C
oJ
III
Q
10
pF LOAD
Figure 5. Output Delay VI. Capacitance
5-D20
8284
ABSOLUTE MAXIMUM RATINGS·
Temperature Under Bias ....•..•••....... O·C to 70·C
Storage Temperature ..••.•••...•• - 65·C to + 150·C
All Output and Supply Voltages ••.•.... - 0.5V to + 7V
All Input Voltages •......•....••.... - 1.0V to + 5.5V
Power Dissipation •....•••.•....••.......... 1 Watt .
'NO TlCE: Stresses above those listed under "Absolute Maximum Ratings"
may cause permanent damage to the device. This is a stress rating only and
functional operation of the device at these or any other conditions above
those indicated in the operational sections of this specification is no'
implied. Exposure to absolute maximum rating conditions for extended
periods may affect device reliability.
D.C. CHARACTERISTICS
Conditions: T,,=O·C to 70·C; Vcc=5V± 10%
Parameter
Symbol
Min
Forward Input Current
IF
IR
Reverse Input Current
Vc
Input Forward Clamp Voltage
Max
Units
-0.5
mA
50
~
-1.0
V
Test Conditions
VF=0.45V
V A = 5.25V
Ic= -5 mA
Icc
Power Supply Current
140
mA
V ,L
Input lOW Voltage
0.8
V
Vcc= 5.0V
V,H
Input HIGH Voltage
2.0
V
V cc =5.0V
V
Reset Input HIGH Voltage
2.6
V
Vce= 5.0V
V
5 mA
4
2.4
V
V
-1 mA
-1 mA
0.25
V
Vce= 5.0V
'HR
VOL
Output lOW Voltage
VOH
Output HIGH Voltage ClK
Other Outputs
VIHR-V,LA
RES Input HystereSiS
0.45
.-
A.C. CHARACTERISTICS
Conditions: TA=O·C to 70·C; Vcc=5V± 10%
TIMING
REQUIREMENTS
Plramlt.r
Symbol
MIn
MIX
Unit.
T•• t CondItIon.
Y,N
TEHEL
External' Frequency High Time
13
ns
90% - 90%
TELEH
External Frequency low Time
13
ns
10% - 10% Y,N
ns
(Note 1)
TElEl
EFI Period
TEHEL + TElEH .;. d
XTAl Frequency
12
TRtVCl
ROYI. ROY2 Set·Up to ClK
35
ns
TCLA1X
AOYI. ROY2 Hold to ClK
0
ns
TAtVRtV
ArfIT. ArR!Set·Up to ROYI. ROY2
15
ns
TClA1X
AENI. AEN2 Hold to ClK
0
ns
TYHEH
CSYNC Set·Up to EFI
20
ns
TEHYl
CSYNC Hold to EFI
TYHYl
CSYNC Width
TllHCl
RES Set·Up to ClK
IiiES Hold to ClK
TClllH
25
MHz
20
ns
2 TELEL
ns
--
-----
65
ns
(Note 2)
20
ns
(Note 2)
5-021
8284
TIMING RESPONSES
Symbol
Min
Parameter
TClCl
ClK Cycle Period
TCHCl
Mill
Units
T.. t Condition.
125
ns
ClK High Time
(tI3TCLCl) + 2.0
n~
Fig. 3 & Fig. 4
TClCH
ClK low Time
('I3TClCl) - 15.0
ns
Fig. 3 & Fig. 4
TCHICH2
TCl2ClI
ClK Riae or Fan Time
TPHPl
PCLK High Time
TClCl- 20
ns
TPlPH
PCLK low Time
TClCL- 20
ns
TRYlCl
Ready Inacllve to ClK (See Note 4)
-8
ns
Fig. 5 & Fig. 6
TRYHCH
Ready Active
('hTCLCl)-15.0
ns
Ftg. 5 & Fig. 6
10
ClK (See Note
10
3)
40
ns
TClil
ClK to Reset Delay
TClPH
ClK to PClK High Delay
TClPl
ClK to PClK low Delay
22
na
TOlCH
OSC to ClK High Delay
-5
12
na
TOlCl
OSC to ClK low Delay
2
20
na
f----
ns
22
na
Notal! 1. ~ .. EFI rlae (5 na max) + EFI fan (6 na max).
2. ee! up and hold only necesaary to guarantee recognition at next clock.
3: Applies only to T3 and TW atates.
4. Applle. only to T2 states.
osc
cu,
0
PClK
0
ROY.2
I --'\----
"EN,.2 I
--+----.
·~~·I~·,:~:.
CSYNC
I
m
I
TEHYL
_TClltH-~TItHCl-
TYHYL
"ESET 0
ALL TIMING MEASUREMENTS ARE MADE AT 1 S YOLTS. UNLESS OTHERWISE HOTED
A.C. TEST CIRCUITS
5-D22
------
--
10V to 3.5V
.--
8286/8287
ABSOLUTE MAXIMUM RATINGS·
Temperature Under Bias ................. O·C to 70·C
Storage Temperature ............. -65·C to + 150·C
All Output and Supply Voltages ........ - 0.5V to + 7V
All Input Voltages .................. - 1.0V to + 5.5V
power Dissipation .......................... 1 Watt
'NO TICE: Stresses above those listed under "Absolute Maximum Ratings"
may cause permanent damage to the device. This is a stress rating only and
functtonal operatton of the device at these or any other conditions above
those indicated In the operattonal sections of this specification is not
Implied. Exposure to absolute maximum rating conditions for extended
periods may affect deVice reliability
D.C. CHARACTERISTICS FOR 828618287
Conditions: Vcc
= 5V ±10% TA = DoC to 70°C
Parameter
-
Test Conditions
Max
Units
Ve
Input Clamp Voltage
-1
V
Icc
Power Supply Current-8287
-8286
130
160
mA
mA
IF
Forward Input Current
-0.2
mA
V F = 0.45V
IR
Reverse Input Current
50
IJ.A
VR=5.25V
VOL
Output Low Voltage -BOutputs
-A Outputs
.45
.45
V
V
IOL
IOL
VOH
Output High Voltage -BOutputs
-AOutputs
V
V
IOH=-5 rnA
IOH= -1 rnA
10FF
IOFF
Output Off Curwnt
Output Off Current
VIL
Input Low Voltage
VIH
Input High Voltage
GIN
Input Capacitance
Symbol
Note: 1. B Outputs -
1m
= 32 mA,
Min
2.4
2.4
le= -5 mA
VOFF =0.45V
VOFF=5.25V
IF
IR
-A Side
-8 Side
IOH = -5 mA. CL = 300 pF
0.8
0.9
2.0
12
A Outputs -
1m
= 16 mAo
IOH
= 32 rnA
= 16 rnA
= -1
V
V
Vee= 5.0V, See Note 1
Vee= 5.0V, See Note 1
V
Vee= 5.0V, See Note 1
pF
F= 1 MHz
V BIAS = 2.5V, Vee= 5V
TA =25·C
mAo CL = 100 pF
A.C. CHARACTERISTICS FOR 828618287
Conditions: Vee = 5V ±10%, TA = O°C to 70°C
Loading: B Outputs A Outputs -
IOL = 32 rnA, IOH = -5 rnA, CL = 300 pF
IOL = 16 mA, IOH = -1 rnA, CL = 100 pF
Symbol
Parameter
TIVOV
Min
Input to Output Delay
Inverting
Non,lnverting
TEHTV
Transmit/Receive Hold Time
TTVEL
Transmit/Receive Setup
TEHOZ
Output Disable Time
TELOV
Output Enable Time
Max
Units
22
30
ns
ns
TEHOZ
ns
ns
30
10
Note: 1. See waveforms and test load circuit on following page.
5-D23
18
ns
30
ns
Test Conditions
(See Note 1)
8286/8287
INPUTS
\'1
J~
/
OE
\
J
-Tivov-
- __
TEHOZ !
,.-_________--+
VOH - .1V TELOV-
1-.
OUTPUTS
\V
l
______
J~'--_ _ _ _ _ _ _ _ _-+j_ _ ,
1·---
C=
~'-----I
VOL+.1V
TEHTV
----I
,-.
TTVEL
__________________________________________~r-------------NOTE: 1. ALL TIMING MEASUREMENTS ARE MADE AT 1.5V UNLE'SS OTHERWISE NOTED.
8286/8287 TIMING
5-024
8286/8287
50
50
8217
40
40
u
III
30
z
~
II
20
10
pF LOAD
pF LOAD
Figure 4. Output Delay VI. Capacitance
OUT
~'"
r
OUT
~H'
OUT
~~"
r
r
300PF
OUT
2.14V
1.5V
1.5V
300PF
100PF
SWITCHING
3-STATE TO VOL
3·STATE TO VOL
B OUTPUT
A OUTPUT
B OUTPUT
1.5V
1.SV
2.28V
~'U'
I
300 pF
OUT
~r
100PF
)'STATE TO VOH
)'STATE TO VOH
B OUTPUT
A OUTPUT
Figure 5. Telt Load Circuit'
5-D25
OUT
~"~
r
100PF
SWITCHING
A OUTPUT
8288
ABSOLUTE MAXIMUM RA TlNGS·
'NOTICE: Stresses above those listed under "Absolute Maximum Ratings"
may cause permanent damage to the device. This is a stress rating only and
functional operation of the device at these or any other conditions above
those indicated in the operational sections of this specification is not
implied. Exposure to absolute maximum rating conditions for extended
periods may affect device reliability.
.O·C to 70·C
... -65·Cto +150·C
- 0.5V to + 7V
. ..... -1.l. to +5.5V
. 1.5 Watt
Temperature Under Bias
Storage Temperature
All Output and Supply Voltages.
All Input Voltages.
power Dissipation.
D.C. CHARACTERISTICS
Conditions: Vcc = 5V ± 10%, T A
Symbol
=O·C to 70·C
Min
Parameter
Vc
Input Clamp Voltage
~_________p_ow_e_r_S~up~p_l~y~C_U_r_re_nt_______________+------~----------+_----____~-----------.------~
IF
Forward Input Current
IR
Reverse Input Current
VOL
Output low Voltage-Command Outputs
Control Outputs
2.4
2.4
Output High Voltage- Command Outputs
Control Outputs
Input low Voltage
VIH
Input High Voltage
2.0.
---------~--~--~----~----------------~----~~---------4----~---t------------------1
_ _IO_F_F____ L-.
Output Off Currenl
A.C, CHARACTERISTICS
Conditions: Vcc=5V ±10%, T A =0·Ct070·C
TIMING REQUIREMENTS
.-- ------
- --. r-'---
------------------,--------,--------------,---------r-----------------------"
Loading
Parameter
Min
Ma.
Unit
Symbol
~----------_1-------------------.---~-------~--------------~-------~----------~-------1
ns
TCLCl
ClK Cycle Period
125
f----.-----f-----------------------~------~I_--------------+---------_+_-------------------I
TClCH
CLK low Time
TCHCl
t--- -
ClK High Time
- - - - - -----.---- -
TSVCH
ns
40
----------I-----~- ------~·--------t__--·-----------------I
Status Active Setup Time
t - - -------1----------------TCHSV
ns
66
--·-----jf---------------~--------+_---------·---------I
f - - - - - - - - - - - - - ---
65
ns
-----t-------+-------+----f------------I
Status Active Hold Time
10
-.-- f---
------ - - - - - - 1 - - - - - - - -
ns
. t-
TSHCL
Status Inactive Setup Time
55
ns
TCLSH
Status Inactive Hold Time
10
ns
t - - - - - - - - - - - - - - - - - . - - - - - - - -----:- --t-------------+--------+--------------------I
- - - - - -----._. - - - - - - _ .. - ----_._--_._---
TIMING RESPONSES
Symbot
TCVNV
TCVNX
~ClLH.
Min
Paremeter
Control Active Delay
Control Inactive Delay
TClMCH
10
ALE MCE Active Delay (from ClK)
t--- TSV~!i:. ~~~~_._ f-~MCE A:_tl::':E.e~y (from ~~~u:.::s.:...) ___
Mu
Unit
45
ns
45
ns
15
ns
Loading
ns
+-______+-_____:::..____---<~.....:..:~......,
15
TCHll
ALE Inactive Delay
15
ns
TClMl
Command Active Delay
10
35
ns
TClMH
Command Inactive Delay
10
35
ns
TCHDTl
Direction ContrOl Active Delay
50
ns
TCHDTH
Direction Control Inactive Delay
30
ns
TAElCH
Command Enable Time
40
ns
TAEHCZ
Command Disable Time
40
ns
TAElCV
Enable Delay Time
200
ns
115
TAEVNV
AEN to DEN
20
ns
TCEVNV
CEN to DEN, PDEN
20
ns
TCElRH
CEN to Command
TCLML
ns
5-D26
IORC
"ROC }
MWTC
IOWC
INTA
AMWC
AlOWC
IOL=32 mA
IOH= -5 mA
CL=300 pF
I Inl = 16mA
Other
{ 10H= -1 mA
CL=80 pF
8288
STATE
- - - T 4 - ---T1
T2------+
-TClCl~
-
In~
eLK
-
I-
TCHSV-
\
TSVCH
I--TClCH-
~
-
V\
L riL-J
lL-
- ~y"'~ l7
TSHCl
TCHCl~
1\
T4 -------
lr'\
\
1\
--. 1--- ---
T3--
I
I
,.~
x~
L
ADDRESS/DATA
rr
TCLlH-
ALE
WRITE
DATA VALID
VALID
CD
TSVlHI.-TCHLl
-
rTCLMH
I
/
-
~
-TClML
i--TCLMl
\
\ IV
r\
-
v
I
I-- TCVNV
V
)
J
)
TCVNX-
1---
I
)
)
f--
TCVNV-
I
\
V
DEN (WRITE)
1\
J
-
i
f--- TCVNX
V
!
PDEN(WRITE)
I
I
I
___0-
TCHDTH-
DT/Ii (READ)
(INTA)
I
J-:::'
IJ
----
I
MCE
-
J1
I/li@
r~ I
TClMCH-1
J
;-j
.--TSVMCH
TCHDTL
\
TCHDTH~
f-
TCVNX
1. ADDAESSIDATA IM1$IS SHOWN ONLY FOR AEFEREfojCE PURPOSES
2. UAOtHG EDGE OF ALE ...... 0 Met: IS OETERMIH[O 8Y' THi FALLING EDGE Of CUt OR STATUS OOING ACTIVE. Wt1tCI1EV£A OCCURS LAST
So AU TlWINQ trUASUftEM[NTS AM ""DE AT LSv V"'tlESS $P£CIFIEO OTHERWISE
8288 Timing Diagram
5-027
L
8288
eEN
DEN
DEN, PDEN Qualification Timing
CO~~::~------------------------------~
CEN----------------------------~T~C~EL~A~H-JI
NOTE: CEN MUST IE lOW OR VALID PRIOR TO T2 TO PREVENT THE COMMAND FROM BEING GENERATED.
8288 Addre.. Enable (AEN) Timing (3-State Enable/Disable)
1.SV
OUT
~'~
I*PF
3·STATE TO HIGH
1.SV
OUT
2.28V
2.14V
~n'
OUT
I*PF
~Q"
r*PF
COMMAND OUTPUT
TEST LOAD
'·STATE TO LOW
Test Load Circuit. - 3 State Command Output Te.t Load
5-028
OUT
~""
r~PF
CONTROL OUTPUT
TEST LOAD
Chapter 6
THE ZILOG Z8000 SERIES
The Z8000 series of microprocessors represent Zilog's first 16-bit products. The Z8000 is the second of the new
16-bit microprocessor generation. Inters 8086. described in Chapter 5. was the first product to appear. Motorola's
MC68000 will likely be the next.
Two Z8000 series microprocessors have been announced. The l8002 is a 40-pin package device capable of addressing up to 65.536 bytes of external memory. The l8001 is a 48-pin package.device capable of addressing up to
eight million bytes of external memory. organized as segments of 65.536 bytes. A third device. the Z801 0 Segmentation and Memory Manager. is a companion to the Z8001. The Z801 0 allows memory segments to be dynamically al/ocated under program control anywhere within the eight million bytes of addressable memory.
The Z8000 series microprocessors are upward compatible at the source program level with the 8080A
~nd
the
l80.
The following is a comparison of interesting Z8000 and 8086 innovations:
1)
Z8000 microprocessors do not pipeline instruction object codes. but under some circumstances they do overlap
the next instruction's fetch with the prior instruction's execute. In contrast. the 8086 has a 6-byte object code
pipeline. which. with associated instruction fetch overlap timing. effectively eliminates instruction fetCh times.
2)
The Z8001 and the Z8002 can be visualized as supporting complex and simple microcomputer configurations.
respectively. In contrast. a single 8086 can operate either in complex mode. comparable to the Z8001. or in simPle
mode. comparable to the Z8002.
3)
Both the Z8001 and the Z8002 have built-in logic to handle bus access priorities,in multi-CPU configurations. The
8086 has equivalent logic.
4)
In multi-CPU configurations. each Z8000 series CPU can have its own local memory. while simultaneously sharing
common memory. The common memory may be shared by all CPUs or by selected CPUs. In this respect. the 8086
and the Z8000 series are comparable.
5)
The Z8001 can address up to eight million bytes of external memory. With the help of the Z801 0 Segmentation
and Memory Management Device. this large external memory can be accessed as up to 128 relocatable segments.
where each segment can have up to 65.536 bytes of external memory. The 8086 offers similar relocatable segments without relying on an additional memory management device; however. the 8086 can directly address only
one million bytes of external memory and can only manipulate four segments at a time.
6)
Both Z8000 series microprocessors can be operated in separate "System" and "Normal" modes. Certain privileged
instructions. including all I/O instructions. can be executed in System mode only. System and Normal modes have
separate Stacks. with separate Stack Pointers. Thus. in program-intensive applications. systems software. executed
in System mode. can be separated from application programs. executed in Normal mode. The 8086 offers no
equivalent logic.
7)
The Z8000 has sixteen 16-bit registers that can alternatively be accessed as 8-bit or 32-bit registers. Fifteen of the
16-bit registers can function as index registers. The 8086. in contrast. has four 16-bit registers. plus three separate
16-bit index registers.
The prime sou rce for the Z8000 series is:
ZILOG. INC.
10460 Bubb Road
Cupertino. CA 95014
6-1
Second sources include:
ADVANCED MICRO DEVICES
901 Thompson Place
Sunnyvale. CA 94086
SGS-ATES COMPONENTI ELETTRONICI SPA
20019 Castelletto d i Settimo
Agrate (Milano)
Italy
The Z8000 series microprocessors are manufactured using N-channel silicon gate MOS technology. The Z8001 is
packaged as a 48-pin DIP. The Z8002 is packaged as a 40-pin DIP. Both devices require a single +5 V power supply. All
signals are TTL-level compatible.
The Z8000 requires an external clock with up to 4 MHz frequency. Instructions execute in a minimum of three clock
periods. The maximum number of clock periods is approximately 20; however. a number of instructions require more
time to execute a variety of complex operations.
6-2
THE Z8001 AND Z8002 CPU'S
Because these two versions of the Z8000 CPU are so similar. we will describe them together. Functions implemented by Z8000 series microprocessor chips are. in terms of our general illustration. equivalent to those of the
8086. as illustrated in Figure 5-1.
Z8001 AND Z8002 PROGRAMMABLE REGISTERS
Programmable registers for the Z8001 and Z8002 microprocessors are illustrated in Figures 6-1 and 6-2,
respectively.
Registers RO through R16 can be used as general purpose accumulators. Registers R1 through R16 can. in addition. function as index registers. Register RD is the only general purpose register which cannot function as an index
register.
Both the Z8001 and the Z8002 can be operated in System mode or Normal mode. A
status flag setting determines the mode of operation. System mode will normally be used by
operating system software; Normal mode will be used by application programs. A number of
instructions. including all I/O instructions. are privileged. and consequently can be executed in
System mode only. System and Normal modes have separate Stack Pointers. These are
shown in Figures 6-1 and 6-2 by Sand N suffixes. which represent "System" and "Normal"
modes. respectively.
Z8000
SYSTEM AND
NORMAL
MODES
Z8000 STACK
POINTERS
For the ZaOD2. the single 16-bit register R15 serves as the Stack Pointer.
For the ZaDD" two 16-bit registers are needed to implement a Stack Pointer. since memory addresses may be up to 23
bits wide. Registers R14 and R15 are used.
Instructions that access 16-bit registers do not make any special allowance for R15 and/or R14 functioning as Stack
Pointers. Thus. the Stack Pointer can be accessed as a general purpose register/accumulator, or it can be used as
the Index register for indexed memory addressing. The fact that there are separate System and Normal mode Stack
Pointers is inconsequential when these registers are being accessed as accumulators or index registers. Depending on
the currently selected mode. one or the other Stack Pointer will be accessible. This may be illustrated as follows:
Normal
Mode
System
Mode
.----1 R11
.-----1 R12
.----1 R13
R14 (or R14N)
~--~
L.-_ _......
1-----1
R11
.-----1 R12
R13
~----I
R14 (or R14S)
.-----1 R15S
R15N
Whenever two 16-bit registers provide a memory address for the Z8001. register bits
are utilized as follows:
15
14
13 12
11
10
9
8
7
6
5
4
3
2
Z8001 32-BIT
ADDRESS
REPRESENTATION
0 " ' - B i t No.
r-o""'T'---s-e-gm-e-nt-N-o-.---"I~o"l~o"lr--o-ll""'""o""'l-0""'1-0""'1-0""'-0-1 Register RN holds the
_
....._ _ _ _ _ _ _ _ _ _....................................._
...._.&.._.&.._
..........
7-bit segment number in
bits 8-14. Other register bits are O.
Offset
..._ _ _ _ _ _ _ _ _ _ _
____________
RN+ 1 holds the
16-bit offset. or
I Register
~_
address within the
segment identified in
Register RN.
6-3
15
14 13 12
11
10
9
S
7
6
5
4
3
2
o ~ Bit No.
(for all registers)
RO - Accumulator
R1
R2
R3
R4
R5
R6
R7
Accumulators and
Index Registers
RS
R9
R10
R11
R12
R13
R14S
R14N
R15S
System and Normal Stack Pointers.
Accumulators and Index Registers
R15N
b.b.b.b.b.b.b.b.~~~.lli~b..lli~b.~iill }
~~~~~~~iJi.i.~;;';';~,;,J
Flags and Control Wo",
} P"'.n1m Counte'
} New P",• .am Status A'ea Po;n'e'
I
Refresh Counter
Figure 6-1. Za001 Microprocessor Programmable Registers
6-4
15
14 13 12
11
10
9
8
7
6
5
4
3
2
o 4----
Bit No. (for all registers)
RO - Accumulator
R4
R5
R6
R7
R8
Accumulators and
Index Registers
R9
R10
R11
R12
R13
R14
R 15S } System and Normal Stack Pointers,
R 15N
Accumulators and Index Registers
Flags and Control Word
Program Counter
New Program Status Area Pointer
Refresh Counter
Figure 6-2.
Z8002 Microprocessor Programmable Registers
6-5
The segment number and offset translate into a 23-bit memory address as follows:
A22} Memory
Segment
.,....-_~A16
.-+-~~ A 15 }
16-bit
23-bit
Memory
Address
Offset
'1---~AO
Thus. the Z8001 Stack Pointer is shown in Figure 6-1 with bits 8 through 14 of Register R14 providing the segment number. while the whole of Register R15 provides the segment offset. The
Z8002 Stack Pointer. shown in Figure 6-2. is a simple 16-bit address register.
Z8000
STACK
POINTER
The Program Counter is a simple 16-bit register for the Z8002. but for the Z8001 two 16-bit words
are used. with the 23-bit address divided into a segment number and an offset. as illustrated
above.
Z8000
PROGRAM
COUNTER
The Z8000 addresses memory as bytes; however 16-bit words must originate on even byte address boundaries. That is why the Z8001 uses two 16-bit words to generate extended memory addresses. even though only 23 bits
of address are required. 23-bit addresses could be implemented in three bytes. rather than in two 16-bit words;
however. this would complicate pushing and popping memory addresses. Were the addresses implemented as three
bytes. all Stack operations would require three byte pushes or three byte pops. By making all addresses occupy two 16bit words. Stack operations are reduced to two word pushes or two word pops which require no more time than three
byte pushes or pops.
The Flags and Control Word provide. the Z8001 and the Z8002 with Status and Control bits.
Bits are interpreted as follows:
15
14 13 12 11
0
0
0
~EG ~/N 0
V
IE
NV
IE
0
.~
0
.
~
.~
10
9
8
0
0
0
0
7
6
0
0
0
C
A~
5
4
3
0
0
0
0
l
S
~
~
P/O DA
.~
o '-"-BitNo.
2
---
Reserved word. l8001 only
0
0
0
H
0
0 ~ Flags and Control Word (FCW)
.f
Auxiliary Carry
Decimal Adjust
Parity or Overflow
Sign
lero
Carry
Non-vectored interrupt enable/disable
Vectored interrupt enable/disable
System/normal mode select
Segmentation mode select (Z8001 only)
Always 0 for Z8002
6-6
The Parity, Overflow, Sign, Zero, and Carry statuses are absolutely standard.
Parity and Overflow share a status bit.
The Parity status is modified by logical instructions which test the parity of byte data. This status is set to 1 for even
parity; it is cleared for odd parity.
The Overflow status is equal to the Exclusive OR of carries out of the high-order and penultimate bits following
arithmetic and logical operations.
The Sign status is set to the value of the high-order result bit following arithmetic operations.
The Zero status is set to 1 when the result of an operation is 0; it is reset to 0 otherwise.
The Carry status reports carries out of the high-order bit following arithmetic operations. This status is also used by
most shift and rotate instructions.
Most microprocessor instructions routinely modify status bits. whether or not such modifications are relevant to the
operation performed. zaooo status logic generally follows the PDP-11 minicomputer. but the zaooo has a few
anomalies. You should therefore consult Table 6-3. which summarizes the zaooo instruction set. in order to determine how a particular status is affected by the execution of any specific instruction.
The Auxiliary Carry and Decimal Adjust status flags differ somewhat from normal use. These flags are modified by
byte arithmetic instructions in order to make binary coded decimal arithmetic possible. You cannot set or reset these
two Status flags using any of the status bit control instructions. and reading the value of these flags provides little
useful information. The assembly language programmer should ignore these two flags.
NVIE and VIE are used to enable and disable non-vectored interrupts and vectored interrupts, respectively. You
enable interrupts by setting the appropriate status bit to 1. and you disable interrupts by resetting the appropriate
status bit to O.
The SIN status flag is used to switch between System and Normal modes. When this bit is 1. zaooo
microprocessors operate in System mode. When this bit is O. zaooo microprocessors operate in Normal mode. Recall
that System and Normal modes have their own separate Stack Pointers; also. certain privileged instructions can only be
executed in System mode.
The SEG status is used by the Z8001 microprocessor only. When this bit is set to 1, the Z8001 operates in Segmented mode; when this bit is set to 0, the Z8001 operates in Nonsegmented mode.
In Segmented mode. all Za001 addresses are computed 23 bits wide. using two 16-bit memory
Z8001
words as previously illustrated. Za001 Nonsegmented. Normal mode is directly equivalent to
SEGMENTED
Za002 Normal mode operations. Za001 Nonsegmented System mode is not exactly equivalent to
MODE
Za002 System mode; differences occur in interrupt acknowledge stack handling. as explained
later in this chapter. Thus Za002 Normal mode programs can be executed within any single segment of Za001 memory.
The Za001 carries an unused word as a companion to the Flag and Control Word. since all Za001 automatic Stack
operations push and pop data in word pairs. Status in the Flag and Control Word must also be pushed and popped as a
32-bit unit - hence the unused companion word.
6-7
The New Program Status Area Pointer is used by interrupt logic. It consists of one or two 16-bit
words. as illustrated in Figures 6-1 and 6-2.
Z8000 NEW
Following any interrupt acknowledge. a vector address is created using the New Program Status
Area Pointer and a 9- or 1O-bit displacement provided by interrupt acknowledge logic. as follows:
STATUS AREA
POINTER
PROGRAM
ze001
15 14 13 12 11 10 9
8 7
6
5
4
3
2
o .--BitNo.
P-~----------~------~~~~~--~~--~~~
o
Segment No.
15 14 13
SN6
12 11
10
9
8
7
6
5
4
2
0
AO
A11 A10
SNO A15
3
Interrupt Vector
Address
zeOO2
15 14
13
12
11
10
9
8
7
6
5
4
3
2
0 4--BitNo.
Upper Offset
Displacement
A15
A10 A9
AO
,/
Interrupt Vector
Address
Although the Z8002 uses just one byte of its New Program Status Area Pointer. 16 bits are provided. since all Z8002
automatic Stack operations push and pop 16-bit words. Likewise. the Z8001 New Program Status Area Pointer uses
two 16-bit words. where a Single 16-bit word would suffice. to accommodate automatic Stack handling logic which
pushes and pops data in 32-bit increments.
The Refresh Counter will be described later. along with memory refresh.
6-8
Z8000 REGISTER DESIGNATIONS
Z8000 series microprocessor Instructions access a-bit, 16-bit, or 32-blt registers, as illustrated in Figure 6-3.
Register designations used by Zilog assembly language mnemonics are shown in this figure.
Byte Instructions access sixteen 8-bit registers. illustrated in Figure 6-3 by RHO through RL7.
Z8000 BYTE
REGISTERS
Word Instructions access the sixteen 16-bit registers RO through R16.
Z8000 16-BIT
REGISTERS
Long word instructions access general purpose registers in pairs. Eight 32-bit registers are
therefore available. shown in Figure 6-3 as RRO through RR14.
Z8000 32-BIT
REGISTERS
Most Z8000 series instructions that access memory or registers have a word version and a byte
version. A limited number of instructions have a long word version.
Multiplication and division Instructions sometimes use 64-bit registers, Shown in Figure 6-3 as RQO through
RQ12.
Z8001 AND Z8002 MEMORY ADDRESSING MODES
Most Z8001 memory addreSSing modes have two forms: one for Nonsegmented mode, the other for Segmented mode.
When operating in Nonsegmented mode. all Z8001 memory reference instructions compute nonsegmented memory
addresses: the offset address is modified. but the segment number is not altered.
When operating in Segmented mode. l8001 memory reference instructions compute segmented memory addresses.
provided the instruction also has a segmented memory addressing option. But there are some memory reference instructions that have no segmented option: these instructions compute nonsegmented memory addresses. even for a
l8001 operating in Segmented mode.
A segmented memory reference Instruction computes new values for the segment number and offset addre.s.
The Z8002 Program Counter is a Single. 16-bit register. equ iva lent to the l8001 Program Counter Offset register.
l8002 memory reference instructions therefore compute nonsegmented memory addresses only.
In the discussion which follows. we will illustrate l8000 memory addressing options for Segmented and Nonsegmen ted modes.
In Segmented mode the base address always specifies the segment. The base address may
occupy two 16-bit words:
Word or Register Number
15 14 13 12 11
10
9
a
Segment No.
7
6
5
1 0 10 I 0
I
4
3
0 , 0
I
2
1
0 -4--Bit No.
0 , 0 '0
I}
1__________1_6_-B_i_t_o_ff_s_et_ _ _ _ _ _ _ _ _ _..1
n + 1 ..
Instruction word
15 14 1 3 12 11
p
p+1
10
9
a
7
6
5
4
3
2
1
'I
16-Bit Offset
Long form
segmented mode
base address
(not in instruction)
0 " - Bit No.
I...- -........se-gm-e-nt-No-.--'-0-'-0-'-0-'-0-'-0....'-0....'-0-'.....
0
I
Z8001
BASE
ADDRESS
I
Long form
segmented mode
base address
(in instruction)
or it may occupy a single 16-bit instruction word:
15 14 13 12 11 10 9 a
7 6
5 4
3 2
0 - 4 - - Bit No.
r -o.....
'---S-e-g-m-e-n-t-N-O-.--.....,r------a--b-it-O-ff-se-t----..
Short form segmented
I•
.
mode base address
6-9
Z8001
Z8002
o ...--Bit
8 7
15
'RO}
No.--"15
r:~rS~!lrs {RO
8
7
~~
o
RRO
Rl
R1
~::}
ROO
RR2
{::
---------------------------::}
RR4
{::~
R04
::}
RRS
{::
---------------::
}
RR8 {
::
R08
~~~~~~..::~} RR,O {R:::"""""""",!!"""",!!,,!,!,
----------------------------
~R12
}RR12{
R12~
R13 h±~~~;lliiilll
......................................................................"""R13
R012
~~~~JI} C:~~::"'!.,
[_:':~':i o i o·:
..
_~G_i'...:' :"....,, ....'..,'"...';
...
;: ...
!:""":·:_:;'"","...
" '.....' '....
...
{
1} P",.,am Co,n'., { :::~~::______
'_1,2"",::,;:,:,·':;;;""'·....
' • ....
.'1"....
:}....
.....____________......... . .,;,......... . .,;,. . . ,;,. . . ,;,.__.;.;:.;a}
iOOiOOi
N;;:.~;::.m {~~~~. . .
Figure 6-3. Various Register Designations for the Za001 and Za002 Microprocessors
6-10
The index or displacement portion of a memory address never specifies the memory segment. The index is always a
single 16-bit value for any Z8000 microprocessor. operating in any mode. The displacement may be an 8-bit. 12-bit. or
16-bit value: but once again. it is the same for all Z8000 microprocessors. and all modes of operation.
Thus. in Segmented mode. there is a clear difference between a base address and an index or displacement. In Nonsegmented mode. there being no segment number. there is no difference between a base address and an index.
Most Z8000 series memory reference instructions access data memory using implied. direct. or indexed memory addressing.
Z8000
Z8002 implied memory addressing may be illustrated as follows:
16-bit
Registers
Memory
IMPLIED
MEMORY
ADDRESSING
....._ _...... RN-1
xxxx
RN -4--lnstruction
specifies
1---+--1
memory address
.....-+_......-RN+2
in register RN.
Memory byte
xxx x is accessed.
RN+1
xxxx-l
xxxx
1----1
1-----1
---Bytes
x represents any hexadecimal digit
The Z8001 uses only nonsegmented implied memory reference instructions to access data memory. The Z8001 does
not use either short or long segmented implied memory addressing to access data memory. Z8001 implied memory addressing may therefore be illustrated as follows:
16-bit
Registers
1----1
Memory
ss xxxx-1
RN-1
RN ..._Instruction specifies
1-~f---IRN+1
memory address is
register RN. Memory
byte ss xxxx is
accessed.
ss xxxx
1----1
t ____
~
I
ss xxxx+3
Bytes
s and x represent any hexadecimal digits
6-11
Segment number ss
is current contents
of Program Counter
Segment register.
Some Za001 program memory reference instructions (such as the subroutine call) use long segmented implied memory
addressing. which may be illustrated as· follows:
16-Bit
Registers
~----I
Program
Memory
RRN-l
RRN -4- Instruction specifies
memory address in
Register RN. Memory
RRN+l
1---\---1
byte xxxx in segment
ss is accessed.
ss 00
..
ss xxxx-l
~--ss xxxx
t-----I
~
Bytes
s and x represent any hexadecimal digits.
RRN is a 32-bit register designation.
We will now examine direct memory addressing.
Z8002
For the Za002. direct memory addressing may be illustrated as follows:
DIRECT
MEMORY
ADDRESSING
Memory
The direct address xxxx is
pppp "'___ _ _-I
PPPP+ 1 r
provided by the instruction
object code
pppp+2
xx
pppp+3 t----:x;-x_--t
I
Instruction code
Direct Address
..
xxxx
~---
xxxx+l
xxxx+2
.----1
t-----11
x and p represent any hexadecimal digits
Note that the direct address xxxx. being a 16-bit value. must start at a memory byte with an even address. This requirement is illustrated above by the address pppp+2.
Furthermore. the high-order byte of a 16-bit memory word is at the lower address. preceding the low-order byte:
Program
Memory
PPPP
t-----I
High-order
byte
-- -hh
PPPP+ 1 t - - - - - t
pppp+2
hh
1 - - - - -.....
pppp+3
pppp+4
II
t-----I
6-12
Low-order
byte
II
16-bit data from memory
A Za001 direct memory address may be nonsegmented, short segmented, or long segmented. Nonsegmented
Za001 direct memory address logic is as illustrated above for a Za002 direct memory address, except that the most recently computed segmented number is output from the Program Counter Segment register via the seven Za001 segment number lines.
Long segmented Za001 direct memory addressing may be illustrated as follows:
Za001 LONG
SEGMENTED
DIRECT MEMORY
Memory
ADDRESSING
tt pppp
} Instruction code
tt pppp+1
The direct address
xxxx, in segment
ss, is provided by
the instruction
object code
tt pppp+2
tt pppp+3
ss
00
tt pppp+4
xx
tt pppp+5
xx
••
J
Long ",gment.d
direct address
••
SSXXXX_1~
ss xxxx
ss xxxx+ 1
sand t are any hexadecimal digits that specify a segment number in the range 00 through 7F 16.
P and x are any hexadecimal digits that specify a memory address within a segment.
We can illustrate a short segmented Za001 direct memory addessing as follows:
Memory
Segment Number
Z8002 SHORT
SEGMENTED
DIRECT MEMORY
ADDRESSING
~
tt pppp
}
t - - - - - I Instruction code
ss'- tt pppp+ 1
Offset " ' " ' tt pppp+2 ~--.. } Short segmented
ss
OOxx_ tt pppp+3
xx
direct address
The direct address
OOxx, in segment
ss, is provided by
the instruction
object code
tt pppp+4 I-~~""""
•
•
SSOOXX_1~'
•
ss OOxx
ss 00xx+1
sand t are any hexadecimal digits that specify a segment number in the range 00 through 7F 16·
P and x are any hexadecimal digits that specify an offset, or memory address within the segment.
6-13
Z8001 and Z8002 indexed memory addressing adds the contents of an index register to a direct address. 16-bit
registers Rl through R15 can function as index registers. Register RO cannot function as an index register. The direct
address provides the base to which an index is added.
Z8002 indexed addressing may be illustrated as follows:
l6-bit
Registers
RN-l
RN
RN+l
Z8002
INDEXED
ADDRESSING
Memory
pppp
t----f
yyyy
} Instruction code
pppp+l
t---\:-oof
xxxx
+ yvyy
zzzz
pppp+2
xx
pppp+3
xx
..
} Direct
Address
zzzz-l
zzzz
zzzz+l
P. x. y and z represent any hexadecimal digits. The selected memory address zzzz is the sum of the
direct address xxxx. which is provided by the instruction ob ject code. and the index yyyy.
The instruction code specifies Register RN as the Index register.
The illustration applies also to nonsegmented laOOl indexed addressing. but for the laOOl a segment number (ss)
would precede the computed address zzzz. Since no segment is computed by the laOOl in Nonsegmented mode. ss
would be the current contents of the Program Counter Segment register.
Here is an illustration of Z8001 short segmented indexed addressing:
l6-bit
Registers
Memory
tt pppp
RN-l
RN
tt pppp+l
tt pppp+2
yyyy
RN+l
Z8001 SHORT
SEGMENTED
INDEXED
ADDRESSING
} Instruction code
ss
Short segmented
} direct address
xx
.....--,.---t
ss OOxx
+ yyyy
ss zzzz-l
~zzzz ~
:=J
.....
--~
ss zzzz .....- - - f
ss zzzz+l
Segment digits are not 4
affected by this addition.
Any carry out of high-order
digit addition will be lost.
The effective memory address ss zzzz is not the simple sum of ss OOxx and yyyy. The segment number ss is output
directly on the segment number pins. bypassing the address addition. OOxx and yyyy are added to create zzzz. the
offset. which is output on the Address Bus. In the event that adding OOxx to yyyy generates a carry out of the highorder bit. this carry is lost. Thus the offset addition wraps around from FFFF16 to 000016. without incrementing the
segment number.
6-14
Long segmented Z8001 indexed addressing uses a four-byte direct address. with a 16bit offset. as follows:
Z8001 LONG
SEGMENTED
INDEXED
ADDRESSING
The computed offset zzzz becomes the sum of xxxx and yyyy.
Note that long segmented indexed addressing offers the same addressing range as short segmented indexed
addressing: the index, on its own, can address the entire offset space of 65,536 bytes. Therefore, the one-byte short
segmented base address offset is no handicap. Suppose, for example, you use indexed addressing to access a data table in the middle of a segment. Using long segmented indexed addressing, the base of the data table might be provided
by the direct address offset. while the Index register provides the displacement into the selected table:
Memory
Table
Origin xxxx from--~t----t xxxx
base address
Displacement into
table from Index register
yyyy
..
But you could just as easily have the index originate at the base of the segment:
Memory
Table
n
11
0000
Base address
offset = 00
I
I
:
:
Start of Table
--Start of Segment
Displacement into
table, from segment
origin, in Index register
xxxx
yyyy
A few l8001 and l8002 instructions access data memory using base relative addressing,
wherein the contents of an Index register are added to a base address, which is also held in
CPU registers. Therefore, l8000 base relative addressing might also be called "implied, indexed" addressing.
6-15
Z8000 BASE
RELATIVE
ADDRESSING
Z8000 IMPLIED
INDEXED
ADDRESSING
Z8002 base relative addressing may be illustrated as follows:
16-bit
Registers
Memory
......-.......
xxx x
RM-1
RM
xxxx
I---~-I
T---------
zzzz-1
RM+1 ......___.......
zzzz - - - - . , . zzzz 1 - - -.......
zzzz+1
I---....-t
RN-1~_...........
RN
YYYY
Index
RN+1
"-v-"
I----t
Byte
x, Y and z represent any hexadecimal digits
The instruction object code must specify the register number from which the base address and the index are taken. In
Nonsegmented mode there is no difference between a base address and an index; each is a single l6-bit value. The
effective memory address zzzz is simply the sum of any two l6-bit registers' contents.
The illustration above applies also to nonsegmented laOOl base relative addressing. However, for the la001 the
memory segment ss, currently in the Program Counter Segment register, is output via the seven la001 segment number lines.
Consider next Z8001 segmented base relative addressing. The base address specifies the
segment. thus the base address and the index differ. Short segmented base relative addressing
may be illustrated as follows:
16-bit
Registers
RM-1
RM
Z8001 SHORT
SEGMENTED
BASE RELATIVE
ADDRESSING
Memory
......- - - - 1
ss xx
ss OOxx
1--_--1
RM+1~__~~____- -
+ yyyy
S5
zzzz-1
1 - - -.......
~~~sszzzzl-_ _--1
ss zzzz+1
RN-1
1 - - -.......
I-~,,"....-t
RN
yyyy
Index
RN+1 ......- -.......
~
Byte
Segment number
Offset
x, y and z represent any hexadecimal digits specifying offsets, or addresses within memory segment ss.
ss can have any value in the range 00 through 7F16'
The index is added to the base address USing the same mechanism described earlier for short segmented indexed addreSSing. The discussion of addressing range given for Short segmented indexed addressing applies also to short segmented base relative addressing.
6-16
Long segmented base relative addressing may be illustrated as follows:
16-bit
Registers
. Z8001 LONG
SEGMENTED
BASE RELATIVE
ADDRESSING
Memory
R::~ l~-ss-o-o-'"
ss zzzz-1
xxxx
ss xxx x
+ yyyy
RRM+1
ss zzzz
....
~--
sszzzz+1 ......_ _--1
~~ ----"
RN-1
~
......---jl--o--l
RN
vyyy
Index
Byte
......- - - - 1
RN+1
......- - - - 1
Segment number
Offset
x, y and z represent any hexadecimal digits specifying an offset address within memory segment ss.
RRM designates a 32-bit register, while RN designates a 16-bit register.
These registers are specified by the instruction object code.
Some program memory reference instructions use program relative addressing. A displacement provided by the instruction object code is added, as a signed binary number, to the
contents of the Program Counter. For the Z8002 this may be illustrated as follows:
Program
. Memory
Z8002
PROGRAM
RELATIVE
ADDRESSING
"",,,+4 - d d d d t 2 s
•
I
I
I
lop~.
xxxx
xxxx+1
C
dd
xxxx+2
xxxx+3
Program relative
addreSSing range
signed binary
displacement
dd
xxxx+4
1--_--
xxxx+4 + dddd ...
I
I
I
I
XXXX+4+ddddt2~
In the illustration above. dddd is divided by 2 to show the addressing range of a program relative address; this is
because dddd is a signed binary number. Therefore. half of the possible values will increase the address in the Program
Counter (xxxx+4); the other half will decrease this address.
Depending on the instruction. dddd may be an 8. 12. or 16-bit signed binary value. This displacement is added to the
contents of the Program Counter after the Program Counter has been incremented to address the first byte of the next
sequential instruction. This is illustrated above as location xxxx+4.
Some program relative instructions (such as Jump Relative) have the displacement included in the op-code word. saving memory space and execution time for short displacements. In these instructions the first byte of the next sequential
instruction would be xxxx+2. instead of xxxx+4 as illustrated above.
6-17
za001 program relative addressing, in Segmented or Nonsegmented mode, follows the illustration above; however, the Program Counter also specifies the memory segment. The displacement is added to the Za001 Program Counter Offset register. The Za001 Program
Counter Segment register is not changed. Thus the displacement for a program relative memory address cannot cross a segment boundary.
Z8001 PROGRAM
RELATIVE
ADDRESSING
A few Z8000 jump instructions use indirect memory addressing. That is to say, the contents of the addressed memory location are loaded into the Program Counter. Very few
microprocessors provide indirect addressing. See Volume 1, Chapter 6 for a detailed description of this addressing mode.
Z8000 INDIRECT
MEMORY
ADDRESSING
The Z8000 instruction set includes a number of memory reference instructions with
auto-increment and auto-decrement. An implied memory address held in a 16-bit register is
incremented or decremented following an instruction's execution, thus leaving the address
pointing to the next sequential memory location in a table or string.
Z8000 AUTOINCREMENT
AND DECREMENT
The Z8000 Stack decrements the Stack Pointer before a push and increments the Stack
Pointer following a pop. In other words, the beginning of the Stack has the highest memory
address, and the current top of Stack has the lowest memory address.
Z8001 AND Z8002 PINS AND SIGNALS
Signals and pin assignments for the two Z8000 series microprocessors are illustrated in Figure 6-4.
ADO-AD16 is a multiplexed 16-bit Data/Address Bus. AS is an address strobe which is pulsed low while an address is being output. OS is a data strobe which is pulsed low while data is being output or input.
STO-ST3 are four machine cycle status Signals whose output levels further Iyyyyl
Abort if IPpppqqqq\ > Ixxxxyyyyl
6-38
The multiply instruction also has word and long word versions. The multiplicand is held in CPU
registers. The multiplier may be held in data memory. in CPU registers. or it may be provided immediately by the multiply instruction. The product is returned in CPU registers. The word option
may be illustrated as follows:
16-bit
Registers
{R:~~pppp
RR~
Z8000
MULTIPLY
INSTRUCTION
16-bit Register. or
2 memory bytes
zm emz . ~
yyvy
yyyy
ppppqqqq
P. q. y. and z represent any hexadecimal digits.
RR represents a 32-bit register pair. as illustrated in Figure 6-3.
Long word multiplication may be illustrated as follows:
16-bit
Registers
32-bit Register. or
4 memory bytes
xxxx
yyyy
P. q. s. t. v. w. x. and y represent any hexadecimal digts.
RQ represents a 64-bit register. as illustrated in Figure 6-3.
The multiply instruction treats all numbers as signed binary values. Status flags are adjusted as follows:
Carry (CI. C is set for overflow or underflow. For word multiplication. underflow occurs if the answer is less than -2 15 ;
overflow occurs if the answer is 2 15 or more. For long word multiplication. underflow occurs if the product is less than
-2 31 ; overflow occurs if the product is 231 or more. Carry is cleared if there is no underflow or overflow.
Zero (Z). The Zero status is set if the result is 0; it is cleared otherwise.
Sign (S). The Sign status is set for a negative result; it is reset otherwise.
The Overflow status is always cleared.
The LOPS instruction is somewhat unusual in that it loads both the Program Counter and the Flag
and Control Word. Data is taken from memory as follows:
Non-Segmented
(Memory bytes)
Segmented
(Memory bytes)
FCW (HI)
00
00
FCW (LO)
PC (HI)
FCW (HI)
PC (LO)
FCW (LO)
PCSEG(HI)
PCSEG(LO
PCOFF(HI)
PCOFF(LO)
The LDPS jump instruction uses indirect memory addressing.
6-39
Z8000 LDPS
INSTRUCTION
Z8000
SUBROUTINE
CALL
A subroutine CALL can use segmented implied memory addressing:
RR
{
RN~
RN+'~
The System Call instruction generates an interrupt acknowledge sequence. You will recall
from the discussion of Z8000 interrupt acknowledge logic given earlier in this chapter that an interrupt identifier is pushed onto the Stack during every interrupt acknowledge sequence. For the
System Call instruction, this identifier is the System Call instruction object code; the low-order
byte is an 8-bit immediate data value which you specify in the instruction operand. This may be illustrated
SC
Z8000
SYSTEM
CALL
as follows:
xx
~
7F
xX16
Note that the JP conditional jump instruction can use segmented implied memory addressing. As we might expect from a Zilog high-end microprocessor, the Z8000 has a
large number of block transfer and search instructions. These instructions come in
groups of eight. For each type of instruction there are four word versions and four comparable
byte versions. The four versions include an increment. an increment and repeat. a decrement.
and a decrement and repeat. See our earlier discussion of block transfer I/O instructions for a
general description of these four variations.
Z8000
CONDITIONAL
JUMP
INSTRUCTIONS
Z8000 BLOCK
TRANSFER
INSTRUCTIONS
The LDM block transfer instructions move data between a number of 16-bit registers
and memory. You can transfer data from memory words to 16-bit registers or from 16-bit registers to memory words.
You can transfer from 1 to 16 words in a Single execution. Register addressing is wrap-around. For example, the instruction:
LDM R13.THERE,6
will transfer six words of data from memory to registers. in the following sequence:
Memory
THERE
R13
THERE+2
R14
THERE+4
R15
THERE+6
RO
THERE+8
R1
THERE+10
R2
Among the block transfer and search instructions there is a group of translate instructions. These are table look-up instructions; they work as follows:
TRxB
@RM.@RN.RW
16-bit Registers
RM
I.
8-bit Memory
ppp-1
popp
.....- -.......
•
RN
•
~
----I
xxxx
.--r xxxx
-----~
+ yy
zzzz---......~
zzzz+1
.....- -.......
RW is decremented and RM is incremented or decremented, depending on the instruction.
6-40
As illustrated above. the contents of the destination memory location serve as an 8-bit index into a memory table. The
contents of the addressed table byte replace the index. RH1 is used. and left with an undefined value. Translate instructions are typically used to convert characters from one code to another. For example. a single translate instruction
could convert an EBCDIC character to an ASCII character. In this case the EBCDIC character code would constitute an
index into a 256-byte ASCII code table. At the location specified by the EBCDIC code. you would store the ASCII
equivalent. Executing a translate instruction would then cause the ASCII equivalent of an EBCDIC code to be loaded
into the memory location in which the EBCDIC code had been stored.
A variation of the Translate instruction is a Translate-and-Test. which loads the addressed table byte into Register RH1.
but leaves it there. The Z flag is set if RH1 is O. and Overflow is set if the counter decrements to O.
~!""""""'~~........-
..
There are a deceptively large number of shift Instructions listed in the register operate
Z8000 SHIFT
group. In fact, the only difference between an arithmetic and a logical left shift lies in the
INSTRUCTIONS
Overflow status. For an arithmetic shift this status is set if the high-order (Sign) bit changes
following the shift; the Overflow status is cleared otherwise. Following a logical shift the Overflow status is undefined.
For right shifts the Sign bit is replicated for arithmetic shifts. while zeros are filled in for logical shifts.
The only difference between a dynamic shift and a non-dynamic shift is in the location of the shift bit count. A dynamic
shift takes its bit count from a CPU register.
A non-dynamic shift takes its bit count from immediate data provided by the instruction operand.
Note from our earUer discussion of the Z8000 Stack that the Stack address is incremented for a pop and decremented for a push. In other words. the bottom of the Stack has
the highest memory address. and the top of the Stack has the lowest memory address.
Four instructions control the MI input and MO output signals. They are MBIT, MREQ,
MRES, and MSET.
MBIT simply inverts the level of the
Mi
input and returns it in the Sign status.
MRES outputs a high signal via MO. while MSET outputs a low signal via MO.
6-41
r--------.
Z8000 STACK
INSTRUCTIONS
Z8000
MIANDMO
INSTRUCTIONS
MREQ uses MI and MO to request external access. This instruction uses Zero and Sign statuses. MREQ execution logic
may be illustrated as follows:
z=o
Yes
5=0
MO= 1
MO=O
Decrement
Counter
to 0
Yes
5=0
5=1
MO=O
Z=l
When the MREQ instruction begins execution, the Zero status is cleared; it is set to 1 after MREQ has completed execution if a request was signalled.
The MI input is tested to see if the external resource being arbitrated is available. If MI is low, then the resource is not
available; MO is output high and the Sign status is reset to O.
If MI is being input high, then the external resource is available. MO is output low, then a time delay is inserted by
decrementing the contents of a 16-bit register to O. This delay gives external logic time to receive and propagate MO.
External logic must acknowledge the resource request by inputting MI low. In response to MI low, MREQ sets the SJ.9.!!
status and the Zero status to 1. But if MI is still high after the counter has decremented to 0, then MREQ outputs MO
high, resets the Sign status to 0, and sets the Zero status to 1. Therefore, following execution of the MREQ instruction,
CPU logic interprets results as follows:
Sign
Zero
MO
o
o
o
o
o
o
o
1
1
~} :::U:::s:~ede but not granted
1
1
1 Request made and granted
1
1
1
1
o
o
1
No request made
Not possible
6-42
THE BENCHMARK PROGRAM
The Z8000 can execute our benchmark program using just three instructions. We assume the following memory
map:
l6-bit
Memory
Locations
NLOC
NLOC+2
NLOC+4
IOBUF
t------I
FREE
t------I
·
COUNT
I
IOB~~
I
I
I
•
•
•
Words
IOBUF+NB
•
•
·
TABLE§
I
•
'-
These words
must be
moved
·•
.•
'REE'a';'" ,. .
•
I
s
I
•
Tobie Wo",
,
•
Using the LDM Multiple Register Load instruction. we can initialize the addresses and word count in appropriate
registers for an LDIR Block Transfer and Repeat instruction. Finally. we update the address of the first free table word.
Here is the necessary instruction sequence:
!LOAD 10BUF STARTING ADDRESS INTO R1. TABLE STARTING
!ADDRESS INTO R2. WORD COUNT INTO R3
NLOC
LDM
LDIR
LD
R1. NLOC. 3
@R2. @R1. R3
NLOC+2. R2
DA
DA
DA
10BUF
FREE
COUNT
!MOVE DATA BLOCK
!UPDATE ADDRESS OF FIRST FREE WORD IN
TABLE
!I/O BUFFER BASE ADDRESS
!DATA TABLE FIRST FREE WORD ADDRESS
!WORD COUNT
6-43
The nomenclature used to identify Z8000 registers is given in Figure 6-3.
The following abbreviation. are u.ed in Table. 6-3, 6-4, and 6-5:
addr
addrls
addrss
adrsx
b8
b16
cc
data8
data16
data32
disp
FCW
FLAGS
flag
int
ioaddr
(I/O)
MI
MO
MSB
n16
NSPOFF
NSPSEG
PC
PCOFF
PCSEG
PSAPOFF
PSAPSEG
rb
rbd
rbs
REFRESH
ri
rid
ris
rid
rls
rqd
rw
rwd
rws
SP
any 16-bit nonsegmented address
any 32-bit long segmented address
any 16-bit short segmented address
one of six standard memory addressing modes
immediate value in the range 0-7
immediate value in the range 0-15
condition codes. as summarized in Table 6-2
8-bit immediate data value
16-bit immediate data value
32-bit immediate data value
address displacement
the Flags and Control Word
low-order byte of FCW
any or all of C. S. P. O. Z
either or both of NVI. VI
an I/O device 16-bit address
an identifier specifying that the prior address is an I/O address
the MI signal input level
the MO signal output level
the most significant (high-order) bit of any data value
immediate value in the range 1-16
Normal Stack Pointer offset
Normal Stack Pointer segment
Program Counter
Program Counter offset
Program Counter segment
Program Status Area Pointer offset
Program Status Area Pointer segment
any byte register
any byte register serving as a destination
any byte register serving as a source
Refresh Counter
any 16-bit index register
any 16-bit register providing implied destination address
any 16-bit register providing implied source address
any 32-bit register serving as a source
any 32-bit register serving as a source
any 64-bit register serving as a destination
any 16-bit register
any 16-bit register serving as a destination
any 16-bit register serving as a source
Stack Pointer (R15 or RR14)
6-44
Object Code
b - immediate value corresponding to b8 or b16
c - condition code (see Table 6-2)
d - destination register
f - code for flags operated on: CZSP/V
i-index or implied register. If i = 0 no register is specified
n - immediate value corresponding to n16
pppp - 16-bit address word or most significant word of 32-bit address
qqqq - least significant word of 32-bit address
r - register
0000 = RO. RRO. ROO. or RHO
0001 = Rt or RH1
0111 = R7. or RH7
1000 = R8. RR8. R08. or RLO
Statuses
1110 = R14. RR14. or RL6
1111 = R15 or RL 7
s - source register
ttttttt - 7-bit unsigned displacement
vv - code for interrupts (VI and/or NV\)
xx - 8-bit address displacement
xxx - 12-bit address displacement
xxxx - 16-bit address displacement
yy - 8-bit immediate data
yyyy - 16-bit immediate data or most significant word of 32-bit data
zzzz - least significant word of 32-bit immediate data
The Z8000 has the following status flags:
C - Carry status
Z - Zero status
S - Sign status
P - Parity status
o - Overflow status
D - Decimal-Adjust
H - Half-Carry
The following symbols are used in the Status columns:
x - flag is affected by operation
(blank) - flag is not affected by operation
1 - flag is set by operation
o - flag is reset by operation
U - flag is unknown after operation
[ [] ]
[]
V
A
E9
contents of the memory location or I/O port whose address is contained in the designated register
contents of memory location. I/O port. or register
data is transferred in the direction of the arrow
data is exchanged between the designated locations on both sides of the arrows
logical OR
~g~aIAND
logical Exclusive OR
6-45
Instruction Mnemonics:
The fixed part of an assembly language instruction is shown in UPPER CASE. the variable part (immediate data. register
name. etc.) is shown in lower case.
Instruction Object Codes:
Instruction words are shown as hexadecimal digits with 4-bit variable fields indicated by lower case letters (e.g .. 67ib).
Instruction words with variable fields that are not multiples of 4 bits are shown as a pair of hexadecimal digits followed
by 8 binary bits (e.g., 7C 000001 vv)
Instruction Execution Times:
Tables 6-3 and 6-4 list instruction execution times in clock cycles. Real time is obtained by dividing the number of
clock cycles by the clock speed.
When several possible execution times are indicated (i.e .. 15-19) the number of clock cycles depends on addressing
and segmentation modes. The relationship is as follows:
Clock Cycles = x-y
Addressing Mode
2)
3)
4)
5)
6)
7)
Clock Cycles
x
y-3
y-1
x+1
y-3
y
addr
addrss
addrls
addr(ri)
addrss(ril
addrls(ril
For two execution times (i.e .. 10, 15) the first is for Nonsegmented mode. the second for Segmented mode.
Instruction times which depend on condition flags are indicated with a slash (i.e., 10. 15/7) with the first time(s) for
condition met and the second for condition not met.
Table 6-2. Condition Codes Used by the
zaooo Assembly Language Instruction Set
Meaning
Status Conditions
Code
CC Value
-
8
Always true
Any
C
ULT
7
Carry
Unsigned less than
C=1
EQ
Z
6
Equal
Zero
Z=1
GE
9
Signed greater than or equal
S$O=O
GT
A
Signed greater than
Z V (S
LE
E9 0) = 0
2
Signed less than or equal
Z V (S E9 0) = 1
LT
1
Signed less than
S{90=O
MI
5
Minus
S=1
NC
UGE
F
No Carry
Unsigned greater than or equal
C=O
NE
NZ
E
Not equal
Not zero
Z=O
NOV
PO
C
No overflow
Parity is odd
PIO = 0
PE
OV
4
Parity is even
Overflow
PIO = 1
PL
0
Plus
S=O
UGT
B
Unsigned greater than
CVZ=O
ULE
3
Unsigned less than or equal
CVZ=1
6-46
Table 6-3. A Summary of the Z8000 Instruction Set
Type
m
~
-...I
g
Mnemonic
Operand(s)
Op Code
>-=
ID
Clock
Cycles
Status
H
D
0
IW
IN"
rwd,@rw
rwd,ioaddr
30sd
3Bd4
pppp
2
4
10
12
INB"
INB"
rwd,@rw
rwd,ioaddr
3Csd
3Ad4
pppp
2
4
10
12
INO"
@rid,@ris,rw
3Bs8
Ord8
4
21
X
INOB"
@rid,@ris,rw
3As8
Ord8
4
21
X
INOR"
@rid,@ris,rw
3Bs8
OrdO
4
21/10""
X
INORB"
@rid,@ris,rw
3As8
OrdO
4
21/10""
X
INI"
@rid,@ris,rw
3BsO
Ord8
4
21
X
INIB"
@rid,@ris,rw
3AsO
Ord8
4
21
X
INIR"
@rid,@ris,rw
3BsO
OrdO
4
21/10""
X
INIRB"
@rid,@ris,rw
3AsO
OrdO
4
21/10""
X
OTOR"
@rid,@ris,rw
3BsA
OrdO
4
21/10""
X
P
Operation Performed
S
Z
C
[rwdl - [[rwllU/O)
[rwdl- [ioaddrlU/O)
Input to 16-bit register rwd a data word from the 1/0 port addressed
directly by ioaddr, or implied by rw.
[rbdl- [[rwllU/O)
[rbdl - [ioaddrlU/O)
Input to 8-bit register rbd a data byte from the 1/0 port addressed
directly by ioaddr, or implied by rw.
[[ridll- [[risllU/O). [ridl- [ridl- 2. [rwl- [rwl- 1
If [rwl = 0 then 0 = 1. Otherwise 0 = O.
[[ridll- [[risllU/O). [ridl - [ridl - 1. [rwl - [rwl - 1
If [rwl = 0 then 0 = 1. Otherwise 0 = O.
Input a 16-bit data word (for INO) or a data byte (for INOB) from the
1/0 port implied by ris to the memory word (for INO) or byte (for INOB)
implied by rid. Decrement the implied memory address in rid by 2 (for
INO) or by 1 (for INOB). Decrement the 16-bit counter rw by 1. If rw
contains 0, set the Overflow status.
[[ridll - [[risl](I/O). [ridl - [ridl- 2. [rwl - [rwl - 1
If [rwl = 0 then 0 = 1. Otherwise reexecute.
[[ridll- [[risllU/O). [ridl- [ridl- 1. [rwl- [rwl- 1
If [rwl = 0 then 0 = 1. Otherwise reexecute. INOR and INORB are
identical to INO and INOB, respectively, except that INOR and INORB
are reexecuted until [rwl = O.
[[ridll- [[risllU/O). [ridl- [ridl + 2. [rwl- [rwl-l
If [rwl = 0 then 0 = 1. Otherwise 0 = O.
[[ridll- [[risllU/O). [ridl- [ridl + 1. [rwl- [rwl- 1
If [rwl = 0 then 0 = 1. Otherwise 0 = O. INI and INIB are identical to
INO and INOB, respectively, except that rid is incremented.
[[ridll- [[risllU/O). [ridl- [ridl + 2. [rwl- [rwl- 1
If [rwl = 0 then 0 = 1. Otherwise reexecute.
[[ridll - [[risllU/O). [ridl - [ridl + 1. [rwl - [rwl - 1
If Irwl = 0 then 0 = 1. Otherwise reexecute. INIR and INIRB are identical to INO and INOB, respectively, except that rid is incremented;
also INIR and INIRB are reexecuted until [rwl = O.
[[ridllU/O) - [[risll. [risl- [risl- 2. [rwl- [rwl- 1
If [rwll=O then reexecute. If [rwl = 0 then 0 = 1 and end execution.
~
" Privileged instruction - can be executed only in system mode.
"" Number of clock cycles depends on the number of repetitions for n/m""; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )I"m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
III
Type
Mnemonic
Operand(s)
Op Code
!>-
III
Clock
Cycles
Status
Operation Performed
H
D
0
OTORB'
@rid,@ris,rw
3AsA
OrdO
4
21110"
X
OTIR'
@rid,@ris,rw
3Bs2
OrdO
4
21/10"
X
OTIRB'
@rid,@ris,rw
3As2
OrdO
4
21/10"
X
OUT'
OUT"
@rw,rws
ioaddr,rws
3Fds
3BsS
2
pppp
4
10
12
OUTB'
OUTS'
@rw,rbs
ioaddr,rbs
3Eds
3AsS
pppp
4
10
12
OUTO'
@rid,@ris,rw
3SsA
Orda
4
21
X
OUTOB'
@rid,@ris,rw
3AsA
Orda
4
21
X
OUTI'
@rid,@ris,rw
3Bs2
Orda
4
21
X
OUTIB'
@rid,@ris,rw
3As2
Orda
4
21
X
~
..
G
~
0)
c
c
~
00
0
g
2
g
.
..
P
S
Z
C
[[ridl)(l/O) - [[ris]]. [risl- [risl- 1. [rwl- [rwl- 1
If [rwlI=O then reexecute. If [rwl = 0 then 0 = 1 and end execution.
Output a block of 1S-bit words (for OTOR) or 8-bit bytes (for OTORB)
from memory to an I/O port. rw specifies the number of words or
bytes. Memory is addressed, using implied memory addressing, by
1S-bit register ris, which is decremented after each output. 1S-bit
register rid specifies the 110 port.
[[ridIW/O) - [[risll. [risl - [risl + 2. [rwl - [rwl - 1
If [rwll= 0 then reexecute. If [rwl = 0 then 0 = 1 and end execution.
[[rid]]U/O) - [[ris]]. [risl- [risl + 1. [rwl- [rwl- 1
If [rwll=O then reexecute. If Irwl = 0 then 0 = 1 and end execution.
OTIR and OTIRB are identical to OTOR and OTORB, respectively, except that OTIR and OTIRB increment the memory address in ris.
[[rwlHl/O) - Irwsl
[ioaddr)(l/O) - [rwsl
Output the data word from 1S-bit register rws to the I/O port addressed directly by ioaddr or implied by rw.
[[rwl)(l/O) - [rbsl
[ioaddr](l/O) - [rbsl
Output the data byte from 8-bit register rbs to the I/O port addressed
directly by ioaddr or implied by rw.
[[ridIW/O) - [[ris]]. [risl- [risl- 2. [rwl- [rwl- 1
If [rwl = 0 then 0 = 1; otherwise 0 = O.
[[ridIW/O) - [[ris]]. [risl- [risl- 1. [rwl- [rwl- 1
If [rwl = 0 then 0 = 1; otherwise 0 = O. Output a data word (for
OUTO) or byte (for OUTOS) from the memory location addressed by
16-bit register ris to the I/O port addressed by 16-bit register rid.
Oecrement ris by 2 (for OUTO) or 1 (for OUTOS). Oecrement the
counter 16-bit register rw.
[[ridl)(l/O) - [[ris]]. [risl- [risl + 2. [rwl- [rwl- 1.
If [rwl = 0 then 0 = 1; otherwise 0 = O.
[[ridIW/O) - [[ris]]. [risl - Iris] + 1. [rwl - [rwl - 1.
If [rwl = Othen 0 = 1; otherwise 0 = O. OUTI and OUTIS are identical to OUTO and OUTDS, respectively, except that the memory address in ris is incremented.
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )'m.
- - - -
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
Type' Mnemonic
t•
OpCode
Operand Is)
III
1
~
c
c
~
0
g
g
en
~
co
pppp
rwd,ioaddr
rbd,ioaddr
@rid,@ris,rw
@rid,@ris,rw
. @rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
ioaddr,rws
ioaddr,rbs
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
3Bd5
3Ad5
3Bs9
3As9
3Bs9
3As9
3Bsl
3Asl
3Bsl
3Asl
3BsB
3AsB
3Bs3
3As3
3Ss7
3As7
3SsS
3AsB
3Bs3
3As3
LO
LO
rwd,adrsx
rwd,@ris
61id
21id
pppp
LO
rwd,risldisp)
31id
LO
rwd,rislrw)
LOS
LOS
rbd,adrsx
rbd,@ris
SIW
SINB·
SIND·
SINOB·
SINOR·
SINORB·
SINI"
SINIB·
SINIR·
SINIRB·
SOTOR·
SOTORS·
SOTIR·
SOTIRB·
SOUrSOUTS·
SOUTO·
SOUTOB·
SOUTI
SOUTIS
4
4
4
4
4
4
4
4
4
4
4
4
4
pppp
Ord8
Ord8
OrdO
OrdO
Ord8
Ord8
OrdO
OrdO
OrdO
OrdO
OrdO
OrdO
pppp
pppp
Ord8
Ord8
Orda
Ord8
4
4
4
4
4
4
4
qqqq
4/6
Clock
Cycles
12
12
21
21
21/10··
21110··
21
21
21/10··
21/10··
21/10··
21110··
21/10··
21/10··
12
12
21
21
21
21
2
9-13
7
xxxx
4
14
71id
OrOO
4
14
60id
20id
pppp
4/6
9-13
7
G)
u
cG)
~
II:
~
0
E
G)
~
~
III
.5
~
.
..
qqqq
2
Status
Operation Performed
H
D
0
X
X
X
X
X
X
X
X
X
X
X
X
P
S
Z
C
These instructions output "special 1/0" status via STO - ST3; otherwise, they are identical to 1/0 instructions as follows:
SIN - IN(l)
SINB - INBll)
SIND -INO
SINOB -INOB
SINOR -INOR
SINORB - INORB
SINI-INI
SINIB -INIB
SINIR -INIR
SINIRB - INIRB
SOTOR - OTOR
OOTORB - OTO'iB
SOTIRB - OTiMo
SOTIR - OTIR
SOUT - OUTll)
SOUTB - OUTBll)
SOUTOB - OUTOB
SOUTO - OUTO
SOUTI - OUTI
SOUTIS - OUTIB
(1 )Only the direct addre~sing option exists for the special 110 version of,
this instruction.
X
X
X
X
[rwdl - [adrsxl
[rwd] - Urisll
Load data from the 1 6-bit memory word addressed by adrsx or Iris]
into 16-bit register rwd.
[rwd] - Uris] + disp]
Load into 16-bit register rwd the contents of the 16-bit memory word
addressed using base relative addressing.
[rwd]- Uris] + [rwll
Load into 16-bit register rwd the contents of the 16-bit memory word
addressed using implied, indexed addressing.
[rbdl - [adrsxl
[rbd] - [[ris]]
Load into 8-bit register rbd the contents of the memory byte addressed by adrsx or Iris].
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m··; n =minimum number of clock cycles and m = number of clock cycles added for each additional repeti tion
of operation. The number of clock cycles for an instruction which repeats k times is n + Ik-l ).m.
- - - - -
----
----
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
Type
-:ae
Mnemonic
Operand Is)
Op Code
t'"
III
Clock
Cycles
LOB
rbd,risldisp)
30id
xxxx
4
14
I..,OB
rbd,ris(rw)
70id
OrOO
4
14
LOL
LOL
rld,adrsx
rld,@ris
54id
14id
pppp
LOL
rld,ris(disp)
35id
xxx x
4
17
LOL
rld,ris(rw)
75id
OrOO
4
17
LO
LO
adrsx,rws
@rid,rws
6Fis
2Fds
pppp
LO
rid(disp),rws
33is
xxxx
4
14
LO
rid(rw),rws
73is
OrOO
4
14
LOB
LOB
adrsx,rbs
@rid,rbs
6Eis
2Eds
pppp
LOB
rid(disp),rbs
32is
xxxx
qqqq 4/6
2
Status
H
D
0
P
Operation Performed
S
Z
[rbd] - [[ris] + disp]
Load into 8-bit register rbd the contents of the memory byte addressed using base relative addressing.
[rbd]- [[ris] + [rwll
Load into 8-bitregister rbd the contents of the memory byte ad.dressed using implied, indexed addressing.
[rid] - [adrsx]
[rld]-Urisll
Load into 32-bit register rid the contents of the 32-bit memory location addressed by adrsx or [ris].
[rid] - [[ris] + disp]
Load into 32-bit register rid the contents of the 32-bit memory location addressed using base relative addressing.
[rid] - Uris] + [ridll
Load into 32-bit register rid the contents of the 32-bit memory location addressed using implied, indexed addressing.
[adrsx) - [rws)
[[ridll - [rws]
Store data from 16-bit register rws into memory word addressed by
adrsx or [rid].
[[rid) + disp) - [rws)
Store data from 16-bit register rws into memory word addressed
using base relative addressing.
[[rid] + [rwll- [rws]
Store data from 16-bit register rws into memory word addressed
using implied, indexed addressing.
[adrsx] +- [rbs]
[[ridll +- [rbs]
Store data from 8-bit register rbs into memory byte addressed by
adrsx or [rid].
[[rid) + disp) +- [rbs)
Store data from 8-bit register rbs into memory byte addressed using
base relative addressing.
12-16
11
:=
!c
0
g
eu
c
Cf
UI
o
i
a:
qqqq 4/6
2
11-15
8
~
E
e
:E
~
as
.5
Ii:
qqqq 4/6
2
11-15
8
4
14
C
" Privileged instruction - can be executed only in system mode.
"" Number of clock cycles depends on the number of repetitions for n/m""; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n
+
(k-1) om.
A Summary of the
zeooo Instruction Set (Continued)
...>
Clock
Cycles
Status
ID
4
14
Table 6-3.
I/)
Type
'::a
Mnemonic
Operand(s)
Op Code
CD
H
LOB
rid(rw),rbs
72is
OrOO
LOL
LOL
adrsx,rls
@rid,rls
50is
1Ods
pppp
LOL
rid(disp),rls
37is
xxxx
4
17
LOL
rid(rw),rls
77is
OrOO
4
17
LOA
rwd,adrsx
76id
pppp
4
12-13
LOA
rld,adrsx
76id
pppp
LOA
LOA
rwd,ris(disp)
rld,ris(disp)
34id
34id
xxx x
xxxx
4
4
15
15
LOA
LOA
rwd,ris(rw)
rld,ris(rw)
74id
74id
OrOO
OrOO
4
4
15
15
LOAR
rwd,disp16
340d
xxx x
4
15
LOAR
rld,disp16
340d
xxxx
4
15
LOR
LORB
rwd,disp16
rbd,disp16
310d
300d
xxx x
xxxx
4
4
14
14
qqqq 4/6
2
D
0
P
Operation Performed
S
Z
C
[[rid] + [rwll- [rbs]
Store data from 8-bit register rbs into memory byte addressed using
implied, indexed addressing.
[adrsx] - [rls]
[[ridll - [rls]
Store data from 32-bit register rls into 32-bit memory long word addressed by adrsx or [rid].
[[rid] + disp] - [rls]
Store data from 32-bit register rls into 32-bit memory long word addressed using base relative addressing.
[[rid] + [rwll- Iris]
Store data from 32-bit register rls into 32-bit memory long word addressed using implied, indexed addressing.
[rwd] - adrsx
Load the unsegmented address into 16-bit register rwd.
[rid] - adrsx
Load the segmented address, in segmented address format, into 32bit register rid.
[rwd] - Iris] + disp
[rid] - Iris] + disp
Load the base relative address into 1 6-bit register rwd (nonsegmented mode) or 32-bit register rid (segmented model.
[rwd]- Iris] + [rw]
[rid] - Iris] + [rw]
Load the implied, indexed memory address into 16-bit regiser rwd
(nonsegmented mode) or 32-bit register rid (segmented mode).
[rwd] - [PC] + disp16
Load the program relative memory address into 16-bit register rwd.
[rld]- [PC] + disp16
Load the program relative segmented memory address, in segmented
format. into 32-bit register rid.
[rwd] - [[PC] + disp16]
[rbd) - [[PC] + disp16]
14-18
11
CD
~
'Sc
0
y
CD
u
c
...~
C11
qqqq 4/6
15-16
f
.;a:
~
0
E
CD
~
~
III
.E
A:
...
I
I
I
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repeti1tion
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-l ).m.
-
- - -
----
-----
--
i
Table 6-3. A Summary of the
III
Type
Mnemonic
Operand(s)
;.
Op Code
CD
Clock
Cycles
LDRL
rld.disp16
350d
xxx x
4
17
LOR
LDRB
LDRL
disp16.rws
disp16.rbs
disp16.rls
330s
320s
370s
xxxx
xxxx
xxxx
4
4
4
14
14
ADD
ADD
rwd.@ris
rwd.adrsx
01id
41id
pppp
qqqq 4/6
ADDB
ADDB
rbd.@ris
rbd.adrsx
OOid
40id
pppp
qqqq 4/6
7
9-13
AOOL
ADDL
rld.@ris
rld.adrsx
16id
56id
pppp
2
qqqq 4/6
14
15-19
AND
AND
rwd.@ris
rwd.adrsx
07id
47id
pppp
qqqq 4/6
ANDB
ANDB
rbd.@ris
rbd.adrsx
06id
46id
pppp
qqqq 4/6
CLR
CLR
@rid
adrsx
ODd8
4Di8
pppp
qqqq 4/6
CLRB
@rid
OCd8
zaooo Instruction Set (Continued)
Status
H
0
0
P
Operation Performed
S
Z
C
[rldl- [[PCl + disp16l
Load the memory word (for LOR). byte (for LOR B) or long word (for
LDRU addressed using program relative addressing into the 16-bit.
8-bit or 32-bit register.
([PC] + disp16] - [rws]
[[PC] + disp16] - [rbsl
([PC] + disp16] - [rls]
Load the register word (for LOR). byte (for LDRB) or long word (for
LDRL) into the memory location addressed using program relative addressing.
"Iii
~!
o c
E '+I
• c
~~
.
~-;
U
E C
III
.~
17
~
A.o!
II:
m
("
N
2
2
7
9-13
X
X
0
0
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
G
U
C
!
....•
•
II:
~
0
E
•
~
~
2
7
9-13
III
"e0
U
•
U)
...
2
2
2
7
9-13
8
11-15
8
X
X
[rwdl - [rwdl + [[risll
[rwdl - [rwdl + [adrsxl
Add the contents of the addressed memory word to the 1 6-bit
destination register.
[rbdl - [rbd] + [[risll
[rbd] - [rbdl + [adrsxl
Add the contents of the addressed memory byte to the 8-bit destination register.
[rid] - [rid] + [(risll
[rid] - [rid] + [adrsx]
Add the contents of the addressed memory long word to the 32-bit
destination register.
[rwd] - [rbd] AND [[risll
[rwdl - [rbd] AND [adrsx]
AND contents of destination 16-bit register with contents of memory
word.
[rbdl - [rbdl AND [[risll
[rbdl - [rbdl AND [adrsx]
AND contents of destination 8-bit register with contents of memory
byte.
[[ridll-O
[adrsxl-O
Clear the memory word.
([(ridll - 0
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-l ).m.
Table G-3. A Summary of the Z8000 Instruction Set (Continued)
Type
Mnemonic
Operand(s)
!•
Op Code
>
III
;;
•
~
-sc
0
9
•cu
CU~B
adrsx
4Ci8
pppp
qqqq 4/6
COM
COM
@rid
adrsx
ODdO
4DiO
pppp
qqqq 4/6
COMB
COMB
@rid
adrsx
OCdO
4CiO
pppp
qqqq 4/6
rwd,@ris
rwd,adrsx
rbd,@ris
rbd,adrsx
rld,@ris
rld,adrsx
OBid
4Bid
OAid
4Aid
10id
50id
@rid,data 1 6
adrsx,data 16
ODdl
4Di1
@rid,data8
adrsx,data8
OCd1
4Cil
CP
CP
CPB
CPB
CPL
CPL
2
2
Clock
Cycles
Status
H
D
0
P
Operation Performed
S
Z
11-15
12
15-19
X
X
12
15-19
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
2
qqqq 4/6
2
qqqq 4/6
2
qqqq 4/6'
9-13
14
15-19
X
X
X
X
X
X
4
6/8
11
14-18
X
X
X
X
X
X
4
6/8
11
14-18
X
X
X
X
X
X
pppp
2
qqqq 4/6
X
X
X
qqqq 4/6
X
X
X
X
X
X
X
pppp
11
13-17
11
13-17
X
X
pppp
pppp
pppp
C
7
9-13
7
X
X
X
X
X
X
e
0)
a,
Co)
i
II:
~
E
•
~
~
III
"u
c
CP
CP
CPB
CPB
0
•
yyyy
yyyy
pppp
yyOO
yyOO
pppp
qqqq
qqqq
(I)
DEC
DEC
DECB
DECB
.
.-
@rid,n16
adrsx,n16
@rid,n16
adrsx,n16
2Bdn
6Bin
2Adn
·6Ain
X
X
[adrsx) - 0
Clear the memory byte.
[[ridlJ - [[ridlJ
[adrsx) - [adrsx)
Ones complement the memory word.
([ridlJ - [[ridlJ
[adrsx) - [adrsx)
Ones complement the memory byte.
[rwd) - [[rislJ
[rwd) - [adrsx)
[rbd) - [[ris)J
[rbd) - [adrsx)
[rid) - [[rislJ
[rid) - [adrsx)
Compare contents of register and memory location. Do not modify
contents of register or memory location, but set status flags. Use 16bit register/memory word for CP, 8-bit register/memory byte for CPB,
32-bit register/memory long word for CPL.
[[ridlJ - data 16
[adrsx) - data 1 6
2
X.
X
[[ridlJ - data8
[adrsx) - data8
Compare contents of memory location with immediate data. Do not
modify memory location, but set status flags. Use 16-bit memory
word for CP, 8-bit memory byte for CPB.
[[ridlJ - [[ridlJ - n 1 6
[adrsx) - [adrsx) - n 16
[[ridlJ - [[ridlJ - n 16
[adrsx) - [adrsx) - n 16
Subtract the immediate value n 1 6 from the memory word (for DEC)
or memory byte (for DECB) addressed by adrsx or [rid). Values in the
range 1-1 6 are subtracted.
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m--; n =minimum number of clock cycles and m = number of clock cycles added for each additional repeti.tion
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )-m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
(I)
Type
Mnemonic
Operand(s)
Op Code
S
>-
III
DIV
DIV
DIVL
DIVL
EX
EX
EXB
EXB
rld.@ris
rld.adrsx
rqd.@ris
rqd.adrsx
rwd.adrsx
rwd.@ris
rbd.adrsx
rbd.@ris
lBid
5Bid
lAid
5Aid
6Did
2Dsd
6Cid
2Csd
INC
INC
INCB
INCB
@rid.n16
adrsx.n16
@rid.n16
adrsx.n16
29dn
69in
28dn
68in
MULT
MULT
MULTL
MULTL
NEG
NEG
NEGB
NEGB
rld.@ris
rld.adrsx
rqd.@ris
rqd.adrsx
@rid
adrsx
@rid
adrsx
19id
59id
18id
58id
ODd2
4Di2
OCd2
4Ci2
OR
OR
rwd.@ris
rwd.adrsx
05id
45id
ORB
rbd.@ris
04id
Clock
Cycles
Status
Operation Performed
H
D
0
P
S
Z
C
2
qqqq 4/6
2
qqqq 4/6
qqqq 4/6
2
qqqq 4/6
2
note 1
note 1
note 1
note 1
15-19
12
15-19
12
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
2
qqqq 4/6
2
qqqq 4/6
11
13-17
11
13-17
X
X
X
X
X
X
X
X
X
X
X
X
note 2
note 2
note 2
note 2
12
15-19
12
15-19
0
0
0
0
'X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
pppp
2
qqqq 4/6
2
qqqq 4/6
2
qqqq 4/6
2'
qqqq 4/6
X
X
X
pppp
,2
qqqq 4/6
7
9-13
pppp
pppp
pppp
pppp
:a
"c
~
'i0
g
"c
Co)
C»
I
U'I
~
pppp
pppp
e
"
';
a::
~
E
"
~
~
III
'a
c
0
Co)
"
U)
pppp
pppp
pppp
2
X
X
X
7
X
X
X
X
X
X
X
O;y'do
}
Divide
see text for a discussion of these instructions
Divide long
Divide long
[rwd) - - [adrsx)
[rwd) - - ([ris))
[rbd) - - [adrsx)
[rbd) - - ([ris))
Exchange contents of the addressed memory location with the
selected register. Use 8-bit (for EXB) or 16-bit (for EX) registers and
memory locations.
[[rid)) - [[rid)) + n 1 6
[adrsx) - [adrsx) + n 1 6
[[rid)) - ([rid)) + n 1 6
[adrsx) - [adrsx) + n 1 6
Add the immediate value n 16 to the memory word (for INC) or memory byte (for INCB) addressed by adrsx or [rid). Values in the range 1 16 are added.
Multlp'y
}
Multi I
. . . .
py
see text for a diSCUSSion of these instructions
Multiply long
Multiply long
[[rid)) - - ([rid))
!adrsxl - - !adrsxl
([rid)) - - ([rid))
[adrsx) - - [adrsx)
Replace the contents of the memory word (for NEG) or byte (for
NEGB) addressed by adrsx or [rid) with its twos complement.
!rwd) - [rwd) OR ([ris))
[rwd) - [rwd) OR [adrsx)
OR the contents of the specified 16-bit register and memory word.
Place the result in the 16-bit register.
[rbd) - [rbd) OR ([ris))
Privileged instruction - can be executed only in system mode.
- Number
of clock cycles depends on the number of repetitions for n/m--; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
--
of operation. The number of clock cycles for an instruction which repeats k times is n
+ (k-l )-m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
Type
Mnemonic
Operand Is)
Op Code
..
Clock
>-
Cycles
S
III
ORB
rbd,adrsx
44id
SUB
SUB
SUBB
SUBB
SUBL
SUBL
rwd,@ris
rwd,adrsx
rbd,@ris
rbd,adrsx
rld,@ris
rld,adrsx
03id
43id
02id
42id
12id
52id
TEST
TEST
TESTB
TESTB
TESTL
TESTL
@rid
adrsx
@rid
adrsx
@rid
adrsx
ODd4
4Di4
OCd4
4Ci4
lCdO
5CiO
TSET
TSET
TSETB
TSETB
@rid
adrsx
@rid
adrsx
ODd6
4Di6
OCd6
4Ci6
XOR
XOR
XORB
XORB
rwd,@ris
rwd,adrsx
rbd,@ris
rbd,adrsx
09id
49id
Oaid
48id
pppp
qqqq 4/6
pppp
qqqq 4/6
pppp
qqqq 4/6
pppp
qqqq 4/6
pppp
qqqq 4/6
pppp
qqqq 4/6
pppp
2
qqqq 4/6
pppp
qqqq 4/6
pppp
qqqq 4/6
pppp
qqqq 4/6
pppp
qqqq 4/6
2
2
2
Status
H
D
0
9-13
7
7
9-13
14
15-19
S
Z
X
X
X
X
X
9-13
X
X
1
1
Operation Performed
P
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
C
-:0
•c
~
';0
c
0
9
•c
.;a::
0
!
Cf>
C11
C11
2
2
a
11-15
a
11-15
13
16-20
X
X
~
0
E
:E•
~
III
'a
c
0
0
•
en
::tI
2
2
2
2
11
14-18
11
14-18
X
X
X
X
7
X
X
X
X
9-13
7
9-13
X
X
X
X
X
X
X
X
[rbd] - [rbd] OR [adrsx]
OR the contents of the specified 8-bit register and memory byte.
Place the result in the 8-bit register.
[rwd) - [rwd) - [[ris)]
[rwd) - [rwd] - [adrsx)
[rbd) - [rbd) - [[ris)]
[rbdt - [rbd) - [adrsx)
[rid) - [rid) - [[ris))
[rid) - [rid] - [adrsx)
Subtract the contents of the addressed memory location from the
selected destination register. Use 8-bit (for SUBB), 16-bit (for SUB)
or 32-bit (for SUBL) memory locations and registers.
[[rid)) OR 0
[adrsx] OR 0
[[ridll OR 0
[adrsx) OR 0
[[ridll OR 0
[adrsx) OR 0
OR the specified memory contents with O. Set status flags. Use a 16bit location for TEST, an 8-bit location for TESTB, and a 32-bit location for TESTL.
[s] - [[rid)](MSB). [[rid)] - FFFF
[s) - [adrsx](MSB). [adrsx) - FFFF
[s)- [[rid)](MSB). [[rid)] - FF
[s) - [adrsx](MSB). [adrsx) - FF
Move the most significant bit of the memory word (for TSET) or byte
(for TSETB) to the sign status. Then fill the word or byte with 1 bits.
[rwd) - [rwd) XOR [[ris))
[rwd) - [rwd) XOR [adrsx)
[rbd) - [rbd) XOR [[risll
[rbd) - [rbd) XOR [adrsx)
Exclusive OR the contents of the addressed memory location and
register. Store the result in the register. Use 16-bit memory and
registers for XOR. Use a-bit memory and registers for XORB.
CD
~
}>
c<
'02Q.c:
co!!t3
~(1)t1)
COO)/I.)
.
..
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m··; n =minimum number of clock cycles and m = number of clock cycles added for each additional repet ition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-l )·m.
Table 6-3. A Summary of the
I/)
Type
!III
:aa!
E
.5
Mnemonic
Operand(s)
III
...>-
Clock
Cycles
a!
Op Code
LO
rwd,data16
210d
yyyy
4
7
LO
LO
@rid,data 1 6
adrsx,data 1 6
OOd5
40i5
yyyy
yyyy
pppp
4
6/8
11
14-18
2
5
4
6/8
11
14-18
6
11
2
5
LOB
rbd,data8
Cdyy
LOB
LOB
@rid,data8
adrsx,data8
OCd5
4Ci5
zaooo Instruction Set (Continued)
Status
H
0
0
P
Operation Performed
S
Z
C
[rwd] - data 16
Load 16-bit immediate data into 16-bit register rwd.
[(rid]] - data 1 6
[adrsx] - data 16
Load 16-bit immediate data into memory word addressed by adrsx or
[rid].
[rbd] - data8
Load immediate data byte into 8-bit register rbd.
[(rid]] - data8
[adrsx] - data8
Load immediate data byte into memory byte addressed by adrsx or
[rid].
[rid] - data32
Load 32-bit immediate data into 32-bit register rid.
[rwd]-b16
Load the immediate 4-bit value b16 into the low-order four bits of
rwd. Clear the remaining twelve bits of rwd.
(See Stack operations).
qqqq
yyOO
yyOO
pppp
qqqq
yyyy
zzzz
LOL
rld,data32
140d
LOK
rwd,b16
BOdb
JP
JP
,@rid
,adrsx
1Ed8
5Ei8
JR
,disp
E8xx
2
6
LOPS-
@ris
39s0
2
12
16
C»
a.C»
PUSH
pppp
2
qqqq .4/6
10,15
7-11
a.
.,E
:::J
LOPS-
adrsx
79iO
pppp
qqqq 4/6
16-17
20-23
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
[PC]- [(rid]]
[PC] - [adrsx]
Jump to the specified memory location. This is the same as a conditional jump with cc = always true.
[PC] - [PC] + (disp-2)
Jump program relative. PC is incremented to the next sequential instruction before disp-2 is added as a signed binary number. This is
the same as a conditional jump relative with cc = always true (blank).
[FCW] - [(ris]]. [PC] - [(ris] + 1] (nonsegmented)
[FCWRES1- [[ris]]. [FCW1- [[ris1 + 11. [PCSEG1- [[ris] + 21
[PCOFF] - [[ris] +3 ] (segmented)
[FCW] - [adrsx]. [PC] - [adrsx+ 1] (nonsegmented)
[FCWRES1 - [adrsx1. [FCW1 - [adrsx+ 11. [PCSEG1 - ladrsx+21
[PCOFF]- [adrsx+3] (segmented)
Load program status and jump as described in accompanying text.
I
I
- P,;v;le..d ;n.truc';on - c.n be e,ecuted only ;n .ys.em mode.
-- Number of clock cycles depends on the number of repetitions for n/m--; n =minimum number of clock cycles and m = number of clock cycles added for each additional repeti1tion
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )-m.
Table 6-3. A Summary of the
Type. Mnemonic
Operand(.)
S">-
Op Code
ID
c
a•
II:
CALL
CALL
CALR
@rid
adrsx
disp
1FdO
5FiO
Dxxx
cc
9EOc
pppp
2.
qqqq 4/6
2
Clock
Cycle.
zaooo Instruction Set (Continued)
Statu.
H
D
0
P
Operation Performed
S
Z
C
(1) or (2). [PC] - [[ridll
(1) or (2). [PC] - [adrsx]
(1) or (2). [PC] - [PC] - disp-2 Program relative memory address.
Call the addressed subroutine. saving information on the Stack as
follows:
(1) [SP] - [SP] - 2. [(SPll - [PC] 16-bit PC (nonsegmented)
(2) [SP] - [SP] - 4. [[SPll - [PC] 32-bit PC (segmented)
If cc is "true" then [PC] - [{SPll.[SP] - [SP] + 2 (for nonsegmented)
or [SP] - [~P] + 4. (for segmented).
If the condition code specified by cc is "true". return from subroutine .
System subroutine call. See accompanying text for a description of this
instruction.
10.15
12-21
10.15
.,c
"a
'i
CJ
"c
',j:l
RET
2
10.1317
~
.a
~
U)
data8
7Fyy
ADD
ADDB
ADDL
rwd.data16
rbd.data8
rld.data32
010d
OOOd
160d
yyyy
yyOO
yyyy
AND
ANDB
rwd.data16
rbd.data8
070d
060d
CP
rwd.data16
CPB
CPL
SC
2
33.39
4
4
6
7
7
14
yyyy
yyOO
4
4
7
7
OBOd
yyyy
4
7
rbd.dataS
OAOd
yyOO
4
rld.data32
100d
yyyy
zzzz
rld.data16
rqd.data32
1BOd
1AOd
yyyy
yyyy
zzzz
zzzz
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
7
X
X
X
X
6
14
X
X
X
X
4
6
note 1
note 1
X
X
X
X
X
X
X
·X
0
[rwd] - [rwd] + data 16
[rbd] - [rbd] + data8
[rid] - [rid] + data32
Add immediate data to the destination register. Use 32-bit
data/register for ADDL. 16-bit data/register for ADD. 8-bit
data/register for ADDB.
[rwd] - [rwd] AND data 16
[rbd] - [rbd] AND data8
AND immediate data with destination register contents. Use 16-bit
data/register for AND. 8-bit data/register for ANDB.
[rwd] - data 16
Compare 16-bit register contents with immediate 16-bit data. Do not
modify register contents. but save Status flags.
[rbd] - dataS
Compare 8-bit register contents with immediate 8-bit data. Do not
modify register contents. but save Status flags.
[rid] - data32
Compare 32-bit register contents with immediate 32-bit data. Do not
modify register contents. but save Status flags.
(See secondary memory reference for memory-immediate compare instructions.)
Divide
} see accompanying text for a discussion of these
Divide long instructions
en
I
(J'I
.....
!
X
!
&
0
Sall
is
E
"
.5
CP
DIV
DIVL
X
- Privileged instruction - can be executed only in system mode.
-- Number of clock cycles depends on the number of repetitions for n/m--; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n
- - - - -
- - -
-----------
+
(k-1)-m.
-
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
10
Type
Mnemonic
Operand(s)
S
Op Code
>
ID
rld,data16
rqd,data32
rwd,data16
190d
180d
050d
yyyy
yyyy
yyyy
rbd,data8
040d
SUB
SUBB
SUBL
rwd,data16
. rbd,data8
rld,data32
XOR
XORB
rwd,data16
rbd,data8
DJNZ
rw,disp
DBJNZ
rb,disp
MULT
MULTL
OR
Clock
Cycles
4
6
4
note 2
note 2
yyOO
4
7
030d
020d
120d
yyyy
yyOO
yyyy
4
4
6
7
7
14
090d
080d
yyyy
yyOO
4
4
7
7
Fr
1ttttttt
2
11
Fr
Ottttttt
2
11
zzzz
Status
H
0
0
P
0
0
7
Operation Performed
S
Z
C
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
~
•c:::s
.~
ORB
X
c
0
g
S
t!CD
a.
0
zzzz
X
1
X
X
X
S
«I
=sCD
E
.5
X
en
U,
(X)
X
X
X
'Multiply
} see accompanying text for a discussion of these
Multiply long instructions
[rwd) +- [rwd) OR data 16
OR the contents of the specified 16-bit register with the immediate
data word .
[rbd) +- [rbd) OR data8
OR the contents of the specified B-bit register with the immediate
data byte.
[rwd) +- [rwd) - data 16
[rbd) +- [rbd) - data8
[rid) + - [rid) - data32
Subtract immediate data from the destination register. Use 32-bit
data/register for SUBL, 16-bit data/register for SUB, 8-bit
data/register for SUBB.
[rwd) +- [rwd] XOR data 1 6
[rbd) +- [rbd] XOR data8
Exclusive-OR the contents of the register with immediate data. Store
the result in the register. Use 16-bit register and data for XOR. Use 8bit register and data for XORB.
[rw] +- [rw) - 1. [PC) +- [PC] + 2. If [rw] is not 0, then
[PC) + - [PC] - [disp-2)
[rb] +- [rb] - 1. [PC] +- [PC) + 2. If [rb] is not 0, then
[PC) +- [PC] - [disp-2]
Decrement a 16-bit register (for DJNZ) or an 8-bit register (for DBJNZ). Increment the Program Counter as per normal operation. If the
decremented register contents is not 0, then subtract twice the displacement, as an unsigned binary number, from the incremented Program Counter, causing a branch back to a lower program memory address. If the decremented register contents is 0, continue execution
with the next instruction.
c
.~
;;
c
0
(J
c
0
a.
E
:::s
~
~
u
c
t!
ID
,
- Privileged instruction - can be executed only in system mode.
-- Numba, of clock cvcles depend. on the numba, of ,e.e';';on. fo, n/m"; n =mlnlmum num"'" of clock CVcle. and m = numba, of clock cvcle. added fo, each addltl""al ",pell lion .
of operation. The number of clock cycles for an instruction which repeats k times is n
+
(k-l) om.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
Type
Mnemonic
Operand(s)
Op Code
!"
>
CD
;;
6a,"::~
E c:
~ 0
::!y
.c c:
u 0
Clock
Cycles
JP
JP
cC,@rid
cC,adrsx
1Edc
5Eic
JR
cC,disp
Ecxx
2
6
EX
EXB
rwd,rws
rbd,rbs
AOsd
ACsd
2
2
6
6
LO
LOB
LOL
rwd,rws
rbd,rbs
rld,rls
A1sd
AOsd
94sd
2
2
2
3
3
5
pppp
2
qqqq 4/6
Status
Operation Performed
H
D
0
P
S
Z
C
If cc is true, [PC) - [[rid]]
If cc is true, [PC) - [adrsx)
Jump to the memory location specified by adrsx or [rid] if condition
code cc is true.
If cc is true, [PC) - [PC) + (disp-2)
Jump program relative if condition code is true. PC is incremented to
address the next sequential instruction before disp-2 is added as a
signed binary number.
10,15/7
7-11
Ii :E
'a
~
c:
CD
0
U
CD
>
0
I
i::E
"
~
,- CD
CII ..
CD "
II:
"a.
CD
[rwd] - - [rws)
[rbd] - - [rbs]
Exchange registers' contents for 16-bit (EX) or 8-bit (EXB) registers.
[rwd) - [rws]
[rbd) - [rbs]
[rid) - [rls)
Move data between any two 16-bit registers (for LO), 8-bit registers
(for LOB) or 32-bit registers (for LOll.
II:
Cf>
(,J'I
CD
CPO
rwd,@ris,rw,cc BBs8
Ordc
4
20
X
U
X
U
CPOB
rbd,@ris,rw,cc
BAs8
Ordc
4
20
X
U
X
U
CPDR
CPDRB
rwd,@ris,rw,cc BBsC
rbd,@ris,rw,cc BAsC
Ordc
Ordc
4
4
20/9-20/9--
X
X
U
U
X
X
U
U
.c
~
II
CD
en
'a
c:
II
~
-!c:
!
~
~
S
ii
[rwd] - [[ris]]. If cc true, Z = 1. If cc false, Z = 0
[ris]- [ris]- 2. [rw]- [rw)- 1. If [rwl = 0,0 = 1 otherwise 0 = O.
[rbd] - [[ris]]. If cc true, Z = 1. If cc false, Z = O.
[ris]- Iris) - 1. [rw]- [rw]- 1. If [rwl = 0,0 = 1 otherwiseO = O.
Search a string for a condition. Compare a word in rwd (for CPO) or a
byte in rbd (for CPOB) with the next word (for CPD) or byte (for CPDB)
in a memory string, using implied memory addressing. Register and
memory contents are not modified, nor are Status flags changed, but
status conditions are compared with cc. If cc is true, Z is set to 1;
otherwise Z is reset to O. Decrement the implied memory address in
ris by 2 for CPD, or by 1 for CPDB. Decrement 16-bit counter rw by 1.
If rw is 0, set 0 to 1; otherwise reset 0 to O.
See CPD.
See CPOB.
CPO and CPOB are identical to CPO and CPOB, respectively, except
that instruction execution is repeated until either Z or 0 status is 1. Interrupts will be acknowledged between reexecutions.
- Privileged instruction - can be executed only in system mode.
of clock cycles depends on the number of repetitions for n/m--; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
-- Number
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )-m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
In
Type
Mnemonic
Operand(s)
Op Code
t
Clock
Cycles
CD
-:sII
:s
!c
0
~
Status
Operation Performed
H
D
0
P
S
Z
C
CPI
rwd,@ris,rw,cc BBsO
Ordc
4
20
X
U
X
U
CPIB
rbd,@ris,rw,cc
BAsO
Ordc
4
20
X
U
X
U
CPIR
CPIRB
rwd,@ris,rw,cc BBs4
rbd,@ris,rw,cc BAs4
Ordc
Ordc
4
4
20/9··
20/9··
X
X
U
U
X
X
U
U
CPSO
@rid,@ris,rw,cc BBsA
Ordc
4
25
X
U
X
U
CPSOB
@rid,@ris,rw,cc BAsA
Ordc
4
25
X
U
X
U
CPSDR
CPSDRB
@rid,@ris,rw,cc BBsE
@rid,@ris,rw,cc BAsE
Ordc
Ordc
4
4
25/14··
25/14··
X
X
U
U
X
'X
U
U
~
.,e
~
CI)
o
oa
c
en
II
.,
.!
In
c
!
~
8
iii
...
[rwdl - [[ris]]. If cc true, Z = 1. If cc false, Z = O.
[risl- [risl + 2. [rwl- [rwl- 1. If [rwl = 0, 0 = 1; otherwise 0 = O.
[rbdl - [[ris]]. If cc true, Z = 1. If cc false, Z = O.
[risl - [risl + 1. [rwl - [rwl - 1. If [rwl = 0, 0 = 1; otherwise 0 = O.
CPI and CPIB are identical to CPO and CPOB, respectively, except that
the implied memory address in ris is incremented by 2 for CPI, or by 1
for CPIB.
See CPI.
See CPIB.
CPIR and CPIRB are identical to CPO and CPDB, respectively, except
that the implied memory address in ris is incremented by 2 for CPI, or
by 1 for CPIB, and instruction execution is repeated until either Z or 0
status is 1. Interrupts will be acknowleged between reexecutions.
[[rid]] - [[ris]]. If cc true, Z = 1. If cc false, Z = O.
[ridl- [ridl- 2. [risl- [risl- 2. [rwl- [rwl- 1.
If [rwl = 0,0 = 1; otherwise 0 = o.
[[rid]] - [[risll. If cc true, Z = 1. If cc false, Z = O.
[ridl - [ridl - 1. [risl - [risl - 1. [rwl - [rwl - 1.
If [rwl = 0, 0 = 1; otherwise 0 = O.
Compare two strings for a condition. Compare the next word (for
CPSO) or byte (for CPSOB) in a source string with the next word (for
CPSO) or byte (for CPSOB) in a destination string. Both strings are addressed using implied memory addressing. No memory contents are
modified, nor are any Status flags changed, but status conditions are
compared with cc. If cc is true, Z is set to 1. Otherwise Z is reset to O.
Decrement the implied memory addresses in ris and rid by 2 for
CPSD, or by 1 for CPSOB. Decrement 16-bit counter rw by 1. If rw is
0, set 0 to 1. Otherwise reset 0 to O.
See CPSD.
See CPSDB.
CPSDR and CPSDRB are identical to CPSD and CPSDB, respectively,
except that the instructions are reexecuted until either Z or 0 status is
1. Interrupts are acknowleged between reexecutions.
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m··; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 ).m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
Type I Mnemonic
Op Code
Operand(s)
t
Clock
Cycles
ID
i
Status
Operation Performed
H
DIOIPISlzlC
CPSI
@rid,@ris,rw,cc;1 BBs2
Ordc
4
25
X
CPSIB
l@rid,@ris,rw,cc:1 BAs2
Ordc
4 I
25
X
CPSIR
CPSIRB
I@rid,@ris,rw,ccl BBs6
@rid,@ris,rw,cc BAs6
Ordc
Ordc
4
4
25/14**
25/14**
X
X
LDD
I
BBsS
Ord8
4 I
20
X
LDDB
I @rid,@ris,rw 'BAs9
Ord8
4 I
20
X
LDDR
@rid,@ris,rw I BBs9
OrdO
4
20/9**
LDDRB
@rid,@ris,rw
BAsS
OrdO
4
20/S**
LDI
@rid,@ris,rw I BBs 1
Ord8
4
20
I
~
~
g
~
.I:
f
0)
m
.....
ca
GI
en
~
ca
@rid,@ris,rw
I
~
.;c
~
g
.JJt
ii
*
I
X
U I X I U I [[ridll- [[risll. If cc is true Z = 1. If cc is false, Z = o.
[rid] +- [rid] + 2. [ris] +- [ris] + 2. [rw] +- [rw] - 1.
If [rw] = 0,0 = 1; otherwise 0 = O.
U I X I U I [[ridll - Uris]]. If cc is true, Z = 1 . If cc is false, Z = O.
[rid] - [rid] + 1. Iris] - Iris] + 1. [rw] - [rw] - 1.
If trw] = 0,0 = 1; otherwise 0 = O.
CPSI and CPSIB are identical to CPSD and CPS DB, respectively, except that the implied memory addresses in rid and ris are incremented
by 2 (for CPS!) or by 1 (for CPSIB).
U
U
III
X
. X·
U See CPSI.
USee CPSIB.
CPSIR and CPSIRB are identical to CPSD and CPSDB, respectively,
except that the implied memory addresses in rid and ris are incremented by 2 (for CPSIR) or by 1 (for CPSIRB) and the instructions are
reexecuted until either Z or 0 status is 1. Interrupts are acknowleged
between reexecutions.
[(ridll - [(ris]]. [rid] - [rid] - 2. Iris] - [ris] - 2.
[rw]- [rw]- 1. If [rw] = 0,0 = 1; otherwise 0 = O.
[(ridll - [[risll. [rid] - [rid] - 1. [ris] - [ris] - 1.
[rw]- [rw]- 1. If [rw] = 0,0 = 1; otherwise 0 = O.
Transfer a word (for LDD) or a byte (for LDDB) from the memory location addressed by register ris to the memory location addressed by
rid. Decrement addresses in rid and ris by 2 (for LDD) or 1 (for LDDB).
Decrement the counter rw by 1. If rw contains 0, set the Overflow
status to 1.
[[ridll - [[risll. [rid] - [rid] - 2. Iris] - Iris] - 2.
[rw]- [rw]- 1. If [rw] :1=0, reexecute.
If [rwl = 0, 0 = 1 and end execution.
[(ridll - [[risll. [rid] - [rid] - 1. Iris] - [ris] - 1.
[rwl- [rwl - 1. If [rw] :1=0, reexecute.
If [rw] = 0,0 = 1 and end execution.
LDDR and LDDRB are identical to LDD and LDDB, respectively, except
that LDDR and LDDRB reexecute until rw has decremented to O.
[[ridll - [(risll. [rid] - [rid] + 2. [ris] - [ris] + 2.
[rw]- [rw]- 1. If [rw] = 0 then 0 = 1; otherwise 0 = O.
Privileged instruction - can be executed only in system mode.
Number of clock cycles depends on the number of repetitions for n/m**; n =minimum number of clock cycles and m
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )*m.
= number of clock cycles added for each additional repetition
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
10
Type
Mnemonic
Operand(s)
>
CD
Clock
Cycles
!
Op Code
Status
Operation Performed
H
D
0
LOIB
@rid,@ris,rw
BAs1
Ord8
4
20
X
LOIR
@rid,@ris,rw
BBs1
OrdO
4
20/9"
1
LOIRB
@rid,@ris,rw
BAs1
OrdO
4
20/9"
1
LDM
rwd,@ris.n16
1 Cs1
OdOn
4
14/3"
LDM
rwd,adrsx,n 16
5Ci1
OdOn
pppp
P
S
Z
;;
•
~
·ac
0
Y
.c
2
en
m
i')
•
IV
en
~
..
c
6/8 17-21/3"
qqqq
II
$.
10
c
t!
t~
u
LOM
@ris,rws,n16
1Cd9
OsOn
LDM
adrsx,rws,n 1 6
5Ci9
OsOn
pppp
4
14/3"
0
iii
TROB
@rid,@ris,rw
B8d8
OrsO
6/8 17-2113"
qqqq
4
25
X
U
C
[[ridll - [[ris]). [rid] - [rid] + 1. [ris] - [ris] + 1.
[rw] - [rwl - 1. If [rwl = 0, then 0 = 1; otherwise 0 = O.
LOI and LOIB are identical to LOO and LOOB, respectively, except that
the source and destination addresses ris and rid are incremented by 2
(for LOn or 1 (for LOIB).
[[ridll- [[risll. [ridl- [ridl + 2. [risl- [risl + 2.
[rwl - [rwl - 1. If [rwl #= 0 then reexecute.
If [rwl = 0 then 0 = 1 and end execution.
[[ridll- [[risll. [ridl- [ridl + 1. [risl- [risl + 1.
[rwl- [rwl - 1. If [rwl #=0 then reexecute.
If [rwl = 0 then 0 = 1 and end execution.
LDiR and LOIRB are identical to LOO and LOOB, respectively, except
that the source and destination addresses ris and rid are incremented;
also, LOIR and LOIRB are reexecuted until rw decrements to O.
[rwdl - [[risll do n 16 times incrementing register and memory addresses
[rwd] - [adrsx] do n 1 6 times incrementing register and memory addresses
Move a block of n 1 6 memory words from memory to 1 6-bit registers.
adrsx or @ris addresses the first, lowest addressed memory word.
rwd addresses the first 16-bit register. n16 can have any value from
1 to 16. (See accompanying text for more details.)
([risll - [rwsl do n16 times incrementing register and memory addresses
[adrsxl - [rwsl do n16 times incrementing register and memory addresses
This instruction is identical to the one above, except that data moves
from registers to memory. Register contents are not affected.
([ridll- [[ris] + IIridlll. [rw]- [rwl- 1. [ridl- [ridl- 1.
Translate a memory byte, as described in the accompanying text.
Decrement the destination address in rid and the'byte counter in rw. If
rw = 0, set 0 to 1. If rw #= 0, reset 0 to O. Byte register RH 1 contents
is lost.
I
...
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 ).m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
I)
Type
-:aII
:::s
Mnemonic
Operand(s)
Op Code
CD
>
Clock
Cycles
S
Status
Operation Performed
H
D
0
P
S
Z
TRDRB
@rid,@ris,rw
B8de
OrsO
4
25/14··
1
U
TRIB
@rid,@ris,rw
B8dO
OrsO
4
25
X
U
TRIRB
@rid,@ris,rw
B8d4
OrsO
4
25/14··
1
U
TRTDB
@rid,@ris,rw
B8dA
OrsO
4
25
X
X
TRTDRB
@rid,@ris,rw
B8dE
OrsE
4
25/14··
X
X
TRTIB
@rid,@ris,rw
B8d2
OrsO
4
25
X
X
TRTIRB
@rid,@ris,rw
B8d6
OrsE
4
25/14··
X
'X
~e
0
Y
~
e
II
II
0)
0,
en
Co)
'a
e
II
...
.!
I)
e
!
t.Jt.
S
iii
C
([rid]) - ([ris] + ([rid]]]. [rw] - [rw] - 1. [rid] - [rid] - 1.
If [rw] = 0,0 = 1 and end execution. If [rw] #:0, reexecute.
This instruction is identical to TRDB, except the instruction is reexecuted until [rw] = O.
([rid)) - [[ris] + [[rid]]]. [rw] - [rw] - 1. [rid] - [rid] + 1
TRIB is identical to TRDB except that the destination address in rid is
incremented.
[[rid)) - [[ris] + [[rid]]]. [rw] - [rw] - 1. [rid] - [rid] + 1.
If [rw] = 0, 0 = 1 and end execution. If [rw] #: 0, reexecute.
TRIRB is identical to TRDB except that the destination address in rid
is incremented; also, TRIRB is reexecuted until [rw] = O.
[RH1)- [[ris] + [[rid]]]. [rw] - [rw]- 1. [rid] - [rid)- 1.
Load a table byte into 8-bit register RH 1, as described in the accompanying text. Reset Z status to 0 if [RH 1] #: O. Set Z status to 1 if
[RH 1] = O. Decrement destination address rid and byte counter rw. If
rw = 0, 0 = 1. If rw #: 0, 0 = o.
[RH 1] - [[ris) + [[ridnJ. [rw] - [rw] - 1. [rid] - [rid] - 1
If [rw] = 0,0 = 1.lf [rw] #:0, 0 = O. If [RH1] = 0, Z = 1; otherwise
Z = O. TRTDRB is identical to TRTDB except that TRTDRB is reexecuted until 0 = 1 or Z = O.
[RH 1] - [[ris] + [[rid]]]. [rw] - [rw] - 1. [rid] - [rid] + 1
TRTIB is identical to TRTDB except that TRTIB increments the
destination address in rid .
[RH1] - [[ris] + [[rid]]]. [rw] - [rw) - 1. [rid] - [rid] + 1
If [rw] = 0,0 = 1. If [rw] #:0, 0 = O. If [RH1] = 0, Z = 1; otherwise
Z = O. TRTlRB is identical to TRTDB except that TRTIRB ~ncrements
the destination address in rid and reexecutes until 0 = 1 or Z = O.
:tJ
~
1>
c<
'Ost
Q.c:
00
111
3
!,.iCD
ID
Clock
Cycles
!
Status
Operation Performed
H
D
0
5
X
0
X
0
S
Z
C
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
P
ADC
ADCB
rwd,rws
rbd,rbs
B5sd
B4sd
2
2
ADD
ADDB
ADDL
rwd,rws
rbd,rbs
rld,rls
81sd
80sd
96sd
2
2
2
4
4
AND
ANDB
rwd,rws
rbd,rbs
87sd
86sd
2
2
4
4
CP
rwd,rws
8Bsd
2
4
X
X
X
X
CPB
rbd,rbs
8Asd
2
4
X
X
X
X
CPL
rld,rls
90sd
2
8
X
X
X
X
DIV
OIVL
MULT
MULTL
rld,rws
rqd,rls
rld,rws
rqd,rls
9Bsd
9Asd
99sd
98sd
2
2
2
2
note
note
note
note
X
X
X
X
X
X
X
X
X
X
X
X
X
X
5
8
X
!
I!
II
a.
0
CJ)
0,
~
i
(I)
"m
II
IE:
i
(I)
"m
II
IE:
- Privileged instruction -
--
1
1
2
2
0
0
[rwd) - [rwd) + [rws) + C
[rbd] - [rbd] + [rbs] + C
Add the source register contents plus the initial Carry to the destination register. Use 16-bit registers for ADC. Use 8-bit registers for
ADCB.
[rwd) - [rwd) + [rws)
[rbd) - [rbd) + [rbs)
[rid) - [rid) + [rls)
Add the Source register contents to the Destination register. Use 32bit registers for ADDL, 16-bit registers for ADD and 8-bit registers for
ADDB.
[rwd)- [rwd) AND [rws)
[rbd) - [rbd) AND [rbs)
AND the Source register contents with the Destination register contents. Use 16-bit registers for AND and 8-bit registers for ANDB.
[rwd) - [rws)
Compare 16-bit register contents by subtracting the Source register
from the Destination register values. Do not modify any register contents, but set Status flags.
[rbdl - [rbsl
Compare 8-bit register contents by subtracting the Source register
from the Destination register values. Do not modify any register contents, but set Status flags.
[rid) - [rlsl
Compare 32-bit register contents by subtracting the Source register
from the Destination register values. Do not modify any register contents, but set Status flags.
Divide
}see text for a discussion of these instructions
Divide long
MUltiPlY} see text f or a d'Iscusslon
. 0 f th
.
ese'instructions
Multiply long
can be executed only in system mode.
Number of clock cycles depends on the number of repetitions for n/m--; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )-m.
Table 6-3.
IfI
Type.. Mnemonic
Operand(s)
Op Code
A Summary of the Z8000 Instruction Set (Continued)
;.
Clock
Cycles
III
OR
ORB
rwd,rws
rbd,rbs
85sd
84sd
2
2
4
4
RLDB
rbd,rbs
BEsd
2
9
Status
Operation Performed
HIDIOIPlslZIC
[rwd] - [rwd] OR [rws]
[rbd] - [rbd] OR [rbs]
OR the contents of the Source register with the Destination register
contents. Use 1 6-bit registers for OR and 8-bit register for ORB.
Left rotate BCD digits in two 8-bit registers specified by rbd and rbs.
The same register cannot be specified for rbd and rbs. Digits are rotated as follows:
XIX
XIX
X
U I
IXI X
----
I
Gl
:s
rbd
!c:
o
m
UI
S
!Gl
RRDB
rbd,rbs
BCsd
2
a.
U
9
!
III
-L
'61
~
~
S
IfI
76543210
'61
rbdl
II
II:
•
-..-
rwd,rws
rbd,rbs
B7sd
B6sd
2
2
5
5
x
X
11 1X
X
X
I I I
X
X
•.--~
1+'--------1
Right rotate BCD digits in two 8-bit registers specified by rbd and rbs.
The same register cannot be specified for rbd and rbs. Digits are rotated as follows:
XIX
o
SBC
SBCB
76543210
I I I I I I I I I I I I I I I I I Irbs
g
m
----- +
76543210
~
,
X
X
7
6
_L
543
-'----~+I
2
1 0
I I I I I I I I I I I I I I I I ,rbs
-..---..--
+
[rwd] - [rwd] - [rws] - C
[rbd] - [rbd] - [rbs] - C
Subtract the Source register contents, plus the initial Carry, from the
Destination register contents using twos complement arithmetic. Use
16-bit registers for SBC. Use 8-bit registers for SBCB.
Privileged instruction - can be executed only in system mode.
Number of clock cycles depends on the number of repetitions for n/m··; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 ).m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
Type
Mnemonic
Operand Is)
Op Code
!'"
>
III
!
i
Status
Clock
Cycles
H
X
SUB
SUBB
SUBl
rwd,rws
rbd,rbs
rld,rls
83sd
82sd
92sd
2
2
2
4
4
XOR
XORB
rwd,rws
rbd,rbs
89sd
88sd
2
2
4
4
ClR
rwd
8Dd8
2
7
ClRB
rbd
8Cd8
2
7
COM
rwd
8DdO
2
7
COMB
rbd
8CdO
2
7
DAB
DEC
DECB
rbd
rwd,n16
rbd,n16
BOdO
ABdn
AAdn
2
5
2
2
4
4
EXTS
rid
B1dA
2
11
EXTSB
rwd
BldO
2
11
EXTSL
rqd
Bld7
2
11
8
0
1
0
P
X
X
X
Operation Performed
S
Z
C
X
X
X
X
X
X
X
X
X
X
X
X
X
0_
!1
'" ~
c
'61
II
C
0
I:C
.(,)
S-
X
'"II
'61
I:C
m
en
m
!
f
II
Q.
..
0
!
'"II
'iii
I:C
"
""
X
X
X
X
X
X
X
X
X
[rwd] - [rwd] - [rws]
[rbd] - [rbd] - [rbs]
[rid] - [rid] - [rls]
Subtract the Source register contents from the Destination register.
Use 32-bit registers for SUBl, 16-bit registers for SUB, 8-bit
registers for SUBB.
[rwd] - [rwd] XOR [rws]
[rbd] - [rbd] XOR [rbs]
Exclusive-OR the contents of Source and Destination registers. Store
the result in the Destination register. Use 16-bit registers for XOR.
Use 8-bit registers for XORB.
[rwd]-O
Clear the Selected Word register.
[rbd]-O
Clear the Selected Byte register.
[rwd] - [rWcI]
Complement the Selected Word register.
[rbd] - [rbd]
Complement the Selected Byte register.
Decimal adjust contents of 8-bit register rbd.
[rwd] - [rwd] - n 1 6
[rbd] - [rbd] - n 1 6
Subtract the immediate value n 16 from a 16-bit register (for DEC) or
an 8-bit register (for DECB).
Bits 16 to 31 of [rid] - bit 15 of [rid].
The sign bit of the low-order word of the register pair is copied into
all bits of the high-order word of the register pair.
Bits 8 to 1 5 of [rwd] - bit 7 of [rwd]
The sign bit of the low-order byte of the register is copied into all bits
of the high-order byte of the register.
Bits 32 to 63 of [rqd] - bit 31 of [rqd]
The sign bit of the low-order register pair of the Quadruple register is
copied into all bits of the high-order register pair.
Privileged instruction - can be executed only in system mode.
Number of clock cycles depends on the number of repetitions for n/m""; n =minimum number of clock cycles and m = number of clock cycles added for each additional ~epetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )"m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
III
Type
Mnemonic
Operand(s)
Op Code
III
>
Clock
Cycles
S
Operation Performed
H
D
0
P
S
Z
C
INC
INCB
rwd,n16
rbd,n16
A9dn
A8dn
2
2
4
4
X
X
X
X
X
X
NEG
NEGB
rwd
rbd
8Dd2
8Cd2
2
2
7
7
X
X
X
X
X
X
X
rwd,1
rwd,2
rbd,1
rbd,2
B3d8
B3dA
B2d8
B2dA
2
2
2
2
6
7
6
7
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
RL
RL
RLB
RLB
X
~
GI
:::s
c
c
:;::;
0
0)
g
..,J
!
m
t!
GI
a.
0
...
SIII
.&
GI
a:
•
••
RLC
RLC
RLCB
RLCB
rwd,1
rwd,2
rbd,1
rbd,2
B3dO
B3d2
B2dO
B2d2
2
2
2
2
6
7
6
7
I
Status
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
[rwd] - [rwd] + n 16
[rbd] +- [rbd] + n 1 6
Add the immediate value n 16 to a 16-bit register (for INC) or an 8-bit
register (for INCB).
[rwd] +- - [fwd]
[rbd] - - [rbd]
Replace the contents of the 1 6-bit register (for NEG) or 8-bit register
(for NEGB) with its twos complement.
Left rotate contents of word (for RU or byte (for RLB) register, n bits
(n = 1 or 2), as follows:
~
I I I I I I I I I I I I I I
~
IIIIII
13J
I~
See accompanying text for a discussion of the Overflow status for all
Register Operate shift and rotate instructions.
Left rotate through Calrry contents of word (for RLC) or byte (for
RLCB) register, n bits (n = 1 or 2), as follows:
Lm-i
I I I I I I I I I I I I I I I;J
~
II IIII
P··I
d .
f
an be executed only in system mode.
nVI ege Instruc Ion - c
..
f
Im o ,. n =minimum number of clock
I
Number of clock cycles depends on the number of. repetlt~ons ~~ n
~ats k times is n + (k-1).
eyc es and m
h
of operation. The number of clock cycles for an instruction w IC re
m.
I~
= number of clock cycles added for each additional repeti tion
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
Type
I
Mnemonic
RR
RR
RRB
RRB
I
Operand(s)
rwd,1
rwd,2
rbd,1
rbd,2
1/1
Op Code
B3dC
B3dE
B2dC
B2dE
Status
t
III
Clock
Cycles
2
2
2
2
6
7
6
7
X
X
2
2
2
2
6
7
6
7
H
0
Operation Performed
S
Z
C
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
0
P
Right rotate contents of word (for RR) or byte (for RRB) register, n bits
(n = 1 or 2), as follows:
rn-4
m-41 I
~
G)
=
RRC
RRC
RRCB
RRCB
rwd,1
rwd,2
rbd,1
rbd,2
B3d4
B3d6
B2d4
B2d6
~c
0
en
m
00
~
!
I!
G)
a.
0
"-
!1/1
'61
G)
a::
.SOA
SOAB
SOAL
rwd,rw
rbd,rw
rld,rw
B3dB
B2dB
B3dF
OrOO
OrOO
OrOO
4
4
4
18/3"
18/3"
18/3"
X
X
X
X
X
X
X
X
X
X
X
X
I 1 I~
Right rotate through Carry contents of word (for RRC) or byte (for
RRCB) register, n bits (n = 1 or 2), as follows:
[ffi:F
I§:tl
liJ
III
--1+1
Shift arithmetic the contents of a byte (for SOAB) word (for SOA) or
long word (for SOAL) register. [rwJ specifies the number of shift bit
positions, and the direction (+ for left shift, - for right shift). 0 shift
is allowed; it causes no shift, but sets status. [rwJ value range is -8
to +8 for SOAB. -16 to + 16 for SOA -32 to +32 for SOAL. Bits 0
to 4 of [rwl are active, with bit 15 used for sign. Shifts occur as
follows:
Left
fu
R;ght
I :J--f£] 0--lII:!: ]-r I ~O
~I
7 6
1514
31 30
2
2
2
Bit Numbers
•
liJ
I I
0
1 0
1 0
--SOAB--SOA--SOAL -
7 6
1514
31 30
2
2
2
0
1 0
1 0
Bit Numbers
Privileged instruction - can be executed only in system mode.
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1 )·m.
Copyright
©
1 979 McGraw-Hili, Inc.
Table 6-3, A Summary of the Z8000 Instruction Set (Continued)
Type
I
Mnemonic
SOL
SOLB
SOLL
I
Operand(s)
rwd,rw
rbd,rw
rld,rw
Op Code
B3d3
B2d3
B3d7
OrOO
OrOO
OrOO
=
:-
Clock
Cycles
4
18/3··
4
18/3. 0
4
18/3
0
•
Status
HIDIOIPISIZIC
u
u
u
XIX
X X
X
X
X
X
X
I
Operation Performed
SOL, SOLB and SOLL are equivalent to SOA, SOAB and SOAL,
respectively, but they perform logical right shifts, Left shifts are identical. Shifts may be illustrated as follows:
Right
Left
m~1 I I J.--E£] [§MIl~1
1
'S=
c
o
0)
m
co
g
SLA
SLAB
SLAL
rwd,data16
rbd,data16
rld,data16
B3d9
B2d9
B3dO
yyyy
yyyy
yyyy
4
4
4
16/3··
16/3··
16/3
SLL
SLLB
SLLL
rwd,data16
rbd,data16
rld,data16
B3d1
" B2d1
4
4
4
16/3··
16/3··
B3d5
yyyy
yyyy
yyyy
16/3 00
SRA
SRAB
SRAL
rwd,data16
rbd,data16
rld,data16
B3d9
B2d9
B3dO
yyyy
yyyy
yyyy
4
4
4
16/3
16/3··
16/3··
SRL
SRLB
SRLL
rwd,data16
rbd,data16
rld,data16
B3d1
B2d1
B3d5
yyyy
yyyy
yyyy
4
4
16/3··
16/3··
16/3··
TSET
TSETB
rwd
rbd
80d6
8Cd6
2
TEST
TESTB
TESTL
rwd
rbd
rid
80d4
8Cd4
9CdO
!
!
&
o
...
!III
0
•
00
x
X
X
X
X
X
X
X
X
X
X
X
X
X
o
X
X
u
u
u
X
X
X
X
X
X
X
X
X
X
X
X
u
u
u
X
o
o
X
X
X
7 6
2
0
1514
31 30
2
0
2
0
-SOLB --+
- SOL--+
- SOLL--+
7 6
1514
3130
I~
2
2
2
0
0
0
SLA, SLAB and SLAL are identical to SOA, SOAB and SOAL, respectively, when these instructions are performing left shifts, except that
SLA, SLAB and SLAL specify the shift bit count immediately.
SLL, SLLB and SLLL are identical to SOL, SOLB and SOLL, respectively, when these instructions are performing left shifts, except that
SLL, SLLB and SLLL specify the shift bit count immediately.
X
X
SRA, SRAB and SRAL are identical to SOA, SOAB and SDAL, respectively, when these instructions are performing right shifts, except that
SRA, SRAB, and SRAL specify the shift bit count immediately.
X
X
'61
~
:u
CD
~
}>
c<
'02Q.c:
qo~i
....
coo)'"
4
X
X
2
7
2
2
7
XIXXIXX
13
X
X
[s]- [rwd)(MSB),[rwd]- FFFF
[s]- [rbd](MSB), [rbd] - FF
Move the most significant bit of the 16-bit register (for TSET) or 8-bit
register (TSETB) to the Sign status. Then fill the register with 1 bits.
7
7
2
SRL, SRLB, and SRLL are identical to SOL, SOLB, and SOLL, respectively, when these instructions are performing right shifts, except that
SRL, SRLB and SRLL specify the shift bit count immediately.
X
X
X
X
X
X
I
[rwd] OR 0
[rbd] OR 0
[rid] OR 0
OR the specified register contents with O. Set Status flags based on
the result. Test a 32-bit register for TESTL, a 16-bit register for TEST
and an 8-bit register for TESTB.
• Privileged instruction - can be executed only in system mode.
.. Number of clock cycles depends on the number of repetitions for n/m··; n =minimum number of clock cycles and m = number of clock cycles added for each additional repeti'tion
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1).m,
Table 6-3. A Summary of the
II)
Type
Mnemonic
Operand(s)
Op Code
S
>-
Clock
Cycles
ID
0)
.!.J
o
~
u
S
(I)
LOCTL'
LOCTL'
NSPSEG,rws
rwd,NSPSEG
70sE
70d6
2
2
7
7
LOCTL'
LOCTL'
NSPOFF,rws
rwd,NSPOFF
70sF
70d7
2
2
7
7
POP
POP
POP
rwdl@ris
@rid,@ris
adrsx,@ris
97sd
17sd
57si
2
2
8
12
15-19
POPL
POPL
POPL
rld,@ris
@rid,@ris
adrsx,@ris
95id
15id
55si
PUSH
PUSH
PUSH
PUSH
@rid,rws
@rid,@ris
@rid,adrsx
@rid,data 16
93is
13is
53di
OOd9
@rid,rls
@rid,@ris
@rid,adrsx
91is
11 is
51di
PUSHL
PUSHL
PUSHL
pppp
qqqq 4/6
2
2
pppp
qqqq 4/6
2
2
pppp
yyyy
qqqq 4/6
4
2
2
pppp
qqqq 4/6
12
19
22-26
9
13
13-17
12
12
20
20-24
zaooo Instruction Set (Continued)
Status
H
D
0
P
Operation Performed
S
Z
C
[NSPSEG] - [rws]
[rwd] - [NSPSEG]
Transfer data between a 16-bit register and tahe Z8001 normal
Stack Pointer Segment Address register (R 14N).
[NSPOFF] - [rws]
[rwd] - [NSPOFF]
Transfer data between a 16-bit register and the normal Stack Pointer
Address register (R15N).
[rwd] - [[ris]], Iris] - Iris] + 2
[[rid]] - [[ris]!. Iris] - Iris] + 2
[adrsx] - [[ris]!. Iris] - Iris] + 2
Pop the memory word addressed by ris, the designated Stack Pointer.
Any register with the exception of RO (for non segmented) or RRO (for
segmented) can be designated as the Stack Pointer. The popped word
is loaded into a 16-bit register, or the memory location addressed by
adrsx or [rid].
[rid] - [[ris]], Iris] - Iris] + 4
[[rid]] - [[ris]], Iris] - [risl + 4
[adrsxl - [[ris]], Iris] - Iris] + 4
POPL is identical to POP, except that a 32-bit long word is popped.
[ridl- [rid]- 2.[[rid]]- [rws]
[rid] - [rid] - 2. [[rid]] - [[ris]]
[rid]- [rid] - 2. [[rid]]- [adrsx]
[rid] - [rid] - 2. [[rid]] - data16
Push a 16-bit word onto a memory stack addressed by rid, the designated Stack Pointer. Any register with the exception of RO (for nonsegmented) or RRO (for segmented) can be designated as the Stack
Pointer. The pushed word can come from a register, the memory
word addressed by adrsx or Iris!. or it may be immediate data.
[rid] - [rid] - 4. [[rid]] - [rls]
[rid] - [rid] - 4. [[rid]] - [[ris]]
[rid] - [rid] - 4. [[rid]] - [adrsx]
PUSHL is identical to PUSH except that a 32-bit long word is pushed,
also there is no immediate version of PUSHL.
• Privileged instruction - can be executed only in system mode.
•• Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repeti tion
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-1).m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
.."
III
Type
Mnemonic
Operand(s)
Op Code
BIT
BIT
BIT
BIT
rwd,b16
@rid,b16
adrsx,b16
rwd,rws
A7db
27ib
67ib
270s
BITB
BITB
BITB
BITB
rbd,b8
@rid,b8
adrsx,b8
rbd,rws
A6db
26ib
66ib
260s
RES
RES
RES
RES
rwd,b16
@rid,b16
adrsx,b16
rwd,rws
A3db
23ib
63ib
230s
RESB
RESB
RESB
RESB
rbd,b8
@rid,b8
adrsx,b8
rbd,rws
A2db
22ib
62ib
220s
>m
Clock
Cycles
Status
Operation Performed
H
D
0
P
S
Z
pppp qqqq 4/6
OdOO
4
4
8
10-14
10
X
X
X
X
2
2
pppp qqqq 4/6
OdOO
4
4
8
10-14
10
X
X
X
X
2
2
fII
c
0
0)
.!.J
'+l
!
a.
0"
.
iD
...
2
2
pppp
OdOO
qqqq 4/6
4
2
2
pppp
OdOO
qqqq 4/6
4
4
11
13-17
10
4
11
13-17
10
C
Z - NOT bit b16 of [rwdl
Z - NOT bit b16 of [[rid]]
Z - NOT bit b 16 of [adrsxl
Z - NOT bit [rwsl of [rwdl
Set the Z status to the complement of the specified bit, which may be
in a 16-bit register or memory word. The bit may be specified immediately, or for a register it may be specified by the low-order four
bits of a 16-bit register.
Z - NOT bit b8 of [rbdl
Z - NOT bit b8 of [[rid]]
Z - NOT bit b8 of [adrsxl
Z - NOT bit [rwsl of [rbdl
Set the Z status to the complement of the specified bit, which may be
in an 8-bit register or memory byte. The bit may be specified immediately, or for a register it may be specified by the low-order three
bits of one of the registers RO - R7.
Bit b16 of [rwdl - 0
Bit b16 of [[rid]]- 0
Bit b 16 of [adrsxl - 0
Bit [rwsl of [rwdl +- 0
Clear the specified bit, which may be in a 16-bit register or memory
word. The bit may be specified immediately, or for a register it may
be specified by the low-order four bits of a 16-bit register.
Bit b8 of [rbdl +- 0
Bit b8 of [[rid]] - 0
Bit b8 of [adrsxl - 0
Bit [rwsl of [rbdl - 0
Clear the specified bit, which may be in an 8-bit register or memory
byte. The bit may be specified immediately, or for a register it may be
specified by the low-order three bits of one of the registers AO - A7.
Privileged instruction - cali be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n ~ (k-11.m.
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
1/1
Type
Mnemonic
Operand(s)
;
Op Code
III
~
SET
:s
SET
G>
~c
0
g
1/1
c
~
t!
G>
..
a.
SET
SET
SETB
SETB
SETB
SETB
rwd,b16
@rid,b16
adrsx,b16
rwd,rws
rbd,b8
@rid,b8
adrsx,b8
rbd,rws
A5db
25ib
65ib
250s
A4db
24ib
64ib
240s
pppp qqqq
OdOO
pppp
OdOO
qqqq
Status
Clock
Cycles
2
4
2
11
4/6
4
2
2
4/6
4
13-17
10
4
11
13-17
10
Operation Performed
H
D
0
P
S
Z
C
Bit b16 of [rwd)- 1
Bit b16 of [[rid]) - 1
Bit b 16 of [adrsx] - 1
Bit [rws) of [rwd) - 1
Bit b8 of [rbd]
Bit b8 of [[rid]]
Bit b8 of [adrsx]
Bit [rws] of [rbd]
SET and SETB instructions are equivalent to RES and RESB instructions, respectively, except that the selected bit is set.
0
iii
01'
int
7C
OOOOOOvv
2
6
EI'
int
7C
00000lvv
2
6
7BOO
2
13,16
IRET'
~
......
N
.
a.
E
LOCTL'
LOCTL'
PSAPSEG,rws
rwd,PSAPSEG
70sC
70d4
2
2
7
7
LOCTL'
LOCTL'
PSAPOFF,rws
rwd,PSAPOFF
70s0
70d5
2
2
7
7
!
.s
...
--
X
X
X
X
X
X
X
Oisable the indicated interrupt(s). Either or both of VI and NVI may be
indicated.
Enable the indicated interrupt(s). Either or both of VI and NVI may be
indicated.
[SP)- [SP] + 2. [FCW]- [[SP]]: [SP]- [SP) + 2. [PC]- [[SP]]
[SP) - [SP] + 2. (Nonsegmented)
[SP] - [SP] + 2. [FCW] - [[SP]]. [SP] - [SP) + 2. [PC] - [[SP]]
[SP] - [SP] + 4. (Segmented)
Return from interrupt. Pop and discard identifier word. Pop flag and
control word. Pop Program Counter.
[PSAPSEG] - [rws]
[rwd] - [PSAPSEG]
These two instructions transfer data between the 28001 Program
Status Area Pointer Segment register, and a 16-bit general purpose
register.
[PSAPOFF] - [rws]
[rwd] - [PSAPOFF]
These two instructions transfer data between the Program Status
Area Pointer and a 16-bit general purpose register.
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-l )'m.
Table 6-3.
I/)
Type
~
-....I
W
I/)
~
S
Mnemonic
Operand(s)
Op Code
A Summary of the Z8000 Instruction Set (Continued)
t
CD
Clock
Cycles
COMFLG
flag
80f5
2
7
LOCTL'
FCW,rws
70sA
2
7
LOCTL'
rwd,FeW
70d2
2
7
LOeTLB
FLAGS,rbs
8es9
2
7
LOeTLB
rbd,FLAGS
8ed1
2
7
RESFLG
SETFLG
TCC
flag
flag
cC,rwd
80f3
80f1
AFdc
2
2
2
7
7
5
TeeB
cC,rbd
AEdc
2
5
Status
Operation Performed
H
D
0
P
S
Z
C
X
X
X
X
X
Complement each status named in the operand. Any or all of C, Z, S,
P, or 0 may be named in any order.
[FCW) - [rws)
Load register contents into FCW. Unassigned bits of FCW are not
affected.
[rwd] - [FeW]
Load Few contents into selected register. Unassigned bits of FeW
are reset to 0 in rwd.
[FLAGS] - [rbs]
Load byte register contents into low-order byte of FeW. Bits 0 and 1,
which are unassigned, are not affected.
[rbd] - [FLAGS]
Load the low-order byte of FCW into byte register rbd. Bits 0 and 1 of
rbd are reset to O.
Reset to 0 each status named in the operand.
Set to 1 each status named in the operand.
If cc is "true" then set bit 0 of Register rwd. Otherwise reset bit 0 of
Register rwd.
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
(I)
.
..
If cc is "true" then set bit 0 of Register rbd. Otherwise reset bit 0 of
Register rbd.
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-11·m.
- - -
-
-
Table 6-3. A Summary of the Z8000 Instruction Set (Continued)
CII
Type
Mnemonic
Operand(s)
Op Code
t
III
Clock
Cycles
0
P
S
7AOO
7DsB
2
2
8/3""
LDCTL"
rwd,REFRESH
7Dd3
2
7
7BOA
2
7
X
7BdD
2
12/7""
X
rwd
0
0
CII
~
C»
III
.J:I.
"c
MRES"
7B09
2
5
Go
MSET"
7B08
2
5
e
NOP
8007
2
7
III
j
Z
7
gc
fi
D
REFRESH,rws
MREQ"
~
Operation Performed
H
HALT
LDCTL"
MBIT"
.:.,
Status
X
C
Halt CPU until reset or interrupt
[REFRESH I - [rwsl
Transfer the contents of the specified 16-bit register into the
Dynamic Memory Refresh Control register.
[rwdl-[REFRESHI
Transfer the contents of the Dynamic Memory Refresh Control.
register to the specified 16-bit register.
[SI-Mi
_I
Set Sign status to 1 if MI is input low (1). Reset Sign status to 0 if MI
is input high (01.
[ZI- O. If Mi = 1 then [SI- 0 and [MOI- O.
If Mi = 0 then [MOI- 1. Decrement [rwdl to O.
If Mi is still 0 then [SI - O. [MOl - O.
If Mi is now 1 then [SI - 1.
[ZI-1.
Execute a multi-micro bus request, as described in accompanying
text.
[MOI-O
Output MO high.
[MOI-1
Output MO low.
No operation.
0
" Privileged instruction - can be executed only in system mode.
"" Number of clock cycles depends on the number of repetitions for n/m""; n =minimum number of clock cycles and m = number of clock cycles added for each additional repetition
of operation. The number of clock cycles for an instruction which repeats k times is n + (k-11"m.
Table 6-4. Z8000 Instruction Set Object Codes
Mnemonic
AOC
AOCB
ADD
rwd,rws
rbd,rbs
rwd,adrsx
Object Code
Bytes
B5sd
B4sd
41id
pppp
2
2
4/6
Clock
Cycles
5
5
9-13
Mnemonic
AOOB
rwd,rws
rwd,@ris
rbd,adrsx
010d
yyyy
B1sd
01id
40id
pppp
AOOL
rbd,rbs
rbd,@ris
rld,adrsx
OOOd
yyOO
BOsd
OOid
56id
pppp
160d
yyyy
rld,rls
rld,@ris
rwd,adrsx
96sd
16id
47id
pppp
4
7
2
2
4/6
4
7
9-13
4
7
2
2
4/6
4
7
15-19
ANOB
rwd,rws
rwd,@ris
rbd,adrsx
070d
yyyy
B7sd
07id
46id
pppp
BIT
rbd,rbs
rbd,@ris
adrsx,b16
060d
yyOO
B6sd
06id
67ib
pppp
BITB
adrsx,bB
A7db
27ib
270s
OdOO
66ib
pppp
CALL
adrsx
A6db
26ib
260s
OdOO
5FiO
pppp
6
14
1FdO
2
2
4/6
7
12
15-19
2
2
2
6/B
7
12
7
14-18
4/6
9-13
4
7
2
2
4
4
7
11
6/B
14-18
4/6
9-13
4
7
2
2
4
4
7
11
4/6
15-19
6
14
2
B
14
COM
BCdB
OCdB
40iO
pppp
COMB
rwd
@rid
adrsx
BOdO
OOdO
4CiO
pppp
COMFLG
CP
rbd
@rid
flag
adrsx,data 16
BCdO
OCdO
BOf5
4Di1
yyyy
pppp
2
2
4/6
8
14
9-13
qqqq
qqqq
qqqq
4
7
2
2
4/6
4
7
9-13
4
7
2
2
4/6
4
7
10-14
rwd,adrsx
4Bid
pppp
rwd,data16
OBOd
yyyy
BBsd
OBid
00d1
yyyy
4Ci1
yyOO
pppp
qqqq
rwd,rws
rwd,@ris
@rid,data16
CPB
adrSX,data8
qqqq
rbd,adrsx
4Aid
pppp
rbd,dataB
OAOd
yyOO
BAsd
OAid
OCd1
yyOO
50id
pppp
qqqq
2
2
4
4
8
10
4/6
10-14
2
2
4
4
B
10
rbd,rbs
rbd,@ris
@rid,dataB
4/6
CPL
rld,adrsx
qqqq
rld,data32
100d
yyyy
rld,rls
rld,@ris
90sd
10id
12-21
zzzz
qqqq
@rid
7
B
15-19
rbd
@rid
adrsx
qqqq
rbd,b8
@rid,bB
rbd,rws
2
2
4/6
CLRB
qqqq
rwd,b16
@rid,b16
rwd,rws
7
8
11-15
80d8
OOdB
4Ci8
pppp
qqqq
rbd,dataB
2
2
4/6
rwd
@rid
adrsx
qqqq
rwd,data16
10/15
11-15
Oxxx
40i8
pppp
qqqq
zzzz
AND
2
4/6
disp
adrsx
qqqq
rld,data32
Clock
Cycles
qqqq
qqqq
rbd,dataB
Bytes
CALR
CLR
qqqq
rwd,data16
Object Code
2
10/15
6-75
2
Table 6-4.
Mnemonic
CPO
rwd,@ris,rw,cc
CPOB
rbd,@ris,rw,cc
CPOR
rwd,@ris,rw,cc
CPORB
rbd,@ris,rw,cc
CPI
rwd,@ris,rw,cc
CPIB
rbd,@ris,rw,cc
CPIR
rwd,@ris,rw,cc
CPIRB
rbd,@ris,rw,cc
CPSD
@rid,@riS,rw,cc
CPSDB
@rid,@ris,rw,cc
CPSDR
@rid,@ris,rw,cc
CPSDRB
@rid,@ris,rw,cc
CPSI
@rid,@ris,rw,cc
CPSIB
@rid,@ris,rw,cc
CPSIR
@rid,@riS,rw,cc
CPSIRB
@rid,@ris,rw,cc
DAB
DEC
DECB
'01
DIV
rbd
adrsx,n16
rwd,n16
@rid,n16
adrsx,n16
rbd,n16
@rid,n16
int
rld,adrsx
rld,data16
rld,rws
rld,@ris
.
..
zaooo Instruction Set Object Codes (Continued)
Object Code
Bytes
Clock
Cycles
BBsS
Ordc
BAsS
Ordc
BBsC
Ordc
BAsC
Ordc
BBsO
Ordc
BAsO
Ordc
BBs4
Ordc
BAs4
Ordc
BBsA
Ordc
BAsA
Ordc
BBsE
Ordc
BAsE
Ordc
BBs2
Ordc
BAs2
Ordc
BBs6
Ordc
BAs6
Ordc
BOdO
6Bin
pppp
qqqq
ABdn
2Bdn
6Ain
pppp
qqqq
AAdn
2Adn
7C OOOOOOvv
5Bid
pppp
qqqq
1BOd
yyyy
SBsd
1Bid
4
20
4
20
4
20/9"
4
20/9"
4
20
4
20
4
20/S"
4
20/9"
4
25
4
25
4
25/14"
4
25/14"
4
25
4
25
4
25/14"
4
25/14"
Mnemonic
DIVL
rqd,adrsx
rqd,data32
Object Code
Bytes
Clock
Cycles
5Aid
pppp
qqqq
1AOd
yyyy
4/6
note 1
6
note 1
2
2
2
2
2
4/6
note 1
note 1
11
11
6
15-19
2
2
4/6
6
12
15-19
2
2
2
2
2
2
4
6
12
11
11
11
S/3"
12
2
4
10
12
2
4/6
10
13-17
2
2
4/6
11
13-17
2
2
4
11
21
4
21
4
21110"
4
21110"
4
21
zzzz
2
4/6
2
2
4/6
2
2
2
4/6
DJNZ
DBJNZ
'EI
EX
rqd,rls
rqd,@ris
rw,disp
rb,disp
int
rwd,adrsx
EXB
rwd,rws
rwd,@ris
rbd,adrsx
EXTS
EXTSB
EXTSL
'HALT
'IN
rbd,rbs
rbd,@ris
rid
rwd
rqd
rwd,ioaddr
'INB
rwd,@rw
rbd,ioaddr
INC
rbd,@rw
adrsx,n16
INCB
rwd,n16
@rid,n16
adrsx,n16
'INO
rbd,n16
@rid,n16
@rid,@ris,rw
'INDB
@rid,@ris,rw
'INOR
@rid,@ris,rw
5
13-17
4
11
13-17
4
11
6
note 1
4
note 1
'INORB
@rid,@riS,rw
2
2
note 1
note 1
'INI
@rid,@ris,rw
9Asd
1Aid
Fr Ottttttt
Fr 1ttttttt
7C 00OOO1vv
6Did
pppp
qqqq
ADsd
2Dsd
6Cid
pppp
qqqq
ACsd
2Csd
B1dA
81dO
B1d7
7AOO
3Bd4
pppp
3Dsd
3Ad4
pppp
3Csd
69in
pppp
qqqq
A9dn
29dn
6Sin
pppp
qqqq
A8dn
2Sdn
3Bs8
Ord8
3As8
Ord8
3BsS
OrdO
3As8
OrdO
3BsO
Ord8
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m"; n = minimum number of clock cycles and
m = number of clock cycles added for each 'additional repetition of operation.
The number of clock cycles for an instruction which repeats k times is n + (k- H ·m.
6-76
4
4
Table 6-4. Z8000 Instruction Set Object Codes (Continued)
Mnemonic
·INIB
@rid,@ris,rw
·INIR
@rid,@ris,rw
·INIRB
@rid,@ris,rw
·IRET
JP
cC,adrsx
JR
LO
cC,@rid
cC,disp
adrsx,data 16
Object Code
Bytes
Clock
Cycles
3AsO
Ord8
3BsO
OrdO
3AsO
OrdO
7BOO
5Eic
pppp
4
21
4
21/10··
4
21/10··
2
4/6
13,16
7-11
2
2
6/8
10,15/7
6
14-18
Mnemonic
LOB
(Cont.)
rbd,@ris
rid(disp),rbs
rid(rw),rbs
@rid,data8
qqqq
1Edc
Ecxx
40i5
yyyy
pppp
rbd,ris(rw)
LOL
@rid,rbs
adrsx,rls
6Fis
pppp
61id
pppp
rwd,data16
210d
yyyy
A1sd
31id
xxx x
71id
OrOO
21id
33is
xxx x
73is
OrOO
OOd5
yyyy
2Fds
4Ci5
yyOO
pppp
rwd.ris(rw)
rwd,@ris
rid (displ.rws
rid (rwl.rws
@rid,data 16
LOB
@rid.rws
adrsx,data8
6Eis
pppp
60id
pppp
.
..
Cdyy
AOsd
30id
xxxx
2
4
7
14
4
14
4
11
2
4/6
8
14-18
4/6
12-16
6
11
2
4
5
17
4
17
2
4
11
17
4
17
2
4/6
11
13-16
4
15
4
15
4
12-13
4
15
4
15
4
15
4
15
qqqq
rld,data32
140d
yyyy
4/6
9-13
rld,rls
rld,ris(disp)
4
7
rld,ris(rw)
2
4
3
14
rld,@ris
rid(disp),rls
4
14
rid(rwl,rls
2
4
7
14
94sd
35id
xxx x
75id
OrOO
14id
37is
xxxx
77is
OrOO
10ds
76id
pppp
4
14
zzzz
LOA
@rid.rls
rld,adrsx
qqqq
rld,ris(disp)
4
11
2
6/8
8
14-18
rld.ris(rw)
rwd.adrsx
rwd.ris(disp)
4/6
11-15
4/6
9-13
rwd.ris(rw)
LOAR
rld,disp
rwd.disp
qqqq
rbd,data8
rbd,rbs
rbd,ris(disp)
14
11-15
qqqq
rbd,adrsx
4
4/6
qqqq
adrsx,rbs
70id
OrOO
20id
32is
xxxx
72is
OrOO
OCd5
yyOO
2Eds
50is
pppp
54id
pppp
qqqq
rwd,rws
rwd,ris (disp)
Clock
Cycles
qqqq
qqqq
rwd,adrsx
Bytes
rld,adrsx
qqqq
adrsx,rws
Object Code
2
2
4
34id
xxxx
74id
OrOO
76id
pppp
34id
xxxx
74id
OrOO
340d
xxxx
340d
xxxx
5
3
14
Privileged instruction - can be executed only in system mode .
Number of clock cycles depends on the number of repetitions for n/m··; n = minimum number of clock cycles and
m = number of clock cycles added for each additional repetition of operation. The number of clock cycles for an instruction
which repeats k times is n + (k-1 ).m.
6-77
Table 6-4. Z8000 Instruction Set Object Codes (Continued)
Mnemonic
'LDcn
LOO
FCW,rws
NSPOFF,rws
NSPSEG,rws
PSAPOFF,rws
PSAPSEG,rws
REFRESH,rws
rwd,FCW
rwd,NSPOFF
rwd,NSPSEG
rwd,PSAPOFF
rwd,PSAPSEG
rwd,REFRESH
FLAGS,rbs
rbs,FLAGS
@rid,@ris,rw
L008
@rid,@ris,rw
LOOR
@rid,@ris,rw
LOORB
@rid,@ris,rw
LOI
@rid,@ris,rw
LOIB
@rid,@ris,rw
LOIR
@rid,@ris,rw
LOIRB
@rid,@ris,rw
LOK
LOM
rwd,b16
adrsx,rws,n 1 6
LOClLB
rwd,adrsx,n 1 6
rwd,@ris,n 16
@rid,rws,n16
'LOPS
adrsx
LOR
@ris
disp,rws
rwd,disp
,
"
Object Code
Bytes
Clock
Cycles
70sA
70sF
70sE
70s0
70sC
70sB
70d2
70d7
70d6
70d5
70d4
70d3
8Cs9
8Cd1
BBs9
Ord8
BAs9
Ord8
BBs9
OrdO
BAs9
OrdO
BBs1
Ord8
BAs 1
Ord8
BBs1
OrdO
BAs1
OrdO
BOdb
5Ci9
OsOn
pppp
qqqq
5Ci1
OdOn
pppp
qqqq
1Cs1
OdOn
1Cd9
OsOn
79iO
pppp
qqqq
39s0
330s
xxxx
310d
xxxx
2
2
2
2
2
2
2
2
2
2
2
2
2
2
4
7
7
7
7
7
7
7
7
7
7
7
7
7
7
20
4
20
4
20/9"
4
20/9"
4
20
4
20
4
20/9"
4
20/9"
2
6/8
17-21/3*·
Mnemonic
LORB
disp,rbs
rbd,disp
LORL
disp,rls
rld,disp
'MBIT
'MREQ
'MRES
'MSET
MULT
rwd
rld,adrsx
rld,data16
MULTL
6/8
rld,rws
rld,@ris
rqd,adrsx
rqd,data32
NEG
rqd,rls
rqd,@ris
adrsx
NEGB
rwd
@rid
adrsx
5
~7-21/3*'
rbd
@rid
NOP
OR
rwd,adrsx
4
14/3"
4
14/3"
rwd,data16
4/6
16-23
rwd,rws
rwd,@ris
2
4
12,16
14
4
14
Object Code
320d
xxxx
300d
xxxx
370s
xxxx
350d
xxxx
7BOA
7BdO
7B09
7B08
59id
pppp
qqqq
190d
yyyy
99sd
19id
58id
pppp
qqqq
180d
yyyy
zzzz
98sd
18id
40i2
pppp
qqqq
80d2
00d2
4Ci2
pppp
qqqq
8Cd2
OCd2
8007
45id
pppp
qqqq
050d
yyyy
85sd
05id
Bytes
Clock
Cycles
4
14
4
14
4
17
4
17
2
2
2
2
4/6
7
12/7"
5
5
note 2
4
note 2
2
2
4/6
note 2
note 2
note 2
6
note 2
2
2
4/6
note 2
note 2
15-19
2
2
4/6
7
12
15-19
2
2
2
4/6
7
12
7
9-13
4
7
2
2
4
7
Privileged instruction - can be executed only in system mode.
Number of clock cycles depends on the number of repetitions for n/m"; n = minimum number of clock cycles and
m = number of clock cycles added for each additional repetition of operation. The number of clock cycles for an instruction
which repeats k times is n + (k-1 I,m.
6-78
Table 6-4. Z8000 Instruction Set Object Codes (Continued)
Object Code
Bytes
Clock
Cycles
rbd,adrsx
44id
pppp
4/6
9-13
rbd,data8
040d
yyOO
84sd
04id
3BsA
OrdO
3AsA
OrdO
3Bs2
OrdO
3As2
OrdO
4
7
2
2
4
4
7
21/10"
4
21/10"
3Bs6
pppp
3Fds
3As6
pppp
3Eds
3BsA
Ord8
3AsA
Ord8
3Bs2
Ord8
3As2
Ord8
57si
pppp
4
12
2
4
10
12
2
4
10
21
4
21
4
21
Mnemonic
ORB
Object Code
Bytes
Clock
Cycles
adrsx,b16
63ib
pppp
4/6
13-17
rwd,b16
rwd,rws
A3db
230s
OdOO
23ib
62ib
pppp
2
4
4
10
2
4/6
11
13-17
2
4
4
10
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
4
11
7
10,13/7
6
7
6
7
6
7
6
7
9
6
7
6
7
6
7
6
7
9
5
5
33,39
18/3"
4
18/3"
4
18/3"
4
18/3"
4
18/3"
4
18/3"
Mnemonic
RES
qqqq
'OTOR
rbd,rbs
rbd,@ris
@rid,@ris,rw
'OTORB
@rid,@ris,rw
'OTIR
@rid,@ris,rw
'OTIRB
@rid,@ris,rw
'OUT
ioaddr,rws
'OUTB
@rw,rws
ioaddr,rbs
'OUTO
@rw,rbs
@rid,@ris,rw
'OUTOB
@rid,@ris,rw
'OUTI
@rid,@ris,rw
'OUTIB
@rid,@ris,rw
POP
adrsx,@ris
POPL
rwd,@ris
@rid,@ris
adrsx,@ris
PUSH
rld,@ris
@rid,@ris
@rid,adrsx
95id
15id
53di
pppp
@rid,data1 6
OOd9
yyyy
93is
13is
51di
pppp
qqqq
4
21/10"
4
21/10"
.
Privileged instruction -
91is
11 is
rbd,rw
SOAL
rld,rw
SOL
rwd,rw
RLOB
RR
4
21
RRB
4/6
15-19
RRC
2
2
8
12
22-26
RRCB
4/6
12
19
13-7
4
12
2
2
9
13
20-24
2
2
4/6
qqqq
@rid,rls
@rid,@ris
SOAB
RLC
qqqq
PUSHL
RROB
SBC
SBCB
SC
SOA
@rid,b8
flag
cc
rwd,1
rwd,2
rbd,1
rbd,2
rwd,1
rwd,2
rbd,1
rbd,2
rbd,rbs
rwd,1
rwd,2
rbd,1
rbd,2
rwd,1
rwd,2
rbd,1
rbd,2
rbd,rbs
rwd,rws
rbd,rbs
data8
rwd,rw
RLB
qqqq
@rid,rws
@rid,@ris
@rid,adrsx
qqqq
RLCB
4/6
2
2
@rid,b16
adrsx,b8
rbd,b8
rbd,rws
RESFLG
RET
RL
qqqq
97sd
17sd
55si
pppp
RESB
A2db
220s
OdOO
22ib
80f3
9EOc
B3d8
B3dA
B2d8
B2dA
B3dO
B3d2
B2dO
B2d2
BEsd
B3dC
B3dE
B2dC
B2dE
B3d4
B3d6
B2d4
B2d6
BCsd
B7sd
B6sd
7Fyy
B3dB
OrOO
B2dB
OrOO
B3dF
OrOO
B3d3
2
orOo
SOLB
rbd,rw
SOLL
rld,rw
B2d3
OrOO
B3d7
Oroa
12
20
can be executed only in system mode .
,. Number of clock cycles depends on the number of repetitions for n/m"; n = minimum number of clock cycles and
m = number of clock cycles added for each additional repetition of operation. The number of clock cycles for an instruction
which repeats k times is n + (k-11.m.
6-79
Table 6-4. Z8000 Instruction Set Object Codes (Continued)
Object Code
Bytes
Clock
Cycles
adrsx,b16
65ib
pppp
4/6
13-17
rwd,b16
rwd,rws
A5db
2505
OdOO
25ib
64ib
pppp
2
4
4
10
2
4/6
11
13-17
Mnemonic
SET
qqqq
SETB
@rid,b16
adrsx,b8
Mnemonic
'SOTIA
@rid,@ris,rw
'SOTIAB
@rid,@ris,rw
'SOUT
ioaddr,rws
'SOUTB
ioaddr,rbs
'SOUTO
@rid,@ris,rw
qqqq
rbd,b8
rbd,rws
SETFLG
'SIN
@rid,b8
flag
rwd,ioaddr
'SINB
rbd,ioaddr
'SIND
@rid,@ris.rw
'SINDB
@rid,@ris,rw
'SINDA
@rid,@ris,rw
'SINDAB
@rid,@ris,rw
'SINI
@rid,@riS,rw
'SINIB
@rid,@ris,rw
'SINIA
@rid,@ris.rw
'SINIAB
@rid.@riS,rw
SLA
rwd,data16
SLAB
rbd,data16
SLAL
rld,data16
SLL
rwd.data16
SLLB
rbd,data16
SLLL
rld.data16
'SOTDA
@rid,@riS,rw
'SOTDAB
@rid,@ris.rw
A4db
240s
OdOO
24ib
8Df1
3Bd5
pppp
3Ad5
pppp
3Bs9
Ord8
3As9
Ord8
3Bs9
OrdO
3As9
OrdO
3Bs1
Ord8
3As1
Ord8
3Bs1
OrdO
3As1
OrdO
B3d9
yyyy
B2d9
yyyy
B3dC
yyyy
B3d1
yyyy
B2d1
yyyy
B3d5
yyyy
3BsB
OrdO
3AsB
OrdO
2
4
4
10
'SOUTDB @rid,@ris,rw
2
2
4
11
7
12
4
12
4
21
4
21
4
21/10"
4
21/10"
4
21
4
21
4
21/10"
rwd,data16
4
21/10"
4
16/3"
rwd,rws
rwd.@ris
rbd,adrsx
4
16/3"
'SOUTI
@rid,@ris,rw
'SOUTIB
@rid,@ris,rw
SAA
rwd,data16
SAAB
rbd,data16
SAAL
rld,data16
SAL
rwd,data16
SALB
rbd,data16
SALL
rld.data16
SUB
rwd,adrsx
Object Code
Bytes
Clock
Cycles
3Bs3
OrdO
3As3
OrdO
3Bs7
pppp
3As7
pppp
3BsB
OrdB
3AsB
OrdB
3Bs3
OrdB
3As3
OrdB
B3d9
yyyy
B2d9
yyyy
B3dD
yyyy
B3d1
yyyy
B2d1
yyyy
B3d5
yyyy
43id
pppp
4
21/10"
4
21/10"
4
12
4
12
4
21
4
21
4
21
4
21
4
16/3"
4
16/3"
4
16/3"
4
16/3"
4
16/3"
4
16/3"
4/6
9-13
4
7
2
2
4/6
4
7
9-13
4
7
2
2
4/6
4
7
15-19
6
14
2
2
2
8
14
5
qqqq
SUBB
qqqq
rbd,data8
4
16/3"
4
16/3"
4
16/3"
4
16/3"
4
21/10"
030d
yyyy
83sd
03id
42id
pppp
SUBL
rbd,rbs
rbd,@ris
rld,adrsx
020d
yyOO
82sd
02id
52id
pppp
qqqq
4
rld,data32
120d
yyyy
rld,rls
rld,@ris
cC,rwd
92sd
12id
AFdc
zzzz
21/10"
TCC
, Privileged instruction - can be executed only in system mode.
"
Number of clock cycles depends on the number of repetitions for n/m"; n = minimum number of clock cycles and
m = number of clock cycles added for each additional repetition of operation. The number of clock cycles for an instruction
which repeats k times is n + (k-1 I,m.
6-80
Table 6-4. Z8000 Instruction Set Object Codes (Continued)
Mnemonic
TCCB
TEST
cC,rbd
adrsx
Object Code
Bytes
Clock
Cycles
AEdc
4Di4
pppp
2
4/6
5
11-15
Mnemonic
Object Code
Bytes
Clock
Cycles
B8d2
OrsO
B8d6
OrsE
4Di6
pppp
4
25
4
25/14*-
4/6
14-18
2
2
4/6
7
11
14-18
2
2
4/6
7
11
9-13
4
7
2
2
4/6
4
7
9-13
4
7
2
2
4
7
TRTIB
@rid,@ris,rw
TRTIRB
@rid,@ris,rw
TSET
adrsx
TSETB
rwd
@rid
adrsx
8Dd6
ODd6
4Ci6
pppp
XOR
rbd
@rid
rwd,adrsx
8Cd6
OCd6
49id
pppp
rwd,data16
090d
yyyy
89sd
09id
48id
pppp
qqqq
rwd
@rid
adrsx
TESTB
8Dd4
ODd4
4Ci4
pppp
2
2
4/6
7
8
11-15
qqqq
qqqq
rbd
@rid
adrsx
TESTL
8Cd4
OCd4
5CiO
pppp
2
2
4/6
7
8
16-20
qqqq
qqqq
rid
@rid
@rid,@ris,rw
TRDB
TRDRB
@rid,@ris,rw
TRIB
@rid,@ris,rw
TRIRB
@rid,@ris,rw
TRTDB
@rid,@ris,rw
TRTDRB
@rid,@ris,rw
*
*"
9CdO
1CdO
B8d8
OrsO
B8dC
OrsO
B8dO
OrsO
B8d4
OrsO
B8dA
OrsO
B8dE
OrsE
2
2
4
13
13
25
qqqq
4
25/14--
4
25
4
25/14**
4
25
rbd,data8
4
25/14-*
rbd,rbs
rbd,@ris
XORB
rwd,rws
rwd,@ris
rbd,adrsx
qqqq
080d
yyOO
88sd
08id
Privileged instruction - can be executed only in system mode.
Number of clock cycles depends on the number of repetitions for n/m**; n = minimum number of clock cycles and
m = number of clock cycles added for each additional repetition of operation. The number of clock cycles for an instruction
which repeats k times is n + {k-1 )·m.
Note 1
DIVL
DIV
Divisor
adrsx
All Others
Not
Aborted
Divisor is
Zero
Dividend
Too Large
Not
Aborted
Divisor
is Zero
Dividend
Too Large
96-100
14-18
26-29
724-728
31-35
52-56
95
13
25
723
30
51
Note 2
MULl
Multiplier
adrsx
All Others
MULTL
Normal
Multiplier
is Zero
Normal
Multiplier
is Zero
71-75
19-22
283 + 7-m - 287 + 7*m
31-35
70
18
282 + 7*m
30
6-81
Table 6-5.
Object Code
OOOd yyOO
OOid
010d yyyy
01id
020d yyOO
02id
030d yyyy
03id
040d yyOO
04id
050d yyyy
05id
060d yyOO
06id
070d yyyy
07id
080d yyOO
08id
090d yyyy
09id
OAOd yyOO
OAid
OBOd yyyy
OBid
OCdO
OCd1 yyOO
OCd2
OCd4
OCd5 yyOO
OCd6
OCd8
ODdO
ODd1 yyyy
ODd2
ODd4
ODd5 yyyy
ODd6
ODd8
ODd9 yyyy
100d yyyy zzzz
toid
11 is
120d yyyy zzzz
12id
13is
140d yyyy zzzz
14id
15id
160d yyyy zzzz
16id
17sd
180d yyyy zzzz
18id
190d yyyy
19id
1AOd yyyy zzzz
zaooo Object Codes
ADDB
ADDB
ADD
ADD
SUBB
SUBB
SUB
SUB
ORB
ORB
OR
OR
ANDB
ANDB
AND
AND
XORB
XORB
XOR
XOR
CPB
CPB
CP
CP
COMB
CPB
NEGB
TESTB
LOB
TSETB
CLRB
COM
CP
NEG
TEST
LD
TSET
CLR
PUSH
CPL
CPL
PUSHL
SUBL
SUBL
PUSH
LDL
LDL
POPL
ADDL
ADDL
POP
MULTL
MULTL
MULT
MULT
DIVL
Instruction
Object Code
Instruction
rbd.data8
rbd.@ris
rwd.data16
rwd.@ris
rbd.data8
rbd.@ris
rwd.data16
rwd.@ris
rbd.data8
rbd.@ris
rwd.data16
rwd.@ris
rbd.data8
rbd.@ris
rwd.data16
rwd.@ris
rbd,data8
rbd,@ris
rwd,data16
rwd,@ris
rbd,data8
rbd,@ris
rwd,data16
rwd.@ris
@rid
@rid.data8
@rid
@rid
@rid.data8
@rid
@rid
@rid
@rid.data 1 6
@rid
@rid
@rid.data 1 6
@rid
@rid
@rid,data1 6
rld,data32
rld,@ris
@rid.@ris
rld,data32
rld.@ris
@rid,@ris
rld.data32
rld.@ris
@rid,@ris
rld.data32
rld,@ris
@rid,@ris
rqd.data32
rqd,@ris
rld,data16
rld.@ris
rqd.data32
6-82
1Aid
1BOd yyyy
1Bid
1CdO
1Cs10dOn
1Cd90s0n
1Dds
1Edc
1FdO
20id
210d yyyy
21id
220s OdOO
22ib
230s OdOO
23ib
240s OdOO
24ib
250s OdOO
25ib
260s OdOO
26ib
270s OdOO
27ib
28dn
29dn
2Adn
2Bdn
2Csd
2Dsd
2Eds
2Fds
300d xxxx
30id xxxx
310d xxx x
31 id xxxx
320s xxxx
32is xxxx
330s xxxx
33is xxxx
340d xxxx
DIVL
DIV
DIV
TESTL
LDM
LDM
LDL
JP
CALL
LOB
LD
LD
RESB
RESB
RES
RES
SETB
SETB
SET
SET
BITB
BITB
BIT
BIT
INCB
INC
DECB
DEC
EXB
EX
LOB
LD
LDRB
LOB
LOR
LD
LDRB
LOB
LOR
LD
LDAR
34id xxxx
LOA
350d xxxx
35id xxx x
370s xxxx
37is xxxx
39s0
3AsO OrdO
3AsO Ord8
3As1 OrdO
3As1 Ord8
3As20rdO
3As20rd8
3As30rdO
LDRL
LDL
LDRL
LDL
LOPS
INIRB
INIB
SINIRB
SINIB
OTIRB
OUTIB
SOTIRB
rqd,@ris
rld.data16
rld.@ris
@rid
rwd,@ris.n 16
@rid.rws,n1 6
@rid,rls
cc.@rid
@rid
rbd,@ris
rwd,data16
rwd,@ris
rbd,rws
@rid,b8
rwd,rws
@rid.b16
rbd,rws
@rid,b8
rwd,rws
@rid,b16
rbd.rws
@rid,b8
rwd.rws
@rid,b16
@rid,n16
@rid,n16
@rid,n16
@rid,n16
rbd.@ris
rwd,@ris
@rid,rbs
@rid,rws
rbd,disp
rbd.ris(disp)
rwd,disp
rwd,ris (disp)
disp,rbs
ridldisp),rbs
disp.rws
ridldisp).rws
rld,disp
rwd,disp
rld.risldisp)
rwd.ris(disp)
rld,disp
rld,ris Idisp)
disp,rls
ridldisp).rls
@ris
@rid,@ris.rw
@rid.@ris,rw
@rid.@ris,rw
@rid,@ris.rw
@rid,@ris.rw
@rid.@ris.rw
@rid.@ris,rw
Table 6-5.
Object Code
3As30rdS
3Ad4 PPPP
3Ad5 pppp
3As6 PPPP
3As7 pppp
3AsS OrdO
3AsS OrdS
3AsS OrdO
3AsS OrdS
3AsA OrdO
3AsA OrdS
3AsB OrdO
3AsB OrdS
3BsO OrdO
3BsO OrdS
3Bs1 OrdO
3Bs1 OrdS
3Bs20rdO
3Bs20rdS
3Bs30rdO
3Bs30rdS
3Bd4 pppp
3Bd5 pppp
3Bs6 pppp
3Bs7 pppp
3BsS OrdO
3BsS OrdS
3BsS OrdO
3BsS OrdS
3BsA OrdO
3BsA OrdS
3BsB OrdO
3BsB OrdS
3Csd
30sd
3Eds
3Fds
40id pppp qqqq
41 id pppp qqqq
42id pppp qqqq
43id pppp qqqq
44id pppp qqqq
45id pppp qqqq
46id pppp qqqq
47id pppp qqqq
4Sid pppp qqqq
4Sid pppp qqqq
4Aid pppp qqqq
4Bid pppp qqqq
4CiO pppp qqqq
4Ci1 yyOO
pppp qqqq
4Ci2 pppp qqqq
4Ci4 pppp qqqq
4Ci5 yyOO
pppp qqqq
Z8000 Object Codes (Continued)
Object Code
Instruction
SOUTIB
INB
SINB
OUTB
SOUTB
INORB
INOB
SINORB
SINOB
OTORB
OUTOB
SOT ORB
SOUTOB
INIR
INI
SINIR
SINI
OTIR
OUTI
SOTIR
SOUTI
IN
SIN
OUT
SOUT
INOR
INO
SINOR
SIND
OTOR
OUTO
SOTOR
SOUTO
INB
IN
OUTB
OUT
AOOB
ADD
SUBB
SUB
ORB
OR
ANOB
AND
XORB
XOR
CPB
CP
COMB
CPB
@rid.@ris.rw
rbd.ioaddr
rbd.ioaddr
ioaddr.rbs
ioaddr.rbs
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
rwd.ioaddr
rwd.iqaddr
ioaddr.rws
ioaddr.rws
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
@rid.@ris.rw
rbd.@rw
rwd.@rw
@rw.rbs
@rw.rws
rbd.adrsx
rwd.adrsx
rbd.adrsx
rwd.adrsx
rbd.adrsx
rwd.adrsx
rbd.adrsx
rwd.adrsx
rbd.adrsx
rwd.adrsx
rbd.adrsx
rwd.adrsx
adrsx
adrsx.dataS
NEGB
TESTB
LOB
adrsx
adrsx
adrsx.dataS
6-83
Instruction
4Ci6 pppp qqqq
4CiS pppp qqqq
40iO pppp qqqq
40i1 yyyy
pppp qqqq
40i2 pppp qqqq
40i4 pppp qqqq
40i5 yyyy
pppp qqqq
40i6 pppp qqqq
40iS pppp qqqq
50id pppp qqqq
51 di pppp qqqq
52id pppp qqqq
53di pppp qqqq
54id pppp qqqq
55si pppp qqqq
56id pppp qqqq
57si pppp qqqq
5Sid pppp qqqq
5Sid pppp qqqq
5Aid pppp qqqq
5Bid pppp qqqq
5CiO pppp qqqq
5Ci1 OdOn
pppp qqqq
5CiS OsOn
pppp qqqq
50is pppp qqqq
5Eic pppp qqqq
5FiO pppp qqqq
60id pppp qqqq
61 id pppp qqqq
62ib pppp qqqq
63ib pppp qqqq
64ib pppp qqqq
65ib pppp qqqq
66ib pppp qqqq
67ib pppp qqqq
6Sin pppp qqqq
6Sin pppp qqqq
6Ain pppp qqqq
6Bin pppp qqqq
6Cid pppp qqqq
60id pppp qqqq
6Eis pppp qqqq
6Fis pppp qqqq
70id ora a
71id OrOO
72is OrOO
73is OrOO
74id OrOO
75id OrOO
76id pppp qqqq
LOL
LOA
TSETB
CLRB
COM
CP
adrsx
adrsx
adrsx
adrsx.data 16
NEG
TEST
LO
adrsx
adrsx
adrsx.data 1 6
TSET
CLR
CPL
PUSHL
SUBL
PUSH
LOL
POPL
AOOL
POP
MULTL
MULT
OIVL
OIV
TESTL
LOM
adrsx
adrsx
rld.adrsx
@rid.adrsx
rld.adrsx
@rid.adrsx
rld.adrsx
adrsx.@ris
rld.adrsx
adrsx.@ris
rqd.adrsx
rld.adrsx
rqd.adrsx
rld.adrsx
adrsx
rwd.adrsx.n 16
LOM
adrsx.rws.n 16
LOL
JP
CALL
LOB
LO
RESB
RES
SETB
SET
BITB
BIT
INCB
INC
OECB
DEC
EXB
EX
LOB
LO
LOB
LO
LOB
LO
LOA
adrsx.rls
cc.adrsx
adrsx
rbd.adrsx
rwd.adrsx
adrsx.bS
adrsx.b16
adrsx.b8
adrsx.b16
adrsx.bS
adrsx.b16
adrsx.n16
adrsx.n16
adrsx.n16
adrsx.n16
rbd.adrsx
rwd.adrsx
adrsx.rbs
adrsx.rws
rbd.ris(rw)
rwd.ris(rw)
rid(rw).rbs
rid(rwl.rws
rld.ris(rw)
rwd.ris(rw)
rld.ris(rw)
rld.adrsx
rwd.adrsx
Table 6-5.
SOsd
Slsd
S2sd
S3sd
S4sd
S5sd
S6sd
S7sd
S8sd
S9sd
SAsd
SBsd
SCdO
SCdl
SCd2
SCd4
SCd6
SCdS
SCs9
8007
SOdO
SOfl
SOd2
SOf3
SOd4
80f5
80d6
SOd8
90sd
91is
92sd
93is
94sd
95id
LOL
LOPS
HALT
IRET
MSET
MRES
MBIT
MREQ
01
EI
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
LOCTL
SC
AOOB
ADD
SUBB
SUB
ORB
OR
ANOB
AND
XORB
XOR
CPB
CP
COMB
LDCTLB
NEGB
TESTB
TSETB
CLRB
LDCTLB
NOP
COM
SETFLG
NEG
RESFLG
TEST
COMFLG
TSET
CLR
CPL
PUSHL
SUBL
PUSH
LDL
POPL
Instruction
Object Code
Instruction
Object Code
77is OrOO
79iO PPPP qqqq
7AOO
7BOO
7B08
7B09
7BOA
7BdO
7C OOOOOOvv
7C 000001vv
7Dd2
70d3
7Dd4
70d5
70d6
70d7
70sA
70sB
70sC
70s0
70sE
70sF
7Fyy
zaooo Object Codes (Continued)
rid(rw),rls
adrsx
96sd
97sd
98sd
99sd
9Asd
9Bsd
9CdO
9EOc
AOsd
Alsd
A2db
A3db
A4db
A5db
A6db
A7db
ASdn
A9dn
AAdn
ABdn
ACsd
AOsd
AEdc
AFdc
BOdO
BldO
Bld7
BldA
B2dO
B2dl yyyy
B2dl yyyy
B2d2
B2d30rOO
B2d4
B2d6
B2d8
B2d9 yyyy
B2d9 yyyy
B2dA
B2dB OrOO
B2dC
B2dE
B3dO
B3dl yyyy
B3dl yyyy
B3d2
B3d30rOO
B3d4
B3d5 yyyy
B3d5 yyyy
B3d6
B3d70rOO
B3d8
B3d9 yyyy
B3d9 yyyy
B3dA
B3dB OrOO
rwd
int
int
rwd,FCW
rwd,REFRESH
rwd,PSAPSEG
rwd,PSAPOFF
rwd,NSPSEG
rwd,NSPOFF
FCW,rws
REFRESH,rws
PSAPSEG,rws
PSAPOFF,rws
NSPSEG,rws
NSPOFF,rws
data8
rbd,rbs
rwd,rws
rbd,rbs
rwd,rws
rbd,rbs
rwd,rws
rbd,rbs
rwd,rws
rbd,rbs
rwd,rws
rbd,rbs
rwd,rws
rbd
rbd,FLAGS
rbd
rbd
rbd
rbd
FLAGS,rbs
rwd
flag
rwd
flag
rwd
flag
rwd
rwd
rld,rls
@rid,rls
rld,rls
@rid,rws
rld,rls
rld,@ris
6-84
AODL
POP
MULTL
MULT
OIVL
OIV
TESTL
RET
LOB
LO
RESB
RES
SETB
SET
BITB
BIT
INCB
INC
OECB
DEC
EXB
EX
TCCB
TCC
DAB
EXTSB
EXTSL
EXTS
RLCB
SLLB
SRLB
RLCB
SOLB
RRCB
RRCB
RLB
SLAB
SRAB
RLB
SDAB
RRB
RRB
RLC
SLL
SRL
RLC
SOL
RRC
SLLL
SRLL
RRC
SOLL
RL
SLA
SRA
RL
SOA
rld,rls
rwd,@ris
rqd,rls
rld,rws
rqd,rls
rld,rws
rid
cc
rbd,rbs
rwd,rws
rbd,b8
rwd,b16
rwd,b8
rwd,b16
rbd,b8
rwd,b16
rbd,n16
rwd,n16
rbd,016
rwd,n16
rbd,rbs
rwd,rws
cC,rbd
CC,rwd
rbd
rwd
rqd
rid
rbd,l
rbd,data16
rbd,data16
rbd,2
rbd,rw
rbd,l
rbd,2
rbd,l
rbd,data16
rbd,data16
rbd,2
rbd,rw
rbd,l
rbd,2
rwd,l
rwd,data16
rwd~data16
rwd,2
rwd,rw
rwd,l
rld,data16
rld,data16
rwd,2
rld,rw
rwd,l
rwd,data16
rwd,data16
rwd,2
rwd,rw
Table 6-5.
Object Code
B3dC
B3dO yyyy
B3dO yyyy
B3dE
B3dF OrOO
B4sd
BSsd
B6sd
B7sd
B8dO OrsO
B8d20rsO
B8d40rsO
B8d60rsE
B8d80rsO
B8dA OrsO
B8dC OrsO
B8dE OrsE
BAsO Ordc
BAs1 OrdO
BAs1 Ord8
BAs20rdc
BAs40rdc
BAs60rdc
BAs80rdc
BAs90rdO
Z8000 Object Codes (Continued)
RR
SLAL
SRAL
RR
SOAL
AOCa
AOC
SBCB
SBC
TRIB
TRTIB
TRIRB
TRTIRB
TROB
TRTOB
TRORB
TRTORB
CPIB
LOIRB
LOIB
CPSIB
CPIRB
CPSIRB
CPOB
LOORB
Instruction
Object Code
Instruction
rwd,1
rld,data16
rld,data16
rwd,2
rld,rw
rbd,rbs
rwd,rws
rbd,rbs
rwd,rws
@rd,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw
rbd,@ris,rw,cc
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw,cc
rbd,@ris,rw,cc
@rid,@ris,rw,cc
rbd,@ris,rw,cc
@rid,@ris,rw
BAs90rd8
BAsA Ordc
BAsC Ordc
BAsE Ordc
BBsO Ordc
BBs1 OrdO
BBs1 Ord8
BBs20rdc
BBs40rdc
BBs60rdc
BBs80rdc
BBs90rdO
BBs90rd8
BBsA Ordc
BBsC Ordc
BBsE Ordc
BCsd
BOdb
BEsd
Cdyy
Oxxx
Ecxx
Fr Ottttttt
Fr 1ttttttt
6-85
LOOB
CPSOB
CPORB
CPSORB
CPI
LOIR
LOI
CPSI
CPIR
CPSIR
CPO
LOOR
LOO
CPSO
CPOR
CPSOR
RROB
LOK
RLOB
LOB
CAlR
JR
OBJNZ
DJNZ
@rid,@ris,rw
@rid,@ris,rw,cc
rbd,@ris,rw,Cc
@rid,@ris,rw,cc
rwd,@ris,rw,cc
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw,Cc
rwd,@ris,rw,cc
@rid,@ris,rw,cc
rwd,@ris,rw,cc
@rid,@ris,rw
@rid,@ris,rw
@rid,@ris,rw,cc
rwd,@ris,rw,cc
@rid,@ris,rw,cc
rbd,rbs
rwd,b16
rbd,rbs
rbd,data8
disp
cC,disp
rb,disp
rw,disp
DATA SHEETS
This section contains specific electrical and timing data for the following devices:
•
•
za001 CPU
Za002 CPU
6-D1
Z8001,Z8002
Composite
AC Timing
Diagram
x
RESET
ii,
~
~
~
->
NYI
->
--®---- ~I~
----®-l ~--®-
·0
~
~
I~
~
WAIT
--@--I
This composite timing diagram does not show actual
timing sequences. Refer to
this diagram only for the
detailed timing relationships
of Individual edges. Use the
preceding illustrations as an
explanation of the various
timing sequences
~
~
c:
I~
r%
~ ~
~
(6~.--
1-L(6~
40V
2.0V
2.0V
:::'V
Clock
Output
Input
Floal
C
X
X
Timing measurements are
made at the follOWing
voltages
High
Low
c,
)
CLOCK
--"
~
~~
---q.
-~
:P-1
1>--
ADDRESS
~
)
@
DATA OUT
1=
-~J-
-®- -r---
'J:
--®-
~-
:x
20
MEMORY WRITE
~
INPUT/OUTPUT
~
- ~r
32
21
:=jt=
NOAM"USY~
BYTElWOAD
K:
-(
"1"' ... _ _
r-~
--®t~
...
,Jo!'~
...
'38
--®
~~
---@-I
J/
X
-
Data sheets on pages 21-02 through 21-04 are reprinted by permission of Zilog, Incorporated.
_-
_-
1-----
y-:: ::::/""-- t>-"
I
6-02
~-
-®I-
~~
t~~
r-®f--@-
~
~(
-r ... __
. . . . r"I
~
'(
~
--@rl
-~~ ~
REA:,~RT;.:
1-
f2'i'
-®-
INTERRUPT~
ACKNOWLEDGE
I> ~i
~
D
/j6'
I---- ~
-® ~~~/ -®
~
®--
~
_
--®I---@-
~
-®-
II
DATA IN
MEMORY READ
O.SV
±O.5V
~
f(
H~-
O.SV
o SV
zaOO1,Z8002
Number Symbol
Parameter
AC
Electrical
Clock Cycle Time
TcC
1
Characteristics
Clock Width (High)
2
TwCh
Clock Width (Low)
TwCl
3
Clock Fall Time
TfC
4
-5--TrC
Clock Rise Time
TdC(SNv)
Clock t to Segment Number Valid (50 pF load)
6
TdC(SNn)
Clock t to Segment Number Not Valid
7
TdC(Bz)
Clock t to Bus Float
8
TdC(A)
Clock t to Address Valid
9
-lO--TdC(Az)-- Clock t to Address Float
TdA(DI)
Address Valid to Data In Required Valid
11
TsDI(C)
Data In to Clock I Setup, Time
12
TdDS(A)
DS t to Address Active
13
14
TdC(DO)
Clock t to Data Out Valid
-15--ThDI(DS)-- Data In to DS t Hold Time
16
TdDO(DS)
Data Out Valid to DS t Delay
17
TdA(MR)
Address Valid to MREQ I Delay
18
TdC(MR)
Clock I to MREQ I Delay
19
TwMRh
MREQ Width (High)
- 20 - - TdMR(A) - - MREQ I to Address Not Active
21
TdDO(DSW)
Data Out Valid to DS I (Write) Delay
22
TdMR(DI)
MREQ I to Data In Required Valid
23
TdC(MR)
Clock I to MREQ t Delay
24
TdC(ASf)
Clock t to AS I Delay
- 25 - - TdA(AS) - - Address Valid to AS t Delay
26
TdC(ASr)
Clock I to AS t Delay
27
TdAS(DI)
AS ! to Data In Required Valid
28
TdDS(AS)
DS t to AS I Delay
29
TwAS
AS Width (Low)
- 30-- TdAS(A) - - AS t to Address Not Active Delay
31
TdAz(DSR)
Address Float to DS (Read) I Delay
TdAS(DSR)
AS! to DS (Read) I Delay
32
TdDSR(DI)
DS (Read) I to Data In Required Valid
33
34
TdC(DSr)
Clock I to DS t Delay
- 35 - - TdDS(DO) - - DS t to Data Out and STATUS Not Valid
36
TdA(DSR)
Address Valid to DS (Read) I Delay
37
TdC(DSR)
Clock t to DS (Read) I Delay
TwDSR
DS (Read) Width (Low)
38
TdC(DSW)
Clock I to DS (Write) I Delay
39
-40--TwDSW--DS (Write) Width (Low)
41
TdDSI(DI)
DS (Input) I to Data In Required Valid
42
TdC(DSf)
Clock I to DS (I/O) I Delay
43
TwDS
DS (I/O) Width (Low)
44
TdAS(DSA)
AS t to DS (Acknowledge) I Delay
-45--TdC(DSA)-- Clock t to DS (Acknowledge) I Delay
DS (Ack.) I to Data In Required Delay
TdDSA(DI)
46
47
TdC(S)
Clock I to Status Valid Delay
48
TdS(AS)
Status Valid to AS t Delay
49
TsR(C)
RESET to Clock t Setup Time
-50--ThR(C)--- RESET to Clock t Hold Time
51
TwNMI
NMI Width (Low)
TsNMI(C)
NMI to Clock t Setup Time
52
53
TsVI(C)
VI, NVI to Clock t Setup Time
54
ThV!(C)
VI, NVI to Clock t Hold Time
-55--TsSGT(C)--SEGT to Clock t Setup Time
56
ThSGT(C)
SEGT to Clock t Hold Time
57
TsMI(C)
Mi to Clock t Setup Time
ThMI(C)
Mi to Clock t Hold Time
58
59
TdC(Mo)
Clock t to Mo Delay
-60--TsSTP(C)--STOP to Clock I Setup Time
61
ThSTP(C)
STOP to Clock I Hold Time
62
TsWT(C)
WAIT to Clock I Setup Time
63
ThWT(C)
WAIT to Clock I Hold Time
64
TsBRQ(C)
BUSRQ to Clock t Setup Time
-65--ThBRQ(C)--BUSRQ to Clock t Hold Time
TdC(BAKr)
Clock t to BUSAK t Delay
66
TdC(BAKf)
Clock t to BUSAK I Delay
67
6-03
MiD
250
105
105
20
400
70
80
0
230
55
190
70
55
330
55
290
70
80
60
0
70
155
80
120
275
160
315
400
960
420
40
180
0
100
140
110
0
70
0
180
0
140
0
70
0
90
0
Max
Unit
ns
2000
2000
ns
ns
2000
20
ns
20--nsns
130
ns
ns
65
ns
100
65--nsns
ns
ns
ns
100
nsns
ns
80
ns
ns
nsns
ns
ns
80
ns
80
nsns
90
ns
ns
ns
nsns
ns
ns
ns
70
nsns
ns
120
ns
95
ns
nsns
ns
120
ns
ns
120--nsns
110
ns
ns
ns
nsns
ns
ns
ns
nsns
ns
ns
ns
120
nsns
ns
ns
ns
nsns
100
ns
100
Z8001,Z8002
Absolute
Maximum
Ratings
Voltages on all inputs and outputs
with respect to GND .......... -0.3 V to + 7.0 V
Operating Ambient
Tempera ture .................. a°e to + 70°C
Storage Temperature ........ -65°C to + 150 °e
Standard
Test
Conditions
The characteristics below apply for the
following standard test conditions, unless
otherwise noted. All voltages are referenced to
GND. Positive current flows into the referenced pin. Standard conditions are as follows:
Stresses greater than those listed under Absolute Maximum Ratings may cause permanent damage to the device.
This is a stress rating only; operation of the device at any
condition above those indicated in the operational sections
of these specifications is not implied. Exposure to absolute
maximum rating conditions for extended periods may affect
device reliability.
V"'.,K
FROM OUTPUT
UNDER TEST
DC
Characteristics
Ordering
Information
[j
GND =
l.j
ooe
::5
aV
TA
::5
=rr!
100',!
L +4.75 V ::5 Vcc ::5 +5.25 V
~ ~~
_
All de pardmeters assume a load capacitance of 100 pF max, except for parameter 6 (50 pF max). Timing references between two
output signals assume a load difference of 50 pF max.
+70 oe
Parameter
Min
Max
Unit
VCH
Clock Input High Voltage
VCC-O.4
VCC+0.3
V
Driven by External Clock Generator
VCL
Clock Input Low Voltage
-0.3
0.45
V
Driven by External Clock Generator
VlH
Input High Voltage
2.0
VCC+0.3
V
-0.3
Symbol
VlL
Input Low Voltage
VOH
Output High Voltage
VOL
Output Low Voltage
IlL
0.8
Condition
V
=
=
V
IOH
0.4
V
IOL
Input Leakage
±10
p.A
0.4 :s VIN :s +2.4 V
IOL
Output Leakage
±10
p.A
0.4 :s VOUT :s +2.4 V
ICC
VCC Supply Current
300
rnA
Part Number
28001 CPU
Z8002 CPU
Temperature
Range
ooe to
ooe to
2.4
-250 p.A
+2.0 rnA
Number
of Pins
Package
+ 70°C
48
CeramiC
Segmented 16-Bit Microprocessor
+70°C
40
CeramiC
Non-Segmented 16-Bit Microprocessor
6-04
Description
Chapter 7
THE MOTOROLA MC68000
The MC68000 microprocessor is Motorola's first 16-bit microprocessor. It is the third of the new generation of these
devices, having been preceded by Intel's 8086 and Zilog's Z8000.
The MC68000 is not program compatible with Motorola's family of 8-bit microprocessors. Motorola has opted for
designing an instruction set which provides maximum power and simplicity rather than compatibility.
The following is a discussion of interesting MC68000 features as compared to similar capabilities of the Z8000 and
8086:
.1)
The MC68000 overlaps the fetching of each instruction's object code with the decoding and execution at
the two prior instructions to obtain a pipeline effect. The Z8000 uses this approach, but only under certain
circumstances. On the other hand, the 8086 performs extensive pipelining using a 6-byte object code
pipeline.
2)
Both the 8086 and the Z8000 family of microprocessors provide methods of operating the devices in a "simple" system configuration or "complex" system configuration. The 8086 accomplishes this within a Single
device by having a number of dual-function pins which serve one function in simple systems and another
function in complex systems. The Z8000, on the other hand, is supplied in two versions: the Z8001 for complex configurations and the Z8002 for simple configurations. The MC68000 is contained in a 64-pin
package and therefore need not attempt to accommodate different complexities of system configurations; it
is always capable of operating in what is, effectively, a "maximum" or "complex" system configuration
mode.
3)
The MC68000 has built-in logic to handle bus access arbitration in multi-CPU configurations. The 8086 and
the Z8000 have equivalent logic.
The MC68000 can directly access up to 16 millio!) (16M) bytes of memory with its 24-bit Address Bus. This
memory space may be expanded to 64M bytes by using the Function Code lines. In comparison, the 8086
can directly address only 64K bytes of memory but can address up to one million bytes using segment
registers. The Z8000 is also limited to 64K bytes of directly addressable memory; however, the Z8001 version can address as many as 48M bytes of memory using internal segment registers and external segmentation in a memory management device.
The MC68000 can be operated in either a "Supervisor" or a "User" mode. Certain privileged instructions can
be executed in Supervisor mode only. Supervisor and User modes also have separate stack pointers. Thus, in
program-intensive applications, systems software (executed in Supervisor mode) can be separated from
applications programs (executed in User mode). The Z8000 series microprocessors provide similar
capabilities. The Supervisor mode of the MC68000 is equivalent to the System mode of the Z8000, while the
User mode of the MC68000 is equivalent to the Normal mode of the Z8000. The 8086 offers no similar operating modes.
.
4)
5)
6)
7)
The MC68000 has seventeen 32-bit registers. Eight of the registers are designated as Data registers and can
be accessed as either 8, 16, or 32-bit registers. The remaining nine registers are designated as Address
registers, with two of these being reserved for use as the stack pointers (Supervisor and User). The Address
registers can be accessed as 16 or 32-bit registers. All of the registers can also function as Index registers. In
contrast. all of the Z8000 registers are 16-bit registers, although they can be paired to operate as 32-bit
registers. The 8086 has only four 16-bit registers plus three separate 16-bit Index registers.
The MC68000 provides separate pins for every data line and address output line. This is possible since the
MC68000 is contained in a 64-pin package and as a result there is no shortage of pin connections. The
Z8000 microprocessors and the 8086 are housed in smaller packages and therefore their data and address
lines must share some pins. Thus the Z8000 and 8086 devices multiplex some of the data and address signals on the same pins, and you must provide external logic to demultiplex these signals.
7-1
The primary source for the MC68000 is:
MOTOROLA SEMICONDUCTOR, INC.
3501 Ed Bluestein Blvd.
Austin, Texas 78721
The MC68000 is manufactured using N-channel HMOS process technology. The device is contained in a dual inline 64pin package. A single +5V power supply is required and all signals are TTL-level compatible.
The MC68000 requires an external clock which can be run at a maximum frequency of 8 MHz. The minimum instruction execution time is four clock periods. The maximum number of clock periods for instruction execution is 158 for
signed division and multiplication.
THE MC68000 PROGRAMMABLE REGISTERS
Figure 7-1 illustrates the registers provided by the MC68000. There are seventeen 32-bit Data and Address
registers, a 32-bit Program Counter (of which only 24 bits are used) and a 16-bit Status register. The most significant difference between the registers provided by the MC68000 and those of other 16-bit microprocessors is that the
Data and Address registers are all 32 bits wide. By comparison, the 8086 and Z8000 microprocessors use 16-bit wide
registers.
The Data registers can be used to handle 8-bit bytes, 16-bit words, or 32-bit long words.
The following illustration shows how the various sized operands are positioned within the
Data registers.
MC68000
DATA
REGISTERS
Data Register
(00-07)
Word Operands - - - - -.......
.....- - - - - - - - - - - - - Long Word Operands
-------------+1
8-bit byte operands occupy bits 0 through 7 of a Data register, while a word operand occupies bits 0 through 15 of a
Data register. A long word operand uses all 32 bits of a Data register. When a Data register is used as a source or
destination operand, only the appropriate low order portion of the register will be altered by the specified operation; the
more significant bits will be unaffected. For example, if you have specified an arithmetic shift left (ASL) instruction with
an operand size of eight bits, then only the least significant eight bits (bits 0-7) of the data register will be shifted: bits 8
through 31 will be unchanged by the instruction execution:
8
31
I
I.
I I I I I
...
- . - - - - - - - - - - - Unchanged
I
7
6
5 4
3
2
1 0 .....-Sit No.
II I I I I II I..
"0"
------------41. ,
To Carry Flag
In addition to being used as the source or destination for instructions the Data registers can also be used as index
registers or data counters.
7-2
3130292827262524232221201918171615141312111098765432 1
I I I I I I I I I I I I I I I
I I I I I I I
O~BitNo.(forall
I I I I I I I IDO
registers)
101
~============~======~====~102
~============~======~====~103
==============~~====~~====~104
~============~======~====~I D5
Data
Registers
~============~~====~~====~'06
__________________________________ ________________ ________________
================~====~~====~
~
~
31
~
~ID7
o
16 15
lAO
~============~============~IA1
~============~~============~'A2
~============~~============~'A3
~============~~============~'A4
~============~~============~IA5
Address
Registers
~====================~~====================~'A6
A7} Stack Pointers
-----------------u;;-Sta;p~Tn;;.----.---·.---------________________
____________ ,___ _
.!u£!'!!i~~.!:!~.!.o~~
2423
I
IProgram Counter
~--------------~------------------15----------------8-7----------------0~
1.._____Sy_s_te_m_B,;"yt_e_____IL..._____U_se_r_B..;,y_te____.....1Status Register
Figure 7-1. MC68000 Programmable Registers
There are seven general purpose Address registers (AO-A6). These registers can handle
MC68000
either 16-bit word or 32-bit long word operands. When you use one of these address registers
ADDRESS
to provide a source operand. either the low order 16 bits will be used (if a word operand has been
REGISTERS
specified) or the entire 32 bits will be used (if a long word operand has been specified). If the
Address register is used as the source for a word operand. then the more significant 16 bits (bits 16-31) will not be
affected. If an Address register is used as the destination operand. however. the contents of the entire register will be
affected. regardless of whether a word or long word operand is specified. If you specify a word destination operand for
an Address register. that word will be automatically sign-extended to 32 bits before it is loaded into the Address
register.
I. I I I
3130292827262524232221201918171615141312111098765432 1 04--BitNo.
I I I I
!.
I I I I I I I I I I
.-
I I I I I I I I I I I r~:~~::)gister
Word Operands
:-
.....- - - - - - - - - - - - - - L o n g Word O p e r a n d s - - - - - - - - - - - - - - - t -
7-3
As we have already pointed out. all of the MC6S000 Data and Address registers are 32 bits wide versus the 16-bit wide
registers of the lSOOO and SOS6. Another significant difference between the MC6S000 registers and those of the SOS6
is the general purpose nature of the MC6S000 registers. This is similar to the approach taken in the lSOOO and provides
the programmer with increased flexibility. Although there are minor differences between the way the Data and Address
(AO-A6) registers handle various data widths. each register type may be used in similar ways. The only dedicated
registers are the Stack Pointer registers (A7. Supervisor and User). the Program Counter and the Status register. Let us
now examine these dedicated registers.
MC68000
The MC68000 can be operated in a Supervisor (or system) mode, or in a User (or normal)
STACK
mode. The state of the S-bit in the Status register determines the mode of operation for the
POINTERS
MC6S000. Supervisor mode will normally be used by operating system software; User mode will
typically be used by application programs. A number of instructions are designated as privileged
and can only be executed when the processor is in Supervisor mode. The Supervisor and User modes also have separate stack pointers as mentioned earlier. As you can see in Figure 7-1, both stack pointers are addressed as
Address register A7. When the MC68000 is operating in the Supervisor state. the User Stack Pointer cannot be
referenced. Conversely. when the MC6S000 is in the User state. the Supervisor Stack Pointer cannot be referenced.
Both the User and Supervisor Stack Pointers operate in the same way: the system stacks are filled from high
memory to low memory. On subroutine calls the Program Counter contents are pushed onto the appropriate system
stack (Supervisor or User). The Program Counter contents will be pulled from the Stack and restored to the Program
Counter on return from subroutines. Since the Program Counter is a 32-bit register, four bytes (two words) of
memory will be required to save the contents of the Program Counter on the Stack. The organization of the Program Counter contents on the System Stack after a subroutine call is illustrated as follows:
Memorv -----,
Address
,
1514131211109
817
6
5 4
3 2 1 0 04--Bit No.
PC low-order word
n
n-2 MSB
LSB
PC high-order word
n - 4
I
- E v e n Byte
Odd B y t e -
Data that is pushed onto the Stack is always written to a word boundary in memory; that is. to a memory location with
an even address. Therefore. when bytes of data must be pushed onto the stack they are written into the high-order half
of the memory word. and the low-order half of that word (corresponding to an odd memory address) will be unchanged.
The MC68000 addresses memory as either 8-bit bytes or as a 16-bit word comprised of two bytes. All words
must be referenced at even address locations. Otherwise. misalignment could occur when the microprocessor
attempts to perform a word operation at an odd-memory address. This same problem exists with any of the 16-bit
microprocessors. but the MC6S000 is the only microprocessor which automatically checks to ensure that all word
references are done at even memory addresses. If a word reference is made to an odd memory address. the MC6S000
begins an exception processing sequence. which will be described later.
The following illustration shows how bytes are organized in memory:
15 14 13 12 11 10 9
2
1 0 4 - - B i t No.
Byte FFFFFE
8
7
6
Byte FFFFFF
5· 4
3
High Memory
Byte FFFFFC
Byte FFFFFD
Byte FFFFFA
Byte FFFFFB
Byte FFFFF8
Byte FFFFF9
Byte FFFfF6
Byte FFFFF7
Byte FFFFF4
Byte FFFFF5
Byte 000006
Byte 000007
Byte 000004
Byte 000005
Byte 000002
Byte 000003
Byte 000000
Byte 000001
7-4
Low Memory
You will note that the first byte in memory (address 000000) occupies the most significant byte half of a memory word.
When words are stored in memory. they are only addressable at even memory addresses. as we have discussed. This
can be illustrated as follows:
15141312 11 10 9
8
7
6
5 4
3 2
1 0 4----Bit No.
Word FFFFFE
High Memory
Word FFFFFC
Word FFFFFA
Word FFFFF8
Word FFFFF6
Word FFFFF4
Word 000006
Word 000004
Word 000002
Word 000000
Low Memory
When 32-bit long words (such as 32-bit addresses) are stored in memory. they occupy two adjacent 16-bit memory
locations or four bytes. The high-order word of the long words is stored at the higher memory location. as illustrated
below:
15 14 13 1 2 11 10 9
8
7
6
5 4
3 2
1 0 " " - - Bit No.
~~_ _ _ _ _ _ !'!2.h..!~!...________
Low-order
LSB
High Memory
~-------------------~---~-----------------
~~-------~~~~--------~
Low-order
LSB
~~---- Long word or address 0------..
LSB
~------------------------------~
Low Memory
The MC68000 provides a 16-bit Status register which is divided into two 8-bit bytes: the
System byte and the User byte. Figure 7-2 shows the bit assignments for the Status
register. The Carry, Overflow, Zero, and Negative bits are the standard ones provided by
most microprocessors.
MC68000
STATUS
REGISTER
The Carry (C) bit is set if there is a carry out of the most significant bit following an addition operation, or if a
borrow is required from the most significant bit during a subtraction. This status bit is also modified by certain shift
and rotate instructions.
The Overflow (V) bit is the exclusive-OR of the carries out of the most significant and next higher-order bits of
the operand following arithmetic operations. The setting of the overflow bit signifies a magnitude overflow since the
result cannot be represented in the specified operand size.
The Zero (Z) bit is set whenever the result otan operation is zero; it is reset otherwise.
The Negative (N) bit is the equivalent of the Sign status bit provided in most microprocessors. The Negative bit is
equal to the value of the most significant result bit following arithmetic operations. If a signed binary arithmetic
operation is being performed. a Negative status of 0 specifies a positive or zero result. whereas a Negative status of 1
specifies a negative result.
The Extend (X) bit is used in multiprecision arithmetic operations. When it is affected by an instruction, it is set
to the same state as the Carry bit.
The three most significant bits (bits 5, 6, and 7) of the User byte of the Status register are not currently
assigned and will always be zero.
7-5
....
System Byte
~~--14
15
13
--~~~------~,
12 11 10 9
8
IT I ISI I
I~
~
..--..
User Byte
'1~--6
5
4~~~--------"
3
2
0 ______ Bit No.
112111 110 I I I IxlNlzlvlCI
"
--
./
j
I
I~
I~
~
~
t
Carry
Overflow
Zero
Negative
Extend
Interrupt Mask
Supervisor/User Mode Select
Trace Mode
Figure 7-2. MC68000 Status Register Bit Assignments
The System byte of the Status register contains status information that is system-related. The User byte, on
the other hand, contains the Condition Code status bits (X, N, Z, V, and C) that are instruction or program
related. Bits in the System byte of the Status register can only be altered when the MC68000 is in the Supervisor mode.
The three least significant bits (bits 8, 9, and 10) of the Status register's System byte form the interrupt mask.
The MC68000 provides seven levels of interrupts. The level of any given interrupt is decoded from the signal's three
interrupt pins, which we will describe later. The interrupt priorities are numbered from 1 to 7, with level 7 having
the highest priority, as shown in the following illustration:
Interrupt
Level
Level 7
Level 6
Level 5
Level 4
Level 3
Level 2
Level'
Level 0
Interrupt Mask
12
1
1
1
1
0
0
0
0
11
,,
,,
0
0
0
0
10
,
,
,
,
0
--
Highest priority
(Non-maskable)
---
Lowest Priority
No Interrupt Request
0
0
0
The level 7 interrupt is nonmaskable and thus cannot be disabled. Level 0 represents a "no interrupt request" condition.
Levels 1 through 6 are the mask-enabled levels. For example, if you set the mask to 100 then only levels 5, 6, and 7 will
be enabled; interrupt levels 1 through 4 are disabled and interrupt requests of those levels will be ignored.
Bit 13 of the Status register is the S-bit which specifies whether the MC68000 is in the Supervisor or User
mode of operation. When this bit is 1, the MC68000 is in the Supervisor mode, and when it is 0 the microprocessor is
in the User mode. Recall that the Supervisor and User modes have their own separate stack pointers; also, certain privileged instructions can only be executed in the Supervisor mode.
The most significant bit of the Status register is the Trace mode (T) flag. If this bit is 0 then the MC68000 operates
normally. If this bit is 1, however, the microprocessor is in the trace mode of operation. The trace mode is the
approximate software equivalent of a hardware implemented single-step mode. After each instruction is executed
in the trace mode, a trap is forced so that a debugging program can monitor the results of that instruction's execution.
7-6
Table 7-1. MC68000 Addressing Mode Summary
Mode
Address Formation
Register Direct Addressing
Data Register Direct
Address Register Direct
EA
EA
Register Indirect Addressing
Register Indirect
Postincrement Register Indirect
Predecrement Register Indirect
Register Indirect with Offset
Indexed Register In':lirect with Offset
EA
(AREGn)
EA
(AREGn); Increment (AREGn)
Decrement (AREGn); EA
(AREGn)
EA
(AREGn) + data 1 6
EA
(AREGn) + (XREGn) + data8
= DREGn
= AREGn
Implied Register Addressing
=
=
=
=
EA = SR. SP. PC
Absolute Addressing
Absolute Short
Absolute Long
EA
EA
= (Next word)
= (Next 2 words)
Program Counter Relative Addressing
Relative with Offset
Relative with Index and Offset
EA
EA
= (PC) + data16
= (PC) + (XREG) + data8
Immediate Data Addressing
Immediate
Quick Immediate
Data
Next word or words
Data inherent in instruction word
=
=
=
EA
Effective Address
DREGn
Any Data Register
AREGn
Any Address Register
XREGn
Any Data or Address Register
used as an Index Register
Contents of
=
=
8-bit offset (displacement)
data8
1 6-bit offset (displacement)
data 16
SR
Status Register
SP
Stack Pointer (User or Supervisor)
PC
Program Counter
=
=
=
=
=
=
() =
MC68000 ADDRESSING MODE SUMMARY
The MC68000 provides six basic types of addressing modes. Variations within these types allow a total of fourteen different modes. as summarized in Table 7-1. At this point. we will look only briefly at the addressing
modes and how they utilize the registers of the MC68000. We will discuss each of the addressing modes in detail
later. just prior to our description of the instruction set.
Most of the addressing modes use the 32-bit Address registers either directly or indirectly to generate the effective
address. The Data registers can be used as sources for addresses in the direct addressing mode. and they can also be
used as Index registers in some of the indirect addressing modes. The indirect addressing modes include post-incrementing or pre-decrementing of an Address register; this capability makes it easy to implement stacks and queues in
memory.
A number of MC68000 instructions use the implied addressing mode; that is. they make implicit reference to either the
Program Counter (PCl. Stack Pointer (SP) or Status Register (SR). For example. Branch. Jump. and Return instructions
will all reference the Program Counter and Stack Pointer during their execution.
Absolute addressing modes do not utilize the Data or Address registers. but instead form the effective address using
data that follows the instruction word in the program. Program Counter relative addressing can use either a displacement or a displacement plus the contents of an Index register to form the effective address. The Index register can be
any of the Data or Address registers.
Most instructions can utilize any of the addressing modes. and address formation is always the same regardless of the
instruction operation itself. These factors do much to enhance the flexibility and power of the instruction set without
making the instruction set difficu It to understand.
7-7
04
03
02
01
DO
AS
UDS
LOS
R/W
OTACK
BG
BGACK
BR
Vee
ClK
GNO
HALT
RESET
VMA
E
VPA
BERR
IPl2
IPL1
IPlO
FC2
FC1
FCO
A1
A2
A3
A4
-
:..
-=-:
:::
-
-::
::
-:::
---
-
-
..
-- -
.::
--
..
:
-
-
::
::
-
-
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
MC68000
64
63
62
61
60
59
58
57
56
55
54
53
52
51
50
49
48
47
46
45
44
43
42
41
40
39
38
37
36
35
34
33
-- :.-
-=--- -=.-
05
06
07
08
09
;, 010
011
012
~ 013
.:~ 014
015
GND
A23
A22
--=.. A21
Vee
A20
A19
A18
.-=. A17
A16
A15
A14
A13
A12
A11
A10
A9
.... A8
A7
A6
A5
-=-
-..
-- -- ..
--------..
:
------.
.
--:.
..
Pin Name
Description
Type
00-015
A1-A23
AS
R/W
UDS. lOS
OTACK
FCO. FC1. FC2
IPLO. IPl1. IPl2
BERR
HALT
RESET
elK
BR
BG
Data Bus
Address Bus
Address Strobe
Read/Write Control
Upper. lower Data Strobes
Data Transfer Acknowledge
Function Code (status) Outputs
Interrupt Requests
Bus Error
Halt Processor Operation
Reset Processor or Reset External Devices
System Clock
Bus Request
Bus Grant
Bus Grant Acknowledge
Enable (Clock) Output
Valid Memory Address
Valid Peripheral Address
Power (+5 V) and Ground
Bidirectional. Tristate
OutPllt. Tristate
Output. Tristate
Output. Tristate
Output. Tristate
Input
Output. Tristate
Input
Input
!nput/Output
Input/Output
Input
Input
Output
Input
Output
Output. Tristate
Input
B'GACi<
E
VMA
VPA
Vee. GNO
Figure 7-3. MC68000 Pins and Signal Assignments
7-8
MC68000 PINS AND SIGNALS
Figure 7-3 illustrates the signals and pin assignments for the MC68000. At this point. we will briefly discuss each
of these signals to provide an overview of how the MC68000 operates. We will defer a detailed discussion of signal and
timing interactions until later in this chapter.
00-015 is the bidirectional 16-bit Data Bus. A 1-A23 is the output 24-bit address bus. Because the MC68000 is
contained in a 64-pin package. the data and address lines need not be multiplexed onto the same pins. as is the case
with the 8086 and Z8000 microprocessors. Note that AO, the least significant bit of the Address Bus, is not output;
this bit is used internal to the MC68000. in conjunction with the data size specification of each instruction. to generate
the UOS and iJ5S signals.
The UOS (Upper Data Strobe) and LOS (Lower Data Strobe) signals determine whether data is being transferred
on either the upper (most significant) byte, the low!!..!!.east significant) byte..J!!' both bytes of the 16-bit data
bus. Table 7-2 defines the significance of the UDS, LOS, and Read/Write (R/W) signals in relation to the data
bus. When ODs is low. data from memory with an even address is accessed and the byte of data is transferred on 08015. When CDs is low. a byte of data located at an odd address is accessed and the transfer occurs on 00-07. When
the MC68000 is transferring a word of data (for example. when fetching an instruction) then both ODS and LOS will be
low and all 16 of the data lines (DO-015) will be used for the transfer.
Table 7-2. MC68000 Oata Bus Control Signal Summary
UDs
LOS
R/W
High
High
-
Low
Low
High
Data bits 8-1 5
Data bits 0- 7
Word Read
High
Low
High
··········;·:.e:·
./·:; .:•.. :.:;:::'"•...::;.:::.q
·:.:i:· i '< /.::'::;::;.,
Data bits 0-7
Byte Read
Low
High
High
Data bits 8- 1 5
00-07
08-015
.,"'.·;r:{"
••.• 'Ji:~'}i""x;"}
""';::,';:'.; .;;~L··; :,~:;'.:;.
c.'
.' .:.''':: ••
Operation
,:,;:.:~{·:Jt"';"C.: k.;L:";;;:<:;:;'..:·::'~::i.:,·Z;I~
".;:;:::,t:.;.•&;.<
t;;/··.·!·:.;.·.·;:i"~;:;·!!.j~:::·.•. . :;;2·;A
Byte Read
Low
Low
Low
Data bits 8-1 5
Data bits 0- 7
Word Write
High
Low
Low
Data bits 0- 7
Data bits 0-7
Byte Write
Low
High
Low
Data bits 8- 1 5
Data bits 8- 1 5
Byte Write
[Z]
No valid data output or input
'.
Table 7-3. MC68000 Function Code Summary
FC2
FC1
FCO
o
o
o
o
o
o
Machine Cycle Type
User data memory access
o
User program memory access
o
o
o
o
1
Supervisor data memory access
o
Supervisor program memory access
Interrupt acknowledge
1>,>,1
Reserved. currently undefined
7-9
Byte Memory with
Even Addresses
Byte Memory with
Odd Addresses
A1-A23 08-015
A 1-A23 00-07
AS - - - - 4
UOS - - - - f
LOS - - - - f
__________
A1-A23~
~~
______________________________
T-~
__
~
00-07
08-015
Figure 7-4. MC6S000 Interface to Memory
The memory interface implied by the UDS and LOS signals is illustrated in Figure 7-4. Byteoriented memory with even addresses will be selected by UOS. and that memory s data lines are
connected to OS-015. LOS references byte memory with odd addresses. and its data will be
applied to 00-07 of the Oata Bus. The AS line. shown in Figure 7-4. is the Address Strobe which is
pulsed low to indicate that a valid data address is being output on the Address Bus (A 1-A23).
MC68000
MEMORY
INTERFACE
DTACK is the Data Transfer Acknowledge input signal. This signal must be asserted by external logic during
every read or write cycle. When the MC6S000 is performing a read or write cycle. it will automatically insert wait
states in the cycle until the OTACK signal is received. This approach is thus the inverse of the logic used by most other
microprocessors: for example. both the Z8000 and SOS6 have a "wait" input which external logic can use to extend a
read or write cycle - if the wait input is not asserted. the read/write cycle will finish normally. The MC6S000 approach
provides for completely asynchronous bus operations that can interface to any type of device regardless of that
device's speed. This approach specifies. however. that all devices in the system must include sufficient logic to
generate the i5TACK signal.
FCO, FC1, and FC2 are the Function Code or processor cycle status outputs. These outputs identify the type of
bus activity currently being performed by the MC68000, as summarized in Table 7-3. The Function Code outputs
are valid whenever AS is true. Five different types of cycles are currently defined: access to either supervisor data
memory, supervisor program memory, user data memory, or user program memory, and interrupt acknowledge
cycles. Whenever the MC6S000 is involved in fetching instructions. it is considered as accessing program memory. All
other memory accesses are identified as data memory accesses. The Function Code outputs could be used to separate
memory into the four different categories - user versus supervisor and program versus data. Thus. by using the FC
outputs an MC68000 system could directly address up to 64 megabytes of memory. with 16 megabytes devoted to
each of the four defined memory categories.
i'PLQ, iPL1, and IPL2 are the interrupt request inputs. These three inputs are decoded internally by the
MC68000 to determine the priority level of the interrupt request. You will recall from our earlier discussion of the
Status register that there is a 3-bit interrupt mask which determines what level of interrupt request will be permitted.
When all three interrupt inputs are low. a non-maskable interrupt (level 7. which is the highest priority) is present. This
level is always recognized by the MC6S000. When all three of the interrupt inputs are high. it indicates that no interrupt
is being requested.
BERR is the Bus Error input. When this signal is low the MC68000 performs a sequence (exception processing
sequence) similar to that which it executes in response to an interrupt request. The purpose of the ftliI!i signal
is to inform the MC68000 when an external device has not responded (using the DTACK input) within an
expected amount of time during a read or write operation. Since the data transfer handshaking approach used by
the MC6S000 requires all external devices to actively respond to every data transfer. the system should include a
mechanism to ensure that the processor is not hung up indefinitely by a device that fails to respond. Thus external logic
should be provided to monitor bus activity and which would utilize the BERR signal to inform the MC6S000 of a "failure
to respond" condition. This logic would separate the preceding cause of a bus error from other causes. such as might
7-10
be generated with a Memory Management Unit (MMU). The MMU would generate BERR if an attempt was made to
access protected memory.
As we have already mentioned. the reaction of the MC68000 to the Bus Error input is similar to the interrupt request
response. We will describe this response. termed "exception processing." in detail later in this chapter. Essentially.
exception processing causes processor status information to be saved. and then allows the processor to execute a program to analyze the cause of the error. The MC68000 also provides a hardware-oriented response to a bus error: if the
HALT signal is asserted in conjunction with the BERR signal. the MC68000 will automatically retry the bus
cycle that produced the error.
The HAi:T signal performs several functions. As we mentioned in the preceding paragraph. it can be used in con·
junction with the BERR signal to initiate rerunning of bus cycles that produced bus errors. When used alone. it
places the MC68000 in a Halt state where the processor is essentially inactive until the HALT signal is negated. This is
the familiar Halt function provided by most microprocessors.
The HALT signal is also used in conjunction with the R'E'SEi" signal to intialize the MC68000. One unusual aspect
of the RESET signal is the fact that it is also an ouput signal; the MC68000 provides a RESET instruction which,
when·executed. causes a low-going pulse to be output on the REffi pin. Thus. you can execute a RESET instruction and use it to initialize other devices in the system without resetting the processor.
HALT. like iiE'S'ET. is an output signal. If the processor ceases executing instructions - for example. if a double
bus fault condition occurs - the MC68000 will output iiAi:i" low. External logic can be then used to detect this
potentially catastrophic condition.
ClK is the single TTL-level compatible clock from which all MC68000 internal timing is derived.
iiR (Bus Request).
BG (Bus Grand, and BGACK (Bus Grant Acknowledge) ar,e all bus arbitration signals. These
signals are used in systems where other devices. such as DMA controllers on other processors. require control of the
System Busses. External devices request access to the System Bus by asserting the BR input. The MC68000 will
then always relinquish the bus after it has completed the current bus cycle. It will also output Bus Grant (BG) low to
let the requesting device know that the bus will become available at the end of the current cycle: However. as
we will see v:..hen we discuss the bus arbitration timing in detail. external devices or logic must monitor more than just
the Bus Grant signal to determine when the bus will actually be available. The Bus Grant Acknowledge (BGACK) signal must be input to the MC68000 by the device requesting the bus once that device takes control of the bus.
BGACK must be held low until the device has completed its bus access operations. Thus BGACK is essentially a "bus
busy" signal that lets the MC68000 (and other devices in the system) know that the bus is unavailable.
The next three signals - E. VPA. and VMA - are provided so that the MC68000 can be easily interfaced to the
standard and widely available 6800 family devices. 6800-based systems use a synchronous method of effecting
transfers of data throughout the system. To accomplish this a system clock Enable (E) signal must be distributed to all
6800 devices in the system so that all relevant data transfers may be synchronized to this clock signal. Thus the Enable (E) signal provided by the MC68000 is the equivalent of the 6800 E signal. The frequency of E is equal to onetenth that of the ClK input to the MC68000: the period for E is equal to 10 ClK periods - E is low for six ClK cycles
and is high for four ClK cycles.
The Valid Peripheral Address (VPA) signal is used by 6800-type devices in the system to inform the MC68000
that a 6800-type data transfer is required. You must provide address decoding logic in the system that determines
when a 6800-type device is being accessed and that generates the WA signal. When the MC68000 receives the
VPA signal, it alters the data transfer timing so that it is synchronous with the Enable (E) signal. The MC68000
will then output the Valid Memory Address (VMA) signal at the appropriate time. VMA is another 6800-type signal and will only be output if the VPA input signal has been asserted at the beginning of a data transfer operation. We
will defer a detailed discussion of these three signals until later when we describe interfacing between the MC68000
and the 6800-family devices.
MC68000 TIMING AND BUS OPERATION
The basic timing for the MC68000 is quite straightforward: instruction execution consists of a combination of
internal cycles and bus access cycles. The total number of clock cycles required for each instruction is defined
in the instruction set summary tables later in this chapter. The number of clock cycles required to perform operations internal to the MC68000 are of little interest to other devices in the system since these operations are transparent
to external logic. It is only when the MC68000 requires access to the system bus for such operations as instruction
fetching. operand fetching. and operand storing that external devices become involved with MC68000 timing.
The MC68000 uses memory mapped I/O. Therefore, bus accesses for data transfers between the MC68000 and
7-11
I
50
I
51
I
52
I
53
I
54
55
I
56
I
57
I
50
I
51
I
52
I
ClK
A1-A23
R/W
08-015
00-07
FCO-FC2
Figure 7-5. MC68000 Read Word Timing
memory are the same as for those between the MC68000 and I/O devices. Data transfers are defined as either
read or write operations, with the transfer of data into the MC68000 defined as a "read" and the transfer of data from
the MC68000 to external logic defined as a "write."
Figure 7-5 illustrates the timing for a read word operation. For purposes of the following timing
discussions. each clock period is sub-divided into two states.
MC68000
READ
TIMING
During state 0 (SO) of the read word cycle. the address and data busses are in the high impedence
state - the MC68000 is not using the System Bus at this point. Address information for the
memory or I/O location is output at the beginning of state 1 (51) on the Address Bus (A 1-A23L Processor cycle status
information is also output at this point on the FCO-FC2 pins. The Address Strobe (AS) signal is asserted at the beginning
of state 2 and can be used by external logic to latch the information on the Address Bus. Simultaneously. the Upper
Data Strobe (05'$') and Lower Data Strobe (Ci5$) signals are asserted to enable selection of both the most significant
byte and least significant byte of a 16-bit word. You will note that these signals are not actually data "strobes" since
there is no data ready to be input or output at this point; it is more accurate to think of them as memory select signals
selecting the upper and/or lower byte of a 16-bit memory word. R/W is normally asserted. so this output does not
change during a read cycle.
The MC68000 now waits for the addressed memory or I/O device to present its data on the Data Bus. When the data is
ready. the external device must assert Data Acknowledge (DT ACK) to the MC68000. The MC68000 expects DT ACK and
the requested data to be present by state 5 (S5). If DTACK is not present by 55. Wait states (SW) will be automatically
7-12
inserted into the read timin~y~as illu~ted in Figure 7-6. Once DTACK is true. the read cycle continues with S5. At
the end of state 6 (S6). the AS. UDS. and lOS signals are negated. At this point the incoming data on DO~Q15 is lat<;:hed
into an internal MC68000 register. External devices can use the negative-to-positive transition ot" AS. 'ODS. or CDS as
the indication that they can remove data from the Data Bus. TheMC68000 maintains the address information and function code information through the end of state 7 (S7) to allow for signal skew within the system. Note that when the
external device senses that the MC68000 has captured the data from the Data Bus (by sensing the high-going transition of AS. DDS. or lOS) that device must return DT ACK high immediately so that it does not interfere with the beginning of the next bus cycle.
If you refer to the Wait state insertion that can occur during read operations. as illustrated in
Figure 7-6. you will see that the Wait states occur between state 4 and state 5. The
MC68000 will maintain valid address output on the address Bus and will hold AS: ODs. and IDS
low during any Wait states for as long as necessary until 5TACK is asserted. You should note that
there will always be an even number of Wait states inserted; all MC68000 operations are based on a complete elK
cycle and there are two "states" per ClK cycle.
I 50 I
51
I
52
I 53 I
54
5W
I 5W I 5W I 5W
S5
I
S6
elK
A1-A23
R/W
00-015
Figure 7-6. MC68000 Wait States During Read Operations
7-13
I 57 I SO
ClK
A1-A23---{~______________~~~______________~~~______________
_--,
\....
\. . .__r
R/W
08-015 -------PV><.,.
DO-07----------------------------~~
H. ._____...)~------
FCO-FC2 - - - {..._ _ _ _ _...
Figure 7-7. MC68000 Read Byte Timing
Timing for a read byte operation is illustrated in Figure 7-7. This figure shows first an even data byte and then an
odd data byte being read by the MC68000. As you can see, the only difference between this timing and that illustrated
for a read word operation in Figure 7-5 is that only 0i5S or
is asserted and only eight lines of the data bus are
utilized when you are reading a byte: UOS is asserted and data is on lines 08-015 when reading a byte located at an
even address and IDS is asserted and data is on lines 00-07 when reading a byte located at an odd address. You
should not be misled by Figure 7-7 into thinking that the MC68000 always reads two consecutive bytes - an even
byte and an odd byte. We have simply shown these two read operations consecutively to illustrate timing for both.
Again, if the MC68000 requires a word of data, it will utilize the entire 16-bit Oata Bus and read the full word in one
operation.
rns
7-14
Timing for a write word operation is illustrated in Figure 7-S. As was the case with read operaMC6S000
tions. the address for the memory location or I/O device is output at the beginning of 51 along
WRITE
with the appropriate function code indicating the current type of processor bus cycle. If the Data
TIMING
Bus was utilized by the MC6S000 in the preceding cycle. the processor returns all of t.h.e data outputs to the high impedence state during 51 and then asserts the Address Strobe (AS) signal and outputs the Read/
Write (R/W) Signal low. Once again. AS can be used to latch the address externally. and the R/W Signal indicates to
memory or I/O devices that the MC6S000 will be placing data onto the Data Bus. No further Signal activity occurs until
the MC6S000 outputs the data on 00-015 at the beginning of state 3 (53), The Upper and Lower Data Strobe signals
(UDS. LOS) are asserted at the beginning of state 4 (54). During write operations. these two signals can be used as
"strobe" signals since they indicate that the data on the Data Bus is valid. If the write operatio~roceed unimpeded. external logic must respond to the data strobe signals by asserting the Data Acknowledge (DT ACK) Signal by the
beginning of state 7 (57). If DTACK is not true by the beginning of 57, Walt states are automatically inserted by
the MC6S000, as illustrated in Figure 7-9. This "slow write" operation is the same as was illustrated for read
operations except that the Wait states are inserted at a different point in the cycle.
The MC6S000 outputs the data on 00-015 through the entire write operation. The Address Strobe (AS) and data
strobes (UDS. LOS) are negated at the beginning of state 9 (59) and the Read/Write (R/W) signal is returned high at the
end of S9. At that point. the Address Bus. Data Bus. and Function Code outputs are all returned to their high impedance
state to free the System Bus for other uses. The external memory or I/O device that was accessed by the write operation
I
50
I
51
I
52
I
53
I
54
I
55
I
56
57
I 58 I
ClK
A1-A23
AIW
08-015
00-07
FCO-FC2
Figure 7-S. MC6S000 Write Word Timing
7-15
59
I
50
I
elK
A1-A23
R/W
00-015 - - - - -.....
Figure 7-9. MC68000 Wait States During Write Operations
must release the Data Acknowledge (DT ACK) signal after it has detected the positive-to-negative transition of the
address or data strobe signals. This ensures that a subsequent bus cycle will not be impeded.
Timing for write byte operation is illustrated in Figure 7-10. As you can see, the only difference between this operation and the write word timing illustrated in Figure 7-8 is the fact that only ODs or L5S is output while a byte is being
written.
7-16
MC68000 READ-MODIFY-WRITE TIMING
The read-modify-write cycle provided by the MC68000 is unusual among microprocessors, although it is frequently provided by minicomputers. The MC68000 uses the read-modify-write cycle only during the execution
of the Test and Set (TAS) instruction. This instruction reads a byte of data. sets condition codes according to the
contents of that byte. sets bit 7 of the byte. and then writes it back into memory. The T AS instruction is intended to be
used as a means of providing "safe" communication between microprocessors in a multi-processor system. Safe communication is ensured with the T AS instruction since the read-modify-write cycle is non-interruptable.
ClK
A1-A23--(..._ _ _ _ _ _ _~H'-
As~..._ _ _ _~1
_______~}\...._------1
R/W
08-015
----04
00-07----------------------------~
. . _______...J}-
FCO-FC2--{..._ _ _ _ _ _ _~H
Figure 7-10.. MC68000 Write Byte Timing
7-17
Figure 7-11 illustrates the timing for a read-modify-write cycle. As you can see, it simply consists of the readbyte cycle followed by a standard write-byte cycle. There is one intervening clock period (58, 59) between the
read and write cycles and it is during this interval that .the byte of data is modified internally for the subsequent
write. Just as was the case with standard read and write. external logic must reply with OT ACK at the proper time or
else Wait states will automatically be inserted to lengthen the read or write operations.
Note that in Figure 7-11 we have shown that either UOS or LOS will be asserted during the read-modify-write operation. Th is is because the T AS instruction always operates on a byte of data. never on a word of data.
Read
Modify
Write
CLK
A1-A23
UOS or
lOS
R/W
00-07or ____________
FCO-FC2
1X~~
~~~__.....I
08-015
LJ
}-
~-------------------------------Figure 7-11. MC68000 Read-Modify-Write Timing
7-18
The MC68000 RESET OPERATION
The MC68000 has an asynchronous reset input. You reset the microprocessor by holding the RESET and HALT signals
low for at least 100 milliseconds. After the RESET and HALT signals are returned high, the MC68000 executes the
following 'operations:
1)
The MC68000 reads the first four words from memory (bytes 000000 through 000007) and uses the contents of these locations to load the System Stack Pointer (SSP) and Program Counter (PC). The contents of
these eight bytes from the beginning of memory are used as follows:
Memory
000000
2
3
o
31
SSP~I__________~____2____~____3____~~____4____~
4
5
6
7
o
31
peL
5
6
8
7
8
.......-
---
000001
000002
000003
000004
000005
000006
000007
000008
2)
The interrupt mask in the Status register is set to all ones so that only level 7 interrupts will be enabled. No
other registers are affected by the reset operation: therefore, when a reset is performed after applying power
to the MC68000, all registers except SSP, PC, and the Status register will contain indeterminate values.
3)
Program execution begins, with the first instruction being fetched from the location indicated by the value
loaded into the Program Counter.
The sequence we just described is the typical externally-initiated reset operation similar to that provided by
most microprocessors. You will recall, however, that the RESET pin is bidirectional; when the MC68000
executes a Reset instruction, a low-going pulse is sent out on the RESET pin. This software RESET pulse is low for
124 ClK cycles. This instruction has no effect on the internal state of the MC68000, therefore none of its internal
registers are affected. In this case, the RESET signal is being used to reset all other devices within the system under the
control of the MC68000.
THE MC68000 HALT STATE
The MC68000 can be forced into a Halt state, at which time its Address Bus, Data Bus, and Function Code outputs (FCO-FC2) are placed in the high-impedance state. This state is similar to the Hold state of the 8086 and the
Stop state of the Z8000. The Halt state can be used to disable the MC68000 and thus free the System Busses for such
activities as direct memory access or mUlti-processor operations. However, since the MC68000 includes an efficient
bus arbitration system, it is more likely that the Halt state will be used to implement a hardware single-step mode.
Figure 7-12 illustrates the timing for the Halt operation. If the MC68000 is in the middle of a bus cycle when the
HALT signal is input low, the bus cycle continues to its normal completion. At the end of the cycle the Address Bus,
Data Bus, and FCO-FC2 signals are all placed in the high impedance state and the MC68000 halts. While it is in this
halted condition, the processor does nothing - it merely waits for the .HAlT signal to return high. Note that the
MC68000 provides no halt acknowledge indication to external logic.
However, while the MC68000 is in the Halt state, its bus arbitration circuitry still operates. Since the MC68000
wiU not be using the bus while it is halted, any bus request made to the MC68000 will be granted immediately. We will
defer a detailed discussion of the bus arbitration circuitry until later.
When the HALT signal is returned high, the MC68000 exits the Halt state within two clock cycles and can then begin
another bus cycle.
The execution of most MC68000 instructions requires multiple bus cycles to fetch the instruction and operands and,
possibly, to store results of the instruction. Since the MC68000 will respond to the HALT input upon completion of
any bus cycle, the halt sequence can occur between two instructions or in the middle of a single instruction.
Therefore, if you are using the HID input to implement a single-step mode of operation, you will be singlestepping by bus cycles rather than single-stepping by instructions. If you want to single-step by instructions, you
7-19
Completion of
Halt State. Address Bus. Data
Initiate next
current bus cycle ---i+--Bus and FCO-FC3 put into-"'~--bus cycle
high impedance state
CLK
A1-A23
R/W
00-015
FCO-FC3
Figure 7-12. MC68000 Halt State Timing
must use the Trace function of the MC68000. This function is implemented by setting the T-bit in the Supervisor byte
of the Status register. We will describe the Trace operation in detail later.
The HALT signal is bidirectional and will be asserted by the MC68000 if it initiates a Halt state rather than having external logic cause the Halt. The MC68000 will automatically enter the Halt state if there is a double-bus fault
(we will discuss bus errors and double-bus faults in detail later). If the MC68000 has automatically entered the Halt
state, the processor will output HALT low and remain in this halted condition until an externally initiated reset operation is performed using RESET. Thus, when HALT is output low by the MC68000, it indicates a catastrophic failure.
THE MC68000 STOP STATE
Following execution of the STOP instruction, the MC68000 microprocessor will enter a Stop state. The STOP
instruction is permitted only when the MC68000 is operating in the Supervisor mode as indicated by the S-bit in the
Status register. The Stop state is similar to the HALT state which we just discussed, since the microprocessor essentially does nothing while in this state. When the STOP instruction is executed, the Status register is loaded with a new
value contained in the instruction. Next. the Program Counter is advanced to point to the next instruction and the
MC68000 stops.
No special signal or status is output by the MC68000 to identify that it is in the Stop state. The Stop state is ended by
one of the exception conditions such as an interrupt request or a RESET. When an exception condition is detected by
the MC68000, it leaves the Stop state and will process the exception condition.
7-20
THE MC68000 BUS CYCLE RERUN TIMING
As we mentioned earlier, the MC68000 can respond in two ways to a System Bus error, indicated by the assertion of iEiiR. It can perform exception processing (which we will describe later), or it can attempt to rerun the
bus cycle which caused the bus error indication. If BERR is asserted by itself, then the exception proces~~.~Jor software) method of handling the bus error is taken. However, if the BERR signal is accompanied by the HALT signal
then the MC68000 recognizes this as a request to rerun the bus cycle.
Figure 7-13 illustrates the timing for the bus cycle rerun operation. In this figure, we have shown a write cycle in
progress, with the MC68000 waiting for the external device to respond with DT ACK so that the cycle can be completed.
Instead of the expected acknowledge signal. external logic forces both the BERR and HALT signals low to indicate that
the cycle was not successfully completed and that the MC68000 should rerun the cycle.
The MC68000 proceeds to complete the cycle that was in progress and then enters the HALT state. The Address Bus,
Data Bus, and Function Code outputs are all placed in the high impedance state and the microprocessor remains halted
until both BERR and HALT are negated. Note that ~ should be negated before HALT is negated to prevent the
MC68000 from interpreting the isolated BERR signal as another bus error - one that is expected to be handled in software. After HALT returns high, the MC68000 will proceed to repeat the cycle that was in progress when the rerun
request was received; i.e., the same address, data, and function code information that was used in the previous bus
cycle will be repeated.
Write Cycle N
Halt
Rerun Write Cycle N
I~I~I~I~I~I~I~I~I~
ClK
A1-A23
~
Address N
l
l
Address N
AS
UOS/lDS
R/W
l
l
OTACK
00-015
FCO-FC2
l
l
l
Function N
l
Oata N
Oata N
Function N
BERR
HALT
Figure 7-13. MC68000 Rerun Bus Cycle Timing
7-21
Figure 7-13 shows the successful completion of the rerun cycle with i5TACR being received in the expected
interval. Of course, this will not always be the case - the attempt to rerun the bus cycle might also result in a bus
error. External logic can continue to request that the cycle be rerun an unlimited number of times, using the combination of BERR and HALT. You should note, however,· that if you are using the software exception processing
method of handling the bus error (BERR asserted alone without HALTl. then two successive bus errors are treated as a
catastrophic error and the MC68000 will automatically enter the Halt state and remain there until reset.
If the MC68000 is performing a read-modify-write cycle and a bus error is encountered, it will not rerun the
cycle. This is done because the read-modify-write is only used during the Test and Set (TAS) instruction. The nature of
this instruction demands complete execution cycle integrity, which might be violated if any of the bus cycles were
repeated. If external logic requests a rerun of the read-modify-write cycle, the MC68000 will instead perform the bus
error exception processing routine, which we will describe later.
MC68000 BUS ARBITRATION LOGIC
The bus arbitration logic provided by the MC68000 is straightforward. The MC68000 does not prioritize
requests for bus accesses by external devices. The processor assumes that it is the lowest priority device in the
system since it always grants bus access to any requesting device so long as the processor is not currently
using the bus itself. Thus the MC68000 allows other devices to utilize the bus between instructions and between bus
cycles of a single instruction. Since there is no built-in arbitration there should be some external bus arbitration logic in
a system of any complexity to prioritize requests for the System Bus so that a high priority device is not superseded by
low priority devices.
There are three signals associated with the bus arbitration logic: Bus Request (BR), Bus Grant (aG)' and Bus
Grant Acknowledge (BGAcK). When the MC68000 is using the System Bus without competition, the input signals BR and BGACK - will be inactive and the BG output will be negated.
Figure 7-14 illustrates the timing for the bus arbitration performed by the MC68000. Bus arbitration commences when an external device pulls the BR input low. When the MC68000 receives a bus request it will respond
by asserting BG one ClK period later. The only exception to this immediate response is when the MC68000 is in the
initial stages of a bus cycle but has not yet asserted AS. In this case the MC68000 waits until one elK period after AS
has been asserted before it asserts BG; the response time in this case will be a maximum of three ClK periods.
Obviously, the Bus Grant signal does not indicate that the bus is available for use by the requesting device at
that point - the MC68000 may still be using the bus to complete its current bus cycle. Therefore the device
requesting the bus must monitor several other signals to determine when the bus is actually available for its
use. First. the external device must wait until AS is negated, indicating that the MC68000 has completed the current
bus cycle. The device requesting the bus must also wait until the DTACK signal is negated, since this indicates that the
device involved in the current MC68000 cycle is no longer using the bus. However, in some systems it may not be
necessary to monitor the DT ACK signal. This is the case when system timing is such that you are always assured that
all external devices will be off the bus when AS is negated. Lastly, the requesting device must check the state of the
BGACK signal. If this signal is true, it indicates that some other device in the system has already been granted use of the
System Bus and has not yet finished with it. Conversely, if BGACK is false, then the System Bus will be available for use
at the end of the current cycle.
After all of the signal conditions we have described are met, the device requesting the bus must assert BGACK.
This informs the MC68000 that the requesting device has ~aken control of the bus. You will note in Figure 7-14
that the MC68000 does not wait for the BGACK signal before it relinquishes control of the bus: the Address and
Data Busses, the Function Code outputs, AS, UDS, lDS, and R/W are all placed in the high impedance state as soon as
the MC68000 has completed the bus cycle that was in progress when the bus request was received. The device that is
using the bus must hold BGACK low for as long as it requires the bus. While an external device has control of the bus,
external logic should prevent bus conflicts by monitoring BGACK; at this point the behavior of BR and BC is unimportant. However, the device using the bus should negate its BR before negating BGACK to avoid an incorrect bus request.
The MC68000 will maintain its output lines in the high impedance state until BGACK is negated, indicating that the
external device is through with the bus. At that point the MC68000 is free to initiate another bus cycle. Note that if
another bus request is pending at that point the MC68000 will acquiesce to that bus request immediately without performing any bus cycles itself.
7-22
MC68000 EXCEPTION PROCESSING LOGIC
All of Motorola's literature on the MC68000 refers to "exception processing" when discussing what we
usually describe as the interrupt system in other microprocessors. They have chosen to use this nomenclature
since the events that can cause "interrupts" in the MC68000 cover a much broader range than those usually
associated with an interrupt request in a typical microprocessor. We will also use the "exception processing"
nomenclature.
MC68000
MC68000 initiates '-------t....--System bus available for use by external devices---+t---completes
a bus cycle
bus cycle
ClK
A1-A23
R/W
00-015
FCO-FC2
Figure 7-14. MC68000 Bus Arbitration Timing
7-23
The MC68000 provides extensive exception processing logic. This logic is similar to that provided by the 8086
and Z8000 in that a jump vector table is used to transfer program control to the appropriate handler program
whenever an exception occurs. The biggest difference between the MC68000's logic and that of the Z8000
and 8086 is that the number of events that can generate an exception in the MC68000 is greater than the number of events that cause interrupts in Z8000. In addition, the MC68000 provides a 7-level priority structure for
external interrupt requests.
Before proceeding to describe the exception processing system, let us discuss the operatMC68000
ing modes of the MC68000, since these affect exception processing. As 'We mentioned preOPERATING
viously. the MC68000 can operate in either a Supervisor mode or a User mode. When the
MODES
MC68000 is reset using the RESET input. it starts operating in the Supervisor mode. The processor remains in Supervisor mode until one of the following instructions is executed: Return from Exception
(RTEL Move to Status Register (MOVE word to SR), AND Immediate to Status Register (ANDI word to SRL and
Exclusive OR Immediate to Status Register (EORI word to SR). None of these instructions automatically causes the transition to User mode of operation - rather. they are capable of changing the state of the S-bit in the Status register. If
one of these instructions resets the S-bit. the MC68000 will begin operating in the User mode.
Once the MC68000 is operating in the User mode, the only thing that can cause a transition back to the Supervisor mode is an exception. All exception processing is performed in Supervisor mode regardless of the current setting of the S-bit of the Status register. When the exception processing has been completed. the Return from Exception
(RTE) instruction allows return to the User mode.
A number of instructions. designated as "privileged." are reserved for the Supervisor mode. An attempt to execute one
of these instructions in the User mode results in a "privilege violation" which is one type of exception. We will discuss
these instructions and the privilege violation response later in this chapter.
MC68000 EXCEPTION TYPES
Exceptions originate in a variety of ways which can be divided into two general categories:
1)
Internally generated exceptions that result from the execution of certain instructions. or from internally
detected errors.
2)
Externally generated exceptions which include bus errors. reset. and interrupt requests.
The response of the MC68000 to the various types of exceptions is similar. Before we describe this response. let us look
at the sources of exceptions since they go well beyond those provided by other microprocessors.
The internally generated exceptions to which the MC68000 responds can be further subdivided into three categories: internally detected errors, instruction traps, and the Trace
function.
MC68000
INTERNALLY
GENERATED
EXCEPTIONS
The following are the internally detected errors which will cause the MC68000 to initiate
exception processing:
1)
2)
3)
Addressing errors. Whenever the MC68000 attempts to access word data. long word data. or an instruction at an odd address. this is an address error since all such accesses must be on even address boundaries.
Privilege violations. Again. some instructions are reserved for use only in the Supervisor mode. Exception
processing will be initiated if you attempt to execute any of the following instructions when in the User
mode: STOP. RESET. RTE. MOVE to SR. AND (word) Immediate to SR. EOR (word) Immediate to SR. OR
(word) Immediate to SR. MOVE USP.
Illegal and unimplemented opcodes. If an instruction is fetched whose bit pattern is not one of the defined
instruction bit patterns for the MC68000. exception processing will be initiated. Two bit patterns are
defined as unimplemented rather than illegal: if bits 15-12 are 1010 or 1111. these are treated as unimplemented instruction opcodes. If these opcodes are fetched. special exception processing is initiated which
can allow you to use these unimplemented instructions in your own software.
Instruction traps are exceptions which are caused by the execution of instructions in your program. There is a
standard TRAP instruction which is similar the Z8000 System Call instruction. There are four other instructions TRAPV, CHK, DIVS, and DIVU - which will cause exception processing to be initiated if certain conditions. such
as arithmetic overflows or divide by zero. are detected.
The third type of internally generated exception occurs when the MC68000 is operating with the Trace function. If the T-bit in the supervisor portion of the Status register is set. exception processing will be performed after each
instruction. The Trace function is used for program debugging since you can analyze. by stepping through the program.
the results of each instruction's execution.
7-24
There are three different types of externally generated exceptions:
1)
2)
3)
Bus errors. When the BERR signal is pulled low by external logic (while HALT is high)
exception processing is initiated.
Reset. When the RESET signal is asserted by external logic, exception processing is
initiated.
Interrupt request. This is the most familiar form of exception processing and is initiated
via the three interrupt request lines (iPLO, IPL 1, and IPL2).
The different types of exceptions have different priorities, and processing of an exception
depends on its priority. The following table lists the types of exceptions according to their
relative priorities, and also defines when processing of each type begins.
Priority
Highest
Lowest
Exception Source
MC68000
EXTERNALLY
GENERATED
EXCEPTIONS
by external logic
MC68000
EXCEPTION
PRIORITIES
Exception Processing Response
RESET
BERR (Bus Error)
Address Error
Abort current cycle, then
process exception
Trace
Interrupt Request
Illegal/Unimplemented Opcode
Privilege Violation
Complete current instruction, then
process exception
TRAP, TRAPV
CHK
Divide-by-zero
Instruction execution initiates
exception processing
The highest priority types of exceptions are Reset. Bus Error, and Address Error. Any of these exceptions will cause
immediate termination of the current instruction, even within a bus cycle. The next group of exceptions - trace, interrupt requests, illegal/unimplemented instructions, and privilege violations - allow completion of the current instruction before initiating exception processing. Note that interrupt requests include an additional prioritization which we
discussed earlier. The lowest priority of exceptions are those that are caused by trap-type instructions. These instructions can initiate exception processing as part of their normal execution. All of the instruction trap exceptions have
equal priority since it is impossible for two of them to generate exceptions simultaneously.
Central to the MC68000 exception processing sequence is a vector table that occupies
MC68000
1024 bytes (512 sixteen-bit words) of memory. This table occupies memory addresses
EXCEPTION
000000 16 through 0003FF 16 . Figure 7-15 illustrates the exception vector table. The table is
VEC:rOR
organized as 256 four-byte vectors. Each vector is a 32-bit address which will be loaded into the
TABLE
Program Counter as part of the exception processing sequence.
As you can see, a number of the vector table entries serve the defined types of exceptions which we have discussed. The remaining entries of the vector table are reserved for use by Motorola and should not be used by your program if compatibility with Motorola software is desired. The first 64 exception vectors have predefined uses; this leaves
192 vectors available to external interrupt requests - this should be more than enough for most applications.
However, the first 64 vector locations are not protected by the MC68000; thus they can be used by external interrupts
if a system requires it.
MC68000 EXCEPTION PROCESSING SEQUENCES
The general sequence of events performed by the MC68000 in response to an exception is the same regardless
of the source of the exception. There are, however, some differences. Let us begin by examining the response
to internally generated exceptions.
If exception processing is initiated as a result of either the Trace function, a TRAP instruction, an illegal or unimplemented opcode, or a privilege violation, the following steps occur:
1)
The Status Register contents are copied into an internal register.
2)
The S-bit in the Status Register is set. thus placing the MC68000 in the Supervisor
mode of operation.
7-25
MC68000
INTERNALLY
GENERATED
EXCEPTION
PROCESSING
Memory
Addresses 1---16 B i t s (Hex)
000000
SSP (High)
000002
SSP (Low)
I'
II
Reset _ Initial SSP
Reset - Initial PC
Vector 2 - Bus Error
Vector 3 - Address Error
Vector 4 - Illegal Instruction
Vector 5 - Divide by 0
Vector 6 - CHK Instruction
Vector 7 - TRAPV Instruction
Vector B - Privilege Violation
Vector 9 - Trace
Vector 10,0 - Opcode 1010 Emulation
I
Vector 11, 0 - Opcode 1111 Emulation
>Vector 12 10
Reserved by
Motorola
I
I
} Vector 23,0
} Vector 24, 0 - Spurious Interrupt
~~~~I)
!jI Vector 25,0 - Levell Interrupt
Vector 26,0 - Level 2 Interrupt
Vector 2710 - Level 3 Interrupt
>ifAuto-Vectors
VPA low
Vector 28 10 - Level 4 Interrupt
I-;..;,o;,__~-H
Vector 29,0 - Level 5 Interrupt
I
Vector 30 10 - Level 6 Interrupt
J
I
Vector31,0- Level 7 Interrupt)
Vector 32,0
I
I
OOOOBC .....-P-C-4-7...(H
..i...
9h..) -1
OOOOBE
PC47 (Low)
OOOOCO
0000C2
t
Vector
PC48 (High)
PC48 (Low)
OOOOFC
PC63 (High)
OOOOFE ......P...C...
6_3...(L_O_W..
)..,
PC64 (HIgh)
PC64 (Low)
I
10
t
PC255 (HIgh)
PC255 (Low)
Vector 64
I
J}
J
1
Reserved by
Motorola
Vector 63
JI
l
0003FC
0003FE [
10
47
Vector 48
I
•
000100
000102
TRAP
Instruction
Vectors
1
) User Vectors
Vector 255
Figure 7-15. MC68000 Exception Vector Table
7-26
3)
The T-bit in the Status Register is reset to disable tracing to allow for continuous execution when debugging using TRACE.
4)
The Program Counter contents are pushed onto the Supervisor Stack. The contents of SSP will be decremented by four since four bytes are required to store the 32-bit contents of Pc.
5)
Status register contents are pushed onto the Supervisor stack; SSP contents are decremented by two, since
the Status register is a 16-bit register.
6)
The new Program Counter contents are taken from the appropriate location in the interrupt vector table.
7)
Instruction execution then begins at the location indicated by the new content of the Program Counter; this
will be the first instruction of the exception processing program you have provided for that particular type
exception.
The way in which the MC68000 responds to an exception caused by a bus error or address
error includes several steps in addition to those described in the preceding paragraphs. First,
recall that either of these errors causes immediate termination of the bus cycle in progress.
The next steps are the following:
MC68000 BUS
AND ADDRESS
ERROR
EXCEPTION
PROCESSING
1)
2)
The contents of the Status register are copied into an internal register.
The S-bit in the Status register is set. placing the MC68000 in the Supervisor mode.
3)
The T-bit in the Status register is reset to disable trace operations.
4)
The contents of the Program Counter are pushed onto the Supervisor stack and the System Stack Pointer
(SSP) is decremented by four.
5)
The contents of the Status register are pushed onto the Supervisor stack and the contents of SSP are decremented by two.
The contents of the MC68000's instruction register, which constitute the first word of the instruction that
was in progress when the bus error occurred, are pushed onto the Supervisor stack and SSP is decremented
by two.
The 32-bit address that was being used for the bus cycle which was terminated is also pushed onto the
Supervisor stack and SSP is decremented by four.
6)
7)
8)
A word which provides information as to the type of cycle that was in progress at the time of the error is
pushed onto the Supervisor stack and SSP is decremented by two.
9)
The Program Counter contents are taken from the appropriate interrupt vector - either the bus error vector
or address error vector of the exception vector table.
10)
Instruction execution resumes at the location indicated by the new contents of the Program Counter.
Figure 7-16 shows the order in which information is pushed onto the Supervisor stack as part of the exception
processing for bus and address errors. The value saved for the Program Counter is advanced two to ten bytes beyond
the address of the first word of the instruction where the error occurred according to the length of that instruction and
its addressing information, if any.
As you can see in Figure 7-16, the five least significant bits of the last word pushed onto the Stack provide information as to the type of access that was in progress when the bus error or address error occurred. The three least
Significant bits are a copy of the Function Code outputs during the aborted bus cycle. Bit 3 indicates the type of proceSSing that was in progress when the error occurred. This bit is set for Group 0 or 1 exception processing and reset for
Group 2 exception and normal instruction processing. Bit 4 indicates whether a read (bit 4 set) or write (bit 4 reset)
cycle was in progress when the error occurred. If an error occurs during the exception processing of a preceding bus
error, address error, or reset operation, the MC68000 will enter the Halt state and remain there.
All of the information that is pushed onto the Supervisor stack as part of the bus and address error exception
processing sequence is intended to aid you in analyzing possible sources of the error. Either of these errors implies
a serious system failure and it is not likely that you will be able to return to normal program execution.
,.....----"'"
MC68000
An external reset causes a special type of exception processing. After the
been pulsed low the following steps occur:
RESET input has
RESET
EXCEPTION
PROCESSING
1)
2)
The S-bit in the Status register is set. placing the MC68000 in the Supervisor mode.
The T-bit in the Status register is reset to disable the trace function.
3)
All three interrupt mask bits in the Status register are set. thus specifying the interrupt priority mask at level
seven.
4)
The Supervisor Stack Pointer (SSP) is loaded with the contents of the first four bytes of memory (addresses
000000-000003) .
7-27
r---------j 0
l
.1
1
0 = Instruction in progress
Exception processing
~{ 1
"+
= Write cycle aborted
= Read cycle aborted
=
Fun,tion Code IFC2. FC 1. FCOI
SSP after
- - - -...
~r_ _ _ _ _ _ _ _ _ _ _ _..!4~3~2~!.1...20~""-Bit No.
exception Lower
Access Type'"
Address t-----C-u-rr-en-t-C-y-C-le;.;.A-d-d-r-e-ss-(h-ig-hL-o-r-ld-e-'r)L.-L--I--I
I 11 I I
Current Cycle Address !low-order)
Instruction Register
Status Register
PC (high-order word)
SSP bef?re - - - - . Higher
PC !low-order word)
exception
Address ...- - - - - - - - - - - - - - - - -.....
-------16-Bit Words-------
Figure 7-16. MC68000 System Stack After Bus Error or Address Error
5)
The Program Counter (PC) is loaded from the next four bytes of memory (addresses 000004-000007).
Instruction execution commences at the address indicated by the new contents of the Program Counter.
which should reference your power-up/reset initialization program.
MC68000
INTERRUPT
The last type of exception processing we will discuss is the sequence initiated by the stanREQUEST
dard interrupt request. An external device requests an interrupt by encoding an interrupt
EXCEPTION
request level on the IPLO-lpL2 inputs. The MC68000 compares these inputs to the interrupt
PROCESSING
mask bits in the Status register. If the encoded priority level is less than or equal to the one
6)
specified by the three-bit mask. the interrupt request will not be recognized by the MC68000.lf the encoded interrupt
level is a higher priority than the level established by the interrupt mask (or if a level seven interrupt request is
input) then the interrupt will be processed. The MC68000 responds to the allowed interrupt request as soon as it
completes the instruction execution currently in progress. Upon completion of the current instruction, the following
steps occur:
1)
The contents of the Status register are saved internally.
2)
The S-bit in the Status register is set. placing the MC68000 in the Supervisor mode.
3)
The T-bit in the Status register is reset to disable the Trace function.
4)
The interrupt mask bits in the Status register are updated to the level of the interrupt request that is
encoded on the IPLO-IPL2 inputs. This allows the current interrupt to be processed without being interrupted by lower priority events.
5)
The MC68000 then performs an interrupt acknowledge bus cycle. This cycle serves two functions; first. the
processor lets the requesting device know that its interrupt request is being serviced. and second. the processor fetches an exception vector byte from the requesting device. Figure 7-17 shows the timing for this
interrupt acknowledge/vector fetch cycle. This cycle is esentially a read cycle with a few minor
differences. First. address lines A 1 through A3 will reflect the states of the IPLO-IPL2 inputs so that external
logic can determine which interrupt request is being processed. All of the other address outputs are set during the interrupt acknowledge cycle. The requesting device responds to the MC68000 by lJlacing a byte of
exception vector data on the lower half of the data bus. The Data Transfer Acknowledge (DTACK) signal is
used to effect this transfer of data just as with a normal read cycle. Throughout the interrupt acknowledge
cycle. the Function Code outputs (FCO-FC2) will be set high since this represents the interrupt acknowledge
function code. After the vector byte has been read from the interrupting device. the MC68000 proceeds
with the following exception processing steps.
The contents of the Program Counter are pushed onto the Supervisor stack and SSP is decremented by four.
6)
7-28
Complete Current ----.+---Interrupt Acknowledge Bus Cycle
Instruction
ClK
A4-A23
A1-A3
R/W
08-015
00-07
FCO-FC2
Interrupt Request
~---- No Request (lPLO = IPl1
= IPL2
= 1)
Figure 7-17. MC68000 Interrupt AcknowledgeiVector Fetch Cycle
7-29
7)
The contents of the Status register are pushed onto the Supervisor stack and SSP is decremented by two.
8)
The Program Counter is loaded with four bytes of data from the appropriate location in the exception vector
table. The address for this location is derived as shown in the following illustration:
Vector No. from
interrupting device--"'1...;'~~L.,..J....l.,~,.l.,..J
Address Outputs
~-------Address
range
= 000000-0003FC----------t
The eig ht bits of data that were read from the requesting device as part of the interrupt acknowledge cycle
are used to form address bits A2 through A9. The two least significant bits and bits A 10 through A23 will all
be set to zero. Thus, addresses 00000016 through 0003FC16 can be generated. If you refer to Figure 7-15,
you will see that these are the upper and lower boundaries of the exception vector table. Under normal circumstances a requesting device should limit itself to producing vectors corresponding to the address range
0000FC16 through 0003FC16 since the lower addresses in the vector table have preassigned uses.
After the Program Counter has been loaded with the new value from the exception vector table, instruction execution
commences at the location indicated by the new contents of the Program Counter; this will be the first instruction of
your interrupt processing routine for the particular device requesting the interrupt.
There are two variations to the interrupt request processing sequence we have just deMC68000
scribed. First, if during the interrupt acknowledge bus cycle the requesting device responds
SPURIOUS
by asserting BERR instead of DTACK, the MC68000 treats this as an indication that the curINTERRUPT
rent interrupt request is a spurious one, and it will use vector 24 in the exception vector table to
load the Program Counter.
MC68000
The second variation on interrupt request processing is the autovector response. If you refer
AUTOVECTOR
to Figure 7-15, you will see that seven vector locations are provided in the exception vector
INTERRUPT
table for autovectors, corresponding to the seven interrupt priority levels. These vectors
RESPONSE
will be used if the device requesting an interrupt r~nds to the interrupt acknowledge bus
cycle by asserting the Valid Peripheral Address (VPA) signal instead of supplying a byte of vector data. If this
occurs, the MC68000 will respond by asserting the Valid Memory Address (VMA) signal. The processor will then use
the appropriate autovector from the exception vector table to obtain a new Program Counter value. This autovector
response was provided specifically to emulate the interrupt timing sequence expected by 6800-family peripheral
devices. The VPA/VMA sequence is the standard 6800 microprocessor interrupt sequence. Of course a non-6800family device in the system could also exploit this autovector capability should it be advantageous.
MC68000 ADDRESSING MODES
The MC68000 utilizes 14 different addressing modes which can be grouped into six basic types. These are:
1)
2)
3)
4)
Direct Register Addressing
a) Data Register Direct
b) Address Register Direct
Direct Memory Addressing
a) Absolute Short
b) Absolute Long
Indirect Memory Addressing
a) Register Indirect
b) Post-increment Register Indirect
c) Pre-decrement Register Indirect
d) Register Indirect with Displacement
e) Register Indirect with Index and Displacement
Implied Register Addressing
7-30
5)
6)
Program Counter Relative Addressing
a) PC-relative with Displacement
b) PC-relative with Index and Displacement
Immediate Data Addressing
a) Immediate
b) Quick Immediate
These addressing modes help create a powerful and efficient instruction set. In particular. two useful features of
the MC68000 addressing are that any address register may be used for direct or indirect addressing. and any register
may be used as an index register.
The general format of a single effective address instruction operation word is shown below. Thetwo least significant 3-bit fields determine the effective address. These fields are the mode field (bits 3-5) and the register field (bits 0-
21.
15
14 13 12
11
10
9
8
7
6
5
4
3
0 ",--Bit No.
2
IXlxlxlxlxlxlxlxlxlxlmlmlmlrlrlrl
TT.
-----Re9ister Field
-
Mode Field
In some cases. the information contained in these two fields may be appended to fully specify the operand. In this
case. one or two additional words are appended onto the instruction. This additional information is called the effective address extension. and its format is:
111-
"E
E
o ::J
~
c: 10
.2 E
r---
~ .~
~ 5
Sxu..0
w-
~
Operation Word
Immediate Operand
(if any. one or two words)
-
Source effective address extension .
(if any. one or two words)
-
Destination effective address extension
(if any. one or two words)
-
--------16 Bits--------
We will now discuss the addressing modes in detail. The following abbreviations are used within this section.
An
CCR
dddd
On
EA
N
PC
pppp
qqqq
xxxx
yyyy
zzzz
Rn
rrr
SP
SR
SSP
ssss
USP
Address register n (0 < n < 7)
Condition code half of the Status register
displacement value
Data register n (0 ~ n :; 7)
effective address
operand size in bytes (1.2.or 4)
Program Counter
any four hex digits
any address or data register n (0 < n ~ 7)
the 3-bit value of n
the active Stack Pointer
Status register
Supervisor Stack Pointer
sign extension digits
User Stack Pointer
7-31
Register Direct Addressing
This addressing mode requires that the operand involved be contained in one of the eight Data
registers or one of the eight address registers (Mode = 0012).
Memory
{: }
............................. } Operation word
xl xl111111010To
Extension word
C
•
I
~II
ssxxyy
Operand (1, 2, or 4 bytes)
+ 1
ssxxyy + 2
ssxxyy + 3
ssxxyy
~
Byte
=
EA
Sign extended value of the extension word
Mode 111 2
Register = 000 2
=
Figure 7-18. MC68000 Absolute Short Direct Memory Addressing
Memory
I,
>Operation word
xlxl11111101011
1<
pp
qq
xx
yy
>First extension word
>Second extension word
I'
Operand (1, 2, or 4 bytes)
ppqqxxyy + 2
ppqqxxyy
+3
EA = Concatenation of extension words
Mode 1112
Register 001 2
=
=
Figure 7-19. MC68000 Absolute Long Direct Memory Addressing
7-32
MC68000
REGISTER
DIRECT
ADDRESSING
Data Register Direct
=
Address Register Direct
=
EA
Dn
Mode 000 2
EA
An
Mode 001 2
=
=
Absolute Data Addressing
There are two forms of this addressing mode. The short form is called absolute short addressing,
while the longer format is called absolute long.
Absolute short. One extension word is necessary for this addressing mode. The address of the
operand is the sign extended value of the extension word. Figure 7-18 illustrates the absolute
short addressing mode.
MC68000
ABSOLUTE
DATA
ADDRESSING
Absolute long. Two words of extension are required for this addressing mode. The address of the operand is the concatenation of the two extension words; the first is the high-order portion, the second is the low-order portion. Figure 719 illustrates the absolute long addressing mode.
Register Indirect Addressing
The five variations of this addressing mode each reference an operand in memory.
Address register indirect. In this mode, the address of the operand is the contents of the
specified Address register. Figure 7-20 illustrates the address register indirect mode.
MC68000
REGISTER
INDIRECT
ADDRESSING
Address register indirect with postincrement. In this mode, the address of the operand is the
contents of the specified Address register. After the instruction using this mode is executed, the contents of this
register are incremented by one, two, or four depending on the size of the operand. If the Address register is A7 (SP)
then the address is incremented by two regardless of the operand size, because the Stack Pointer must be kept on a
word boundary. Figure 7-21 illustrates the Address register indirect with postincrement mode.
Address register indirect with predecrement. This addressing mode is similar to the previous one vvith the exception
that the contents of the specified Address register are decremented before they are used to reference the operand.
Memory
Address Registers
AO
A1
l'--T"T""II"'"'T""T""'Ii""'T"-I} Operation word
~----------------~
A2 ~--------------~
~--------------~
A3
A4 ~--------------~~
xxxxyyyy
A5
A6
~--------------~
(In this example rrr
= 100)
I
xxyyyy
1
xxyyyy
+2
+3
xxyyyy
Operand (1, 2, or 4 bytes)
~
EA
[An]
Mode 010 2
Register
n
=
•
xxyyyy~1
+
~----------------~
A7 ~----------------~
=
•
Byte
=
Figure 7-20. MC68000 Address Register Indirect Memory Addressing
7-33
Memory
Address Registers
AO
A1 ~---------------I
A2 ~-----------------I
A3 ~---------------I
A4 ~-----------------I
I
A5 ~-----------------L~
xxxxyyyy
A6
A7
..
~ 0 ..""" 11. 2. 0.4 bytesl
~---------------I
~-------------(In this example rrrr
= 101)
xxyyyy
+2
xxyyyy + 3
~
=
=
Byte
EA
[An)
[An)
[An) + N
Mode 011 2
Register
n
=
I
=
Figure 7-21. MC68000 Address Register Indirect with Postincrement Addressing
Memory
Address Registers
AO
A1
A2
t-----~x:x:xX~y:y:y~y---:::~~---------------------Jt1~~nr.r.T;1
~
______________-a. .
A3
A4 ~-----------------I
I
~-----------------I
..
A6~______________~
A7
~-------------(In this example rrrr
001)
=
xxyyyy - N
=
I
~ Operand (1, 2, or 4 bytes)
+1
+2
xxyyyy - N + 3
xxyyyy - N
xxyyyy - N
[An) = [An) - N
EA
[An)
Mode 100 2
Register
n
=
I
I
A5~______________~
~
Byte
=
Figure 7-22. MC68000 Address Register Indirect with Predecrement Addressing
Again. if A7 is specified then the address is always decremented by two. Figure 7-22 illustrates the address register
indirect with predecrement mode.
Address register indirect with displacement. One word of extension is required with this addressing mode. The
address of the operand is the sum of the contents of the specified Address register and the sign-extended 16-bit displacement word contained in the extension word. Figure 7-23 illustrates the address register indirect with displacement mode.
Address register indirect with index and displacement. This addressing mode requires one word of extension which
is formatted as shown in Figure 7-24. The operand address is the sum of the specified address register. the sign-extend
7-34
Memory
Address Registers
!
operation word
xfx1110111r1 r1r
.-
xxxxyyyy
~
xxxxyyyy
+ssssdddd~
dd
{
,
zzzzzzzz
-.,.-..-
Displacement word
dd
I
I
~
zzzzzzzz
zzzzzzzz + 1
zzzzzzzz
(In this example rrr = 010)
zzzzzzzz
=
+
+
2
3
I
~
Operand (1, 2, or 4 bytes)
~
EA
[An] + dddd (sign extended)
Mode = 101 2
Register
n
Byte
=
Figure 7-23. MC68000 Address Register Indirect with Displacement Addressing
15 14 13 12 11 10 9
8
7 6
5 4
3
2
lPD-/-A""l-R-e-g-is-t-er-"I'IIW-rL"·I~o-'-lo-""lo-I"---D-is-p-la-ce-m-e-nt-l-n-te-g-e-r
0 . . . - B i t No.
---'1
t.
----------DisPlacement (-128 through + 127)
Not used, always 0
L - - - - - - - - - - - - - - - - - - - - - - - I n d e x size:
o sign extended low-order
integer in index register
1 long value in index
register
=
=
.....- - - - - - - - - - - - - - - - - - - - - - - - - I n d e x register number
.....- - - - - - - - - - - - - - - - - - - - - - - - - - - - Index register indicator:
o = data register
1 address register
=
Figure 7-24. MC68000 Extension Word Format for Indexing
7-35
Memory
Address Registers
AO~______________~
A1
A2~______________~
A3
..I;L..........
~-----------------I
______________~
______________
I
~~
zzzzzz
A6~_____X_x_xx_y_y_y_y~~--I
A7
Extension word
~-"""'---I
A4~
A5~
l
§
operation word
xx110r"r
_ .......fi Oo..........lr
000
p .........
d
d
~-----------------I
zzzzzz
~------------~~
+
1
I
Operand (1._2. or 4 bytes)
zzzzzz + 2
zzzzzz + 3
~
Data Registers
OO~
01
02
Byte
_____________~
ppppqqqq
~--------------~~
Index Size
03~______________-4
04
long (WfL = 1)
~-----------------I
_ _ _ _ _ _ _~
xxxxyyyy
~--- + ssssssdd
~--- + ppppqqqq
zzzzzz
05~
06~______________~
07~______________~
=
short (W fL
= 0)
xxxxyyyy
ssssssdd
ssssqqqq
zzzzzz
or_ -_ _ _ _ _J '
=
(In this example Of A 0 and r2r2r2 001 therefore the
index register is 01 ; r, r, r, = 11 0 which indicates A6 is used)
=
EA
[An) + [Rn)
Mode 1102
Register = n,
=
+ dd
(sign extended)
Figure 7-25. MC68000 Address Register Indirect with Index and Displacement
.A.ddres~ing
displacement integer in the least significant byte of the extension word, and the contents of the Index register. Address
formation for the Address Register Indirect with Index and Displacement is illustrated in Figure 7-25.
Implied Register Addressing
There are some instructions that implicitly refer to a specific register. These registers are the Program Counter (PCl. the Stack Pointer (SP-SSP or USPl. and the status register (SR). Table 7-5
shows those instructions in which a register holding the operand is implied.
MC68000
IMPLIED
REGISTER
ADDRESSING
Program Counter Relative Addressing
There are two formats for PC-relative addressing. Both require one word of extension and both
provide displacement. The second format includes indexing in additional to displacement.
7-36
MC68000
PROGRAM
COUNTER
RELATIVE
ADDRESSING
Memory
lope..,","
[PC]- 2
[PC] - 1 xixl1111110111o
dd
[PC] + 1
dd
~[PC]
OOPPPCCC ~
+ ssssdddd
I
I
zzzzzz
-c:...
Extension word
I
I
''''''~
zzzzzz
+
wend
Operand (1, 2, or 4 bytes)
1
zzzzzz + 2
zzzzzz + 3
EA = [PC] + dddd (sign extended)
Mode = 1112
Register = 010 2
---Byte
Figure 7-26. MC68000 Program Counter Relative Addressing
DO
[PC] - 2
[PC] - 1 x x 1 1 1
01 ~----------------~
~----------------~
02
03
04
~----------------~L.Lppppqqqq
~----------~~~
I
Memory
Data Registers
Operation word
11
[PC]
~ r r r ~ 00 0
[PC] + 1
d
d
__---
I
~--------------~~
05
~--------------~~
06
I
I
ZZZZZZ
07 ~------------~~
Extension word
I
~ Operand (1, 2, or 4 bytes)
zzzzzz + 1
~----------~--~
zzzzzz
zzzzzz
+2
+3
---Byte
+ ssssssdd
+ ppppqqqq
OOPPPCCC
+ ssssssdd
ssssqqqq
_---- -----""./
zzzzzz
zzzzzz
....
'--
(In this example Of A = 0 and rrrr = 011, therefore the selected index register is 03.)
EA = [PC] + [Rn] + dd (sign extended)
Mode=1112
Register = 011 2
Figure 7-27. MC68000 Program Counter-Relative with Index and Displacement Addressing
7-37
Byte .,___o_o_o_o_o_oo_o
____~___o..;..:;.pe_r_an_d_ _ _,,1 Extension word(s)
or
Word ,_ _ _ _ _ _ _ _ p.e.ra.n_d_ _ _ _ _ _ _....
o..
or
Long Word t-____~O~p_e_ra-n~d~(h~i9
..h-or-d-e-rh_a_lf~)_ _ _ _-I
Operand (low order half)
'-~--------------~~~--------------~~
Word
=
EA
Not required; the operand is part of the instruction
Mode =111 2
Register 1002
=
Figure 7-28. MC68000 Immediate Data Addressing Extension Words
The value contained in the Program Counter which is used in address calculation is the address of the extension
word.
PC-relative with displacement. This addressing mode generates an effective address by summing together the value
of the Program Counter and the sign extended value of the extension word. Figure 7 -26 illustrates the PC-relative with
displacement mode.
PC-relative with index and displacement. This mode requires an extension word format similar to that required by
the address register indirect with index and displacement mode (see Figure 7- 24). The address is calculated as shown
in Figure 7-27.
Immediate Data Addressing
The operand for immediate data addressing is the value that immediately follows the instruction
word. Thus. depending on the size of the operand. either one or two extension words will be
necessary. as illustrated in Figure 7-28.
MC6S000
IMMEDIATE
DATA
ADDRESSING
THE MC68000 INSTRUCTION SET
Table 7-6 summarizes the instruction set of the MC6S000. Instruction object codes and execution times are
given alphabetically in Table 7-7. Instruction object codes are given numerically in Table 7-S.
When compared to other microprocessor instruction sets, the MC6S000 instruction set might seem quite large:
over 300 instructions are listed in Table 7-6. However. if you examine this table closely. you will see that slight variations of the same instruction mnemonic may appear several times. These are different forms of the same instruction.
There are actually 56 basic instructions provided in the MC68000. We have listed all the variations of a single
instruction as though they were distinct instructions in order to make our description of the instruction set consistent
with similar ones for other microprocessors.
One of the most significant characteristics of the MC68000 instruction set is its orderliness. Despite its apparent
complexity. this instruction set should be relatively easy to learn. since the variations are consistent and therefore predictable. These variations are due to the different addressing modes available and to the MC68000's ability to handle
five different data types. Since there are really only 56 basic mnemonics that you must learn. it is more likely that you
will use all of the instructions in the way that they were intended and thus obtain the full power of the instruction set.
Let us examine the MC6S000 instruction set by instruction categories, as given in Table 7-6.
One thing to keep in mind is that the MC68000 uses memory-mapped 1/0: therefore there are no separate 1/0
instructions. The primary memory reference instructions will also be used to accomplish I/O.
The basic format of all instructions is the same. The op-code for every instruction is one word. Additional extension
words are required when the addressing modes specified use constants (immediate operands). absolute addresses. or
7-38
displacements. Accordingly. an instruction can be anywhere from 2 to 10 bytes in length. The number of bytes for
each instruction is listed in Table 7-6.
All of the primary memory reference instructions have byte, word, and long word versions. Secondary memory
reference instructions can use most of the memory addressing modes. There are byte, word, and long word versions of most, but not all, of these instructions.
The Move instruction provided by the MC68000 allows data movement between registers. from register to memory.
from memory to register. and directly from one memory location to another. The Move Multiple Registers (MOVEM)
instruction allows all of the MC68000 register contents to be quickly saved in memory or restored from memory.
The MC68000 does not provide a block move instruction such as those available with the 8086 and the Z8000
microprocessors. However, since the Move instructions can move data from one memory location to another. it is
simple to move blocks of data by using the Move instruction in conjunction with the Decrement and Branch
(DBcc) instruction.
Both signed and unsigned multiply and divide instructions are included in the instruction set. In comparison. the
Z8000 provided only unsigned multiplication and division. However. the Z8000 provides 32-bit multiplication and division while the MC68000 can only multiply two 16-bit operands producing a 32-bit result. or divide a 32-bit dividend by
a 16-bit divisor. The divide instructions reference the dividend in one of the Data registers: the divisor may
reside in memory or in another Data register. Both the divisor and the dividend are treated as signed binary numbers
in the DIVS instruction and as unsigned binary numbers in the DIVU instruction. After the division instruction has been
executed. the quotient is returned in the low-order half of the dividend register and the remainder is returned in the
high-order half of the dividend register.
The multiply instructions also have only a word version: there is no long word version. As with the division. there
is a signed (MULS) and unsigned (MULU) version of the multiply instructions. One of the operands must reside in the
least significant half of a Data register while the other operand can be either a memory word. the lower half of another
Data register. or can consist of immediate data included as part of the instruction. Upon completion of the multiply
operation. the 32-bit product is returned in the source operand Data register.
The MC68000 includes standard Jump and Jump to Subroutine instructions (JMP and JSR) which use specific
addresses for loading the Program Counter. There are also the Branch Always and Branch to Subroutine instructions (BRA and BSR) which cause a transfer of program control relative to the Program Counter's current contents.
The Trap instruction is the MC68000's equivalent of the System Call instruction provided by the Z8000. You
will recall from the earlier discussion of the MC68000 exception processing logic that the Trap instruction automatically switches the MC68000 into the Supervisor mode. which utilizes a separate Stack Pointer to isolate the operating
system from application programs.
The MC68000 also provides several instructions that are specifically designed to simplify implementation of
higher level languages. These instructions are unique to the MC68000. The Link (LINK) and Unlink (UNLK)
instructions can be used to maintain a linked list of local data and parameter areas on the Stack and thus
simplify operations where there are frequent interrupts of nested subroutines.
The Link instruction uses the System Stack Pointer (SFPl. one of the other Address registers. as a "frame pointer" and a
displacement value. This instruction will typically be used at the beginning of a subroutine. The Link instruction first
pushes the current value of the frame pointer onto the Stack. The current value of the Stack Pointer is then loaded into
the frame pointer so that it now points to the top of the current Stack. Finally. the displacement value included with the
Link instruction is used to decrement the System Stack Pointer so that it is displaced to clear a space in memory for
storage of such things as local variables and parameters. These variables can then be accessed via the frame pointer.
The Unlink (UNLK) instruction is used to clean up the Stack at the end of a subroutine and would be executed just prior
to returning to a higher level subroutine. The Unlink instruction loads the System Stack Pointer with the contents of the
frame pointer. The frame pointer is then loaded with the address pulled off the Stack. Thus. both the frame pointer and
the System Stack Pointer will be restored to the values they held before the subroutine was called.
7-39
ABBREVIA TIONS
Following are the abbreviations used for instruction formats and operation descriptions.
addr
An
bitb
bitl
cc
Direct address (16 or 32 bits)
Address registers. n = 0-7 (8. 16. or 32 bits. depending on the instruction size)
Bit number of byte 0-7
Bit number of long word 0-31
Condition code:
CC
CS
EO
F
GE
GT
HI
LE
LS
LT
MI
NE
PL
T
VC
VS
CCR
count
dadr
Carry clear
Carry set
Equal
False
Greater than
or equal
Greater than
High
Less than or
equal
Low or same
Less than
Minus
Not equal
Plus
True
No overflow
Overflow
d16(An)
d8(An.i)
addr
On
d8
d16
i
jadr
label
madr
reg-list
rd
rs
0011
1101
1011
0110
1010
0000
1000
1001
Register indirect
Register indirect with
postincrement
Register indirect with
predecrement
Register indirect with
displacement
Register indirect. indexed
Direct address
-(An)
aDn
data3
data8
data16
data32
1110
0010
1111
Condition Code register - the low-order byte of the Status register
Shift count (1-8)
Destination address. which may be any of the following addressing modes:
(An)
(An)+
dAn
0100
0101
0111
0001
1100
Destination Address register. This form is used only when there are two An operands.
Destination Data register. This form is used only when there are two Dn operands.
3 bits of immediate data
8 bits of immediate data
16 bits of immediate data
32 bits of immediate data
Data register. n = 0-7 (8. 16. or 32 bits. depending on instruction size)
8-bit address displacement. Required. even if zero on indexed instructions.
16-bit address displacement
Index register (An or Dn)
Jump address - same as sadr except no (An)+ or -(An)
Address label
Multiple-instruction address - same as dadr except no (An)+ or -(An)
Register list naming one or more registers. each item in the list separated by a comma. Items may
have the form:
Dn
Single data register
An
Single address register
Range of registers
rn l-rn
Destination registers (dDn or dAn)
Source register (sDn or sAn)
7-40
sadr
Source address. which may be any of the following address modes:
(An)
(An)+
Register indirect
Register indirect with
postincrement
Register indirect with
predecrement
Register indirect with
displacement
Register indirect. indexed
Direct address
Program relative
Program relative. indexed
-(An)
d16(An)
d8(An.i)
addr
label
label (j)
sAn
sOn
SR
USP
vector
[[J]
[ ]
Source Address register. This form is used only when there are two An operands.
Source Data register. This form is used only when there are two Dn operands.
Status register (16 bits)
User Stack Pointer. Note that this is Register A7.
Trap address vector. the memory location containing the address of the Trap routine.
The contents of the memory location whose address is contained in the designated register
(indirect memory addressing. or implied addressing).
The contents of a register or memory location (register addressing or direct memory addressing).
For example:
[Dn] -
[[An))
indicates that the contents of the memory location addressed by Register An are loaded into Dn.
whereas:
[On] -
[An]
indicates that the contents of Register An itself are loaded into On.
X
x
+
x
1\
V
¥
Complement the value of x.
Bits y through z of x. For example. On <0-7 > means the low-order byte of Dn. If the z term is omitted. then only the bit selected by y is being referenced. Thus On <0> means the least significant
bit of Dn.
Add
Subtract
Multiply
Divide
Logical AND
Logical OR
Logical Exclusive-OR
Equals
Data moves in the direction of the arrow
Data are exchanged between two locations
INSTRUCTION MNEMONICS
Table 7-6 summarizes the MC68000 instruction set. The MNEMONIC column lists the instruction mnemonic (e.g ..
MOVE. ADD. JMP). The OPERAND(s) column lists the operands used with the instruction mnemonic.
The fixed part of an assembly language instruction is shown in UPPER CASE. The variable part (register number.
address. immediate data. etc.) is shown in lower case.
The BYTES and CLOCK CYCLES are repeated in this table for reader convenience. Refer to "Instruction Object Code
Tables" and the text accompanying Table 7-7 for a description of these entries.
7-41
STATUS
The effect of instruction execution on the status bits is listed in Table 7-6. The status bits are:
T
S
X
N
Z
V
C
Trace mode
Supervisor state
Extend bit
Negative (or Sign) bit
Zero bit
Overflow bit
Carry bit
The following symbols are used in the STATUS columns:
X -
flag
(blank) 1 - flag
o - flag
is affected by operation
flag is not affected by operation
is set by operation
is cieared by operation
OPERATION PERFORMED
This column shows the sequence of operations that occurs when the instruction is executed. (Instruction fetches are
not shown. nor is the incrementing of the Program Counter for the purpose of instruction fetches.) Each operation is
generally shown in the following form:
destination
source
indicating that the source contents moves to the destination. replacing the destination contents. For example. the LEA
instruction operation is:
[An]
jadr
The effective address. which may be any of the jadr forms. is loaded into the specified Address register.
Following the arrow sequence is a description of the operation in words.
Alternate Mnemonics
The MC68000 instruction set allows a choice of mnemonics for many operations. An "I" can be appended to the
instruction mnemonic for an immediate operation. An "A" can be appended to the instruction mnemonic for an
Address register operation. An ".S" can be appended to force a short-form conditional branch instruction.
Mnemonic choices are summarized in Table 7-5 under these headings:
PRIMARY MNEMONIC
ALTERNATE MNEMONIC
Lists the nominal mnemonic form
Lists the alternate choices that can be used in place of
the primary mnemonic.
Shows the operand category to which the primary and
alternate mnemonics apply. xx is any allowed operand
selection.
Identifies the operation.
OPERAND
DESCRIPTION
For simplicity. only the primary mnemonics are shown in the instruction set tables that follow.
Note that there are no mnemonic alternates for the instruction variations X (Extend). M (Multiple). and P (Peripheral
Data). These suffixes cannot be omitted from their respective instruction mnemonics.
Bear in mind that the assembler will select the "Quick" version of an instruction (e.g .. MOVEQ. AOOQ. SUBQ) whenever
possible. Thus you can use the alternates for these mnemonics - the more general MOVE. ADD and SUB - without
sacrificing any opportunities for code shortening.
For example: MOVE.L #40.02
is coded as: MOVEQ #40.02
Another example: ADD #1. DO
is coded as: ADOQ.W #1.00
7-42
MC68000 INSTRUCTION OBJECT CODE TABLES
The object code for each MC68000 instruction is shown alphabetically by instruction mnemonic in Table 7-6. The
object codes are listed in numerical order in Table 7-7.
For instruction words which have no variations, object cades are represented as four hexadecimal digits; for example,
4E71.
For instruction words with variation in one of the two bytes, the object code is shown as a combination of lower case
variables, hex digits, and binary digits. Each byte of an instruction word in Tables 7-7 and 7-8 is subdivided into two
"nibble" fields (1 nibble = 4 bits). If a single digit appears in a nibble field, it is a hexadecimal digit. If four digits, or a
combination of digits and lower-case variables (for example, 1 rrr), appear in a nibble field, each digit represents a single
bit.
Note that some lower-case variables are used to represent hexadecimal digits rather than binary digits. When four of
these hexadecimal variable characters (for example xxxx or yyyy) are used to represent a 16-bit word, they will appear
grouped together in the center of the 2-byte column comprising that word.
INSTRUCTION EXECUTION TIMES
Table 7-7 lists the instruction execution time in clock cycles. Each cycle = 125 nanoseconds (when fCLK =8.0 MHz).
The abbreviations and notations used in the "clock cycles" column are defined as follows:
+ea
Effective address overhead. This is the additional time required to execute the instruction for addressing
modes that take longer to execute than the nominal register indirect address. The following are the additional clock cycles required:
Additional Clock Cycles
Addressing Mode
(An)
(An)+
-(An)
d16(An)
d8(An,j)
addr-16-bit
addr-32-bit
label
label(j)
o
o
2
5
7
5
10
5
7
For shift instructions, the number of shifts. For move multiple instructions, the number of registers being
moved.
The first value is for branch or trap taken, the second is for branch or trap not taken. In the case of Bcc,
the first of the latter numbers is for a two-byte instruction (8-bit displacement), and the second is for a
four-byte instruction (16-bit displacement). In the case of DBcc, the first of the latter numbers is for
branch not taken due to condition true, and the second is for branch not taken due to counter timeout.
Indicates maximum value.
The lower value is for condition false (byte set to all ones); the higher value is for condition true (byte
cleared to all zeroes).
N
The following abbreviations are used in Table 7-7:
a
bbb
bbbbb
ccc
ddd
Operand addressing mode (1 bit)
o = data register to data register
1 = memory to memory
3 bits of immediate data. In bit operations the bit numbers 0 - 7.
Bit numbers a - 31.
Shift count 000 = 8 shift,S
001 = 1 shift
010 = 2 shifts
011 = 3 shifts
100 = 4 shifts
101 = 5 shifts
110 = 6 shifts
111 = 7 shifts
Destination register - same coding as rrr.
7-43
eeeee
Source effective address (6 bits)
Address Mode
(An)
(An)+
-(An)
d16(An)
d8(An,i)
addr-16-bit
addr-32-bit
label
label(j)
[EXT]
MODE/REGISTER
010rrr
011 rrr
100m
101 m
110rrr
111000
111001
111010
111011
xxxx
a iii w 000 xx
pppp
pppp qqqq
xxxx
a iii w 000 xx
hhhhhh
iii
jjjjjj
kkkk
One or two optional words of extension addressing that mayor may not appear, depending on the
addressing mode (see the Addressing Modes description).
Destination effective address - same as eeeeee except no label or label(j).
Destination effective address but in a format with the MODE and REGISTER fields switched (e.g.,
(An)=rrrOl0).
Multiple-destination effective address - same as ffffff except no (An)+ or -(An}.?
Index register - same coding as rrr.
Jump effective address - same as eeeeee except no (An)+ or -(An).
Register mask list for predecrement mode, in the following format (a "1" selects the register):
mmmm
1514131211109876543210
0001 D2D3D4D5D6D7AOAl A2A3A4A5A6A7
Register mask list for non-predecrement modes, in the format (a "1" selects the register):
[EXT]
ffffff
gggggg
1514131211109876543210
A7 A6 A5 A4 A3 A2 A 1 AO 07 06 D5 04 D3 D2 01 DO
pppp
qqqq
rrr
sss
t
vvvv
w
xx
xxxx
yy
yyyy
zzzz
16-bit address word or most Significant word of 32-bit address
Least significant word of 32-bit address
Register 000 = DO or AO
001 = Dl or Al
010 = D2 or A2
011 = 03 or A3
100 = 04 or A4
101 = D5 or A5
110 = D6 or A6
111 = D7 or A7
Source register - same coding as rrr
Type of register 0 = On
1 = An
4-bit vector
Index size. 0 = sign extended, low-order integer in index register
1 = long word value in Index register
8-bit address displacement
16-bit address displacement
8-bit immediate data
16-bit immediate data or most significant word of 32-bit data
Least significant word of 32-bit data
INTERFACING THE MC68000 WITH 6800 PERIPHERALS
Many peripheral components have been developed by Motorola and other manufacturers for the 8-bit 6800
microprocessor. In general. any asynchronous peripheral device can be used with the MC68000 with only a small
amount of external logic needed to meet the interface requirements (handshaking, etc.). However, the 6800-family
components are based on synchronous read/write operations. This imposes certain constraints when you
attempt to use a 6800 peripheral device with an asynchronous processor such as the MC68000. Obviously, it was
in Motorola's interest to design the MC68000 so that it would be able to use both conventional asynchronous devices
and the family of existing synchronous 6800 devices. Therefore they have included logic to simplify interfacing 6800
peripheral devices.
7-44
Again, the MC68000 performs read/write operations asynchronously. The signals involved with these operations are
the strobes (AS, UDS, lOS). the R/W signal. the Data Transfer Acknowledge signal (DT ACK). and of course the address
(A 1-A23) and data (00-015) signals.
Three additional signals are used to perform the synchronous read/write operations required by 6800 peripheral
devices. These signals are Valid Memory Address (ViViA), Valid Peripheral Address (VPA), and Enable (E). Figure
7-29 illustrates the timing of the synchronous read and write cycles. After the MC68000 has output the address on
A 1-A23 and has asserted the Address Strobe (AS). external logic is expected to decode information on the address
lines. If a 6800 peripheral device is being addressed, then the external logic should assert the VPA input to the
MC68000. This causes the MC68000 to emulate the data transfer timing of the 6800 microprocessor. As a
result, the transfer of data is synchronized with the clock signal E. The MC68000 will keep the address outputs
valid throughout this cycle.
During a read cycle, the 6800 peripheral device is expected to place data on tbe Data Bus when the E signal is high.
Note that the Data Transfer Acknowledge (DT ACK) signal is not used since that signal implies an asynchronous transfer
of data. Instead, the falling edge of E indicates that the data transfer (either read or write) has been completed. The
MC68000 then proceeds to complete the cycle in the normal fashion by negating the strobe signals and returning the
Address Bus to the high impedance state.
You will note in Figure 7-29 that there is a difference in the total number of ClK cycles for the read and write
operations. You should not infer from this that all 6800-type read operations take four more ClK cycles than
write operations. That is only the case in the example shown, and has to do with the phase of E when t!:1e read
or write operation was begun. In general. the E signal and the current MC68000 cycle state will not be synchronized
at the outset of a 6800 reference cycle. This is because the E signal has a duty cycle of 40%: E is high for four ClK
periods and low for six ClK periods. The MC68000 instruction cycles, on the other hand, vary in the number of ClK signal periods needed to execute. During the write cycle we have shown in Figure 7-29, the E Signal is in synchronization
with the instruction execution cycle. Thus this particular write cycle takes the minimum possible number of ClK cycles
to execute. Note that the MC68000 automatically inserts wait states after the VPA signal is input. The number of wait
states inserted will depend on how much time is needed in order to synchronize with the signal.
The
VMA
signal is output by the MC68000 in response to the
VPA output.
At the end of the read or write cycle, the 6800 peripheral device or the address decoding logic in the system must
negate the VPA Signal within one clock period after the MC68000 negates AS. Otherwise, the MC68000 will assume
that the following cycle is also supposed to be a 6800-type synchronous cycle.
Figure 7-30 summarizes the timing constraints of 6800 peripherals. It includes the 6800 processor signals for
reference so you can compare them with those associated with the MC68000.
A SIMPLE MC68000/6800 INTERFACE EXAMPLE
Figure 7-31 illustrates a simple interface of two 6800 peripheral devices in an MC68000-based system. In this
example, the address region 000000 16 through 7FFFFF 16 (the lower eight megabytes) is used for asynchronous devices
including memory. The upper eight megabytes is used, albeit inefficiently, for the two synchronous 6800 peripheral
devices. The PIA (Peripheral Interface Adaptor) is assigned addresses 800000 16 through BFFFFF 16, while the ACIA
(Asynchronous Communications Interface Adaptor) is assigned addresses C00000 16 through FFFFFF 16.
Interrupt request signals are connected directly to the IP'i:O and iPU input pins of the MC68000. Note that IPl2
is tied high. In this example, an interrupt from the ACIA causes if5[Q to become active thus generating an interrupt of
level 1 (the lowest priority). Both PIA interrupts are connected to IPL 1. When either of these becomes active, an interrupt of level 2 is generated. If both the ACIA and the PIA request an interrupt simultaneously, an interrupt of level 3
would be generated.
For a detailed description of how the MC68000 responds to interrupt requests, refer to our earlier discussion of
MC68000 exception processing.
We have also included logic that will cause the MC68000 to use its autovector capability during response to an
interrupt request from one of the 6800 family devices. Recall that if the VPA signal is asserted to the MC68000 during an interrupt acknowledge cycle, then no byte of vector data need be supplied by the requesting device; instead, the
MC68000 gets the appropriate autovector from the exception proceSSing vector table.
7-45
Standard Read Cycle
SO
S2
S4 S6
.. I.
ISO
6800 Peripheral Read Cycle
S2
S4 SW SW SW SW SW SW SW SW SW SW S6
-,-
ISO S2
6800 Peripheral Write Cycle
S4 SW SW SW SW SW SW S6
ClK
A1-A23
AS
DM"/lDS
.....
~
R/IN
m
OTACK
00-015
E
VPA
VMA
Figure 7-29. MC68000 Synchronous Read/Write Timing for6aOO Peripherals
•I
ISO
so
52
54
5W
SW
SW
SW
56
so
MC6800 E Clock Freq.
Type
2.0 MHz
B
1.5 MHz
A
1.0 MHz
Std
SW
SW
5W
MC68000 ClK
E
MC6800*
Type B
Type A
Std
MC6800 VMA,
RiW
Peripheral*
~150 os~
B~
180 os
Type
Type A
70 os
140 os
270 os
Std
160 os
~
I
MC6800 Address
~30 os MC6800*
r--10 os Peripheral*
Peripheral*
Type B
Type A
~
180ns~
220 ns
320 ns
F
~~~~~~--~~
MC6800
Read Data
MC6800
Write Data
MC68000
Address
MC68000 (8 MHz)
t-1870s-1
Write Data
* Times are given for different MC6800 device clock frequencies
Figure 7-30. MC68000/6800 Interface Timing Signal Summary
7-47
10 os MC6800·
10 ns Peripheral
+5 V
o
AS~----~--4~;--------------------~-"
VMA~----~---~~'----------~
2"" ':;:,t' ;,,' ,,:;,,',,>: ';,: ", ":'/:,,:L:; :: ':;;""'<:":', ':::O':,::,, :'
00-015
1C~;J',,!';
...
i,
"
....
I~
N
"c
o
c
«
~
MC68000
VPA~ I~••
FCO~
INTAC~
FCl 1 1 - - - - - -......
FC2t---..........",
+5V
C)
,U
r .0
en
u a:
C;;
r~
0
, .,
I~
(J)
(J)
u
I~
a:
Ii
N
I•
1.
(J)
(J)
...
u
~
b
0
c
r
~
,
Ql
0
u
"t:I
(J)
a:
E
~
>
6850 ACIA
III
...
w
I~
a:
I~
I~
w
~
IPL21-----~
~1-------~~~~-_4~
IPLO
RESET~------~.-~----~~~~~-----~---~~a..
E
R/W~--------~-------~~--------~,-~:"
Figure 7-31. A Simple MC68000/6800 Interface Example
7-48
III
Ql
(,)
.s;
III
6821 PIA
I~ I~
~
« «
Ql
.J:.
"0
0
I-
Table 7-4. MC68000 Instructions Which Use Implied Registers
Instruction
Implied Registerls)
Branch Conditional (Bee), Branch Always (BRA)
Branch to Subroutine (BSR)
PC
PC, SP
Check Register against Bounds (CHK)
SSP,SR
T est Condition, Decrement and Branch (DBcc)
PC
Signed Divide (DIVS)
SSP,SR
Unsigned Divide (DIVU)
SSP,SR
Jump (JMP)
PC
Jump to Subroutine (JSR)
PC,SP
Link and Allocate (LINK)
SP
Move Condition Codes (MOVE CCR)
SR
SR
Move Status Register (MOVE SRI
Move User Stack Pointer (MOVE USP)
USP
Push Effective Address (PEA)
SP
Return from Exception (RTE)
PC,SP,SR
Return and Restore Condition Codes (RTR)
PC,SP,SR
Return from Subroutine (RTS)
PC, SP
Trap (TRAP)
SSP,SR
Trap on Overflow (TRAPV)
SSP,SR
Unlink (UNLK)
SP
7-49
Table 7-5. MC68000 Instructions Which Use Implied Registers
Primary
Mnemonic
ADD. B
ADD.W
ADD. L
ADDQ.B
ADDQ.w
ADDQ.L
AND.B
AND.w
AND.L
Bcc
CLR.w
CMP.B
CMP.W
CMP.L
EOR.B
EOR.W
EOR. L
MOVE. W
MOVE. L
MOVEQ
OR. B
OR.W
OR. L
SUB.B
SUBW
SUB.L
SUBQ. B
SUBQ. W
SUBQ.L
Alternate
Mnemonic
ADDI. B
ADD
ADDA.W
ADDI.W
ADDA.L
ADDI.L
ADD.B
ADD
ADD.W
ADD. L
ANDI.B
AND
ANDI.w
ANDLL
Bcc.5
CLR
CMPI.B
CMP
CMPA.w
CMPI.W
CMPA.L
CMPI.L
EORI. B
EOR
EORI.W
EORI. L
MOVE
MOVEA.W
MOVEA.L
MOVE. L
ORI. B
OR
ORI.W
ORI. L
SUBI. B
SUB.
SUBA. W
SUBIW
SUBA. L
SUBI. L
SUB.B
SUB
SUBW
SUB. L
Operand
Description
dataS,xx
XX,XX
xX,An
data16,xx
xX,An
data32,xx
data3,xx
data3,xx
Add
Add
Add
Add
Add
Add
Add
Add
data3,xx
dataS,xx
XX,XX
data16,xx
data32,xx
xx
xx
dataS,xx
XX,XX
xX,An
data16,xx
xX,An
data32,xx
dataS,xx
XX,XX
data16,xx
data32,xx
XX,XX
xX,An
xX,An
dataS,xx
dataS,xx
XX,XX
data16,xx
data32,xx
dataS,xx
XX,XX
xX,An
data16,xx
xX,An
data32,xx
data3,xx
data3,xx
Add Quick Long
AND Immediate Byte
AND Word
AND Immediate Word
AND Immediate Long
Conditional Branch Short
Clear Word
Compare Immediate Byte
Compare Word
Compare Address Register Word
Compare Immediate Word
Compare Address Register Long
Compare Immediate Long
Exclusive OR Immediate Byte
Exclusive OR Word
Exclusive OR Immediate Word
Exclusive OR Immediate Long
Move Word
Move Address Register Word
Move Address Register Long
Move Quick (always Long)
OR Immediate Byte
OR Word
OR Immediate Word
OR Immediate Long
Subtract Immediate Byte
Subtract Word
Subtract Address Register Word
Subtract Immediate Word
Subtract Address Register Long
Subtract Immediate Long
Subtract Quick Byte
Subtract Quick Word
data3,xx
Subtract Quick Long
7-50
Immediate Byte
Word
Address Register Word
Immediate Word
Address Register Long
Immediate Long
Quick Byte
Quick Word
Table 7-6. MC68000 Instruction Set Summary
Mnemonic
Operand(s)
LEA
jadr,An
MOVE.B
(Anl.On
Bytes
2.4.
or 6
\
(An) +.On
-(Anl.On
d16(An),Dn
I
Clock
Cycles
Status
Operation Performed
T
S
X
N
Z
V
C
2(010)+
[An] - jadr
Load effective address into specified address register. The addressing
size is long. although the address loaded may be byte. word. or long.
depending on how it is subsequently used. 2
2
8(2/0)
X
X
0
0
2
8(210)
X
X
0
0
2
10(2/0)
X
X
0
0
4
12(3/0)
X
X
0
0
sadr
d8(An.il.On
a
addr.On
4
14(3/0)
X
X
0
0
40r 6
4(1/0)+
X
X
0
0
4
12(3/0)
X
X
0
0
4
14(3/0)
Dl
:::I
label.On
Q.
I
4'
.....
&.
.....
3'
Dl
-<
labellil.On
3:
~
3
[On<0-7>] - [[An]]
Register indirect
[On<0-7>] - [[An]]. [An] - [An] + 1
Register indirect with postincrement 1
[An] - [An] - 1. [On<0-7>] - [[An)]
Register indirect with predecrement 1
[On <0-7>] - [[AnI + d161
Register indirect with displacement
[On<0-7>1 - [[An] + d8 + [ill
Register indirect. indexed
[On<0-7>] - [addrl
Direct address
[On<0-7>1- [[PC] + d16]
Program relative
[On<0-7>1 - [[PC] + d8 + [ill
Program relative, indexed
Load byte to data register from memory location specified by any of
the addressing modes above. Bits 8-31 of the data register are not
affected .
S!
'<
MOVE.B
:Ill
~
~
;
On(An)
On,(An)+
2
\
I
:::I
9(1/1)
X
X
0
0
2
9(1/1)
X
X
0
0
2
9(1/1)
X
X
0
0
4
13(2/1)
X
X
0
0
4
15(211)
X
X
0
0
40r 6
5(011)+
X
X
0
0
n
~
On,-(An)
On,d 16(An)
On,d8(An, i)
Dn,addr
I
>dadr
[[An]] - [On <0-7>1
Register indirect
[[An)) - [On <0-7>1. [AnJ - [AnI + 1
Register indirect with postincrement 1
[Anl- [AnJ- 1, [[An]] - [On<0-7>1
Register indirect with predecrement 1
[[AnJ + d16J - [On<0-7>1
Register indirect with displacement
[[An] + d8 + [ill- [On <0-7>]
Register indirect, indexed
[addrJ - [On<0-7>J
Direct address
Store byte from data register to memory location specified by any of
the addressing modes above.
MOVE.B
sadr,dadr
2,4
6,8
or 10
5(1/1)+
X
X
0
0
[dadrl - [sadrI
Store byte from specified source memory location to specified
destination memory location. 1
Notes:
1.
2.
3.
4.
Postincrement and predecrement change by 1, unless the address register specified is the Stack Pointer (A 7), where the address is changed by 2 rather than 1 to keep the Stack Pointer
on a word boundary.
The effective address must be on an even word boundary (0000, 0002, 0004, etc.l.
Postincrement and predecrement change by 2.
Postincrement and predecrement change by 4.
I
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
a
II>
:::I
Operand(s)
~
"
I'J
3'
-<
s:
Operation Performed
S
X
N
Z
V
C
X
X
0
0
sadr. Dn
2.4
or 6
4(1/0)+
MOVE.w
sadr.An
2.4
or 6
4(110)+
MOVE.w
rs.dadr
2.4
or 6
5(0/1)+
X
X
0
0
[dadrl - [rs<0-15>1
Store word to memory location from data or address register. 2• 3
MOVE.w
sadr.dadr
2.4
6.8
or 10
5(0/1)+
X
X
0
0
[dadrl - [sadri
Store word from source memory location to destination memory location. 2• 3
MOVE.L
sadr.Dn
2.4
or 6
4(110)+
X
X
0
0
[Dn<0-31>1- [sadrl
Load long word to data register from memory location. 2• 3
MOVE.L
sadr.An
2.4
or 6
8(2/0)+
MOVE.L
rs.dadr
2.4
or 6
10(0/2)+
X
X
0
0
[dadrl - [rs<0-31 >1
Store long word from data or address register to memory location. 2• 4
MOVE.L
sadr.dadr
2.4
6.8
or 10
14(112)+
X
X
0
0
[dadrl - [sadri
Store long word from source memory location to destination memory
location.2. 4
MOVEM.w
jadr.reg-list
4.6
or 8
8 + 4n(2 + n/O)+
II>
CD
Status
T
MOVE.w
CI.
U,
Clock
Cycles
Bytes
3
g
[Dn<0-15>1 - [sadri
Load word to data register from memory location. Bits 16-31 of the
data register are not affected. 2• 3
[Anl<0-15>1
[An <16-31 >1- [An<15>1
Load word to address register from memory location. The sign is
extended to all upper bits of the register. 2• 3
[An <0-31 >1 - [sadrI
Load long word to address register from memory location. 2• 4
-<
[reg 1 <0-15>1 - [[Anll. [reg 1 < 16-31 >1 - [reg 1 < 15>1
[reg2<0-15>1- [[An + 2ll.[reg2<16-31>1- [reg2<15>1
[reg3<0-15>1- [[An + 4ll.[reg3<16-31>1- [reg3<15>1
:II
i-ii
[reg n <0-15>1- [[An + 2n-2)IUregn<16-31 >1 - [regn<15>1
Load multiple words from sequential memory locations to specified
registers. in order 00-07. AO-A 7. The sign is extended to all upper bits
of the register.2
:::I
n
CD
(';
0
3-
S·
c
CD
e:
MOVEM.w
(An)+.reg-list
4
8
+ 4n(2
+ n/O)
[reg 1 <0-15>1 - [[Anll.[reg1 <16-31 >1- [reg 1 UAnl[An + 21
[reg2<0-15>1- [[AnlJ.[reg2<16-31 >1- [reg2<15>J.[Anl[An + 21
[reg3<0-l5>I- [[AnIUreg3<16-31 >1- [reg 3 < 15>1. [Anl[An + 21
[regn <0-15>1 - [[Anll. [regn< 16-31 >1 - [regn< 15>J.[Anl [An + 21
Same as above except with postincrement. 3
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
MOVEMW
Operand(s)
reg-list,madr
Bytes
4,6
or 8
Clock
Cycles
Status
Operation Performed
T
S
X
N
Z
V
C
4 + 5n(1/n)+
[(Anll - [reg 1 <0-15>]
[(An + 211 - [reg2<0-15>]
[(An + 411 - [reg3<0-15>]
[(An + (2n-2)] - [regn<0-15>]
Store multiple words to sequential memory locations from specified
registers, in order 00-07, AO-A7.2
MOVEMW
reg-list,-(An)
4
[An]- [An-2J.[[An11-lregn<0-15>]
4 + 5n(1/n)+
[An] [An] [An] Store
ment
0
QI
:::I
Q.
~
3'
-<
MOVEM.L
jadr,reg-list
QI
~
(11
Co)
(An)+,reg-list
reg-list.madr
3:
3
"g
<
::u
4,6
or 8
4
4,6
or 8
8 + 8n(2 + 2n/0)
8 + 8n(2 + 2n/0)
4 + lOn(l/n)+
th.. _" 32 M' of tho
".;,te~
.re
4
4
4 + lOn(l/n)
d16(Anl,On
16(4/0)
[On<8-15>] - [[An] + d16],[An] - [An] + 2
[On<0-7>] - [[An] + d16]
Load peripheral data bytes from alternate memory locations to data
register word. The address is a byte address. 3
MOVEPW
On,d16(An)
4
18(2/2)
[[An] + d16] - [On<8-15>],[An] - [An] + 2
[[An] + d16] - [On<0-7>]
Store peripheral data bytes from data register long to alternate
memory locations. The address is a byte address. 3
MOVEP.L
dos(Anl.Dn
4
24(6/0)
[On<24-31»- [(An) + d16),[An)- [An) + 2
[On<16-23>]- [[An] + d16],[An]- [An] + 2
[On<8-15>] - [[An] + d16],[An] - [An] + 2
[On<0-7>]- [[An) + d16]
Load peripheral data bytes from alternate memory locations to data
register long. The address is a byte address. 3
MOVEP.L
Dn,d16(An)
4
28(214)
[[An] + d16] - [On <24-31 >],[An] - [An] + 2
[[An) + d16)- [On < 16-23»,[An) -[An] + 2
[[An] + d16] - [On<8-15>],[An] - [An] + 2
[[An] + d16] - [On<0-7>]
Store peripheral data bytes from data register long to alternate
memory locations. The address is a byte address. 3
reg-list,- (An)
CD
:::I
C'I
~
='c
"S:
~pt
s.m." MOVEMW ..
moved. 2,4
MOVEP. W
"
it
"
n
0
}
[An-2],[[An11 - [reg3<0-15>]
[An-2],[[An11 - [reg2<0-15>]
[An-2],[[An11 - [reg 1 <0-15>]
multiple words to sequential memory locations with predecreto specified registers, in order A7-AO, 07-00. 2, 3
Table 1-6.. MC68000 Instruction Set Summary (Continued)
Status
Operand(s)
Bytes
Clock
Cycles
X
N
Z
V
C
ABeD
-(sAnl.-(dAn)
2
19(3/1)
X
u
x
u
X
[sAnl - [sAnl - 1
[dAnl - [dAnl-1
[[dAn)] - [[dAn)] + [[sAn)] + X
Add decimal memory byte to memory byte with carry (Extend bit).
Both addresses are byte 1.
AOO.B
sadr.On
2.4
or 6
4(110)+
X
X
X
X
X
[On<0-7>1 - [On<0-7>1 + [sadri
Add byte to data register from memory location. Bits 8-31 of the data
register are not affected. 1
AOO.B
On.dadr
2.4
or 6
9(1/1)+
X
X
X
X
X
[dadrl - [dadrl + [On<0-7>J
Add byte to memory location from data register. 1
AOO.w
sadr.On
2.4
or 6
4(110)+
X
X
X
X
X
[On<0-15>1 - [On<0-15>1 + [sadrJ
Add word to data register from memory location. Bits 16-31 of the
data register are not affected. 2• 3
AOO.w
sadr.An
2.4
or 6
8(1/0)+
CI.
I»
ADD.W
Dn.Dadr
2.4
or 6
9(1/1)+
X
X
X
X
X
[dadrl - [dadrl + [On<0-15>J
Add word to memory location from data register. 2. 3
3:
CD
3
AOO.L
sadr.On
2.4
or 6
6(1/0)+
X
X
X
X
X
-..J
[On <0-31 >1 - [On <0-31 >1 + [sadri
Add long word to data registers from memory location. 2• 4
U'1
-<
AOO.L
sadr.An
2.4
or 6
6(1/0)+
AOO.L
On.dadr
2.4
or 6
14(1/2)+
X
X
X
X
X
[dadrl - [dadrl + [On <0-31 >1
Add long word to melnory locations from data register.2. 4
AODX.B
- (sAn). -dAn)
19(3/1)
X
X
X
X
X
[sAnJ - [sAnJ - 1
[dAnJ - [dAnl - 1
[[dAn]] - [[dAn]] + [[sAn]] + X
Add memory byte to memory byte with carry (Extend bitl. Both
addresses are byte. 1
Mnemonic
C/I
CD
n
I
Operation Performed
T
5
[An <0-31 >1 - [An <0-31 >1 + [sadrl (sign extended)
Add word to address register from memory location. The sign of the
memory word is extended to a full 32 bits for the operation. 2• 3
0
=
-<
I
~
0
::u
CD
it
iil
~
CD
~
CD
30
2
0
~
AOOX.w
-(sAn).-(dAn)
2
19(3/1)
X
X
X
X
X
~
AOOX.L
-(sAn).-(?An)
2
32(5/2)
I
[An <0-31 >1 - [An <0-31 >1 + [sadri
Add long word to address register from memory location. 2• 4
-<
'tI
CD
I
X
X
X
X
X
[sAnJ - [sAnJ - 2
[dAnJ - [dAnJ - 2
[[dAn]] - [[dAn]] + [[sAn)] + X
Add memory word to memory word with carry (Extend bit). Both
address are word. 2• 3
[sAnl - [sAnl - 4
[dAnl - [dAnl - 4
[[dAn]] - [[dAn]] + [[sAn]] + X
Add memory long word to memory long word with carry (Extend bid.
Both addresses are long word.2. 4
ANO.B
sadr.On
2.4
or 6
4(1/0)+
X
X
0
0
[On<0-7>J - [On<0-7>J < [sadrJ
AND byte to data register from memory location. Bits 8-31 of the data
register are not affected. 1
AND.B
Dn.dadr
2.4
or 6
9(1/1)+
X
X
0
0
[dadrl - [dadrl < [On<0-7>J
AND byte to memory location from data register. 1
I
I
I
I
I
I
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Operand(s)
Bytes
Status
Clock
Cycles
Operation Performed
T
S
X
N
Z
V
C
ANO.w
sadr,On
2,4
or6
4(1/0)+
X
X
0
0
[On<0-15>1 - [On<0-15>1 /\ [sadri
AND word to data register from memory location. Bits 16-31 of the
data register are not affected. 2, 3
ANO.W
On,dadr
2,4
or 6
9(1/1)+
X
X
0
0
[dadrl - [dadrl < [On<0-15>1
AND word to memory location from data register. 2, 3
ANO.L
sadr,Dn
2,4
or 6
6(1/0)+
X
X
0
0
[On <0-31 >1 - [On <0-31 >1 /\ [sadrI
AND long word to data register from memory location. 2, 4
ANO.L
On,dadr
2,4
or 6
14(1/2)+
X
X
0
0
[dadr! - [dadrl < [On <0-31 >!
AND long word to memory location from data register. 2, 4
CLR.B
dadr
2,4
or 6
9(1/1)+
0
1
0
0
[dadrl-O
Clear memory byte to zeroes. 1
CLR.w
dadr
2,4
or 6
9(1/1)+
0
1
0
0
[dadrl - 0
Clear memory word to zeroes. 2, 3
CLR.L
dadr
2,4
or 6
14(1/2)+
0
1
0
0
CMP.B
sa(Jr,On
2,4
or 6
4(1/0)+
X
X
X
X
[On <0-7> I - [sadrl
Compare data register byte with memory byte and set condition codes
accordingly. Register/memory data are not changed on any compares. 1
CMP.w
sadr,On
2,4
or 6
4(1/0)+
X
X
X
X
[On<0-15>! - [sadrl
Compare data register word with memory word and set condition
codes accordingly.2, 3
n
CMP.w
sadrAn
2,4
or 6
6(110)+
X
X
X
X
~
(I
3
[An<0-15>1 - [sadri
Compare address register word with memory word and set condition
codes accordingly.2, 3
CMP.L
sadr,On
2,4
or 6
6(1/0)+
X
X
X
X
[On <0-31 >1 - [sadrI
Compare data register with memory long word and set condition
codes accordingly.2, 4
CMP.L
sadrAn
2,4
or 6
6(1/0)+
X
X
X
X
[An <0-31 >1 - [sadrI
Compare address register with memory long word and set condition
codes accordingly.2, 4
CJ)
CD
n
0
:::I
Co
III
-<
3:
(I
3
......
&.
(71
~
<
:II
!.
CD
CD
:::I
(I
0
-<
0
'C
[dadrl - 0
Clear memory long word to zeroes. 2, 4
(I
Ql
!
n
0
~
CMPM.B
(sAn)+,(dAn)+
2
12(3/0)
X
X
X
X
[[dAnl1 - [[sAn))
[dAn! - [dAn! + 1
[sAn! - [sAn! + 1
Compare memory bytes and set condition codes accordingly. The
memory data are not changed on any compares. 1
CMPM.w
(sAn)+,(dAn)+
2
12(3/0)
X
X
X
X
[[dAn)) - [[sAn))
[dAn I - [dAnl + 2
[sAnl - [sAnl + 2
Compare memory words and set condition codes accordingly.2, 3
CMPM.L
(sAn)+,(dAn)+
2
20(5/0)
X
X
X
X
[[dAn)) - [[sAn))
[dAnl - [dAnl + 4
[sAn! - [sAnl + 4
Compare memory long words and set condition codes accordingly.2, 4
5'
c
(I
!:
- -
-
..
-
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Ul
CD
Operand(s)
Q.
III
3
-<
(J)
it
u,
::II
CD
Operation Performed
T
S
X
N
Z
V
C
2.4
or 6
< 158(1/0)+
X
X
X
0
[On<0-15>1 ~ [On <0-31 >1 +- [sadrI
[On < 16-31 > I ~ remainder
Divide signed numbers. Division by zero causes a TRAP. The source
address is a word address. 2• 3
OIVU
sadr.On
2.4
or 6
~140(1/0)+
X
X
X
0
[On<0-15>1 ~ [On <0-31 >1 +- [sadrI
[On<16-31 >1 ~ remainder
Oivide unsigned numbers. Division by zero causes a TRAP. The source
address is a word address. 2• 3
EOR.B
On.dadr
2.4
or 6
9(1/1)+
X
X
0
0
[dadrl ~ [dadrl ¥ [On<0-7>1
Exclusive-OR byte to memory location from data register. 1
EORW
On.dadr
2.4
or 6
9(1/1)+
X
X
0
0
[dadr] ~ [dadr] ¥ [On<0-15>1
Exclusive-OR word to memory location from data registers. 2• 3
EORl
On,dadr
2.4
or 6
14(1/2)+
X
X
0
0
[dadrl ~ [dadr] ¥ [On<0-31 >1
Exclusive-OR long word to memory location from data register.2. 4
MUlS
sadr.On
2.4
or 6
<70(1/0)+
X
X
0
0
[On <0-31 >1 ~ [On<0-15>1 x [sadrI
Multiply two 16-bit signed numbers. yielding a 32-bit signed product.
The source address is a word address.2. 3
MUlU
sadr.On
2.4
or 6
<74(210)+
X
X
0
0
[On <0-31 >1 ~ [On<0-15>1 x [sadrI
Multiply two 16-bit unsigned numbers. yielding a 32-bit unsigned product. The source address is a word address.2. 3
NBCO
dadr
2.4
or 6
9(1/1)+
X
U
X
U
X
[dadrl ~ 0 - [dadrl - X
Negate decimal memory byte. This operation produces the tens complement if X
0 or the nines complement if X
1.
NEG.B
dadr
2.4
or 6
9(111)+
X
X
X
X
X
[dadrl ~ 0 - [dadrl
Negate memory byte. 1
NEGW
dadr
2.4
or 6
9(1/1)+
X
X
X
X
X
[dadrl ~ 0 - [dadr]
Negate memory word.2. 3
2.4
or 6
14(112)+
X
X
X
X
X
[dadrl ~ 0 - [dadrl
Negate memory long word. 2• 4
2.4
or 6
9(1/1)+
X
X
X
X
X
[dadrl ~ 0 - [dadrl - X
Negate memory byte with Extend bit. 1
2.4
or 6
9(1/1)+
X
X
X
X
X
[dadrl ~ 0 - [dadr] - X
Negate memory word with Extend bit.2. 3
X
0
.....,j
Status
sadr.On
-<
s:CD
Clock
Cycles
OIVS
()
0
::I
Bytes
Cil
~
=
CD
~
CD
3
~
'<
0
"C
CD
NEG.l
dadr
!..
NEGX.B
dadr
~
n
0
~
5'
NEGXW
dadr
c
CD
e:
=
NEGX.l
dadr
2.4
or 6
14(1/2)+
X
X
X
X
[dadr] ~ 0 - Idadrl - X
Negate memory long word with Extend bit.2. 4
NOT.B
dadr
2.4
or 6
9(1/1)+
X
X
0
0
[dadrl ~ Idadr]
Ones complement memory byte. 1
NOTW
dadr
2.4
or 6
9(111)+
X
X
0
0
[dadrl ~ [dadrl
Ones complement memory word.2. 3
NOT.L
dadr
2.4
or 6
14(1/2)+
X
X
0
0
Idadr] ~ [dadr]
Ones complement memory long word. 2• 4
ORB
sadr,On
2,4
or 6
4(1/0)+
X
X
0
0
[On<0-7>1 ~ [Dn<0-7>1 V [sadrl
OR byte to data register from memory location. Bits 8-31 of the data
register are not affected. 1
ORB
On.dadr
2.4
or 6
9(1/1)+
X
X
0
0
Idadrl ~ [dadrl V [On<0-7>1
OR byte to memory location from data register. 1
Table 7-6, MC68000 Instruction Set Summary (Continued)
C'
Mnemonic
Operand(s)
Bytes
k
C o,c
yc es
Status
Operation Performed
T
S
X
N
Z
V
C
ORW
sadr,On
2. 4
or 6
4(1/0)+
X
X
0
0
[On<0-15>1 - [On<0-15>1 V [sadrI
OR word to data register from memory location, Bits 16-31 ofthe data
register are not affected,2, 3
ORW
On,dadr
2,4
or 6
9(1/1)+
X
X
0
0
[dadr] - [dadrJ < [On<0-15>J
OR word to memory location from data register,2, 3
ORL
sadr,On
2,4
or 6
6(1/0)+
X
X
0
0
[On <0-31 >1 - [On <0-31 >] V [sadrI
OR long word to data register from memory location,2, 4
ORL
On,dadr
2,4
or 6
14(112)+
X
X
0
0
[dadrl - [dadr] < [OnVO-31 >]
OR long word to memory location from data register,2, 4
SBeo
-(sAnl,-(dAn)
U
X
U
X
[sAn] - [sAn] - 1
[dAn] - [dAn] - 1
[[dAn]] - [[dAn]] - [[sAn]] - X
Subtract decimal memory byte from memory byte with carry (Extend
bit), Both addresses are byte,1
see
2
19(3/1)
X
dadr
2,4
or 6
9(111)+
SUB,B
sadr,On
2,4
or 6
4(1/0)+
X
X
X
X
X
[On<0-7>] - [On<0-7>] - [sadri
Subtract memory byte from f)yte in data register, Bits 8-31 of the data
register are not affected,1
SUB.B
On,dadr
2,4
or 6
9(1/1)+
X
X
X
X
X
[dadrl - [dadrl - [On<0-7>]
Subtract byte in data register from memory byte. 1
SUBW
sadr,On
2,4
or 6
4(1/0)+
X
X
X
X
X
[On<0-15>1 - [On<0-15>] - [sadri
Subtract memory word from word in data register. Bits 16-31 of the
data register are not aftected,2, 3
SUBW
sadrAn
2,4
or 6
8(1/0)+
X
X
X
X
X
[An <0-31 >] - [An <0-31 >1 - [sadri (sign extended)
Subtract memory word from address register contents. The sign of the
memory word is extended to a full 32 bits for the operation,2, 3
~
SUBW
On,dadr
9(1/1)+
X
X
X
X
X
R
2,4
or 6
[dadrl - [dadr] - [On<015>]
Subtract data register word from memory location word,2, 3
SUB,L
sadr,On
2,4
or 6
6(1/0)+
X
X
X
X
X
[On<0-31>]-[On<0-31>]-[sadr]
Subtract memory long word from data register contentS,2, 4
g.
SUB.L
sadr,An
2,4
or 6
6(1/0)+
X
X
X
X
X
[An<0-31>]-[An<0-31>]-[sadr]
Subtract memory long word from address register contents.2, 4
-
SUB.L
On,dadr
14(1/2)+
X
X
X
X
X
SUBX,B
-(sAn).-(dAn)
2,4
or 6
2
19(3/1)
X
X
X
X
X
[dadr] - [dadr] - [On <0-31 >]
Subtract contents of data register from memory long word.2, 4
[sAn] - [sAnl - 1
[dAn] - [dAn] - 1
[[dAn]] - [[dAn]] - [[sAn]] - X
Subtract memory byte from memory byte with borrow (Extend bit),
Both addresses are byte. 1
SUBX.w
-(sAnl,-(dAn)
2
19(3/1)
X
X
X
X
X
C/l
~
~
91
i
co
5
-..J
&.
-..J
-<
~
;;
iil
~
E
5
-<
=
[dadrJ - [all 1's if cc
TRUE
[dadr] - all O's if cc = FALSE
Set status in memory byte, 1
o
"g
~
g
~
[sAn] - [sAnJ - 2
[dAnl - [dAnl - 2
[[dAn]] - [[dAn]] - [[sAn]] - X
Subtract memory word from memory word with borrow (Extend bitl.
Both addresses are word. 2, 3
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
2i
Operand(s)
Bytes
Clock
Cycles
Status
Operation Performed
T
S
X
N
Z
V
C
X
X
X
X
X
[sAnl - [sAnl - 4
[dAnl - [dAnl - 4
[[dAn]] - [[dAn]] - [[sAn]] - X
Subtract memory long word from memory long word with borrow
(Extend bitl. Both addresses are long word.2, 4
SUBX.L
-(sAnl.-(dAn)
:i" <
TAS
dadr
2,4
or 6
11(111)+
X
X
0
0
[dadr<7>1 - 1
Test status of memory byte and set high-order bit to 1.
-3
TST.B
dadr
2,4
or 6
4(1/0)+
X
X
0
0
[dadrl - 0
Test status of memory byte. The byte value is not changed.
TST.W
dadr
2,4
or 6
4(1/0)+
X
X
0
0
[dadrl - 0
Test status of memory word. The word value is not changed.
TST.L
dadr
2,4
or 6
4(1/0)+
X
X
0
0
[dadrl - 0
Test status of memory long word. The long word value is not changed.
MOVEa
data8,On
2
4(1/0)
X
X
0
0
[On<0-7>1- data8
[On <8-32>1 - [On <7>1
Load immediate data byte to data register. The sign is extended to all
upper bits of the data register.
MOVE.B
data8,On
4
8(2/0)
X
X
0
0
[On<0-7>1 - data8
Load immediate data byte to data register. Bits 8-31 of the data
register are not affected.
MOVE.B
data8,dadr
9(111)+
X
X
0
0
[dadrl - [data81
Load immediate data byte into memory location. 1
MOVE.W
data16,On
4
8(2/0)
X
X
0
0
[On<0-15>1 - data16
Load immediate data word to data register. Bits 16-31 of the data
register are not affected.
'MOVE.W
data16,An
4
8(2/0)
2
32(5/2)
I»
:::I
Co
n~
o 3
~ ~
;Co s:CD
0
-<
:Jl
!!.
CD
(;
:::I
0
CD
-..I
&.
en
4,6
or 8
3'
3
CD
Co
§"
[An <0-15>1 - data16
[An < 16-31 > I - [An < 15> I
Load immediate data word to address register. The sign is extended to
all upper bits of the register.
CD
3'
MOVEW
data 16,dadr
4,6
or 8
9(111)+
X
X
0
0
[dadrl - data16
Load immediate data word into memory location. 2, 3
MOVE.L
data32,On
6
12(3/0)
X
X
0
0
[On <0-31 >1 - data32
Load immediate data long word into data register.
MOVE.L
data32,An
6
12(3/0)
MOVE.L
data32,dadr
AOO.B
data8,On
AOO.B
data8,dadr
4,6
or 8
AOOW
data16,On
4
6,8
or 10
4
[An <0-31 >1- data32
Load immediate data long word into address register.
18(2/2)+
X
X
0
0
[dadrl - data32
Load immediate data long word into memory location. 2, 4
8(2/0)
X
X
X
X
X
[On<0-7>1 - [On<0-7>1 + data8
Add immediate data byte to data register. Bits 8-31 of the data register
are not affected.
13(2/1)+
X
X
X
X
X
[dadrl - [dadrl + data8
Add immediate data byte to memory location. 1
8(2/0)
X
X
X
X
X
[On<0-15>1 - [On<0-15>1 + data16
Add immediate data word to data register. Bits 16-31 of the data
rt)~ter are not affected.
3
CD
Co
iii'
i
0
~
~
CD
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Operand(s)
ADD.w
data16,An
ADD.w
data 16,dadr
AOO.L
data32,On
ADD.L
data32,An
ADD.L
data32,dadr
Bytes
4
Operation Performed
T
S
X
N
Z
V
C
[An <0-31> 1 - [An <0-31> 1 + data 16 (sign extended)
Add immediate data word to address register. The sign of the data
word is extended to a full 32 bits for the operation.
8(210)
X
X
X
X
X
[dadr] - [dadrl + data16
Add immediate data word to memory location. 2, 3
6
16(3/0)
X
X
X
X
X
[Dn<0-31 >1- [Dn<0-31 >1 + data32
Add immediate data long word to data register.
6
16(3/0)
6,8
or 10
22(3/2)+
[An <0-31 >1 - [An <0-31 >1 + data32
Add immediate data long word to address register.
X
X
X
X
X
~
CD
X
X
X
X
X
[Dn<0-7>1- [On<0-7>1 + data3
Add immediate three bits to data register byte. Bits 8-31 of the data
register are not affected.
9(1/0)+
X
X
X
X
X
[dadrl - [dadrl + data3
Add immediate three bits to memory byte. 1
2
4(1/0)
X
X
X
X
X
[Dn<0-15>1 - [Dn<0-15>1 + data3
Add immediate three bits to data register word. Bits 16-31 of the data
register are not affected .
2
4(1/0)
ADDQ.B
data3,dadr
ADDQ.W
data3,Dn
ADDQ.W
data3,An
!;:
AODQ.W
data3,dadr
00
ADDQ.L
data3,Dn
ADDQ.L
data3,An
AOOQ.L
data3,dadr
AND.B
data8,Dn
AND.B
data8,dadr
4,6
or 8
ANO.w
data16,On
ANO.w
data 16,dadr
AND.L
data32,Dn
AND.L
data32,dadr
2
2,4
or 6
~
~.
[An<0-15>1 - [An<0-15>1 + data3
Add immediate three bits to address register word. Bits 16-31 of the
address register are not affected.
0
og
;3.
:i"
c
..
[dadrl + data32
Add immediate data long word to memory location. 2, 4
3
U1
[dadr] -
4(1/0)
data3,Dn
..
..
Status
13(211)+
4,6
or 8
AODQ.B
3"
Clock
Cycles
9(1/1)+
X
X
X
X
X
[dadrl - [dadrl + data3
Add immediate three bits to memory word. 2, 3
2
8(1/0)
X
X
X
X
X
[Dn<0-31>1- [Dn<0-31 >1 + data3
Add immediate three bits to data register long word.
2
8(1/0)
2,4
or 6
[An <0-31> 1 - [An <0-31> 1 + data3
Add immediate three bits to address register long word.
1:
X
X
X
X
[dadrl - [dadrl + data3
Add immediate three bits to memory long word. 2, 4
8(2/0)
X
X
0
0
[Dn<0-7>1 - [Dn<0-7>1 A data8
AND immediate data byte to data register. Bits 8-31 of the data
register are not affected.
13(2111+
X
X
0
0
[dadrl - [dadrl A data8
AND immediate data byte to memory byte. 1
4
8(2/0)
X
X
0
0
[On<0-15>1 - [On<0-15>1 A data16
AND immediate data word to data register. Bits 16-31 of the data
register are not affected.
4,6
or 8
13(2/1)
X
X
0
0
[dadrl - [dadrl A data16
AND immediate data word to memory word. 2. 3
6
16(3/0)
X
X
0
0
[Dn<0-31>1- [Dn<0-31 >1 A data32
AND immediate data long word to data register.
22(3/2)+
X
X
0
0
[dadrl - [dadrl < data32
AND immediate data long word to memory.2, 4
2,4
or 6
4
6,8
or 10
14(1/2)
X
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Operand(s)
Bytes
Clock
Cycles
Status
Operation Performed
N
Z
V
C
8(210)
X
X
X
X
[Dn <0-7> 1 - data8
Compare data register byte with immediate data byte and set condition codes accordingly. Register data are not changed on any compares.
8(2/0)+
X
X
X
X
[dadrl - data8
Compare memory byte with immediate data byte and set condition
codes accordingly.1
T
S
X
CMP.B
data8.Dn
CMP.B
data8.dadr
4.6
or 8
CMPW
data16.Dn
4
8(2/0)
X
X
X
X
[Dn<0-15>1- data16
Compare data register word with immediate data word and set condition codes accordingly.
CMPW
data16.An
4
8(210)
X
X
X
X
[An < 0-15 > 1 - data 16
Compare address register word with immediate data word and set
condition codes accordingly.
CMPW
data 16.dadr
4.6
or 8
8(2/0)+
X
X
X
X
[dadrl - data 16
Compare memory word with immediate data word and set condition
codes accordingly.2. 3
CMP.l
data32.Dn
6
14(3/0)
X
X
X
X
[Dn<0-31 >1 - daJa32
Compare data register with immediate data long word and set condition codes accordingly.
CMP.l
data32.An
6
14(3/0)
X
X
X
X
[An<0-31>1- data32
Compare address register with immediate data long word and set condition codes accordingly.
CMP.l
data32.dadr
12(3/0)+
X
X
X
X
[dadrl - data32
Compare memory long word with immediate data long word and set
condition codes accordingly.2. 4
DIVS
data16.Dn
4
:s; 162(210)
X
X
X
0
[Dn<0-15>1 ~ [Dn<0-31 >1 -;- data16
[Dn < 16-31 > 1 ~ remainder
Divide signed numbers. Division by zero causes a TRAP.
DIVU
data16.Dn
4
:s; 148(210)
X
X
X
0
[Dn<0-15>1 ~ [Dn<0-31 >1 -;- data16
[Dn< 16-31 >1 ~ remainder
Divide unsigned numbers. Division by zero causes a TRAP.
EORB
data8.Dn
4
8(2/0)
X
X
0
0
[Dn<0-7>1 ~ [Dn<0-7>1 ¥data8
Exclusive-OR data byte to data register. Bits 8-31 of the data register
are not affected.
EOR.B
data8.dadr
4.6
or 8
13(2/1)+
X
X
0
0
[dadrl ~ [dadrl !oJ. data8
Exclusive-OR data byte to memory byte. 1
EORW
data16.Dn
4
8(210)
X
X
0
0
[Dn<0-15>1 ~ [Dn<0-15>1 !oJ. data16
Exclusive-OR data word to data register. Bits 16-31 of the data register
are not affected.
EORW
data 16.dadr
13(211)+
X
X
0
0
[dadrl ~ [dadrl !oJ. data16
Exclusive-OR immediate data word to memory word. 2• 3
EORl
data32.Dn
16(3/0)
X
X
0
0
[Dn<0-31 >1 ~ [Dn>0-31 >1 ¥ data32
Exclusive-OR immediate data long word to data register.
EORl
data32.dadr
22(3/2)+
X
X
0
0
[dadrl ~.[dadrl ¥ data32
Exclusive-OR immediate data long word to memory.2. 4
4
3'
3
...,J
eno
~
Q.
~
~
0
'C
~
!
6.8
or 10
~
n
~
~.
~
e:
4.6
or 8
6
6.8
or 10
I
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
a,
--
Bytes
Clock
Cycles
Status
T
S
X
Operation Performed
N
Z
V
C
MULS
data16,On
4
::;74(210)
X
X
0
0
[On<0-311 >1 - [On<0-15>1 x data16
Multiply two 16-bit signed numbers, yielding a 32-bit signed product.
MULU
data16,On
4
~74(2/0)
X
X
0
0
[On <0-31 >1 - [On<0-15>1 x data16
Multiply two 16-bit unsigned numbers, yielding a 32-bit unsigned product.
OR.B
data8,On
4
8(2/0)
X
X
0
0
[On<0-7>1 - [On<0-7>1 V data8
OR immediate data byte to data register. Bits 8-31 of the data register
are not affected.
OR.B
data8,dadr
4,6
or 8
13(2/1)+
X
X
0
0
[dadrl - [dadrl V data8
OR immediate data byte to memory byte. 1
ORW
data16,On
4
8(210)
X
X
0
0
[On<0-15>1 - [On<0-15>1 V data16
OR immediate data word to data register. Bits 16-31 of the data
register are not affected.
ORW
data 16.dadr
13(2/1)+
X
X
0
0
[dadrl - [dadrl V data16
OR immediate data word to memory word. 2, 3
OR.L
data32,On
16(3/0)
X
X
0
0
[On <0-31 >1 - [On <0-31 >1 V data32
OR immediate data long word to data register.
OR.L
data32,dadr
22(3/2)+
X
X
0
0
[dadrl - [dadrl V data32
OR immediate data long word to memory.2, 4
SUB.B
data8,On
~
SUB.B
data8,dadr
4,6
or 8
n
0
SUBW
data16,On
SUBW
data 16,An
SUBW
data 16,dadr
SUB.L
data32,On
SUB.L
data32.An
SUB.L
data32,dadr
SUBQ.B
data3,On
SUBQ.B
data3.dadr
§'
3
-...J
Operand(s)
11>
c.
~
4,6
or 8
6
6,8
or 10
8(2/0)
X
X
X
X
X
[On<0-7>1- [On<0-7>1 - data8
Subtract immediate data byte from data register. Bits 8-31 of the data
register are not affected.
13(2/1)+
X
X
X
X
X
[dadrl - [dadrl - data8
Subtract immediate data byte from memory byte. 1
4
8(2/0)
X
X
X
X
X
[On<0-15>1 - [On<0-15>1 - data16
Subtract immediate data word from data register. Bits 16-31 of the
data register are not affected.
4
8(2/0)
4
0
'0
CD
11>
~
;0'
c
11>
e:
[An <0-31>1- [An<0-31>1- data16 (sign extended)
Subtract immediate data word from address register. The sign of the
data word is extended to a full 32 bits for the operation.
13(2/1)+
X
X
X
X
X
[dadrl - [dadrl - data16
Subtract immediate data word from memory word. 2, 3
6
16(3/0)
X
X
X
X
X
[On <0-31 >1 - [On <0-31 >1 - data32
Subtract immediate long word from data register contents.
6
16(3/0)
4.6
or 8
6.8
or 10
2
2.4
or 6
[An <0-31 >1 - [An <0-31 >1 - data32
Subtract immediate data long word from address register.
22(3/2)+
X
X
X
X
X
[dadrl - [dadrl - data32
Subtract immediate data long word from memory word. 2, 4
4(1/0)
X
X
X
X
X
[On <0-7>]- [On <0-7> I - data3
Subtract immediate three bits from data register byte. Bits 8-31 of the
data register are not affected.
9(1/1)+
X
X
X
X
X
[dadrl - [dadr] - data3
Subtract immediate three bits from memory byte. 1
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Operand(s)
Bytes
Clock
Cycles
Status
Operation Performed
T
S
X
N
Z
V
C
X
X
X
X
X
SUBQ.W
data3.0n
2
4(1/0)
SUBQ.W
data3.An
2
4(1/0)
..
SUBQ.W
data3.dadr
no
SUBQ.L
data3.0n
:i'
c
SUBQ.L
data3.An
1:
SUBQ.L
data3.dadr
2.4
or 6
14(112)+
BRA
label
2 or 4
10(2/0)
[PC] -label
Branch unconditionally (short).
JMP
jadr
2.4
or 6
4(110)+
[PC]- jadr
Jump unconditionally.
BSR
label
2 or
4
10.8(1/0)
10.12(210)
[A7] - [A7] - 2
[[A71l- [PC]
[PC]-Iabel
Branch to subroutine (short).
JSR
jadr
2.4
or 6
14(1/2)+
[A7] - [A7] - 2
[[A7]]- [PC]
[PC]- jadr
Jump to subroutine.
3'
3
CD
a.
~.
[An<0-15>]- [An<0-15>] - data3
Subtract immediate three bits from address register word. Bits 16-31
of the address register are not affected.
CD
0
"C
CD
CD
3-
9(1/1)+
X
X
X
X
X
[dadr] - [dadr] - data3
Subtract immediate three bits from memory word. 2. 3
2
8(1/0)
X
X
X
X
X
[On<0-31>]- [On<0-31>]- data3
Subtract immediate three bits from data register contents.
2
8(1/0)
2.4
or 6
[An<0-31>]- [An<0-31>]- data3
Subtract immediate three bits from address register contents.
CD
~ ~
~ 3:
o 'U
::t •
-.J
enN
VI
c
IT
(;
S.
:i'
CD
0
»
r
..
r
~
~
0
0
:::I
a.
;:;:
o·
:::I
X
[dadr] - [dadr] - data3
Subtract immediate three bits from memory long word.2. 4
RTR
2
20(5/0)
[SR<0-4>] - [[A7<0-4>ll
[A7] - [A7] + 2
[PC]- [[A7]]
[A7] - [A71 + 2
Restore condition codes and return from subroutine.
10.8(1/0)
10.12(2/0)
[PCI -label
Branch if condition met.
12(210)
10(210).
14(3/0)
If cc then no further action.
[On<0-15>1 - [On<0-15>1 - 1
If [On < 0-15 > I
-1 then no further action.
[PCI -label
Test condition. decrement and branch. Loop until the specified condition is true or until the loop count is exhausted.
Bee
label
OBcc
On.label
iii
g
X
[PC]- [[A7]]
[A7] - [A7] + 2
Return from subroutine
z
~
X
16(4/0)
l:I
01
X
2
l:I
m
X
RTS
:::I
a.
c:
[On<0-15>] - [On] - data3
Subtract immediate three bits from data register word. Bits 16-31 of
the data register are not affected.
2 or
4
4
=
--
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Operand(sl
Bytes
Clock
Cycles
Status
Operation Performed
T
S
X
N
Z
V
C
MOVE.B
sOn,dOn
2
4(1/01
X
X
0
0
(dOn <07> 1 - [sOn <0-7> 1
Move one byte of any data register to any data register. Bits 8-31 of
the destination register are not affected.
MOVEW
rS,On
2
4(1/01
X
X
0
0
[On<0-15>1 - [rs<0-15>1
Move one word of any data or address register to any data register.
Bits 16-31 of the destination register are not affected.
MOVEW
rS,An
2
4(1/0)
MOVE.L
rS,On
2
4(1/01
MOVE.L
rS,An
2
4(1/01
ABCO
sOn,dOn
2
6(1/01
X
u
X
U
X
[dOn <0-7>1 - [dOn <0-7>1 + [sOn <0-7>1 + X
Add decimal source data register byte to destination data register byte
with carry (Extend bitl. Bits 8-31 of the destination data register are
not affected.
AOO.B
sOn,dOn
2
4(1/0)
X
X
X
X
X
[dOn <0-7>1 - [dOn <0-7>1 + [sOn <0-7>1
Add byte from data registers to data register. Bits 8-31 of the destination data register are not affected.
AOOW
rs,On
2
4(1/01
X
X
X
X
X
[On<0-15>1 - [On<0-15>1 + [rs<0-15>1
Add word from source register to data register. Bits 16-31 of the
destination data register are not affected.
AOOW
rS,An
2
8(1/0)
i
AOO.L
rS,On
2
8(1/01
ca
AOO.L
rS,An
2
8(1/0)
AOOX.B
sOn,dOn
2
4(1/01
X
X
X
X
X
[dOn <0-7>1 - [dOn <0-71 + [sOn <0-7>1 + X
Add source data register byte to destination data register byte with
carry (Extend bid. Bits 8-31 of the destination data register are not
affected.
AOOXW
sOn. dOn
2
4(1/0)
X
X
X
X
X
[dOn<0-15>1- [dOn<0-15>1 + [sOn<0-15>1 + X
Add source data register word to destination data register word with
carry (Extend bit). Bits 16-31 of the destination data register are not
affected.
AOOKL
sOn,dOn
2
8(1/0)
X
X
X
X
X
[dOn <0-31 >1- [dOn <0-31 >1 + [sOn <0-31 >1 + X
Add source data register long word to destination data register long
word with carry (Extend bitl.
ANO.B
sOn,dOn
2
4(1/01
X
X
0
0
IdOn<0-7>1 - [dOn <0-7>1 < [sOn<0-7>1
ANO byte from data register to data register. Bits 8-31 of the destination data register are not affected.
:D
CD
ca
~
!t
21
[An<015>1 - [rs<0-15>1
[An < 16-31 > ] - [An<15>]
Move one word of any data or address register to any address register.
The sign is extended to all upper bits of the address register.
CD
ca
i
!t
s::
0
X
X
0
0
<
CD
en"w
[An<0-31>1- [rs<0-31 >1
Move the contents of any data or address register to any address
register.
[An <0-15>1 - [An <0-15>1 + [rs<0-15>1 (sign extendedl
Add word from source register to address register. The sign of the
source word is extended to a full 32 bits for the operation.
:D
CD
ca
0;'
X
X
X
X
X
21
CD
0;'
i
0
'C
[On <0-31> 1 - [rs<0-31 > 1
Move the contents of any data or address register to any data register.
[On<0-31>1- [On <0-31 >1 + [rs<0-31 >1
Add long word from source register to data register.
[An <0-31 >1 - [An <0-31 >1 + rS<0-31 >1
Add long word from source register to address register.
CD
~
CD
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Operand(s)
Bytes
Clock
Cycles
Status
Operation Performed
T
S
X
N
Z
V
C
ANO.w
sOn.dOn
2
4(1/0)
X
X
0
0
[dOn<0-15>1 - [dOn<0-15>1 < [sOn<0-15>1
AND word from data register to data register. Bits 16-31 of the
destination data register are not affected.
ANO.l
sOn.dOn
2
8(110)
X
X
0
0
[dOn<0-31>1- [dOn <0-31 >1 < [sOn <0-31 >1
AND long word from data register to data register.
CMP.B
sOn.dOn
2
4(1/0)
X
X
X
X
[dOn<0-7>1 - [sOn<0-7>1
Compare data register bytes and set condition codes accordingly.
Register data are not changed on any compares.
CMP.w
rs.On
2
4(1/0)
X
X
X
X
[On<0-15>1 - [rs<0-15>1
Compare data register word with register word and set condition
codes accordingly.
CMP.w
rs.An
2
6(1/0)
X
X
X
X
[An <0-15>1 - [rs<0-15> I
Compare address register word with register word and set condition
codes accordingly.
CMP.L
rs.On
2
6(1/0)
X
X
X
X
[On <0-31 >1 - [rs<0-31 >1
Compare data register with register and set condition codes accordingly.
CMP.l
rs.An
2
6(1/0)
X
X
X
·X
[An <0-31> I - [rs<0-31 > 1
Compare address register with register and set condition codes accordingly.
OIVS
sOn.dOn
2
~158(1/0)
X
X
X
0
[dOn<0-15>1- [dOn <0-31 >1 + [sOn<0-15>1
[dOn < 016-31> I - remainder
Divide signed numbers. Division by zero causes a TRAP.
OIVU
sOn.dOn
2
~140(1/0)
X
X
X
0
[dOn<0-15>1- [dOn <0-31 >1 + [sOn<0-15>1
[dOn<16-31>1- remainder
Divide unsigned numbers. Division by zero causes a TRAP.
EOR.B
sOn.dOn
2
4(1/0)
X
X
0
0
[dOn <0-7>1 - [dOn<0-7>1 ¥ [sOn <0-7>1
Exclusive-OR byte from data register to data register. Bits 8-31 of the
destination data register are not affected.
EORW
sOn.dOn
2
4(110)
X
X
0
0
[dOn<0-15>1 - [dOn<0-15>1 ¥ [sOn<0-15>1
Exclusive-OR word from data register to data register. Bits 16-31 of the
destination data register are not affected.
EaRL
sOn.dOn
2
8(1/0)
X
X
0
0
[dOn <0-31 >1 - [dOn <0-31 >1 ¥ [sOn <0-31 >1
Exclusive-OR long word from data register to data register.
EXG
rs.rd
2
6(110)
MUlS
sOn.dOn
2
::;,70(110)
X
X
0
0
[dOn <0-31 >1 - [dOn<0-15>1 x [sOn<0-15>1
Multiply two 16-bit signed numbers. yielding a 32-bit signed product.
MUlU
sOn.dOn
2
::;70(1/0)
X
X
0
0
[dOn<0-31>1- [dOn<0-15>1 x [sOn<0-15>1
Multiply two 16-bit unsigned numbers. yielding a 32-bit unsigned product.
ORB
sOn.dOn
2
4(1/0)
X
X
0
0
[dOn<0-7>1- [dOn<0-7>1 V [sOn<0-7>1
OR byte from data register to data register. Bits 8-31 of the 'destination
dat register are not affected.
:ID
CD
cg
iii'
';"l
en
~
i
:II
CD
cg
!a'
~
0
'CI
~
~
CD
R
0
~
:i"
c
CD
~
[rdl-~
[rsl
Exchange the contents of two registers. This is always a long word
operation.
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
:a
CD
Operand(s)
Bytes
Clock
Cycles
Status
Operation Performed
T
S
X
N
Z
V
C
OR.W
sOn.dOn
2
4(1/0)
X
X
0
0
[dOn<0-15>1 - [dOn<0-15>1 V [sOn<0-15>1
OR word from data register to data register. Bits 16-31 of the destination data register are not affected.
ORL
sOn.dOn
2
8(1/0)
X
X
0
0
[dOn <0-31 >1 - [dOn <0-31 >1 V [sOn <0-31 >1
OR long word from data register to data register.
SBeO
sOn.dOn
2
6(1/0)
X
U
X
U
X
[dOn<0-7>1- [dOn<0-7>1 - [sOn <0-7>1 - X
Subtract decimal source· data register byte from destination data
register byte with carry (Extend bit). Bits 8-31 of the destination data
register are not affected.
SUB.B
sOn.dOn
2
4(1/0)
X
X
X
X
X
[dOn<0-7>1- [dOn<0-7>1 - [sOn <0-7>1
Subtract data register bytes. Bits 8-31 of the destination data register
are not affected.
SUB.W
rs.On
2
4(1/0)
X
X
X
X
X
[On<0-15>1 - [On<0-15>1 - [rs<0-15>1
Subtract register words. Bits 16-31 of the destination data register are
not affected.
SUBW
rs.An
2
8(1/0)
SUB.L
rs.On
2
8(1/0)
SUB.L
rs.An
2
8(1/0)
SUBX.B
sOn.dOn
2
4(1/0)
X
X
X
X
X
[dOn<0-7>1 - [dOn<0-7>1 - [sOn <0-7>1 - X
Subtract source data register byte from destination data register byte
with borrow (Extend bit). Bits 8-31 of the destination data register are
not affected.
SUBXW
sOn.dOn
2
4(1/0)
X
X
X
X
X
[dOn<0-15>1- [dOn<0-15>1 - [sOn<0-15>1 - X
Subtract source data register word from destination data register
word with borrow (Extend bit). Bits 16-31 of the destination data
registers are not affected.
SUBX.L
sOn.dOn
2
8(1/0)
X
X
X
X
X
[dOn<0-31>1- [dOn<0-31>1- [sOn<0-31>1- X
Subtract source data register long word from destination data register
long word with borrow (Extend bit).
CLRB
On
2
4(1/0)
0
1
0
0
[On<0-7>1 - 0
Clear data register byte to zeroes. Bits 8-31 of the data register are not
affected.
CLRW
On
2
4(1/0)
0
1
0
0
[On<0-15> - 0
Clear data register word to zeroes. Bits 16-31 of the data register are
not affected.
CLRL
On
2
6(1/0)
0
1
0
0
[On<0-31>1-0
Clear data register to zeroes
EXTW
On
2
4(1/0)
X
X
0
0
[On<8-15>1 - [On<7>1
Extend sign bit of data byte to data word size. Bits 16-31 of the data
register are not affected.
EXT.L
On
2
4(1/0)
X
X
0
0
[On < 16-31 > I - [On < 15> I
Extend sign bit of data word to long data word size.
ca
iii'
;
1
CD
ca
iii'
i
[An<0-15>1 - [An<0-15>1 - [rs<0-15>1 (sign extended)
Subtract source register word from address register. The sign of the
source word is extended to a full 32 bits for the operation.
0
'C
CD
iil
;
-..J
m
U1
n
0
a
X
X
X
X
X
[An <0-31 >1 - [An <0-31 >1 - [rs<0-31 >1
Subtract source register long word from address register.
:i'
c
CD
S:
:a
CD
ca
iii'
i
0
'C
CD
iil
;
[On <0-31> I - [On <0-31> 1 - [rs<0-31 > 1
Subtract source register long word from data register.
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
I
0)
0)
Bytes.
Status
Clock
Cycles
Operation Performed
T
S
X
N
Z
V
C
NBCD
On
2
6(1/0)
X
U
X
U
X
[On<0-7>1 - [On<0-7>1 - X
Negate decimal register byte. Bits 8-31 of the data register are not
affected.
NEG.B
On
2
4(1/0)
X
X
X
X
X
[On1 - 0 - [On<0-7>1
Negate register byte. Bits 8-31 of the data register are not affected.
NEG,W
On
2
4(1/0)
X
X
X
X
X
[On<0-15>1 - 0 - [On<0-15>1
Negate register word. Bits 16-31 of the data register are not affected.
NEG.l
On
2
6(1/0)
X
X
X
X
X
[On <0-31 >1- 0 - [On <0-31 >1
Negate register long word.
NEG.B
On
2
4(1/0)
X
X
X
X
X
[On<0-7>1 - 0 - [On <0-71 - X
Negate register byte with Extend. Bits 8-31 of the data register are not
affected.
NEG,W
On
2
4(1/0)
X
X
X
X
X
[On<0-15>1 - 0 - [On<0-15>1 - X
Negate register word with Extend. Bits 16-31 of the data register are
not affected.
:D
NEG.L
On
2
6(1/0)
X
X
X
X
X
[On <0-31 >1 - 0 - [On <0-31 >1 - X
Negate register long word with Extend.
i
NOT.B
On
2
4(1/0)
X
X
0
0
[On<0-7>1 - [On<0-7»
Ones complement data register byte. Bits 8-31 of the data register are
not affected.
NOT.W
On
2
6(1/0)
X
X
0
0
•iil
[On<0-15>1 - [On<0-15»
Ones complement data register word. Bits 16-31 of the data register
are not affected.
NOT.l
On
2
6(1/0)
X
X
0
0
00
[On<0-3>1- [On <0-31 >1
Ones complement data register contents.
Sec
On
2
9(1/1)
SWAP
On
2
4(1/0)
X
X
0
0
[On<0-15>1 -~ [On< 16-31 >1
Exchange the two 16-bit halves of a data register.
TAS
On
2
4(1/0)
X
X
0
0
[On<7>1-1
Test status of data register byte and set bit 7 to 1.
TST.B
On
2
4(1/0)
X
X
0
0
[On<0-7>1 - 0
Test status of data register byte. The data register contents are not
changed.
TST,W
On
2
4(1/0)
X
X
0
0
[On<0-15>1 - 0
Test status of data register word. The data register contents are not
changed.
TST.l
On
2
4(1/0)
X
X
0
0
[On<0-31» - 0
Test status of data register long word. The data register contents are
not changed.
=•iii'
'-I
Operand(s)
~
=•iii'
i
0
'D
·..
:::I
I
[On<0-7>1 - all l's if cc = TRUE
[On <0-1 >1 - all O's if cc = FALSE
Set status in data register byte.
:t.
:::I
c
e:•
I
--- ---
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
ASL
Operand(s)
dadr
Bytes
2,4
or 6
Clock
Cycles
9(1/11+
Status
Operation Performed
T
S
X
N
Z
V
C
X
X
X
X
X
15
0
101
:r'
Arithmetic shift left one bit of memory word. A zero is shifted into bit
O. Bit 15 is shifted into both Carry and Extend bits. 2, 3
ASL.B
count,On
On,dOn
2
2
6 + 2N(1/0)
6 + 2N(1/0)
X
X
X
X
X
X
X
X
X
X
a-
31
8
I
7
0
HJ
Arithmetic shift left of data register byte. The number of shifts is
specified as a direct count 11-8) or in a data register (1-63). Zeroes are
shifted into bit 0. Bit 7 is shifted into both Carry and Extend bits.
ASL.W
count,On
On,dOn
2
2
6 + 2N(1/0)
6 + 2N(1/0)
X
X
X
X
X
X
X
X
X
X
...,j
en.....
31
16
15
0
a~
I
en
:r
f--.{2J
;:
As ASL.B except shifts are for one word.
ASLL
count,On
On,dOn
2
2
8 + 2N(1/0)
8 + 2N(1/0)
X
X
X
X
X
X
X
X
X
X
31
0
:r'
t---m
As ASL.B except shifts are for entire register.
ASR
dadr
2,4
or 6
9(1111+
X
X
X
X
X
c5
-1:
Arithemtic shift right one bit of memory word. Bit 15 is propagated to
bit 14. Bit 0 is shifted into both Carry and Extend bits.
ASR.B
count,On
On,dOn
2
2
6 + 2N(1/0)
6 + 2N(1/0)
X
X
X
X
X
X
X
X
X
X
31
I
8
7
°
cs----l!
Arithmetic shift right of data register byte. The number of shifts is
specified as a direct count 11-8) or in a data register (1-63). Bit 7 is propagated to the right. Bit 0 is shifted into both Carry and Extend bits.
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
ASRW
Operand(s)
count.Dn
Dn.dDn
Clock
Cycles
Bytes
2
2
6
6
+ 2N(1/0)
+ 2N(1I0)
Status
Operation Performed
T
S
X
N
Z
V
C
X
X
X
X
X
X
X
X
X
X
31
0
16 15
I
-1:
EJ
As ASRB except shifts are for one word.
ASRL
count.Dn
Dn.dDn
2
2
8
8
+ 2N(1/0)
+ 2N(1/0)
X
X
X
X
X
X
X
X
X
X
31
0
-1:
L5
As ASRB except shifts are for entire register.
LSL
-..J
m
CXl
dadr
2.4
or 6
9(1/1)+
X
X
X
0
:r"
15
X
0
10'
Logical shift left one bit of memory word. A zero is shifted into bit O.
Bit 15 is shifted into both Carry and Extend bits. (Note that LSL is identical to ASL except for the Overflow condition.)2. 3
en
::r
;:
n
0
LSL.B
count.Dn
Dn.dDn
2
2
6
6
+ 2N(1I0)
+ 2N(1/0)
X
X
X
X
X
X
0
0
X
X
,
a-HID
31
8
a
:i'
c
CD
S:
7
0
Logical shift left of data register byte. The number of shifts is specified
as a direct count (1-8) or in a data register (1-63). ZeroeS are shifted
into bit O. Bit 7 is shifted into both Carry and Extend bits.
LSL.W
count.Dn
Dn.dDn
2
2
6
6
+ 2N(1/0)
+ 2N(1/0)
X
X
X
X
X
X
0
0
X
X
31
16 15
0
£l
I
~
As LSL.B except shifts are for one word.
LSL.L
count.Dn
Dn.dDn
2
8
8
+ 2N(1/0)
+ 2N(1I0)
X
X
X
X
X
X
0
0
X
X
:r.
31
As lSL.B except shifts are for entire register.
0
t---[Q)
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
LSR
Operand(s)
dadr
Clock
Cycles
Bytes
2,4
9(1/1)+
Status
Operation Performed
T
S
X
N
Z
V
C
X
X
X
0
X
'1:
~
or 6
Logical shift right one bit of memory word. A zero is shifted into bit 15.
Bit 0 is shifted into both Carry and Extend bits.
LSR.B
count,On
On,dOn
2
2
6 + 2N(1/0)
6 + 2N(1/0)
X
X
X
X
X
X
0
0
X
X
7
0
r-t:
I
Logical shift right of data register byte. The number of shifts is
specified as a direct count 11-8) or in a data register (1-63). Zeroes are
shifted into bit 7. Bit 0 is shifted into both Carry and Extend bits.
LSR.W
'"co
m
countOn
On,dOn
2
2
6
6
+ 2N(1/0)
+ 2N(1/0)
X
X
X
X
X
X
0
0
X
X
=
0
'1:
-1:
~
As LSRB except shifts are for one word.
en
::T
n
0
15
I
LSRL
~
count,On
On,dOn
2
2
8 + 2N(1/0)
B + 2N(1/0)
X
X
X
X
X
X
0
0
31
X
X
:i"
0
I2l---f
c
CD
S.
As LSRB except shifts are for entire register.
ROL
dadr
2,4
9(111)+
X
X
0
X
15
0
1
~.
or 6
Rotate left one bit of memory word. Bit 15 is shifted into bit 0 and into
the Carry.
ROL.B
count,On
On,dOn
2
2
6 + 2N(1/0)
6 + 2N(1/0)
X
X
0
X
X
X
0
X
31
I
8
7
0
I-'l
EJ. t
Rotate left of data register byte. The number of shifts is specified as a
direct count (1-8) or in a data register (1-63). Bit 7 is shifted into bit 0
and into the Carry.
I
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
ROL.W
Operand(s)
count.Dn
On.dOn
Clock
Cycles
Bytes
2
2
6
6
+ 2N(1/0)
+ 2N(1/0)
Status
Operation Performed
T
S
X
N
Z
V
C
X
X
X
X
0
0
X
X
16 15
31
0
I.
I
1
m1- t
As ROL. B except shifts are for one word.
ROL.L
count.On
On.dOn
2
2
8 + 2N(1/0)
8 + 2N(1/0)
X
X
X
X
0
0
31
X
X
0
~.
~
As ROL.B except shifts are for entire register.
ROR
oadr
2.4
9(111)+
X
X
0
r
0
15
X
or 6
-~
Rotate right one bit of memory word. Bit 0 is shifted into bit 15 and
into the Carry.
i"
ROR.B
"
count.On
On.dOn
2
2
6
6
+
+
2N(1/0)
2N(1/0)
X
X
X
X
0
0
X
o
8
31
X
7
0
I-~
{
I
Rotate right of data register byte. The number of shifts is specified as a
direct count (1-8) or in a data register (1-631. Bit 0 is shifted into bit 7
and into the Carry.
ROR.W
count.On
On.dOn
2
2
6
6
+
+
2N(1/0)
2N(1/0)
X
X
X
X
0
0
X
X
31
16 15
I
0
-:rm
I
f
As ROR.B except shifts are for one word.
ROR.L
count. On
On.dOn
2
2
8 + 2N(1/0)
8 + 2N(1/0)
X
X
X
X
0
0
X
X
r
31
0
.~
As ROR.B except shifts are for entire register.
ROXL
dadr
2.4
or 6
9(111)+
X
X
0
X
r'
15
0
~ ~
Rotate left one bit of memory word and Extend one bit. Bit 15 is shifted
into both Extend and Carry bits. The Extend bit is shifted into bit O.
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
ROXL.B
Operand lsI
count,On
On,dOn
Clock
Cycles
Bytes
2
2
6
6
+
+
2NI1/01
2NI1/01
Status
Operation Performed
T
S
X
N
Z
V
C
X
X
X
X
0
0
X
X
31
0
7
8
I
I-~
I
~
Rotate left of data register byte with Extend. The number of shifts is
specified as a direct count 11-81 or in a data register 11-631. Bit 15 is
shifted into both Extend and Carry bits. The Extend bit is shifted into
bit O.
ROXL. W
count,On
On,dOn
2
2
6
6
+ 2NI1/01
+ 2NI1/01
X
X
X
X
0
0
X
X
31
16
0
15
I
I
~ )l
II
I
As ROXL.B except shifts are for one word.
ROXL.L
count,On
On,dOn
2
2
8
8
+ 2NI1/01
+ 2NI1/01
X
X
X
X
0
0
X
X
r
0
31
~ ~
4
As ROXL.B except shifts are for entire register.
""'"
.!.J
....
en
ROXR
dadr
::T
2,4
or 6
911/11+
X
X
0
15
=
n
0
~
S·
c
.,
r
X
0
~ ~
Rotate right one bit of memory word and Extend. Bit 0 is shifted into
both Extend and Carry bits. The Extend bit is shifted into bit 15.
CD
S:
ROXR.B
count,On
On,dOn
2
2
6
6
+ 2N(1/0)
+ 2N(1/0)
X
X
X
X
0
0
X
X
31
8
I
0
7
I-I
t
~ ~
Rotate right of data register byte with Extend. The number of shifts is
specified as a direct count 11-8) or in a data register (1-63). Bit 0 is
shifted into both Extend and Carry bits. The Extend bit is shifted into
bit 7.
ROXR.W
count.On
On,dOn
2
2
6 + 2N(1/0)
6 + 2N(1/0)
X
X
X
X
0
0
X
X
31
I
16
15
I
~
0
•I
As ROXR.B except shfits are for one word.
ROXRL
count,On
On,dOn
2
2
8 + 2N(1I0)
8 + 2N(1/01
X
X
X
0
X
X
0
X
r
31
~ ~
0
•• ~ ~
As ROXR.B except shifts are for entire register.
Table 7-6. MC68000 Instructim Set Summary (Continued)
Mnemonic
BTST
BTST
Operand(s)
bitl.On
On,dOn
bitb,dadr
On,dadr
~
BSET
3:
III
bitl,On
On,dOn
bitb,dadr
:::I
is'
c
!o·
:::I
On,dadr
BClR
bitl.On
On,dOn
bitb,dadr
-..J
.!.J
On,dadr
N
BCHG
bitl.On
On,dOn
bitb,dadr
On,dadr
Bytes
Clock
Cycles
Status
T
S
X
N
Operation Performed
Z
V
C
[l]
[On]
[l] - [dOn < [On] > ]
Test a bit of a data register and reflect status in lero bit. The bit to be
tested may be specified directly or in a data register (bit 0-31 in either
case).
4
2
10(2/0)
6(1/0)
X
4,6
or 8
2,4
or 6
8(2/0)+
X
[l] -
4(110)+
X
[l] -
4
2
4,6
or 8
2,4
or 6
12(2/0)
8(1/0)
13(2/11+
X
X
X
[ l ] - [On 1. [On] - 1
[l] - [dOn < [On] > 1. [dOn < [On] > - 1
[l] - [dadr < bitb > 1. [dadr < bitb >] - 1
9(1/11+
X
[l] -
4
2
4,6
or 8
2,4
or 6
14(2/0)
8(1/0)
13(2/1)+
X
X
X
[l] - [On1. [On] ~- 0
[l)·- [dOn<[On»l. [dOn<[On») - 0
(ll - [dadr1. (dadr ]-O
9(1/11+
X
[l] -
4
2
4,6
or 8
2,4
or 6
12(2/0]
8(1/0)
13(211)
X
X
X
[l] -
9(1/1)
X
[l] -
X
[dadr < bitb >]
[dadr<[On]>]
Test a bit of a memory byte and reflect status in Zero bit. The bit to be
tested may be specified directly or in a data register (bit 0-7 in either
case). 1
(dadr<[OnJ>1. [dadr<[On]>] ~- 1
Test a bit as (BTST) and then set the specified bit.
(dadr«On]>l. [dadr«Onl>] - 0
Test a bit (as BTST) and then clear the specified bit.
[l] [l] -
[On l. [On]- [On]
[dOn - [On] > 1. [dOn < [On] > 1. [dOn <[dOn] >l
[dadr < bitb > 1. [dadr < bitb >] - [dadr < bitb > ]
[dadr<[OnJ>l. [dadr<[On]>]- [dadr<[OnJ>]
Test a bit (as BTST) and then complement the specified bit.
MOVE
An,USP
2
4(1/0)
[USP]- [An]
Move contents of address register to User Stack Pointer. This is a privileged instruction.
MOVE
USP,An
2
4(1/0)
[An] - [USP]
Move contents of User Stack Pointer to address register. This is a privileged instruction.
LINK
An,d16
4
18(212)
[A7] - [A7] - 2
[[A711- [An]
[An] - [A7]
[A7] - [A7] + d16
Save the contents of the specified address register on the Stack, load
the current Stack Pointer to the specified address register, and set the
Stack Pointer to point beyond the temporary stack storage area.
VI
Dr
n
;I;
PEA
jadr
2,4
or 6
10(1/2)+
I
[A7] - [A7] - 2
[[A711- jadr
Compute long word address and push address onto the Stack. 3
I
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Ci
~.::I n~
;
Operand Is)
Bytes
Clock
Cycles
UWLK
An
2
12(3/0)
CHK
data16.0n
4
On.dOn
2
Status
Operation Performed
T
S
X
N
Z
V
C
[A7] - [An]
[An] - [[A7]]
[A7] - [A7] + 2
Store the contents of the specified address register to the Stack
Pointer (A 7) and load the specified address register from the stack.
~
!:
~
!!
2
~
X
U
U
U
X
U
U
U
X
U
U
U
sadr.On
2.4
or 6
vector
2
37(4/3)
[A7] - [A7] - 2
[[A711- [PC]
[A7] - [A7] - 2
[[A711- [SR]
[PCl - vector
Initiate exception processing through specified vector .
TRAPV
2
37(5/3).
4(110)
If Overflow
1 then TRAP
Initiate exception processing through Overflow vector if the Overflow
bit is on.
RTE
2
20(5/0)
X
X
X
X
X
[SR] - [[A7]1. [A7] - [A7] + 2
[PC] - [[A7]l. [A7] - [A7] + 2
Return from exception.
2
12(210)
X
X
X
X
X
[SR<0-4>] - [On <0-4]
Move status data from data register to condition codes.
12(210)+
X
X
X
X
X
[SR<0-4>] [sadr<0-4>]
Move status data from memory location to condition codes. The
source address is a word address. 2• 3
X
X
X
X
X
[SR] - data8<0-4>
Move immediate status data to condition codes.
TRAP
III
:::I
CI.
~
iil
';oJ
'C
.....
Co)
f
If [On <0-15>] < 0 or [On <0-15>] > data16
then [PC] - CHK interrupt vector
If [dOn <0-15>] < 0 or [dOn <0-15>] > [On <0-15>]
then [PC] - CHK interrupt vector
If [On<0-15>] <0 or [On<0-15>] > [sadr]
then [PC] - CHK interrupt vector
Check register against bounds and initiate Check interrupt processing
if register word is out of bounds. The upper bound is a twos complement integer specified as immediate data. in a data register. or in a
memory word. 2. 3
49(6/3).
12(210)
45(5/3),
8(1/0)
45(5/3).
8(1/0)
=
MOVE
On.CCr
MOVE
sadr.CCR
MOVE
data8.CCR
4
16(3/0)
MOVE
On.SR
2
12(2/0)
X
X
X
X
X
X
X
[SR]- [On<0-15>]
Moves status word from data register to Status register. This is a privileged instruction.
MOVE
sadr.SR
12(2/0)+
X
X
X
X
X
X
X
[SR] - [sadr]
Move status word from memory location to Status register. This is a
privileged instruction. The source address is a word address. 2• 3
MOVE
data16.SR
4
16(3/0)
X
X
X
X
X
X
X
[SR] - data 16
Move immediate status word to Status register. This is a privileged
instruction.
MOVE
SR.On
2
6(1/0)
2.4
or 6
2.4
or 6
[On<0-15>] - [SR]
Move contents of Status register to data register. Bits 16-31 of the
data register are not affected.
Table 7-6. MC68000 Instruction Set Summary (Continued)
Mnemonic
Operand(s)
Bytes
Clock
Cycles
MOVE
SR.dadr
2.4
or 6
9(1/1)+
AND.B
dataS.SR
4
20(3/0)
AND.w
data 16.SR
4
20(3/0)
0
EORB
dataS.SR
4
20(3/0)
.
EORW
data 16.SR
4
20(3/0)
ORB
dataS.SR
4
20(3/0)
ORW
data 16.SR
4
20(3/0)
C/I
g
Status
Operation Performed
T
S
X
N
Z
V
C
[dadrl - [SRI
Move contents of Status register to memory location. The destination
address is a word address. 2. 3
X
X
X
X
X
X
X
[SR<0-7>1- [SR<0-7>1 /\ dataS
AND immediate data byte to low-order Status register byte.
X
X
X
X
X
[SRI - [SRI ./\ data16
AND immediate data with Status register. This is a privileged
instruction.
X
X
X
X
X
[SR<0-7>1 - [SR<0-7>1 ¥ dataS
Exclusive-OR immediate data byte to low-order Status register byte.
X
X
X
X
X
[SRI - [SRI ¥ data 16
Exclusive-OR immediate data with Status register. This is a privileged
instruction.
X
X
X
X
X
!SR<0-7>1- !SR<0-7>1 < dataS
OR immediate data byte to low-order Status register byte.
X
X
X
X
X
[SRI - [SRI V data 16
OR immediate data with Status register. This is a privileged instruction.
Ii
Ci
aS·
c
e:
~
NOP
2
4(110)
3:
RESET
2
132(1/0)
()
STOP
4
S(2/0)
-..J
~
iii'
!!.
iii
..
::I
0
Ii
n
~
2.
data16
X
X
X
X
No operation.
Reset. This is a privileged instruction.
X
X
X
X
X
[SRI - data16
Stop processor. This is a privileged instruction.
Table 7-7. MC68000 Instruction Object Codes
ABCO
AOO.B
AOO.L
AOO.W
....,
.!.J
(J1
AOOO.B
AOOQ.L
AOOO.w
AOOX.B
AOOX.L
AOOX.w
ANO.B
AND.L
AND.w
-(SAn),-(dAn)
sOn,dOn
dataS,dadr
dataS,On
On,dadr
sadr,On
sOn,dOn
data32,An
data32,dadr
data32,On
Dn,dadr
rS,An
rS,On
sadr,An
sadr,On
data16,An
data 16,dadr
data16,On
On,dadr
rS,An
rS,On
sadr,An
sadr,On
data3,dadr
data3,On
data3,An
data3,dadr
data3,On
data3,An
data3,dadr
data3,On
-(sAn),-(dAn)
sOn,dOn
-(sAn),-(dAn)
SOn,dOn
-(sAn),-(dAn)
sDn,dDn
dataS,dadr
dataS,On
dataB,SR
Dn,dadr
sadr,On
sOn,dOn
data32,dadr
data32,Dn
Dn,dadr
sadr,Dn
sOn,dOn
data 16,dadr
data16,Dn
data16,SR
On,dadr
sadr,Dn
C
C
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
5
5
5
5
5
5
5
0
0
0
0
0
D
0
0
0
C
C
C
0
0
C
C
C
0
0
0
C
C
2
2
4,6,orS
4
2,4,or6
2,4,or6
2
6
6, S,or 10
6
2,4,or6
2
2
2,4,or6
2,4,or6
4
4,6,orS
4
2,4,or6
2
2
2,4,or6
2,4,or6
2,4,or6
2
2
2,4,or6
2
2
2,4,or6
2
2
2
2
2
2
2
4,6,orB
4
4
2,4,or6
2,4,or6
2
6,B,orl0
6
2,4,or6
2,4,or6
2
4,6,orS
4
4
2,4,or6
2,4,or6
19(3/1)
6(1/0)
13(2/1)+
S(2/0)
9(1/1)+
4(1/0)+
4(110)
16(3/0)
22(3/2)+
16(3/0)
14(1/2)+
S(1/0)
S(1/0)
6(1/0)+
6(1/0)+
S(2/0)
13(2/1)
8(2/0)
9(11)+
S(1/0)
4(110)
8(1/0)+
4(1/0)+
9(1/0)+
4(1/0)
8(1/0)
14(1/2)+
B(1/0)
4(1/0)
9(1/1)+
4(1/0)
19(3/11
4(1/0)
32(5/2)
B(1/0)
19(3/1)
4(1/0)
13(2/1)+
B(2/0)
20(3/0)
9(1/1)+
4(1/0)+
4(1/0)
22(3/2)+
16(3(0)
14(1/2)+
6(1/0)+
B(1/0)
13(2/1)+
B(2/0)
20(3/0)
9(1/1)
4(1/0)
Table 7-7. MC68000 Instruction Object Codes (Continued)
ASl
ASL.B
ASL.l
ASL.W
ASR
ASRB
ASRl
ASRW
BCC
BCHG
BClR
-.oJ
.:...
0)
BCS
BEQ
BGE
BGT
BHI
BlE
BlS
BlT
BMI
BNE
BPl
BRA
BSH
sOn,dOn
dadr
count,On
On,dOn
count,On
On,dOn
count,On
On,dOn
dadr
count,On
On,dOn
count,On
On,dOn
count,On
On,dOn
label
label
bitb,dadr
bitl,On
On,dadr
On,dOn
bitb,dadr
bitl,On
On,dadr
On,dOn
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
bitb,dadr
bitl,On
On,dadr
On,dOn
2
~~m6
2
2
2
2
2
2
~~m6
2
2
2
2
2
2
4
2
~~m8
4
~~m6
2
~~m8
4
~~m6
2
4
2
4
2
4
2
4
2
4
2
4
2
4
2
4
2
4
2
4
2
4
2
4
2
~~m8
4
~~m6
2
4(1/0)
9(1/1)+
6 + 2N(1/0)
6 + 2N(1/0)
8 + 2N(1/0)
8 + 2N(1/0)
6 + 2N(1/0)
6 + 2N(1IO)
9(1/1)+
6 + 2N(1/0)
6 + 2N(1/0)
8 + 2N(1/0)
8 + 2N(1/0)
6 + 2N(1/0)
6 + 2N(1/0)
10,12(2/0)
10,18(1/0)
13(2/11+
12(2/0)
9(111)+
8(1/0)
13(2/1)+
14(2/0)
9(111)+
8(1/0)
10,12(2/0)
10,18(1/0)
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0),
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0)
10,12(2/0)
10,8(1/0)
10(2/0)
10(110)
13(2/1)+
12(2/0)
9(1/1)+
8(1/0)
Table 7-7. MC68000 Instruction Object Codes (Continued)
Byte 4
Instruction
BSR
BTST
BVC
BVS
CHK
CLR.B
ClR.l
ClRW
CMP.B
".:..
"
CMP.l
CMP.W
CMPM. B
CMPM. l
CMPM.W
OBCC
OBCS
OBEQ
OBF
OBGE
OBGT
OBHI
OBlE
OBlS
OBlT
OBMI
OBNE
OBPl
label
label
bitb,dadr
bitl,On
On,dadr
On,dOn
label
label
label
data16,On
On,dOn
sadr,On
dadr
On
dadr
On
dadr
On
dataB,dadr
dataB,On
sadr,On
sOn,dOn
data32,An
data32,dadr
data32,On
rS,An
rS,On
sadr,An
sadr,On
data16,An
data16,dadr
data16,On
rS,An
rS,On
sadr,An
sadr,On
(sAn)+,(dAn)+
(sAn)+,(dAn)+
(sAn) +, (dAn) +
On,label
On,label
On,label
On,label
On,label
On,label
On,label
On,label
On,label
On,label
On,label
On,label
On.label
6
6
0
0
0
0
6
6
6
6
4
4
4
4
4
4
4
4
4
0
0
B
B
B
0
0
B
B
B
B
B
0
0
B
B
B
B
B
B
B
5
5
5
5
5
5
5
Byte 5
Byte 6
Byte 7
Byte 8
Byte 9 Byte 10
Bytes
Clock Cycles
4
2
4,6,orB
4
2,4,or6
2
4
2
4
2
4
2
2,4,or6
2,4,or6
2
2,4,or6
2
2,4,or6
2
4,6,orB
4
2,4,or6
2
6
6, B, or 10
6
2
2
2,4,or6
2,4,or6
4
4,6,orB
4
2
10,12(2/0)
10, B(1/0)
B(2/0)+
10(2/0)
4(1/0)
6(1/0)
10,12(2/0)
10, B(1/0)
10,12(2/0)
10,B(1/0)
49(6/3),12(2/0)
45(5/3),B(1/0)
45(5/3)+,Bl/0)
9(1/1)+
4(1/0)
14(1/2)+
6(110)
9(1/1)
4(1/0)
B(2/0)
B(2/0)
4(1/0)+
4(1/0)
14(3/0)
12(3/0)+
14(3/0)
6(1/0)
6(1/0)
6(1/0)+
6(1/0)+
B(2/0)
B(2/0)+
B(2/0)
6(1/0)
4(1/0)
6(1/0)+
4(1/0)+
12(3/0)
20(5/0)
12(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12(2/0),10(2/0),14(3/0)
12 (2/0), 10(2/0), 14(3/0)
2
2,4, or6
2,4,or6
2
2
2
4
4
4
4
4
4
4
5
5
4
4
4
5
4
5
5
4
4
5
Table 7-7. MC68000 Instruction Object Codes (Continued)
Instruction
OBRA
OBT
OVC
OVS
DIVS
OIVU
EOR.B
EOR. L
EOR. W
...,J
.!.J
EXG
CO
EXT.L
EXTW
JMP
JSR
LEA
LINK
LSL
LSL.8
LSL.L
LSL.W
LSR
LSR.B
LSR.L
LSRW
MOVE
On.label
On.label
On.label
On.label
data16,Dn
sadr,On
sOn,dOn
data16.0n
sadr.On
sOn,dOn
data8.dadr
data8,On
data8,SR
On.dadr
sOn,dOn
data32,dadr
data32,On
Dn,dadr
sOn,dOn
data 16,dadr
data16,Dn
data16,Sr
On.dadr
sOn,dDn
An,An
An.On
On, An
On,On
On
On
jadr
jadr
jadr.An
An.d16
dadr
count. On
On.dOn
count. On
On.dOn
count.On
Dn.dOn
dadr
count. On
On.dOn
count. On
On.dOn
count. On
On.dOn
An.USP
data8.CCR
data16.SR
Dn.CCR
On.SR
5
5
5
8
8
8
8
8
8
0
0
0
4
4
4
4
2,4,or6
2
4
2,4.or6
2
4.6.or8
4
4
2A.or6
2
6,8,or10
6
2.4.or6
2
4,6,or8
4
4
2,4,or6
2
2
2
E
E
E
E
E
E
E
E
E
E
E
E
A
4
6
6
F
F
C
C
2
2
2
2,4.or6
2.4.or6
2.4.or6
4
2.4.or6
2
2
2
2
2
2
2.4.or6
2
2
2
2
2
2
2
4
4
2
2
12(2/0).10(2/0).14(3/0)
12(2/0).10(2/0).14(3/0)
12 (2/0).10(2/0),14(3/0)
<162(2/0)
<15811/0)+
<158(1/0)
<148(2/0)
<140(1/0)+
<140(1/0)
13(2/1)+
8(2/0)
20(3/0)
9(1/1)+
4(1/0)
22(3/2)+
16(3/0)
14(1/2)+
8(110)
13(2/1)+
8(2/0)
20(3/0)
9(1/1)+
4(1/0)
6(1/0)
6(110)
6(1/0)
4(1/0)
4(1/0)
4(1/0)+
14(1/2)+
2(0/0)+
18(2/2)
9(1/1)+
6 + 2N(1/0)
6+
8+
8+
6+
6+
2N(1I0)
2N(1I0)
2N(1/0)
2N(1/0)
2N(1/0)
9(1/1)+
6 + 2N(1/0)
6 + 2N(1I0)
8 + 2N(1/0)
8 + 2N(1/0)
6 + 2N(1/0)
6 + 2N(1/0)
4(1/0)
16(3/0)
16(3/0)
12(2/0)
12(2/0)
Table 7-7. MC68000 Instruction Object Codes (Continued)
MOVE.S
MOVE. L
MOVE.W
~
....,
<0
MOVEM. L
MOVEM. W
MOVEP. L
MOVEP. W
MOVEQ
MULS
MULU
NBCO
NEG.B
sadr,CCR
sadr,SR
SR,dadr
SR,On
USP,An
dataS,On
dataS,dadr
On,dadr
sOn,dOn
sadr,dadr
sadr,On
An,dadr
data32,An
data32,dadr
data32,Dn
On,dadr
rS,An
rS,On
sadr,An
sadr,dadr
sadr,On
An,dadr
data16,An
data 16,dadr
data16,On
On,dadr
rS,An
rs,On
sadr,An
sadr,dadr
sadr,On
(An)+,reg-list
jadr,reg-list
reg-list, - (An)
reg-list,madr
(An) + ,reg-list
jadr,reg-list
reg-list, - (An)
reg-list,madr
d16(An),On
On,d16(An)
d16(An),On
On,d16(An)
dataS,On
data16,On
sadr,On
sOn,dOn
data16,On
sadr,On
sOn,dOn
dadr
On
dadr
4 il~~llee_
: Ill.; ~ ~ ~II
4;;~C"
l
4 it"JJ 6 . '
1.3_
1.11./¥,
1 ''>'''::001111
'1 ee_
~I~ti
0111
1"'"
2 .'
2
2 !
. 3
G;,~'<
W~,
00181
. ' :gg4 . ; ;
~ lo~_.
~>,y*
2
2 ,'"
II
ggee.
OOee'" +
ooWij
,gg7'''1441
'" • K~
gg P<0':;;;
3
3
. 311 fii~
.·C¢!+~
3 lliP;; gg Od'liil
4
3
1
.011
33'~Olee;m
3 .gg
33 •
III
III
III:;
ee_
i
11~1
II
liiil~:.
4
!C:!tf~
10' .
:
.1
O~! 4 ' i I I
o ;1\1;
C
°8°ill
o •. ...~~
1.:,0.0.:
~
S
11:eel
C'IIQ'C.!
g~+~
:4~00ff1
lif) O~ff;" '~jl
Bytes
Clock
2,4,or6
2,4,or6
2,4,or6
2
2
4
4,6,orS
2,4,or6
2
2, 4, 6, S or 10
2,4,or6
2,4,or6
6
6,S,or 10
6
2,4,or6
2
2
2,4,or6
2,4,6, S, or, 10
2,4,or6
2,4,or6
4
4,6,orS
4
2,4,or6
2
2
2,4,or6
2,4, 6, S, or 10
2,4,or6
4
4,6,orS
4
4,6,orS
4
4,6,orS
4
4,6.orS
4
4
4
4
2
4
2.4,or6
2
4
2,4,or6
2
2,4,orS
2
12(2/0)+
12(2/0)+
9(111)+
6(1/0)
2.4.or6
4(110)
S(2/0)
9(1/1)+
5(0/1)+
4(1/0)
5(111)+
4(1/0)+
10(0/2)+
12(3/0)
lS(2/2)+
12(3/0)
10(0/2)+
4(110)
4(110)
S (2/0) +
14(112)+
4(1/0)+
5(0/1)+
S(2/0)
9(111)+
S(2/0)
5(0/1)+
4(1/0)
4(1/0)
4(1/0)+
5(0/1)+
4(1/0)+
S + Sn(2 + 2n/0)
S + Sn(2 + 2n/0)+
4 + 10n(1/n)
4 + 10n(1/n)+
S + 4n(2 + n/O)
S + 4n(2 + n/O)+
4 + 5n(1/n)
4 + 5n{1/n)+
24(6/0)
2S(2/4)
16(4/0)
lS(2/2)
4(110)
<74(2/0)
<70(1/0)+
<70(1/0)
<74(2/0)
<70(1/0)+
<70(1/0)
9(1/1)+
6(1/0)
9(1/1) ...
Table 7-7. MC6SOOO Instruction Object Codes (Continued)
Bytes
Clock Cycles
4
2
4
4
~~m6
4
~~m6
4
4
4
~~m6
4(1/0)
14(1/2)+
6(1/0)
9(111)
4(1/0)
9(111)+
4(110)
14(112)+
6(1/0)
9(1/1)+
4(1/0)
4(1/0)
9(1/1)+
4(1/0)
14(1/2)+
6(110)
9(1/1)+
6(1/0)
13(2/1)+
8(2/0)
20(3/0)
9(111)+
4(1/0)+
4(1/0)
22(3/2)+
16(3/0)
14(1/2)+
6(1/0)+
8(1/0)
13(2/1)+
8(2/0)
20(3/0)
9(1/1)+
4(110)+
4(110)
10(1/2)+
132(1/0)
9(1/1)+
6 + 2N(1/0)
6 + 2N(1/0)
8 + 2N(1I0)
8 + 2N(1/0)
6 + 2N(1/0)
6 + 2N(1/0)
9(1/1)+
6 + 2N(1/0)
6 + 2N(1/0)
8 + 2N(1/0)
8 + 2N(1/0)
6 + 2N(1/0)
6 + 2N(1/0)
9(1111+
6 + 2N(1/0)
Instruction
NEG.L
NEG.w
NEGX.B
NEGX.L
NEGX.w
NOP
NOT.B
NOT.L
NOT.W
OR. B
....,
00
o
OR. L
OR. W
PEA
RESET
ROL
ROL.B
ROL.L
ROL.W
ROR
ROR.B
ROR.L
RORW
ROXL
ROXL.B
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
data8.dadr
data8.Dn
data8.SR
Dn..dadr
sadr.Dn
sDn.dOn
data32.dadr
data32.Dn
Dn.dadr
sadr.Dn
sOn.dOn
data 16.dadr
data16.0n
data16.SR
Dn.dadr
sadr.Dn
sDn.dDn
jadr
dadr
count. On
On.dOn
count. On
Dn.dOn
count. On
Dn.dOn
dadr
count. On
Dn.dDn
count.Dn
Dn.dDn
count.On
On.dOn
dadr
count. On
4
4
4
4
4
4
4
4
4
4
4
2
2
2
~~m6
2
~~m6
2
2
2.4.or6
2
2.4.or6
2
2.4.or6
2
o
o
o
4.6.or8
4
8
8
8
2.4.or6
2.4.or6
4
2
o
o
6.8, or 10
8
8
8
2.4.or6
2.4.or6
2
4.6.or8
o
o
o
8.
8
8
4
4
6
4
4
2.4.or6
2.4.or6
2
2.4.or6
2
2.4.or6
2
2
2
2
2
2
2.4or6
2
E
E
E
E
E
E
2
2
2
2
2
2.4.or6
2
Table 7-7. MC68000 Instruction Object Codes (Continued)
Instruction
ROXl.L
ROXl.W
ROXR
ROXR.B
ROXR.L
ROXR.W
RTE
RTR
RTS
SBeo
see
ses
SEQ
-...I
~
SF
SGE
SGT
SHI
SLE
SLS
SLT
SMI
SNE
SPL
ST
STOP
SUB. B
SUB. L
On,dOn
count.On
On,dOn
count, On
On,dOn
dadr
count.On
On,dOn
count.On
On,dOn
count, On
On,dOn
-(sAn),-(dAn)
sOn,dOn
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
dadr
On
data16
dataB,dadr
dataB,On
On,dadr
sadr,On
sOn,dOn
data32,An
data32,dadr
Byte 1
E
E
E
E
E
E
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
Byte 2
Byte 3
Byte 4
Byte 5
Byte 6
Byte 7
Byte 8
Byte 9 Byte 10
Bytes
Clock Cycles
2
2
2
2
2
2,4,or6
2
2
2
2
2
2
2
2
2
2
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2,4,or6
2
2, 4,or 6
2
2,4,or6
2
4
4,6,orB
4
2,4,or6
2,4,or6
6 + 2N(1/0)
B + 2N(1/0)
B + 2N(1/0)
6 + 2N(1/0)
6 + 2N(1/0)
9(1/1)+
6 + 2N(1/0)
6 + 2N(1/0)
B + 2N(1/0)
B + 2N(1/0)
6 + 2N(1/0}
6 + 2N(1/0)
20(5/0)
20(5/0)
16(4/0)
2
6
6,B,or 10
19(3/1)
6(1/0)
9(1/1)+
6,4(110)
9(1/1)+
6,4(1/0)
9(1/1)+
6,4, (1/0)
9(111)+
6,4(1/0)
9(111)+
6,4(1/0)
9(111)
6,4(1/0)
9(111)+
6,4,(110)
9(111)+
6,4(1/0)
9(1/1)+
6,4(1/0)
9(111)+
6,4(110)
9(1/1)+
6,4(110)
9(1/1)+
6,4(110)
9(111)+
6,4(1/0)
9(1/1)+
6,4(110)
B(2/0)
13(2/1)+
B(2/0)
9(1/1)+
4(1/0)+
4(1/0)
16(3/0)
22(3/2)+
Table 7-7. MC68000 Instruction Object Codes (Continued)
SUBW
SUBQ.B
SUBQ.L
SUBQ.w
SUBX.B
"c»
SUBX.L
N
SUBXW
svc
SVS
SWAP
TAS
TRAP
TRAPV
TST.B
TST.L
TSTW
UNLK
data32,On
On,dadr
rS,An
rS,On
sadr,An
sadr,On
data1S,An
data 1S,dadr
data16,On
On,dadr
rS,An
rS,On
sadr,An
sadr,On
data3,dadr
data3,On
data3,An
data3,dadr
data3,On
data3,An
data3,dadr
data3,On
-(sAn), -(dAn)
sOn, dOn
-(sAn),-(dAn)
sOn,dOn
- (sAn), - (dAn)
sOn,dOn
dadr
On
dadr
On
On
dadr
On
vector
dadr
On
dadr
On
dadr
On
An
2,4,orS
2
2
2,4,orS
2,4,or6
4
4,6,or8
4
2,4,or6
2
2
2,4,orS
2,4,or6
2,4,or6
2
2
2,4,or6
2
2
2,4,or6
2
2
2
2
2
2
2
2,4,or6
2
2,4,orS
2
2
2,4,or6
2
2
2
2,4,or6
2
2,4,or6
2
2,4,orS
2
2
16(3/0)
14(112)+
8(1/0)
8(1/0)
S(1/0)+
6(110)+
8(2/0)
13(2/1)+
8(210)
9(1/1)+
8(1/0)
4(1/0)
8(1/0)+
4(1/0)+
9(111)+
4(1/0)
8(1/0)
14(1/2)+
8(1/0)
4(1/0)
9(111)+
4(110)
19(3/1)
4(110)
32(5/2)
8(1/0)
19(3/1)
4(110)
9(111)+
6,4(1/0)
9(111)+
S,4(1/0)
4(110)
11(111)+
4(1/0)
36(4/3)
37(5/3),4(1/0)
4(1/0)+
4(110)
4(1/0)+
4(1/0)
4(1/0)+
4(1/0)
12(3/0)
Table 7-8. MC68000 Object Codes in Numerical Order
ORB
ORB
ORB
OR.w
OR.w
OR.w
ORl
ORl
BTST
MOVEP.w
BTST
BCHG
MOVEP.l
BCHG
BClR
MOVEP.w
BClR
BSET
MOVEP.l
BSET
ANo.B
ANo.B
ANo.B
ANo.w
ANo.w
ANo.w
ANo.l
ANo.l
SUB.B
SUB.B
SUB.w
SUB.w
SUB.l
SUB.l
ADo.B
ADo.B
ADo.w
Aoo.w
Aoo.l
Aoo.l
BTST
BTST
BCHG
BCHG
BClR
BClR
BSH
BSH
EOR.B
EOR.B
EORB
EOR.w
EOR.w
EOR.w
EORl
EORl
CMP.B
CMP.B
CMP.w
CMP.w
CMP.l
CMP.l
MOVE.B
MOVE.B
MOVE.B
MOVE.B
MOVE.B
MOVE.B
MOVE.l
MOVE.l
MOVE.l
MOVE.l
MOVE.l
7-83
data8,Dn
data8,dadr
data8,SR
data16,on
data 1 6,dadr
data16,SR
data32,on
data32,dadr
on,don
d16{Anl.on
on,dadr
on,don
d16{An),on
on,dadr
on,don
on,d16(An)
on,dadr
on,don
on,d16{An)
on,dadr
data8,on
data8,dadr
data8,SR
data16,on
data 16,dadr
data16,SR
data32,on
data32,dadr
data8,on
data8,dadr
data16,on
data 16,dadr
data32,on
data32,dadr
dataS,on
dataS,dadr
data16,on
data 16,dadr
data32,on
data32,dadr
bitl,on
bitb,dadr
bitl.on
bitb,dadr
bitl.on
bitb,dadr
bitl.on
bitb,dadr
dataS,on
dataS,dadr
dataS,SR
data16,on
data 16,dadr
data16,SR
data32,on
data32,dadr
dataS,on
dataS,dadr
data16,on
data 16,dadr
data32,on
data32,dadr
son,don
sadr,on
dataS,on
on,dadr
sadr,dadr
dataS,dadr
rS,on
sadr,Dn
data32,Dn
rS,An
sadr,An
Table 7-8. MC68000 Object Codes in Numerical Order (Continued)
Instruction
MOVE.L
MOVE.L
MOVE.L
MOVE.L
MOVE.w
MOVE.w
MOVE.w
MOVE.W
MOVE.w
MOVE.w
MOVE.w
MOVE.w
MOVE.w
NEGX.B
NEGX.B
NEGX.w
NEGX.w
NEGX.L
NEGX.L
MOVE
MOVE
CHK
CHK
CHK
LEA
CLR.B
CLR.B
CLR.w
CLR.w
CLR.L
CLR.L
NEG.B
NEG.B
NEG.w
NEG.W
NEG.L
NEG.L
MOVE
MOVE
MOVE
NOT.B
NOT.B
NOT.w
NOT.w
NOT.L
NOT.L
MOVE
MOVE
MOVE
NBCO
NBCO
SWAP
PEA
EXT.w
MOVEM.w
MOVEM.W
EXT.L
MOVEM.L
MOVE. L
TST.B
TST.B
TST.w
TST.w
TST.L
TST.L
TAS
TAS
MOVEM.W
MOVEM.w
MOVEM. L
MOVEM. L
MOVEM. L
TRAP
2
2
2
2
3
3
3
3
4
3~01ee
3
.
7
3." " ,',' , gOO t$$S:'
3~mggee
3'
gll
a
4
40';00ft
40::,; .," 4
40'
Olff
40'~ B
40':" <.;,1 Oft
4C;h,fl
C
40,:; {llff
4
B
4,
10ee
4'
B
4
l1jj
4~,>~.
0
42'"'/
4
4~ ,':; ,00ff
Q~
f!:ft::
~q
42": ',Olff
4!i:;'ir B
4~<::,:
10ft
44',i/
0
44~:·' :'" OOff
4~'::i 4
44 ,>;:, Olff
44·, B
44"
10ft
44:.
440
c
;':,:llee
44··;'
46
46
F
..
00ij.d:4
OOft
f,{ff.
46:,4~Q
46, ' ; 01 ft
46.
B
.
4ij, ~'10ff
"it>
46,,', C
4~> 'Jlee~~
4$.
'F
48
, 0
4tf ,
OOft
4fit, , ' 4
48 ,,'. 01jj
48:' . B
48
48,"
4$'
.
ftf!·:;
Om:
.
.10hhhlitlh
A
C
48····. :'llh~
48': ,': E ',",.
4A
o~.
4A
OOft ffff' .
41\
4 Om
4A
Olff ffff
41\
8
4A . 1Off ftff
4A,
C
4A
l1ff ffff·
4C
'10jj
4C
A 9$5$
4C
E Q'~s
4C,
l1jj
4C
E ~~.~
4E.
4 Ww
, (EXT)
Orrr
Orrr
W.:
{E~T)
f~T)
ijjj:.,
7-84
data32,An
rS,dadr
sadr,dadr
data32,dadr
rS,On
sadr,On
data16,On
rS,An
sadr,An
data16,An
rS,dadr
sadr,dadr
data16,dadr
On
dadr
On
dadr
On
dadr
SR,On
SR,dadr
On,dOn
sadr,On
data16,On
jadr,An
On
dadr
On
dadr
On
dadr
On
. dadr
On
dadr
On
dadr
On,CCR
sadr,CCR
dataB,CCR
On
dadr
On
dadr
On
dadr
On,SR
sadr,SR
data16,SR
On
dadr
On
jadr
On
reg-list.madr
reg-list,-(An)
On
reg -list,madr
reg-list.-(An)
On
dadr
On
dadr
On
dadr
On
dadr
jadr,reg-list
(An) +,reg-list
(An)+.reg-list
jadr,reg-list
(An)+,reg-list
vector
Table 7-8. MC68000 Object Codes in Numerical Order (Continued)
LINK
UNlK
MOVE
MOVE
RESET
Nap
STOP
RTE
RTS
TRAPV
RTR
JSR
JMP
ADDQ.B
ADDQ.B
ADDQ.w
ADDQ.w
ADDQ.w
ADDQ.l
ADDQ.l
ADDQ.l
ST
DBT
ST
SUBQ.B
SUBQ.B
SUBQ.w
SUBQ.w
SUBQ.w
SUBQ.l
SUBQ.l
SUBQ.l
SF
DBF
SF
SHI
DBHI
SHI
SlS
DBlS
SlS
see
DBee
see
ses
DBes
ses
SNE
DBNE
SNE
SEQ
DBEQ
SEQ
sve
Dve
sve
SVS
DVS
SVS
SPl
DBPl
SPl
SMI
DBMI
SMI
SGE
DBGE
SGE
SlT
DBlT
SlT
SGT
DBGT
7-85
An.d16
An
An.USP
USP.An
data16
jadr
jadr
data3.Dn
data3.dadr
data3.Dn
data3.An
data3.dadr
data3.Dn
data3.An
data3.dadr
On
Dn.label
dadr
data3.Dn
data3.dadr
data3.Dn
data3.An
data3.dadr
data3.Dn
data3.An
data3.dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
dadr
On
Dn.label
Table 7-8. MC68000 Object Codes in Numerical Order (Continued)
Instruction
SGT
SlE
OBlE
SlE
BRA
BRA
BSR
BSR
BHI
BHI
BlS
BlS
Bee
Bee
Bes
Bes
BNE
BNE
BEQ
BEQ
Bve
Bve
BVS
BVS
BPl
BPl
BMI
BMI
BGE
BGE
BlT
BlT
BGT
BGT
BlE
BlE
MOVEQ
OR. B
OR.B
OR. W
OR.W
OR.l
OR.l
OIVU
OIVU
OIVU
SBe
SBeo
OR.B
OR.W
OR.l
OIVS
OIVS
OIVS
SUB.B
SUB.B
SUBW
SUBW
7-86
dadr
On
On.label
dadr
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
label
dataS.On
sOn.dOn
sadr.On
sOn.dOn
sadr.On
sOn.dOn
sadr.On
sOn.dOn
sadr.On
data 16.0n
sOn.dOn
-(sAn).-(dAn)
On.dadr
On.dadr
On.dadr
sOn.dOn
sadr.On
data16.0n
sOn.dOn
sadr.On
rs.On
sadr.On
F
Id
01
:y
d
m
II
li
e
5
"
)1
iE
THE 2901, 2901 A, AND 2901 B MICROPRO'
The 2901, 2901A, and 29018 are identical except for execution speeds. The
than the 2901 ; the 2901 B is about 25% faster than the 2901 A. For details see the de
The 2901 provides a 4-bit slice through the arithmetic and logic unit of a Centl
the Central Processing Unit's registers may also be generated out of 2901 Ie
Figures 8-1 and 8-2 functionally illustrate 2901 logic.
Figure 8-1 is a variation of Figure 4-3 from Volume 1 : it illustrates 2901 logic
description given in Chapter 4 of Volume 1. Figure 8-2 is a more accurate repr
paths. Note that all logic and data paths in Figure 8-2 are four bits wide.
2901 logic consists of an arithmetic and logic unit. a local. two-part read/write mer
and logic unit performs addition. subtraction. and the standard Boolean operati
receives two inputs and generates one output. The local read/write memory storE
results. In addition to the local read/write memory there is a "Q register" which is
double-length operations. You may compare the 2901's 16 registers to a CPU VI
usually implement a CPU's accumulator in the Q register. and you mayor may not i
registers in local RAM (in general. however. you will implement these registers in
You will frequently see obvious parallels between 2901 logic and CPU logic. You I
understand 2901 logic; however. do not assume that these parallels translate intc
The many data paths within the 2901 have been selected to link the ALU. local read
output in a functionally efficient manner. Shift logic has been inserted at selected
combination of data paths with ALU and shift logic minimizes the number of steps n
cessing Unit functions.
The few 2901 enhancements over the prior 6701 were designed specifically to redl
implement typical CPU operations; and these few enhancements were sufficient t
2901 MICROPROCESSOR SLICE PINS AND SIGNALS
Pins and signal assignments for the 2901 are illustrated in Figure 8-3. We will
each of these signals superficially before examining device operations in detail.
We may divide 2901 signals into these three categories:
1)
2)
3)
Control inputs that are generated by a microinstruction
Control signals connecting 2901 slices
Data and status outputs
First consider microinstruction-generated inputs.
AO-A3 and 80-83 are two 4-bit addresses which select locations within the 29
10-18 is a 9-bit instruction code which determines data flow and arithmetic/logic.
9-bit control code can be divided into three 3-bit fields as follows:
8
7
654
3
o
2
I I I I I I I I I
'--BitNo.
~Instruction coe
'L4'-_ALU
l
ALU
ALU
8-2
sourceid
operation
destinatic
Chapter 8
2900 SERIES CHIP SLICE PRODUCTS
Chip slice products represent a radical departure from the single-chip Central Processing Units that we have described up to this point. Chip slice products are. in fact. the building blocks for many Central Processing Units: they
are also used to build intelligent controllers.
There are·a variety of chip slice-type products on the market today; however, the 2900 series products are the
clear leaders in terms of sales and customer acceptance. The 2900 series is an enhancement of the older 6700
series chip slice products, which are not described since they are now obsolete.
Chip slice products are described conceptually in Chapter 4 of Voume 1 (in fact, the "general case" product described in Volume 1, Chapter 4 is a thinly disguised variation of the 2901 microprocessor slice). Therefoer, the
discussion which follows assumes that you have a conceptual understanding of chip slice devices and
microprogramming. If you do nto have this background, see Chapter 4 of Volume 1 before reading any further.
In this chapter we will describe the following 2900 series parts:
~
•
•
•
•
The
The
The
The
The
2901. 2901 A. and 2901 B microprocessor slices
2902A Look-Ahead Carry Generator
2903 Enhanced Microprocessor Slice
2909A. 2910. and 2911A Microprogram Sequencers
2930 and 2932 Program Control Units
All 2900 series devices use bipolar LSI technology. 2900 series microinstruction execution times vary with manufacturer and device. Consult the data sheets at the end of this chapter for details.
The primary source for the 2900 series chip slice products is:
ADVANCED MICRO DEVICES
901 Thompson Place
Sunnyvale. California 94086
Secondary sources for the 2900 series include:
MOTOROLA SEMICONDUCTOR
Box 20912
Phoenix. Arizona 85036
RAYTHEON SEMICONDUCTOR
350 Ellis Street
Mountain View. Caiifornia 94042
NATIONAL SEMICONDUCTOR
2900 Semiconductor Drive
Santa Clara. California 95050
FAIRCHILD CAMERA & INSTRUMENT CORPORATION
464 Ellis Street
Mountain View. California 94042
8-1
THE 2901, 2901A, AND 29018 MICROPROCESSOR SLICE
The 2901, 2901A, and 29018 are identical except for execution speeds. The 2901A is approximately 30% faster
than the 2901; the 2901 B is about 25% faster than the 2901 A. For details see the data sheets at the end of this chapter.
The 2901 provides a 4-bit slice through the arithmetic and logic unit of a Central Processing Unit. Some or all of
the Central Processing Unit's registers may also be generated out of 2901 logic.
Figures 8-1 and 8-2 functionally illustrate 2901 logic.
Figure 8-1 is a variation of Figure 4-3 from Volume·1; it illustrates 2901 logic in terms of the general chip slice
description given in Chapter 4 of Volume 1. Figure 8-2 is a more accurate representation of 2901 logic and data
paths. Note that all logic and data paths in Figure 8-2 are four bits wide.
2901 logic consists of an arithmetic and logic unit a local. two-part read/write memory, and shift logic. The arithmetic
and logic unit performs addition, subtraction, and the standard Boolean operations. The arithmetic and logic unit
receives two inputs and generates one output. The local read/write memory stores data, which may be operands or
results. In addition to the local read/write memory there is a "0 register" which is used as a temporary register or for
double-length operations. You may compare the 2901's 16 registers to a CPU with 16 accumulators. You will not
usually implement a CPU's accumulator in the 0 register, and you mayor may not implement a CPU's general purpose
registers in local RAM (in general. however, you will implement these registers in .Iocal RAM).
You will frequently see obvious parallels between 2901 logic and CPU logic. You may use these parallels to help you
understand 2901 logic; however, do not assume that these parallels translate into CPU implementation.
The many data paths within the 2901 have been selected to link the ALU, local read/write memory, data input and data
output in a functionally efficient manner. Shift logic has been inserted at selected points along data paths so that the
combination of data paths with ALU and shift logic minimizes the number of steps needed to create typical Central ProceSSing Unit functions.
The few 2901 enhancements over the prior 6701 were designed specifically to reduce the number of steps required to
implement typical CPU operations; and these few enhancements were sufficient to render the 6701 obsolete.
2901 MICROPROCESSOR SLICE PINS AND SIGNALS
Pins and signal assignments for the 2901 are illustrated in Figure 8-3. We will summarize functions performed by
each of these signals superficially before examining device operations in detail.
We may divide 2901 signals into these three categories:
1)
2)
3)
Control inputs that are generated by a microinstruction
Control signals connecting 2901 slices
Data and status outputs
First consider microinstruction-generated inputs.
AO-A3 and 80-83 are two 4-bit addresses which select locations within the 2901 local 16 X 4 bit RAM.
10-18 is a 9-bit instruction code which determines data flow and arithmetic/logical operations within the 2901. This
9-bit control code can be divided into three 3-bit fields as follows:
8
7
654
3
o .--BitNo.
2
I I I I I I I I I
~t
_
L
-
f.-,nstruction code
'4"'--___
8-2
ALU source identifier
ALU operation identifier
ALU destination identifier
Data In
16 x 4 bits of
t--4
I
bits wide_!
I
I
I
I
4-bit wide shifter
4-bit wide
Complementer.
Addition and
Boolean logic
(ALU Block)
t wide shifter
A (Direct connection from A-A to Y-Y)
Figure 8-1. The 2901 Microprocessor Slice
Volume 2
Rev. A. Update 2
8-3
1-79
,....----,1>-........0
RAM3
CL~~D-_-D--I
D;,octtg~JklmRr1TfD~EJJ
Data
01
Inputs
DO
ControlD-------_t-....+__...J
LeHers separated by a btoken line relate
this figure 10 Figure 22-1
DE
Figure 8-2.
2901 4-Bit Slice Logic
8-4
A3
A2
A1
AO
16
18
17
RAM3
RAMO
(+5V)VCC
F
10
11
12
CP
Q3
BO
B1
B2
B3
..
.-..
...
-
--...-...
..
--.. --...
-- -..--- --...
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
2901
DIP
-- -..
-...-
..
...-
--..
P
OVR
C(N+4)
-- G
-..
-..
---
OE
Y3
Y2
Y1
YO
~
F3
GND
CN
14
15
13
DO
01
02
03
QO
18
17
RAM3
nc
RAMO
(+5V) VCC
F
10
11
12
CP
nc
Q3
BO
B1
B2
B3
QO
03
02
01
---
--
..-
--
,-
-- ...--..
-...
-- -...
--...
--..
-- --.....
..-
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
2901A
FLAT
PACK
Pin Name
Description
Type
DO - 03
YO - Y3
OE
AO - A3
BO - B3
10 -18
RAMO, RAM3
QO,Q3
CN
C(N+4)
Data input
Data output
Output enable
Local RAM A address
Local RAM B address
Instruction code
Local RAM shift logic I/O
Q register shift logic I/O
Carry in
Carry out
Carry generate/propagate
Zero status
Sign status
Overflow status
Clock
Power, ground
Input
Output. Tristate
Input
Input
Input
Input
Bidirectional
Bidirectional
Input
Output
Output
Output. Open collector
Output
Output
Input
G,P
F
F3
OVR
CP
VCC,GND
Figure 8-3.
2901 A Microprocessor Slice Pins and Signal Assignments
8-5
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
----
--..
--..
..-.
-------
-...
16
AO
A1
A2
A3
OE
Y3
Y2
Y1
YO
P
OVR
C(N+4)
G
F3
GND
CN
14
15
13
DO
00-03 is a data input port. All data entering a 2901 must be input via DO-D3. We include these four pins among the
microinstruction-generated group since a microinstruction could indeed generate immediate data (in macro assembly
language terms) to be input via DO-D3. A more common alternative might be to generate this data out of an external
buffer. using microinstruction bits to enable a single output as follows:
2 to 4
Decoder
' - - - -..... D3. D2. D1. DO I--.....~ DO - D3
Enable 1
Enable 2
Enable 3
Buffer 1
Buffer 2
Buffer 3
OE is an output enable control. When OE is low. the 2901 can output data via YO-Y3. When OE is high. YO-Y3 is
floated. A microinstruction must anticipate microcycles within which data output is to occur and must generate a low
OE control at this time. When no data output is anticipated YO-Y3 should be floated.
Let us now examine data and status output signals.
As indicated in the previous paragraph. a 2901 outputs the results of internal operations via the four pins YO-V3.
In addition, separate Overflow (OVR) and Zero (F) status indicators are output. These indicators are used to generate standard Overflow and Zero statuses - as we will describe later.
Carry Status logic and associated signals are not simple status outputs; they are more accurately represented as interslice connecting signals. CN is the carry in used by addition and.!ubtra£,tion. C(N+4) is the carry out generated
by addition and subtraction. Carry Look-Ahead logic uses the P and G signals, together with the 2902 Carry
Generator. in order to compute the carry for an arithmetic operation occurring in parallel at two or more 2901 slices.
This carry logic has been described in Volume 1. Chapter 4; it is summarized later in this chapter when we describe the
2902 Carry Look-Ahead Generator.
The 2901 has two sets of internal shift, logic. For multislice shifts. bits shifted -out from one slice must be shifted
into the adjacent slice. QO and Q3 are the shift pins used by one set of shift logic; RAMO and RAM3 are the shift
pins used by the other set of shift logic.
CP is the master clock signal used to control and synchronize event sequences within the 2901.
8-6
2901 LOGIC
We will now examine 2901 logic in detail.
The best place to start understanding 2901 logic is at the read/write memory (local RAM):
RAM3
RO/Li
-"-
~--\
RAMO
LO/RI
./
?~
~
"--
31N
MUX
""'--
'--
--
31N
MUX
I-
I-
31N
MUX
31N
MUX
I~
.A-
l
Z--r----r----r----r- Z
AWord
Add ress
roGA1
01
DO
D--
A2o--
C .... New>
-
03
16-Bit by 4-Bit 2-Port RAM
(Local RAM)
A3D--
CLOCK
CP
02
-~
~ ~
~ - - A .~
C .... New>
m m all:C
B- i - -
C .... New>
----B
B Latch
A Latch
E
H ...-E
C· .... New>
'
;;
:;:,
~};
'"
,"
w
'"
:::
r
, ii
A MUX
SMUX
t
t
'"
,
'"
"
";cc
"".:;'
ALU
':
'.'
>"''" , ,',,,,,,.,,""",',,','
",' . .,•••.J"...
Figure 8-8.
."
2901 Destination Code 4 Data Paths
8-20
,.w
"",
'"
'"
"'"
'"
,",
"+'"~'''''''%'''''c'''
",,,
"
_f
n
3%"'~:MUX
3-IN
MUX
§jffrft down
t
I
Q
Local RAM
t
t
A Latch
B Latch
,
Data
Register
r
t
II
R MUX
S MUX
t
t
r
ALU
~
~
Y
~
nu)(
Figure
:::'c,""""
M,":::"
,''''''
,",',<:'
8~9.
:"',','0«
..
"',,,.... ,",
i,'L'"
@<'i
2901 Destination Code 5 Data Paths
8-21
:,':
.',,:,"'"
":',.
...: :~'::.:""
Q
Local RAM
Register
A Latch
B Latch
R MUX
S MUX
Data--+---....
ALU
Figure 8-10.
2901 Destination Code 6 Data Paths
8-22
i'" '" "
,'""",',,
" "
"n'"", ,,,, , , ' , ' , , ' ,'"
',M'",'
,",' ,,',';, "';;, ,,:,;, ",,' x
,,",
,'h
,H"
" " " " " " " , " / ) : ' .":\
"",;,,;,:,;;;'1",:::;:;:;::,;,
;
~
3-INauX
ii"
: E'
,
,
3-IN
MUX
shifiiJl,
l
I
Q
Local RAM
Register
;
;
A Latch
B Latch
A MUX
S MUX
t
t
Data--+-----,
t
ALU
y ~,':J~:;,
"
1!'"
Figure 8-11.
',""',",'
',W ,'""
',w',,;, " " " ,
H' " " " " " , " ' "
,,' ,
",w"", ,," "',,'oW'",'"
,"'E""
"v';;;;',c; \"'",;,,'; ;W',""'W;i"
2901 Destination Code 7 Data Paths
8-23
'
'" ""'w"'>' ""
,
W ;""':"
Let us now examine status logic of the 2901. You can generate Zero, Sign, Overflow and
Carry statuses. The Zero, Sign and Overflow statuses are easy to understand, so we will
look at them first.
2901
STATUS
LOGIC
Every 2901 generates an Overflow status at the OVR pin. This status is the exclusive-OR of carries out of the penu Itimate bit and the high-order bit. This may be illustrated as follows:
2901
OVERFLOW
STATUS
3
C3
2
0 ",,--Bit No.
C2
C2 = Carry from bit 2 to bit 3
C3 = Carry out of bit 3
OVR = C2
e
C3
Every 2901 generates an Overflow status; however, in a multi-2901 configuration only the high-order (or most significant) 2901 Overflow status is usually used. Lower-order 2901 Overflow status outputs can be ignored. For an 8-bit configuration this may be illustrated as follows:
High-order
:
Low-order
2901
I
2901
I
7
6
5
4 I 3
2
0 ~Bit No.
00 $5
11
Ignore low-order
OVR
High-order OVR
becomes CPU status
The Sign status wh ich is output at pin F3 is the level of the high-order ALU output bit. like the
Overflow status, the Sign status is output by every 2901 in a multislice configuration; however,
only the high-order 2901 Sign status is significant. For an 8-bit configuration this may be illustrated as follows:
High-order
2901
SIGN
STATUS
Low-order
2901
2901
I
6
5
4
I3
2
I I
I I
F3 = level
of bit 7.
Use as CPU
sign status
F3 = level
of bit 3.
o
~BitNo.
The Zero status is the NOR of the four ALU output lines, FO, F1, F2, and F3. If all fou r of these
outputs are 0, then the Zero status output is 1. The Zero status is an open collector Signal;
therefore in multislice configurations Zero status outputs can be wire-ANDed. The AND of all Zero
status outputs in a multi-2901 configuration generates the Zero status for the CPU
(Zero = FO • F1 • F2 • etc.).
2901
ZERO
STATUS
2901 Carry status logic is not straightforward because in a multi-2901 configuration an arithmetic
2901
operation (such as addition) should occur in parallel at each slice; but the carry from a low-order
CARRY
slice will not be generated in time to be accounted for by a parallel operation occurring at a highSTATUS
er-order slice. This problem has been described in detail in Chapter 4 of Volume 1, therefore we
will not dwell on it at this time. For now it suffices to note that you can use the CN and-C(N+4) pins of a 2901 to
generate carry if you allow ample time between clock cycles for the carry to ripple up through the slices. But if you
8-24
want to perform the entire arithmetic operation optimally. you must use the propagate (Pi and generate (G) signals. in
addition to CN and C(N+4). These signals are processed by the 2902 Carry Look-Ahead device. which is described
later in this chapter.
Table 8-10. given in the 2902 Carry Look-Ahead device discussion. summarizes the exact logic used by the 2901 to
generate P. G. CN and C(N+4).
The 2901 can generate a Half-Carry status. The Half-Carry status is needed by microprocessors
that use binary arithmetic with decimal adjust to generate binary coded decimal logic. In an 8-bit
configuration the C(N+4) output from the low-order 2901 becomes the Half-Carry status.
2901
HALFCARRY
STATUS
SOME 2901 OPERATIONS
In order to illustrate 2901 logic inaction, we will now show how various operations can be
performed for a Central Processing Unit created using two 2901 slices. We will show the
microcode for each operation, based on the following 34-bit microinstruction code:
2901
SAMPLE
MICROCODE
l1XX10000000XOXXXX
X
X
X
X
X
X
X
XBBBBAAAA
21201F1E1D1C1B1A191B1716151413121110 OF OEOD OC OB OA 09 OB0706050403020100 ~Microinstructionbit
CYCXSI SOOEIB 17 16 15 1413121110 T1 TOM1 Me DH3DH2DHI DHODL3DL2DL1 DLOB3B2Bl BOA3A2Al AO
I
1---'
-~TTTTT
·Local RAM A address
Local RAM B address
Low-order 2901 immediate data in
High-order 2901 immediate data in
Source select
o 0 - Immediate data from miCrocode
o 1 - Data from buffer 1
1 0 - Data from buffer 2
1 1 - Data from buffer 3
Destination select (Four arbitrary destinatIons)
Instruction code
Output enable
Shift/rotate logic control
Carry in control
00 - 0 in
01 - 1 In
1 0 - source 1 in
1 1 - source 2 in
The fields of the illustrated microinstruction are all self-evident. and have been described in the preceding text. with
the exception of CY. CX. 51 and SO. These four bits are used by shift and rotate logic. which we are about to describe.
Note that all fields. with the exception of the immediate data fields. are shared by the two 2901 devices. This is
because 2901 devices operate in parallel and must perform identical operations at any instant The immediate 4-bit
data fields differ since an 8-bit data field is unlike 4-bit halves.
Consider shift and rotate logic (in macroprogram terms) for one or more 2901 slices. Figure
2901 SHIFT
AND ROTATE
22-12 shows one possibility using 25LS253 Dual 4 Input Multiplexers to select the correct conOPERATIONS
nections for RAMO. RAMN. 00. and ON. We refer to the high-order pins as "ON" and "RAMN"
since one or more slices may be present. For a single slice. RAMN and ON would become RAM3
and 03. respectively. For two 2901 slices. RAMN and ON would become RAM7 and 07. respectively
The key to Figure 8-16 lies in the 17 signals which are input to the 1G and 2G pins of the 25L5253 device. Recall that
17 is one of three control signals input to the 2901 destination control logic. 17 controls shift logic at the local RAM and
register 3-IN MUX inputs. 17 is always high when a downshift occurs. 17 is always low when an upshift occurs. Thus
in Figure 8-12. 17 conditions one 25LS253 device to output data. while disabling the other device.
o
8-25
IT sr1 rT 1
1 yy
T1 1 1
i7
lG
CD
~
CJ)
GND- lCO
+5V- lCl
lC2
GND- lC3
2CO
GND+5V-=- 2Cl
2C2
2C3
--.
A
B
lG
2G
lY
25LS253
2Y
00
03
--
-
04
0(N-4)
Loworder
2901
RAMO
--
-
-..
0(N-3)
ON
-
.-
RAM4
RAM
(N-4)
- -
lY
High-
order
2901
RAM3
A
RAM
(N-3)
25LS253
RAMN
FN
I
Figure 8-12. 2901 Shift and Rotate Logic
2Y
B
2G
lCO ~GND
lCl 1--+5V
lC2
lC3 ~
2CO ~GND
2Cl ~+5V
2C2
2C3
-
r4-
Shifts do not always occur at the 2901 local RAM or Q register inputs (see Figures 8-4 through 8-11). But that is not
a problem. If the low-order to high-order 25LS253 device is enabled by 17. but no shift is to occur. then the 2901 will ignore the active 25LS253 output.
When a shift is specified by 16. 17. and 18. then the SO and S1 inputs control the output of the selected 25LS253 device
- which determines the kind of shift or rotate that will occur.
In this discussion of shifts and rotates, the sense of a "Ieft" or "right" shift can cause confusion since all vendor
2900 literature illustrates bit positions from right to left:
Low-order
bit
High-order
bit
We have elected to make our illustrations compatible with vendor literature so that you will have less trouble connecting descriptions of the same parts. But in macro assembly language terms a left shift normally implies multiplication:
7
6
54
32
High-order
bit
0
Low-order
bit
while a right shift implies division:
765432
0
After right shift
High-order
bit
Low-order
bit
8-27
Given the bit numbering system used by 2900 vendor literature. the opposite shift logic would apply. That is to say. a
left shift would become a divide:
0234567
Low-order
bit
High-order
bit
while a right shift becomes a multiply:
o
234567
Before right shift
After right shift
Low-order
bit
High-order
bit
In order to avoid confusion, we shall refer to "upshifts" and "downshifts". An "upshift" causes multiplication.
while a "downshift" causes division. An upshift becomes a left shift in macrolanguage terms. but looks like a right shift
in 2900 vendor literature. and in the illustrations of this chapter. A downshift becomes a right shift in macrolanguage
terms. but looks like a left shift in 2900 vendor literature. and in the illustrations of this chapter. We have elected to live
with this confusion since it is smaller than the confusion which would arise if all our 2900 series part descriptions inverted bit numbers and data flows with respect to vendor literature.
Beginning with the simplest case. consider a simple downshift where zero is loaded into the high-order bit and the loworder bit is lost. In effect the number has been divided by two.
Figure 8-13 illustrates a downshift occurring in local RAM only. 17 is-low. therefore the right-most 25LS253 device
(as illustrated in Figure 8-13) is enabled. while the left-most 25LS253 device is disabled. SO and S 1 are both low.
therefore 1CO is output at 1Y and 2CO is output at 2Y. Thus 0 is loaded into RAMN - and it is assumed that the three
bits 16. 17. and 18 cause the downshift to occur at the local RAM 3-IN MUX logic.
Note that a 0 downshift will occur in Figure 8-13 at the same time as the local RAM downshift - if 16.17. and 18 codes
have enabled the Q register 3-IN MUX downshift logic. For clarity we have not shown both downshifts occurring.
Were 17 high. then in Figure 8-13 an upshift would occur with 0 loaded into RAMO. and thence DO.
When executing a down- or upshift, as illustrated in Figure 8-13. you could shift in 1. rather than O. by inputting SO
high and S1 low. This causes 1C1 to be output at 1Y and 2C1 to be output at 2Y.
Next consider a down rotate; this operation is illustrated in Figure 8-14.
The only difference between the down rotate illustrated in Figure 8-14 and the downshift illustrated in Figure 8-13 is
the source for the Y2 output. A high input at S 1 with a low input at SO causes 1C2 to be output at 1Y and 2C2 to be
output at 2Y. 1C2 and 2C2 receive their inputs from 00 and RAMO of the low-order 2901. respectively: hence a down
rotate is achieved.
8-28
17 50
51
±±b
m
lG A
Q314
...
--I Q(N-3~i9h_
Q4
Q(N-4)
14
QN
order
2901
I.
1Y
..
25L5253
~.gO.i. . ~_G_+~_~_.,
1C3 ~----.
GND
2C01
,. ;;:tC'1;
h-l1q~;~~~~~m~ili.~~:;::_~I;B~.~C::;~~:I'ii;:· :::;:_I'l)tj~f;0~~;~~n~.jj~: ::.:t: 2y ;'r;:;"!' "~g~ 14-
co
FN
t..J
1
co
~
o
:E
t;t~4
";
"(N~::~J
DA.M
" " .... ,'
'lJ~~!'!·"'''''ffl'uN
tN_31~i.1 'M'!
.. "
" .' • ,' "
l
"
~
o
~
M.q-
~-~
LJ I
ecce
c
~
M
~
~-~
! 1__
.q-
ic
2V
~
0
:El«-~
MN ....
iii~
ecce
Figure 8-14. A 2901 Down Rotate
11V
17
SO
o
o
2G
1CO
1C1
19~
2C1
' "" 20'2:,,,~~~
12V:f';;f;<;'lt~'
4
;il~.'
FN
;.1(,.",;
B
25L5253
2901
A.M.
co
W
o
A
U U W
0314
100
25L5253
2VI
51
!±b
!±b
1G
50
51
You can generate an up rotate by inputting 17 high -which disables the right-most 25LS253 (as illustrated in Figure
8-14) and enables the left-most 25LS253.
We need to stress again the fact that 17 has been chosen to enable the left-most 25LS253 when high. and the rightmost 25LS253 when low. because this conforms to the way in which shift logic within the 2901 is controlled.
Let us now examine arithmetic shifts. The difference between an arithmetic shift and a logical shift lies in the highorder bit of a binary number. which arithmetic shift logic treats as a sign bit the sign bit must be excluded from the
shift. For arithmetic shifts the logic illustrated in Figure 8-12 concatenates local RAM with the Q registers to generate
a double length number. For two 2901 slices this may be illustrated as follows:
High-order
2901
6
7
5
Low-order
2901
3
4
2
0 ....--Overall bit number
I
Local RAM
3
2
0
3
2
0 . . . . . - Local bit number
3
2
0
3
2
0 . - - Local bit number
Ia
II
15 14 13 12
11
10
9
register
8 '--"Overall bit number
+
Sign bit
When an arithmetic downshift occurs. the high-order ALU output bit - which is the signed bit - is recycled back to
RAMN. thus ensuring that it is preserved through the downshift. The remainder of the arithmetic number is shifted
down one bit position. with the low-order local RAM bit (output via RAMO) becoming the high-order Q register bit (via
ON). This may be illustrated as follows:
6
7
4
5
I I
f
a
1
I
2
3
RAMN
15
f-.
00 - - . Lost
Re~ister
ON 4
2t3~
0
1C3
1Y
Local RAM
~
I
14 13
12 I 11
10
9
t
RAMO
8
1
High-order
2901
I
I
Low-order
2901
As illustrated in Figure 8-12. an arithmetic downshift is generated by 17=0. SO=1 and S1 =1.
8-31
An arithmetic upshift causes 0 to be shifted into 00 while ON, the high-order a register bit. is shifted into RAMO. This
may be illustrated as follows:
I
I
6
7
I
~
I
4:3
5
2
0
~OO"'O
I
a register
I
I
~
2(:3
ON
I
local RAM
2Y
~
lost .... RAMN.-t
~RAMO
~1-5~1-4~1-3--1~2~1-1-1~1-0~9~'-'
8
I
High-order
2901
I
low-order
2901
Note that this arithmetic upshift does not preserve the high-order sign bit. Therefore the arithmetic upshift is really a
double length logical upshift.
You can easily generate double length down and up logical rotates by concatenating the a registers with the local
RAM. Connect the 1C3 input of the left side 25LS253 device to the RAMN output. Connect the 2C3 input of the left
side 25LS253 device to the ON output. Connect the 1C3 input of the right side 25LS253 device to the RAMO output.
Connect the 2C3 input of the right side 25LS253 device to the 00 output.
All of the shift and rotate logic functions we have just described, as well as the Status register and carry-in multiplexer,
are contained in the 2904 Status and Shift Control Unit. This device eliminates most of the MSI. such as the two
25LS253s around the 2901 s.
8-32
Let us now look at the simple problem of loading data into a local RAM location. If the data is
immediate - that is to say, if it is provided by the microinstruction itself - then the following
single microinstruction will load eight bits of data into the local RAM location addressed by the B
address:
2901
DATA INPUT
r - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - O u t p u t result to RAM
- - - - - - - - - - - - - - - - - - - - - - - - - - - - O R Data with 0
r - - - - - - - - - - - - - - - - - - - - - - - - - - I n p u t Data and 0 to ALU
Select immediate data
ffg ,. ____1. __-_,:LI
ill
~~,'='~
;.0 w",'" "" ""-
XXXXX011011111XXOO
D D
D
D D D D DBBBBXXXX
21201F1E1D1C1B1A191B1716151413121110 OF OE OD OC OB OA 09 080706050403020100~Microinstructionbit
CYCXS1 SOOE IB 17 16 15 1413 12 11 10 T1 TOM1 MO DH3DH2DH1 DHODL3DL2DL1 DLOB3B2B1 BOA3A2A1 AO
•
"--~.
TT-CT'------LocaIRAMAaddress
Local RAM B address
Low-order 2901 immediate data in
High-order 2901 immediate data in
L-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Source select
o 0 - Immediate data from microcode
o 1 - Data from buffer 1
1 0 - Data from buffer 2
1 1 - Data from buffer 3
L . . . - - - - - - - - - - - - - - - - - - - , - - - - - - D e s t i n a t i o n select (Four arbitrary destinations)
L - - - - - - - - - - - - - - - - - - - - - - - - - - - - I n s t r u c t i o n code
L . . . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - O u t p u t enable
L . . . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - S h i f t / r o t a t e logic control
' - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - C a r r y in control
00 - 0 in
01 - 1 in
1 0 - source 1 in
1 1 - source 2 in
If the data which is to be loaded into local RAM comes from an external buffer. and we arbitrarily assume that it comes
from external data buffer number 2. then the following single microinstruction will transfer the data from external
buffer 2 to the local RAM location selected by the B address:
r - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - O u t p u t result to RAM
- - - - - - - - - - - - - - - - - - - - - - - - - - - - O R Data with 0
r - - - - - - - - - - - - - - - - - - - - - - - - - I n p u t Data and 0 to ALU
Select buffer 2 as data source
,
±fE
n~
I
,=tr--I------~~~tl~:;;ion into which data is loaded
XXXXXO 11 0 1 1 1 1 1 XX 1 0 X X
X
X
X X X X BBBBXXXX
21201F 1E 1D 1C 1B 1A 19 1817 16 15 14 13 1211 10 OF OE OD OC OB OA 09 08 07060504030201 00 -4--Microinstruction bit
CYCXS1 SOOE 18 17 16 15 1413 12 11 10 T1 TOM1 MO DH3DH2DH1 DHODL3DL2DL1 DLOB3B2B1 BOA3A2A1 AO
~.--
TTTT'-----LocalRAMAaddress
Local RAM B address
Low-order 2901 immediate data in
High-order 2901 immediate data in
L . . . - - - - - - - - - - - - - - - - - - - - - S o u r c e select
o 0 - Immediate data from microcode
o 1 - Data from buffer 1
1 0 - Data from buffer 2
1 1 - Data from buffer 3
L - - - - - - - - - - - - - - - - - - - - - - - D e s t i n a t i o n select (Four arbitrary destinations)
' - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - I n s t r u c t i o n code
' - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - O u t p u t enable
L - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -_ _ _ Shift/rotete logic control
' - - - - - - - - - - - - - - - - - - - - - - - - - - - - -_ _ _ _ _ _ _ _ Carry in control
00 - 0 in
01 - 1 in
1 0 - source 1 in
1 1 - source 2 in
8-33
We described logic associated with microinstruction bits M1 and MO earlier.
An arithmetic or logic operation performed on two sources taken from local RAM, with the result being output via Y to
external destination number 1, requires the following single microinstruction:
OutputresultviaY
Select ALU operation
Select A and B latches as ALU operands
Y output destination select
Local RAM B address
Local RAM A address
,
.
.
---j
il,....--..-l....-..=t =L -,, "
±fffi
X X X X 0 0 0 1 1 1 1 0 0 1 0 1 XX
X
X
X
X
X
X X
XBBBBAAAA
2120 1F 1E 10 1C 1B 1A 191817161514131211 10 OF OE 00 OC OB OA 09 OB 070605040302 01 OO~Microinstruction bit
CYCX 51 SO OE 18 17 16 15 14 13 12 11 10 T1 TO M1 MO DH3DH2 DHt DHO DL3 Dl2 DL1 DLO B3 B2 B1 BO A3 A2 A1 AO
""'"
- - -'-1TTl~T
I
-Local RAM A address
-Local RAM B address
-Low-order 2901 immediate data in
High-order 2901 immediate data in
-Source select
o 0 - Immediate data from microcode
o 1 - Data from buffer 1
1 0 - Data from buffer 2
1 1 - Data from buffer 3
Destination select (Four arbitrary destinations)
Instruction code
Output enable
Shift/rotate logic control
Canry in control
00 - 0 in
01 - 1 in
1 0 - source 1 in
1 1 - source 2 in
Now consider the same operation performed on one operand taken from local RAM (as addressed by A), while the other
operand is provided by the microinstruction as immediate data: the result is returned to the local RAM location addressed by B. Here is the necessary microinstruction:
it
I
t
t
t
,
-
"
- - _1_
Disable Y output
Output ALU result to 8
Specify AL U operation
Select immediate data and A latch as ALU operands
Destination select
Immediate data. second operand
Result address in local RAM
Operand address in local RAM
Don't care
XX X X 1 0 1 1 I I 1 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 8 B B 8 A A A A
21201F1E1D1C181A191B1716151413121110 OF OE 00 OC 08 OA 09 080706050403020100...-Microinstructionbit
CYCX 51 SO OE 18 17 16 15 14 13 12 11 10 T1 TO M1 MO DH3DH2 DH1 DHO DL3 DL2 DL 1 OLO 83 B2 B1 80 A3 A2 A1 AO
'----~----TLL- tL-----
'----------------------
L..-._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
L..-._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
L..-._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
L..-._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Local RAM A address
Local RAM 8 addr~ss
Low-order 2901 immediate data in
High-order 2901 immediate data in
Source select
o 0 - Immediate data from microcode
o 1 - Data from buffer 1
1 0 - Data from buffer 2
1 1 - Data from buffer 3
Destination select (Four arbitrary destinations)
Insiruction code
Output enable
Shift/rotate logic control
~----------------------------------- Carry in control
00
01
10
1 1
8-34
-
0 in
1 in
source 1 in
source 2 in
Two microinstructions, with appropriate looping and select logic, is all you need in order to
multiply two 8-bit numbers and generate a 16-bit product. The algorithm needed for this
multiplication initially stores the multiplier in the low-order eight bits of the product space with
the multiplicand in a separate 8-bit storage location as follows:
7
0
15
8
•
Product (initial)
o
7
Q
B
~Product
(final)
I.--Product space
7
o
7
0
A
~Multiplier
(initial)
J...-MultiPlicand
We are going to downshift the contents of the 16-bit product space eight times. After eight shifts. the multiplier will
have been shifted out and lost. Therefore the high-order eight bits of the product space will initially be assigned to the
low-order eight bits of the product. as shown above.
In the typical "shift" and "add" multiplication routine (which we have described in Volume 1) the multiplicand is
upshifted one bit position at a time. and is added to the product whenever there is a 1 in the multiplier bit corresponding to the current upshift. Here is a simple illustration of two 4-bit numbers being multiplied to create an 8-bit product:
1010·0101
00110010
=
Step 1:
00000000
0101
101':e'
Step 2:
00000000
01010
00001010
1 0"110
Step 3:
00001010
010100
n'·10
Step 4:
00001010
0101000
00110010
01::01 0
The multiplicand initially corresponds to the low-order multiplier bit. The multiplicand is subsequently upshifted three
times. corresponding to the three higher-order multiplier bits. Following the first and third upshift. the multiplicand is
added to the product. since bits 1 and 3 of the multiplier are 1.
Now instead of upshifting the multiplicand. as illustrated above. we could downshift the product's space. This may be
illustrated as follows:
Step 1:
Step 2:
00000000
0101
101ib
00000000
0101
00001010
Step 3:
Step 4:
00001010
0101
o 0 0 0 10 10
0101
00110010
8-35
'1>0 1 0
This is the algorithm we are about to use. This algorithm allows the multiplier to be stored in half of the product space,
since this space is slowly shifted out.
Returning to our 8-bit X 8-bit multiplication, after the first shift the 16-bit product space will be shared by the low-order
nine bits of the product and the high-order seven bits of the multiplier:
765432
0765432
0
~sNft
C••,
8
765
432
0765432
Multiplier
, Product
and ultimately the sixteen bits of the product space will be occupied by the 16-bit product - after the entire multiplier
has been shifted out. Each time the contents of the product space are shifted down one bit position, the next low-order
bit of the multiplier will be shifted out and will appear at output pin RAMO. This output is tested. If it is 1, then the
multiplicand is added to the high-order eight bits of the product space {the 0 register) before the next shift occurs. The
carry from the addition must become the next high-order product bit prior to the next addition. Therefore the carry is
shifted into the high-order 0 register bit via 07.
In this fashion, the multiplicand is added to the product in each bit position that corresponds to a 1 in the multiplier.
The logic for this operation may be illustrated as follows:
Since we have discussed multiplication algorithms in some detail in Volume 1, we will not spend more time in this
chapter describing the theory of this multiplication algorithm. Rather. consider the following example:
2C' A4 =1C30
S~rt
0000000000101100
10100100
Step 1:
000000000010110'0-0000000000010110
Step 2:
0000000000010110---0000000000001011
Step 3:
0000000000001011-0000000000000101
10100100
1010010000000101
Step 4:
101001000000010';1:-0101001000000010
10100100
1111011000000010 C=O
Step 5:
1 1 1 1 0 1 1 0 0 0 0 0 0 0 1; 0 - 0 1 1 1 1 0 1 1 0 0 0 0 0 0 0 1
Step 6:
o1 11 10
1 1 0 0 0 0 0 0 01-0 0 1 1 1 1 0 1 1 0 0 0 0 0 0 0
10100100
1110000110000000
Step 7:
1 1 1 0 0 0 0 1 1 0 0 0 0 0 00:- 0 1 1 1 0 0 0 0 1 1 0 0 0 0 0 0
Step 8:
o1 1 10 0 0 0
End
0001110000110000
C=O
C=O
1 1 0 0 0 0 0 -0- 0 0 1 1 1 0 0 0 0 1 1 0 0 0 0 0
The algorithm above starts by downshifting 0 and B registers' contents as a single 16-bit entity. Carry, which must initially be 0, is Shifted into the high-order 0 register bit via 07, while the low-order bit of B appears at RAMO. If RAMO is
0, then must be added to O. If RAMO is 1, then the multiplier in the local RAM location with address A must be added
to O. A second microinstruction accomplishes this addition. If this addition generates a carry, then the carry bit must be
loaded into the next high-order product bit. By connecting the carry to 07 we make sure that any carry is loaded into
the next high-order product bit on the next downshift of 0 and B registers' contents. Necessary logic is illustrated in
Figure 22-15.
°
8-36
D
01-
CP- CK
7y
T
I I I
lG A
(X)
W
--.J
1s, r
S,l
GND- lCO
+5V- lCl
lC2
GND- lC3
GND- 2CO
+5V- 2Cl
2C2
2C3
B
T
C8
2G
lY
00
03
Loworder
2901
25LS253
..
2Y
RAMO
RAM3
-
...
04
07
I I I
lG A
lY
High-
--
25LS253
order
2901
--
RAM4
RAM7
2Y
CY· CX
B
2G
lCO
1Cl -=-+5V
lC2
lC3
2CO
2Cl -+5V
2C2
2C3
--
9
.
FN
I
--- D
O~
CP- CK
4-
~
Figure 8-15. 2901 Connections for Binary Multiplication
In Figure 8-15. CX and CY high is the characteristic combination identifying binary n:tultiplication. As compared to
Figure 8-12. we have modified the 1CO and 2CO inputs to the right-most 25LS253 device so that when CY and CX are
both high. a downshift loads the Carry out into the high-order bit of 07. while 00 is loaded into RAM7 - and thence
into the high-order bit of the local RAM location with address B. Thus a downshift treats the 0 and local RAM locations
as a single 16-bit register. which. following a downshift causes a prior carry to be input at Q7 while the low-order
multiplier bit is output at RAMO.
Before describing the logic surrounding 11 in Figure 8-15. let us look at the two microinstructions which must be executed sequentially within a loop in order to perform the required multiplication. First we execute a downshift
microinstruction. then we execute an add microinstruction. as follows:
-~-L-~I=-r=--===-1==--=. =-=t-=-_=+-_=-==-~?~~~
1Ir----+--1r---r-....--=--L==---
11001100000011XXXX x x X
x x x X XBBBBXXXX
21201F1E1D1C1B1A191B1716151413121110 OF OE 00 OC OB OA 09 080706050403020100~Microinstructionbit
CYCX 51 SO ~ 18 17 16 15 14 13 12 11 10 T1 TO Ml MO DH3DH2 DHI DHO Dl3 Dl2 Dl1 DlO B3 82 Bl BO A3 A2 Al AO
------1TT -t-T
. local RAM A address
local RAM B address
. low-order 2901 immediate data in
,High-order 2901 immediate data in
,Source select
o 0 - Immediate data from microcode
o 1 - Data from buffer 1
1 0 - Data from buffer 2
1 1 - Data from buffer 3
Destination select (Four arbitrary destinations)
Instruction code
Output enable
Shift/rotate logic control
Carry in control
00 - 0 in
01 - 1 in
1 0 - source 1 in
1 1 - source 2 in
Multiplication specified
Disable Y output
Store sum in Q
Add two operands
and Q or A and Q fonm AlU operands
Initial multiplier/final product low-order
eight bits address in local RAM
Local RAM multiplicand address
Don't care
o
1
• •
t1 - I
-
1
l1XX10000000XOXXXX X
X
X
X
X X X
X8888AAAA
2120 IF IE 10 lC 18 lA 191817 16 15 141312 11 10 OF OE 00 OC OB OA 09 08 07060504030201 00 4--Microinstruction bit
CYCX 51 SO
18 17 16 15 14 13 12 11 10 T1 TO Ml MO DH3DH2 DHI DHO Dl3 Dl2 Dl1 DlO 83 82 Bl 80 A3 A2 Al AO
,
or
----'-~TTTTT
8-38
·Local RAM A address
·Local RAM 8 address
,Low-order 2901 immediate data in
,High-order 2901 immediate data in
,Source select
o 0 - Immediate data from microcode
o 1 - Data from buffer 1
1 0 - Data from buffer 2
1 1 - Data from buffer 3
Destination select (Four arbitrary destinations)
Instruction code
Output enable
Shift/rotate logic control
Carry in control
00 - 0 in
01 - 1 in
1 0 - source 1 in
1 1 - source 2 in
The downshift microinstruction is self-evident. Note. howev~r. that RAMO must become the 11 instruction bit for the
addition so that 0 will be added to Q when RAMO outputs O. while the contents of the local RAM location addressed by
A are added to Q when RAMO outputs 1. But when CY and CX are not both high. binary multiplication is not in
progress: therefore 11 comes directly from the microinstruction. The three NAND gates shown in Figure 8-15 provide
the necessary logic.
The multiplication example we have just described is a useful illustration of 2901 logic. but the2903. which we describe next. performs binary multiplication and division automatically.
8-39
THE 2903 MICROPROCESSOR SLICE
The 2903 is a 4-bit microprocessor slice. The 2903 is conceptually similar to the 2901, which we have already
described. The 2903 has more versatile signals than the 2901, and more on-chip functions; however, the 2903
and the 2901 are driven by clocks with approximately equal frequency. But remember, the 2901A and 2901B
are faster than the 2901; therefore, they are faster than a 2903 - excluding special 2903 functions. The 2903A
is currently in development and, when available, will offer faster operation than the' 2903.
The 2903 is not a superset of the 2901. Microprograms written for the 2903 and the 2901 will be completely
different so will external logic supporting the two devices. Nor is the 2903 always the part of choice. as compared to a
2901. If your application uses a lot of complex arithmetic and logic operations. or if your application requires a large
number of local registers. then the 2903 is the part of choice. But if your application stresses execution speed. then the
2901 A or 2901 B may be a better choice.
2901 and 2903 ALU logic also differ sharply. The 2903 performs operations which encompass the simple 2901 ALU
functions; the 2903 also performs a separate set of more complex operations. Furthermore. 2903 ALU logic discriminates between a high-order slice. a low-order slice. and an intermediate slice; 2901 logic makes no such
high/intermediate/low-order distinctions. By discriminating between high-order. low-order. and intermediate slices.
the 2903 is able to perform operations oil non-symmetrical data. For example. a twos complement binary number is
non-symmetrical since the high-order bit is a sign bit subject to different interpretation from other bits. which are magnitude bits. Also. by discriminating between high-order. low-order. and intermediate slices. the 2903 makes double use
of many signals; signals perform secondary functions at slices where the primary function is meaningless. For example.
Carry. Generate. and Propagate signals share pins with Overflow and Sign status. since the Carry. Generate. and Propagate signals are meaningless at the most significant slice. while status signals are meaningful only at the most significant slice.
In the description of the 2903 which follows, we will compare and contrast the 2903 with the 2901. We will
refer to the 2901 description, together with Chapter 4 of Volume 1 for all conceptual information.
The 2903 is packaged as a 48-pin DIP. It uses bipolar LSI technology.
A 2903 FUNCTIONAL OVERVIEW AND COMPARATIVE ANALYSIS
Figures 8-16 and 8-17 functionally illustrate 2903 logic. Figure 8-16 is a variation of Figure 8-1. given earlier in
this chapter. and of Figure 4-3. from Volume 1; it illustrates the 2903 in terms of the general chip slice description
given in Chapter 4 of Volume 1. Figure 8-17 is a more accurate representation of 2903 logic and data paths.
SuperficiaUy the 2903 and the 2901 look very similar. Both have an arithmetic and logic unit which receives two inputs and generates a single output. Both have a 16 x 4-bit. two output-port RAM. additional local data storage in the 4bit Q register. and two sets of shift logic.
The 2903 16 x 4-bit local RAM. like the 2901. receives two 4-bit addresses - the A and B addresses. Data can be written into the 2903 local RAM location addressed by B. but only when the separate WE control input is low. The 2901
has no signal equivalent to WE. Data addressed by A and B is output to the 2903 A and B latches; but the 2903 B latch
has an output enable control. OE . which must be low for the B latch contents to be passed on. The 2901 has no signal
B
equivalent to OE .
B
Both 2901 and 2903 A and B latch outputs are transmitted to the Rand S ALU input multiplexers; but that is the only
similarity between the twosets of ALU input logic. The 2901 uses three instruction code bits to generate eight possible
combinations of Rand S inputs. The 2903 uses one instruction code bit. together with two new control signals to select
substantially different ALU operand combinations. The 2903 then makes up for the lack of operand input options with
additional ALU functions.
Both 2901 and 2903 ALU outputs may go to the Q register. the Y port. or the 16 X 4-bit local RAM.
Like the 2901. the 2903 Q register has shift logic at its input. The 2903 also has shift logic on the local RAM data path;
but 2903 shift logic precedes the Y outputs. and has a separate output enable control Signal OEy like the 2901.
Perhaps the most obvious difference between the 2901 and the 2903 lies in the data input and output ports.
The 2901 has a single data input port. DO-D3. and a single data output port. YO-Y3. The 2903 has the same data input
port. DAO-DA3. but the 2903 has two bidirectional data output ports. DBO-DB3 and YO-Y3.
8-40
4-bit
Shifter
Data in
or out
Figure 8-16. The 2903 Microprocessor Slice
8-41
A Word
Address
AO
A1
A2
{ A3
B1
BWord
BO}
16 x 4-Bit 2-Port RAM
~~
Address
WE
EP
00
4:.
N
~13
~t ~t -t -,
DAO
DAl
DA2
DA3
1m
fA ~ 113
OEy
of.r.
oh
11~""
oA---oA
I
R2
"Z
I
R1
"1
I
RO
"U
I
DBO
DB1
DB2
DB3
I so I
Sl
SU
S1
I
S2
5Z
Arithmetic and
Logic Unit
I
S3 ~Io
53
I
I I III
III
I I
L...-SIOO
o
......CP
Reaister
G/N
P/OVR
Z
C(N+4)
5103 4
I III
.11loloom mm
nLOI03
Figure 8-17. 2903 4-Bit Slice Logic
01°0
EA
DAO
DAl
DA2
DA3
12
13
14
eN
C(N+4)
P/OVR
GND
G/N
OEy
YO
Yl
Y2
Y3
SIOo
SI03
Z
DBO
DBl
2903
48
47
46
45
44
43
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
01°3
B3
B2
B1
BO
CP
10
11
WRITE/MSS
LSS
lEN
WE
VCC (+5V)
15
16
17
18
OEB
AO
Al
A2
A3
DB3
DB2
Description
Type
Data input
Data input/output
Local RAM A address
Local RAM B address
ALU R input select
Local RAM write enable
Data input/output
RAM B output/DBO - DB3 input enable
YO - Y3 output enable
RAM shifter controls
shifter controls
Carry logic input
Carry logic output
Carry look ahead generate/Negative status
Carry look ahead propagate/Overflow status
Zero status/control
Instruction code
Instruction enable
Least significant slice select
Most significant slice select/Write indicator
Power, Ground
Input
Input/output
Input
Input
Input
Input
Input/output
Input
Input
Bidirectional
Bidirectional, Tristate
Input
Output
Output
Output
Bidirectional, Open collector
Input
Input
Input
Bidirectional
Pin Name
DAO - DA3
DBO - DB3
AO - A3
BO - B3
EA
WE
YO - Y3
OEB
OEy
SIOO, SI03
0100, 0103
CN
C(N+4)
G/N
P/OVR
Z
10 - 18
lEN
LSS
WRITE/MSS
VCC,GND
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
a
Figure 8-18. 2903 Microprocessor Slice Pins and Signal Assignments
8-43
2903 MICROPROCESSOR SLICE PINS AND SIGNALS
Pins and signal assignments for the 2903 are illustrated in Figure 8-18. We will summarize functions performed by each of these signals superficially before examining device operations in detail.
2903 signals can be divided into these three categories:
1)
2)
3)
Data inputs and outputs
Instruction and control inputs that are generated by a microinstruction
Control and status signals connecting 2903 slices. and status signals generated by 2903 slices
First consider data inputs, outputs and associated address signals.
AO-A3 and BO-83 are two 4-bit addresses which select locations within the 2903 local 16 x 4-bit RAM. Data
may be written into the local RAM location addressed by B - but only while both WE and the clock signal, CP,
are input low.
While CP is high. the contents of the local RAM location addressed by AO-A3 are written into the A latch -which is
therefore changing continuously. When CP goes low. the A latch contents are stable. holding whatever data was read
from local RAM at the instant that CP made its high-to-Iow transition. The A latch contents are continuously output to
the ALU R input multiplexer.
The B latch output is enabled by the OEB control signal. When this signal is high. the B latch still receives data from the
local RAM location addressed by BO-B3. but the B latch output is floated.
If OE is low and the B output from local RA~s enabled. then DBO-DB3 becomes a 4-bit output. The B output apB
pears at DBO-OB3. as illustrated earlier. When OE is high and the B latch output is disabled. DBO-DB3 becomes a 4B
bit data input. Data input via DBO-DB3 can be selected as the ALU S operand.
DAO-DA3 always functions as a 4-bit data input.
The R input to the ALU may be the A latch output from local RAM, or the DAO-DA3 external
data input. If EA is high, then DAO-DA3 is selected. If EA is low then the local RAM A latch
output is selected.
2903 ALU
INPUT
OPTIONS
The low-order instruction code input (10) determines the ALU S input. If 10 is high. the Q
register output becomes the S input to the ALU. If 10 is low. the B output from the local RAM. or data input via DBO-OB3
becomes the ALU S input. These options are summarized in Table 8-5: logically they may be illustrated as follows:
A latch
outPut-----~
- - - - - - - - - - - - A L U R input
EA ____T
__
DAO-DA3
-
8 latch output--------__1
ALU S input
OE8--.....- - I
D80-D83-----------------I
IO---------------------~__1
Q Register output------------------~~
ALU input options are described in more detail later when we look at 2903 logic.
YO-Y3, which were data output pins of the 2901, are bidirectional 2903 pins (see Figure 8-17).
OEy is a control signal which enables or disables the~LU output. If OEy is low. then ALU output. after passing
through the ALU shifter. appears at the YO-Y3 pins. But if OEy is high. ALU output is disabled and YO-Y3 become input
pins. Data input at YO-Y3 can be written into the local RAM location addressed by BO-B3. provided WE and CP are low.
8-44
The 2903 has a 9-bit instruction code which is input via 10-18. The interpretation of this instruction code differs sharply
from the 2901. Without reference to the 2901. the 2903 instruction code interpretation may be illustrated as
follows:
....- - - - - - - - - - - - - - 1 6 special functions
See Table 22-8
~~
00000. Select special functions
,
.
TI_'__
18 171
6 \ 5 \ 4
13\
2 \
\ 0
~Instruction code with bit number
selectALU S input
See Table 22-5
Not 0 0 0 O. Select ALU simple functions
See Table 22-6
Specify ALU output destination
See Table 22-7
As illustrated above, the 2903 instruction code has two different interpretations.
We can compare 2901 and 2903 instruction codes. but to do so we must include the EA and OES control inputs as instruction code contributors. The two instruction codes may now be compared as follows:
2901:
2903:
18
17
16
15
14
13
12
11
10
l~~
18
17
16
15
14
13
12
11
EA
10
OES
Note that OES and EA do not usually come from the microcode.
2903 instruction code interpretations are quite complex and make use of additional control and status signals.
Therefore we will complete our summary of signals before examining instruction code interpretations in detail.
Let us now examine status and control signals of the 2903. We have already described WE. EA.
The 2903 has logic to discriminate between a most significant slice, a least significant slice,
or an intermediate slice.
OE S'
and
OEy .
2903 SLICE
SIGNIFICANCE
SELECT
When LSS is input low, a 2903 acts as a least significant slice. As a least significant slice,
the WRITE/MSS signal becomes a WRITE output. As such. WRITE/MSS is output low for every
microcycle during which data is written into local RAM. Frequently the WE inputs for all 2903 slices will be connected
to the WRITE output of the least significant 2903 slice. This may be illustrated as follows:
Most
Significant
Intermediate
2903
2903
2903
+5V
LSS
LSS
+5V
r----
W/MSS
WE
~~
2903
+5V
+5V
~
Least
Signficant
Intermediate
f-o
LSS
r--o
+5V
0.- W/MSS
00- W/MSS
~
WE
WE
~~
~
8-45
W/MSS
WE
LSS
~
j
-~
At intermediate and most significant slices, LSS is input high. Now the WRITE/MSS signal becomes an MSS input. A low MSS input selects the most significant slice, while a high MSS input selects an intermediate slice, as
illustrated above.
iEN is described in vendor literature as an "instruction enable"
input. This may lead you to believe that it enables
or disables the 10-18 instruction code inputs. but in fact the effect of iEN is more limited. When low. lEN allows data
to be written into the Q register: it also enables the WRITE output at the least significant slice. When lEN is high. data
cannot be written into the Q register. and the WRITE output at the least significant slice is constantly output high. If the
WE inputs for all slices are connected to the WRITE output of the least significant slice. then lEN high effectively disconnects the instruction code input. since it prevents data from being written into the Q register or local RAM: but it
does not prevent an instruction from being decoded and executed by the ALU. and it does allow data to be output via
the DB and/or Y pins.
2903 ALU logic has the stand~d Ca..':!y In (CN) and Carry Out (C(N+4)) signals. The 2903 also has
2903
the Carry Look-Ahead signals G and P. But if you look at the discussion of Carry Look-Ahead logic
STATUS
given in Chapter 4 of Volume 1 (and later in this chapter for the 2902). you will see that G and P
SIGNALS
outputs are not used at the most significant slice. Conversely. the Sign and Overflow status outputs are meaningful only at the most significant slice. Therefore 2903 pins share G with the Sign status (N) and P
with the Overflow status (OVR). These pins output Sign (N) and Overflow (OVR) statuses at the most significant
slice: they output Carry generate (3) and propagate (P) for intermediate and least significant slices.
The 2903 also has an open-collector Zero status output (Z). This Signal is output high when all ALU outputs are low.
The 2903 makes additional use of its shifter signals (SIOO, 8103, QIOO, Q103) and its status signals (CN.
C(N+4), N, OVR, and Z). These Signals are occasionally used in special ways by ALU operations that do not use the
signals for their primary purpose. For a summary see Table 8-8 and associated text.
SIOO and SI03 are ALU shifter connections. QIOO and Q103, likewise, are Q register shifter connections. These
Signals allow shifts to occur across multiple slices as described for the 2901. These Signals will always be connected as
follows:
Most
Significant
Slice
Intermediate Slices
QI03
0103
5103
Least
Significant
Slice
5100
0100
5103
+5V
CP is the master clock signal used to control and synchronize events within the 2903.
8-46
0100
0100
SIOO
5100
2903 LOGIC
We will now examine 2903 logic in detail.
The best place to start understanding 2903 logic is at the read/write memory lIocal RAM):
DO
A Word
Address
{~~---I"
A2~
A3
01
D2 03
BO
B1 } B Word
B2
~
Address
16 x 4-Bit 2-Port RAM
.
WE
A3 A2 A1
AO
BO
B1
B2 B3
B
B Latch
A Latch
OEB
~~-1-~-----DBO
~-+----. OB1
OAO-----+-+--+~
...
DA3i
OA 1 ----40-+--0
OA2----+...
.......- - - O B 2
....----OB3
The 2903 local RAM consists of sixteen 4-bit locations. You will use pins AO-A3 to identify the location from which
data will be output at the A latch. You use pins 80-83 to identify the 4-bit location from which data may be output to
the 8 latch or into which data may be written via YO-Y3.
Data may be written into the local RAM location addressed by B may be illustrated as follows:
but only while WE and CP are input low. This
CP
Y
BO - B3
Local RAM
Data Stable
Data Stable
Data Changing
RAM Location X
RAM Location Y
X and Yare any two hexadecimal memory addresses
, High WE inhibits write
As illustrated above. the contents of the local RAM location addressed by 8 are changing while WE and CP are both
low. When CP goes high. contents of the addressed RAM location are stable. holding whatever data was input when
CP made its low-to-high transition. If WE is high. local RAM is not accessed and its contents remain stable.
Data is output from the local RAM locations addressed by A and 8 when CP is high. The contents of the local RAM location addressed by A are output to the A latch. The contents of the local RAM location addressed by B are output to the
8 latch. These outputs occur when CP is high; therefore the A and 8 latch contents are continuously changing while
CP is high. but they are stable while CP is low. holding whatever data was input when CP made its high-to-Iow transition.
8-47
The A latch contents are output continuously. We can therefore illustrate A latch output timing as follows:
CP
AD - A3
I
X
\
1
\
1
\
P
X
Q
X
R
Read P
A Latch
Read Q
~
Local RAM
C
E:
Q Stable
~
RAM Location P
C
r
[
Read R
E:
R Stable
~
RAM Location Q
C
E
RAM Location R
P.O. and R are any three hexadecimal addresses. S represents stable data. and C represents changing data in the
selected RAM location.
In the illustration above. the RAM location addressed by A is shown as stable while CP is high and changing while CP is
low. The stable data is output to the A latch while CP is high. The A latch contents subsequently become stable while
CP is low - at which time local RAM contents are changing until RAM access time has elapsed. Thus race conditions
are avoided.
The A latch outputs are continuously enabled.
B latch timing is a little more complex than A latch timing because the B latch has its own output enable control
signal OE ' When OE B is high. the B latch output is floated. But when OE is low. the B latch outputs are enabled. B
B
S
latch timing may be illustrated as follows:
CP
80 - B3
B Latch
I
\___1
I
\
X~_ _ _ _ _L_ _ _ _ _~I _ _ _ _ _ _ _M _ _ _ _ _ _~I
Read L
L Stable
M Stable
\_--'
_______
X
N ______
N Stable
Read N
DEB
B Latch output - - - - - - - - - - - - - - - - - - . . . . . , . , .
DBO - DB3
Input
Input
Output
[M) means "contents of RAM location M".
The VO- V3 input to local RAM may come from the ALU output. or from the VO-V3 pins.
Unlike the 2901. there is no shifter at the local RAM input: rather. the shifter has been moved to
the ALU output. and the shifter output is itself enabled or disabled by the OEy control input. If
OEy isJQw. then the shifter output is enabled: it appears as output at YO-Y3 and at RAM DO-D3.
But if OEy is high. YO-Y3 become input pins providing local RAM with its data input.
The 2903 local RAM. like the 2901. generates a 4-bit slice through selected programmable registers of a
Central Processing Unit. But it is much easier to extend 2903 local RAM using external memory. This is because
the 2903 has one data input port and two bidirectional data ports situated between local RAM and the ALU. The 29705
is used as an expansion RAM.
8-48
CPU register implementation and ALU operand inputs are logically dependent on each other. since the primary function of CPU registers is to store ALU source or destination data. We will therefore explore the ALU operand options
available using a 2903. and see what impact these options have on register implementation .
Turning to the 2903 Arithmetic and Logic Unit. these three aspects of ALU logic are important:
....------....
2903 ARITHMETIC
AND LOGIC UNIT
1)
The operands which are input to the ALU
2)
3)
The ALU operation which is to be performed
The destination for the ALU output. (The destination specification includes any shift operations.)
Instruction code bit 10. together with EA and OE ' controls the data input to the 2903 ALU; instruction code bits 11
S
through 14 specify simple ALU functions. while 15 through 18 specify the destination and shift operation for simple functions. Instruction code bits 15 through 18 may also specify special 2903 functions.
Table 8-5 shows the ALU operand source options that can be specified using 10. EA. and
OE B , Let us now explore these options in detail,
Table 8-5. 2903 ALU Rand S Operand Selections
Control Signal
R Operand
EA
10
OEa
0
0
0
1
1
1
0
0
1
0
0
1
0
1
0/1
0
1
0/1
A latch output
A latch output
A latch output
DAO - DA3 input
DAO - DA3 input
DAO - DA3 input
8-49
S Operand
B latch output
DBO - DB3 input
Q register output
B latch output
DBO - DB3 input
Q register output
2903 ALU
OPERAND
OPTIONS
Beginning with the logically simplest case. we will assume that EA is low. so the A latch output becomes the ALU R input. Any of the ALU S input options could also accompany EA high. in which case DAO-DA3 becomes the ALU R input.
Consider 00-03 providing the ALU S input. while DBO-DB3 is idle:
DO - D3
Local RAM
B Latch
A Latch
1 4 t - - - - - - - OEB = 1
_ - - - - - -... DBO - DB3
DAO - DA3 _ _ _ __
~~~QO-Q3
.....~-IO=l
EA=O---t~
ALU
Data paths illustrated above would probably be used by a complex ALU operation involving one source operand. This
source operand comes from local RAM via the A latch. while the complex ALU operation acts on temporary data held in
the 0 register.
8-50
Now consider the same data paths illustrated above. but with OE input low. so that B latch data is output via OBOB
DB3:
DO - 03
L
Local RAM
I
I
A Latch
B Latch
ril : _
~9
,,-
DAO - DA3 0
EA=O
-
D. "
~
RM
OEB = 1
....
..
f
~
SM
DBO - DB3
00 -03
-..
10
=
1
J,
"4 "
ALU
Data being output via DBO-DB3 will probably come from a CPU register implemented in local RAM. DBO-DB3 could be
connected to external read/write memory within which additional CPU registers are implemented. The direct data path
from local RAM to DBO-OB3 can be used effectively to implement any register-to-register operation within a CPU. If. for
example. an Accumulator or other primary register is implemented in local RAM while secondary registers are held in
external RAM. then the data path illustrated above lends itself readily to register-register data transfers. which may. or
may not. occur in parallel with any other CPU operation.
8-51
Now consider the data paths we just illustrated. but with EA high. so that the ALU R operand comes from the external
data inputs DAO-DA3. This may be illustrated as follows:
DO - D3
Local RAM
A Latch
B Latch
.·1 ......-----_
_-------.I
DAO - DA3 "'"-_______.....
RMUX
S MUX
OEB = 0
DBO - DB3
QO - Q3
............... 10 = 1
ALU
Data input via DAO-DA3 may be immediate data coming from a microinstruction. or non-immediate data taken from
any other source. DAO-DA3 may also receive its input from an external RAM within which additional CPU registers are
held.
8-52
But. moving away from complex operations that may require the ALU S operand to come from the Q register. let us examine some more complex data paths used by simple CPU operations. In the simplest case. the two ALU operands will
come from local RAM. This may be illustrated as follows:
DO - 03
Local RAM
1'/ .•4 - - - - - - - OEB = 0
Ik'}l~--------J DBO - DB3
_---..1 00 - 03
DAO - DA3 c......_ _ _....
EA= 0
·~:,MUX
---I~
....~-IO=O
"<>;:
AlU
Data paths illustrated above show two ALU operands being taken from local RAM locations -
8-53
probably CPU registers.
We can take the S ALU input from DBO-DB3 by inputting OE B high. thus enabling the data from the B latch. This may
be illustrated as follows:
DO - D3
Local RAM
B Latch
A Latch
....~------OEB = 1
l'Iaiiil!!BSB DBO - DB3
_-----J 00 - 03
DAO - DA3 '--_ _ _'"
"'1---10= 0
EA= 0
ALU
8-54
Data entering at DBO-DB3 could be immediate data coming from a microinstruction. or data from an external RAM
location being used to implement additional CPU registers. By inputting EA high. we can take. both the Rand S ALU inputs from external RAM:
DO - 03
Local RAM
A Latch
B Latch
....f--------
DEB = 1
DBa - DB3
DAO - DA3 .......____.......,
EA
=1
,_---..1
RMUX
$MUX
00 - 03
""'-10=0
AlU
In the illustration above you see one of the more significant 2903 advantages. as compared to the 2901. The 2901
allows a single operand to be taken from external RAM. and that reduces the effectiveness of external RAM as a means
of implementing the two-port CPU registers in a 2901 configuration. It limits you to CPU architectures that include a
group of secondary registers. only one of which can provide an ALU operand during the execution of any instruction.
But the 2903. by allowing external data inputs to the Rand S ALU operands. allows you to implement CPU registers in
internal local RAM. or in external RAM like the 29705. without compromising register logic associated with either implementation.
8-55
The 2903 has local RAM addressing. The 2901 allows you to specify just two local RAM addresses within a single microcycle. The A and B addresses identify the two ALU operands while
the B address also identifies the destination address for the ALU product. Thus the ALU operand
specified by the B address must be overwritten if the ALU product is to be returned to local RAM.
But the 2903 allows either two or three local RAM addl'ilsses to be specified within a single
microcycle; you have the option of creating one. or two B addresses within a single microcycle. If
you create one B address, timing may be illustrated as follows:
CP
2903 TWOADDRESS
TIMING
\'---~/
I
\
2903
LOCAL RAM
ADDRESSING
\
\~-_____-II
I
I
I
I
WRITE =WE
Operand R Address
AO - A4
Operand S and Result Address
80 - 84
A and B provide the two local RAM addresses. As described earlier. while CP is high the contents of these two local
RAM locations are output to the A and B latches. Subsequently. when CP is low. data is written back to the local RAM
location addressed by B. since WE is low. In the illustration above. we show WE being driven low at the proper time by
the WRITE output. WE will usually be connected to the WRITE output from the least significant 2903 slice.
We generate three local RAM addresses in a single 2903 microcycle by changing the B address after reading an operand, and before writing back the result. Timing may be illustrated
as follows:
P-----...
2903 THREEADDRESS
MICROCYCLE
CP
AO - A4
I
80 - 84
Operand S
Address
In the illustration above we delay lEN going low until the last quarter of the microcycle. This gives external logic sufficient time to change the B address. While lEN is high. WRITE is held high. Thus. delaying the lEN low pulse delays the
WRITE pulse - which in turn delays the WE low input until a new address is stable at B. You can generate three-address timing. as illustrated above. by changing the lEN waveform from its normal two-address shape:
cp _ _ _
\ ___
lEN
~
i
8-56
\
~~-_\
to the following continuous three-address shape:
CP
I
\
'''--_ _'--'1
I
I
\
I
I
I
JI
I
I
I
I
J.-ts~
4
\
I
I
-I
T
You cannot directly drive WE from external logic in order to create a three-address microcycle since external logic may
not b~le to identify mi~ycles during which....!:!2 write operation~o occur-and WE should be held high. By
using lEN. and connecting WE to WRITE. you use lEN logic to provide WE with its correct shape. while you use WRITE
to discriminate between microcycles within which a write should. or should not. occur.
You use instruction code bits 10 through 14 to distinguish between simple ALU functions and
special 2903 functions. When the five instruction code inputs 10-14 are all low. 15 through 18 are
interpreted by the 2903 as "special function" identifiers. If one or more of the five inputs 10-14
are high. then simple ALU functions are interpreted as summarized in Table 8-6. These simple fLinctions are all self-evident and need no special discussion.
2903
SIMPLE
ALU
FUNCTIONS
Table 8-6. 2903 Simple ALU Function Specifications
Instruction
Code
ALU Dependent Output Signals
ALU Operation and
Output
MSS
1413121110
00000
00001
000 1 X
001 0 X
o0 1 1 X
o 1 OOX
o1o1X
o1 10 X
o1 1 1X
1 OOOX
1 001 X
1 o 1 0 X
1 o 1 1 X
1 1 0 0 X
1 101 X
1 1 1 OX
1 1 1 1 X
See Table 22-8
A" ALU outputs high
S - R - 1 + CN
R - S - 1 + CN
R + S + CN
S + CN
S + CN
R + CN
R + CN
A" ALU outputs low
RAND S
R EXCLUSIVE NOR S
R EXCLUSIVE OR S
RAND S
R NOR S
R NAND S
R OR S
GIN
P/OVR
C(N+4)
0
C(N+4)
C(N+4)
C(N+4)
C(N+4)
C(N+4)
C(N+4)
C(N+4)
0
0
0
0
0
0
0
0
0
OVR
OVR
OVR
OVR
OVR
OVR
OVR
0
0
0
0
0
0
0
0
Other
MSS
.Other
0
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
G
P
P
P
P
P
P
P
0
0
0
0
0
0
0
0
G
G
G
G
G
G
G
G
G
G
G
G
G
G
G
Z
0
z
z
z
z
z
z
z
1
z
z
z
z
z
z
z
R = R operand input
S = S operand input
Rand S are the complements of Rand S operand inputs. respectivelv
CN = Carry in. C(N+4) = Carry out
MSS = Most Significant Slice
Table 8-6 also summarizes output signal levels associated with each ALU operation. Additional signal levels more
closely associated with the ALU destination specification are given in Table 8-7.
8-57
Table 8-7. 2903 Destination and Shift Specifications for Simple ALU Operations
18 17 16 16
o
CIO
Hex
Shift
Code
Result
(= Y if
OEy = 01
Shift
Load
Y3
MSS11
Y3
Other
Y2
MSS1
Y2
Other
YO
All
SI03
MSSlI
SI03
Other
SIOO
All
0103
0103
0100
WRITE
0 0 0
0
DA21
(F/2)A
None
Q
F3
SI03
SI03
F3
F2
F1
I
I
FO
Hi-Z
Hi-Z
0
1
DL3)
(F/2IL
Q
SI03
SI03
F3
F3
F2
Fl
I
I
FO
Hi-Z
Hi-Z
0
001
0
2
DA2)
(F/2)A
None
DL3)
(Q/2)L
F3
SI03
SI03
F3
F2
F1
I
I
FO
I
QO
0
001
1
3
DL3)
(F/2IL
DL31
SI03
SI03
F3
F3
F2
F1
I
I
FO
I
QO
0
o
o
o
o
(Q/2I L
0
4
None
F
None
Q
F3
F3
F2
F2
Fl
FO
I
I
P
Hi-Z
Hi-Z
0
1
5
None
F
DL31
(Q/2)L
F3
F3
F2
F2
F1
FO
I
I
P
I
QO
1
1 1 0
6
None
F
None
F
F3
F3
F2
F2
F1
FO
I
I
P
Hi-Z
Hi-Z
1
1 1 1
7
F
None
F
F3
F3
F2
F2
F1
FO
I
I
P
Hi-Z
Hi-Z
0
1 000
8
9
None
UA4)
(2F)A
None
Q
F3
F2
Fl
Fl
FO
SIOO
F2
F3
I
Hi-Z
Hi-Z
0
UL5)
(2F)L
Q
F2
F2
Fl
Fl
FO
SIOO
F3
F3
I
Hi-Z
Hi-Z
0
UA4)
(2F),A
None
UL5)
(2Q)L
F3
F2
Fl
Fl
FO
SIOO
F2
F3
I
Q3
I
0
0
1
CIO
o Register
ALU Output
000 1
1
1
o
o
100 1
a.
Signal Detail
Shifter/Destination Summary
Instruction
Code Bits
o
o
1 0
A
1 1
B
UL5)
(2F}L
UL5)
(2Q)L
F2
F2
Fl
Fl
FO
SIOO
F3
F3
I
Q3
I
o
o
0
C
None
F
Q
F3
F3
F2
F2
Fl
FO
F3
F3
Hi-Z
Hi-Z
Hi-Z
1
1
D
None
F
None
UL5)
(2Q)L
F3
F3
F2
F2
F1
FO
F3
F3
Hi-Z
Q3
1
1
1 1 1 0
E
None
SIOO
None
Q
SIOO
SIOO
SIOO
SIOO
SIOO
SIOO
SIOO
SIOO
I
Hi-Z
Hi-Z
0
1 1 1 1
F
None
F
None
Q
F3
F3
F2
F2
F1
FO
F3
F3
Hi-Z
Hi-Z
Hi-Z
0
1
1 1
1 1
1)
2)
3)
4)
5)
MSS = Most Significant Slice
DA = Down Arithmetic
DL = Down Logical
UA = Up Arithmetic
UL = Up Logical
I = Input pin
P = Parity of SI03, F3, F2, Fl, FO
HI-Z = High impedance
F3, F2, F1 and FO are the four ALU output bits. F3 is the high-order bit. FO is the low-order bit.
-
With regard to Table 8-6. note that the Carry Out signal. C(N+4). is active for arithmetic operations only.
P/OVR generates an Overflow status (OVR) at the most significant slice. and a Carry propagate signal (P) at other slices.
Like the Carry Out. P/OVR is active only for arithmetic operations. Unlike P/OVA. GIN is active for all ALU operationsarithmetic and logical. The most significant slice outputs the Sign status (N) which is. in fact. the level of the highorder ALU output bit. Other slices output the Carry generate signal @.
For a discussion of the Carry generate and propagate signals
(3
and
P)
refer to the 2902 description.
The Zero status is active for all slices. during all simple ALU operations. The Zero status is output high when all four
ALU output signals are low. The Zero status output is low otherwise.
Let us now examine 2903 destination options.
2903
DESTINATION
Table 8-7 summarizes destination and shift specifications implied by instruction code bits
OPTIONS
16 through 18 for the simple ALU operations summarized in Table 8-6. In Table 8-7 we show
the ALU output and Q register operations. together with a detailed summary of associated
signal levels. The detailed signal summary is given since slice significance and shift specifications combined make
signal levels less than self-evident. If you look at the signal outputs shown in the signal detail section of Table 8-7. and
compare these signal outputs with the illustrations of arithmetic and logic shifts given below. then the table will be
easy to understand.
Note that signals SIOO. 0100. and 0103 are frequently in a high impedance state.
Selected destination specifications hold WRITE high. These specifications give you the option of not writing ALU output into local RAM - assuming that the WE inputs are connected to the least significant slice WRITE output.
Destination code E propagates the SIOO input across all Y outputs. This code is used to extend the sign of a binary
number. as we will describe later.
Destination codes 4. 6. 6. and 7 report parity of the ALU output at the SIOO pin. Parity is reported for the 5-bit binary number given by S103. F3. F2. F1. and FO. Odd parity generates a high output at SIOO while even parity generates a
low output at SIOO.
Parity logic of the 2903 is cascadable across chip slices since the SIOO parity output of each slice becomes the SI03
input for the adjacent. less significant slice. The SIOO output from the least significant slice will always report the
parity for the combined ALU output. We will demonstrate this multislice parity logic for the simple case of 8-bh data
generated using two 2903 slices. This may be illustrated as follows:
Most Significant
Slice
r
A
SI03
F3
F2
F1
Least Significant
Slice
~~---------'~~--------"
S100/S103 F3 F2 F1 FO SIOO
"'
FO
'\J ~. .---..c/
o
1
~?O
Most Significant
Slice
0
1
1
0
0
Least Significant
Slice
~~---------,~~--------,,,
~~--------,~~--------~,
SI03 F3 F2 F1 FO S100/S103
\ J ~. .---..v
o
1
~/1
8-59
F3
F2
F1
FO
SIOO
0
1
1
0
1
The ALU shifter, but not the Q shifter, gives you the option of specifying either an
arithmetic or a logical shift. The Q shifter allows you to specify logical shifts only.
Logical shifts treat all bits in the same way. Thus. an 8-bit upshift may be illustrated as follows:
7
6
5
4
3
2
2903
SHIFT
LOGIC
o 4---Bit No.
Before upshift
After upshift
An 8-bit downshift may be illustrated as follows:
7
6
5
4
3
2
o
4--BitNo.
Before downshift
After downshift
An arithmetic shift assumes that the high-order bit is a sign bit bit arithmetic upshift may be illustrated as follows:
7
6
5
4
3
2
which must be excluded from any shift. Thus an 8-
o
.....-BitNo.
Before upshift
After upshift
An 8-bit arithmetic downshift may be illustrated as follows:
7
6
5
4
3
2
o
4---Bit No.
Before downshift
After downshift
The 2903 can perform arithmetic shifts since you must identify the most significant. least significant. and intermediate
2903 slices in a multislice configuration. Thus. when you specify an arithmetic shift. logic internal to the most significant slice isolates the high-order bit from the shift. while intermediate and least significant slices perform simple logic
shifts.
8-60
The 2903 ALU shifter is located on the ALU output, in front of the Y data input/output port.
In contrast. the 2901 ALU shifter is located at the local RAM input. Also, the 2903 ALU shifter output can be enabled or disabled via the OEy control signal. Thus you have a large number of
microprogram-selectable options for handling ALU output, over and above the destination options summarized in Table 8-7. ALU output may be transmitted to local RAM:
A
YO - Y3 <. ;'';;,T ';;'f,
...
"'>:..;:::." .;..... ;:.:.:::;; . <>:<.::
:.:::\L~/:}!\·!;;··h:i>'
.•••• M···.d
<';';'::'
U
~
.oil
2903 ALU
OUTPUT
DESTINATIONS
D
Local RAM
A
B
A Latch
B Latch
--
WE=O
--
CP = 0
JJ-
D,
DAO - DA3 It
;',
".
it
OEy =0
..-
Shifter
.-
-
EA
---
...
) DBO - DB3
"'
..
~
RMUX
U
;,
S MUX
0 QO - Q3
-
10
•
'<
:r
"<
~
ALU
l'!;";:;,
<'j':.".. .:•. \. /'; ";';':.:.>"'::0:.';.
.
;: ::
:;
:>•..
;:~l
...
) To Q Register
p
8-61
By holding WE high you can output data at the Y pins. but not write the output to local RAM:
A
YO - Y3
~"\''';;:':';;,
'"
;,V;':;
•
}
~
D
Local RAM
A
B
A latch
B latch
--
WE=1
-
CP = X
~I~
D,
DAO - DA3 h
OEy=O
-
Shifter
.>
EA
.
:,
~
R MUX
>
U
SMUX
I
~
OEB
..
~
} DBO - DB3
II 00 - 03
-
10
'< ;,
AlU
>
:,'
"h',)" ,,;:";,;:,'\, ";, ;kf','
,"
: " ' , ,,';, i i
:::'\~":;:;\;'"
~l
",:;{, A..
..
) To Q Register
In either of the above cases the data may be shifted up or down. if so specified by instruction code bits 15-18 (see Table
8-7).
8-62
You can also discard the ALU output and use the Y pins as the data input port to local RAM:
YO - Y3 l:t.< .. :.>
...•.
.
.> •... :.
. '::.<
....
. ....•.....:..
.'.:.":
".:'.:'
..
:
'\
U
~
----
D
Local RAM
A
B
A Latch
B Latch
~~-~
CP = 0
..)
DBO - DB3
r
D, ,
DAO - DA3 h
--
Shifter
EA
-
11 00 - 03
J
~\. ~
R MUX
S MUX
-
10
J
.,.,.~
'<
ALU
Jl
.A.
..
)
To
a Register
If WE and OEy are both high. ALU output to YO-Y3 or local RAM is discarded
You can use the last two ALU destination codes shown in Table 8-7 to extend a sign bit
across one or more parallel 2903 devices within a single microcycle. Since the ALU destination code is used to generate sign extend logic, this operation can occur in conjunction with
any compatible ALU operation specified by instruction code bits 14, 13, 12, and 11.
2903 SIGN
EXTEND
LOGIC
ALU destination code F transmits the ALU output to the Y pins. and pulses WRITE low Assuming that OEy and WE
are both .Iow. the ALU output will appear at the Y pins. and will be written into local RAM while CP is low
ALU destination code E transmits the 5100 input across all four ALU output lines Again. WRITE is pulsed low: if OEy
and WE are both input low. then the 5100 level is output at all Y pins. and is wntten into local RAM while CP is low
8-63
You can use this pair of ALU destination codes to extend a sign bit by applying the level of the sign bit to the SIOO
input of those 2903 slices that are to extend the sign. Consider a 16-bit Central Processing Unit where the sign for the
low-order byte must be extended across the high-order byte. This may be illustrated conceptually as follows:
Least
Most
Significant
Device
Intermediate
Device
2
3
15
I
14
13
12
11
10
9
15
14
13
12
11
10
9
8
Si~nificant
1
0
7
8
zIzIzI z z zIzIz
Intermediate
Device
I
X
6
5
Device
4
3
o ....-BitNo.
2
I y I y I y I y I y I y I y
6
5
432
y
y
y
o
I
Before sign extension
4 - - B i t No.
I I I I I
y
y
y
y
After sign extension
X = sign bit
y = data bit
Z = irrelevant bit
A Central Processing Unit implemented using 2903 slices will automatically generate a sign extended ALU result for
any arithmetic operation. You use sign extend logic to create data, rather than modify results of any computation.
Suppose. for example. an 8-bit data input is received from an I/O port: if a 16-bit CPU is to interpret this data byte as a
signed binary number. then the high-order bit must be propagated through the high-order byte of a 16-bit word as illustrated above.
This is easily done using the E and F ALU destination codes. This is illustrated in Figure 8-19.
Let us examine Figure 8-19. The two low-order 2903 slices are generating real data. These two slices therefore receive
an F ALU destination code via 18-15. This destination code causes the ALU output to appear at the Y pins. and the highorder ALU output bit to appear at 5103. The two high-order 2903 slices generate the high-order byte across which the
sign must be extended. These two 2903 slices therefore receive an E destination code via 18-15. The E destination code
causes the 5100 input to be propagated across the ALU outputs.
8-64
I
I
...
Sign Out
-
z
C(N+4)
OVR
W/MSS
N
~~j
111
--
15(1) = 1
15(0) = 1
-
0100
SI03
SIOO
~
-
SI03
+5V
C(N+4)·
IS
2903
()
SIOO
~
I
!e~
+
4 III
0103
0100
~
-
SI03
SIOO
f4-
+5V
C(N+4)
LSS
2903
",,"-Y
CN
()
W/MSS ~~
W/MSS f----1.
~
z
0100
0103
_X
CN
IS
2903
+5V
LSS
WE
z
C(N+4)
CN
!e~
15(3) = 0
15(2) = 0
--
SI03 2903 SIOO
~
16-18 =
0103
0103 MSS 0100
I
I
z
~
LSS t - WE
!e~
j
I
-
j
LSS
WE
!
CN f - -
W/MSS
~
~
!!!~
+~
~
I-
LSSQ
WE
t -:-
-
I-
--
t~
~"V'7
.,,>
."V.,
'v~
Y3 - YO
XXXX
Y3 - YO
XXXX
Y3 - YO
XYYY
Y3 - YO
YYYY
MSS means Most Significant Slice
IS means Intermediate Slice
LSS means Least Significant Slice
Y = data bit
X = sign bit
Figure 8-19. 2903 Sign Extend Logic
2903 SPECIAL FUNCTIONS
Let us now examine special 2903 functions. These functions are summarized in Table 8-8.
Special functions are implied by instruction codes bits 15 through 18 when instruction code bits 10 through 14 are all O.
Nine special functions are provided; seven special function codes are unused. You should be sure not to use any
of the unspecified special function codes since the 2903 device's response to these unspecified function codes is not
guaranteed.
Table 8-8 summarizes signal outputs and exact ALU operations associated with each of the special functions.
Wherever a signal's primary purpose is meaningful. the signal is so used by a special function. Where a signal's primary
purpose is not meaningful. the special function may generate an output to meet its specific needs.
Do not attempt to understand ALU operations or signal utilization merely by inspecting Table 8-8. Many of the ALU
operations, although absolutely accurate representations of ALU logic, rely on specific external pin connections to
generate the required net effect. Signals, likewise, are used in special ways that depend not only on the special function, but also on the required pin connections which have been arbitrarily selected by the 2903 designers.
ALU logic and signal utilization is described in detail function by function.
We will begin by examining the simpler 2903 special functions, since many of these simple special functions act as accessory commands to the more complex functions.
There are two normalization functions: a single length and a double length normalization. The double length
normalization function is also the first twos complement divide instruction.
8-65
Table 8-8. 2903 Special Functions Summary
Associated Signals
Instruction
Code(l)
ALU Operation
SI03
ALU Function
18 17 16 15
MSS
Other
SIOO
0103
0100
P/OVR
WiiiTE
C(N+4)
Z
GIN
MSS
Other
MSS
Other
MSS
IS
LSS
F = S + CN if Z = L
F = R + S + CN if Z
=H
Hi-Z
I
FO
I
00
0
C(N+4i
OVR
P
N
G
I
I
00
F = S + CN if Z = l
F = R + S + CN if Z
=H
Hi-Z
I
FO
I
00
0
ON+4)
OVA
P
N
G
I
I
00
I
I
P
Hi-Z
Hi-Z
0
C(N+41
OVR
P
N
G
Z
Z
Z
I
I
P
Hi-Z
Hi-Z
0
C(N+4'
OVR
P
(2)
G
53
I
I
Hi-Z
I
FO
I
00
0
CIN+4)
OVR
P
N
G
I
I
00
F = 5 + CN
F3
F3
Hi-Z
03
I
0
(3)
02(!)01
P
03
G
(4)
(4)
(4)
F = S + CN
R3@F3
F3
I
03
I
0
(5)
F2~Fl
P
N
G
(6)
(6)
(6)
F = S + R + CN if Z = l
F = S - R - 1 + CN if Z = H
~
F3
I
03
I
a
CIN+41
OVR
P
N
G
(7)
I
I
F = S + R + CN if Z = l
F = S - R - 1 + CN if Z = H
F3
F3
Hi-Z
03
I
0
CIN+41
OVA
P
N
G
(7)
I
I
0
0
0
0
Unsigned multiply
0
0
0
1
Unused
0
0
1
0
Twos' complement multiply
0
0
1
1
Unused
0
1
0
0
Increment
F = 5 + 1 + CN
0
1
0
1
Sign/Magnitude twos
complement
F = S + CN if Z
F = S + CN if Z
0
1
1
0
Twos complement multiply.
last cycle
F = 5 + CN if Z = l
F = S - R - 1 + CN if Z
0
1
1
1
Unused
1
0
0
0
Single length normalize
1
0
0
1
Unused
1
0
1
0
Double length normalize
1
0
1
1
Unused
1
1
a
a
Twos complement divide
1
1
a
1
Unused
1
1
1
a
Twos complement divide.
final
1
1
1
1
Unused
=l
=H
=H
CD
m
0>
ao, a" a, ,,'" a' •• ,", '00' a ....~ ,."" ""}
Fa. Fl. F2 and F3 are the four AlU output bits.
RO. R1. R2 and R3 are the four R operand bits.
SO. S 1. S2 and S3 are the four S operand bits.
Bit 3 is the high-order bit.
Bit 0 is the low-order bit.
1)
2)
3)
4)
5)
6)
7)
10 - 14 must all be O.
N if Z = O. S3 (i) F3 if Z = 1.
o 3 (!) 02 at MSS. CiN+4; at other slices.
Zero status for 0 register output.
F3 (!) F2 at MSS. CIN+41 at other slices.
Zero status for combined. a-bit 0 register and AlU outputs.
Sign compare output.
Hi-Z
I
P
MSS
IS
lSS
Other
=
=
=
=
=
=
=
Signal floated
Input signal
Parity of S103. Y3. Y2. Yl. YO
Most Significant Slice
Intermediate Slice
least Significant Slice
IS or LSS
2903
The normalization operation upshifts the contents of a data word until the two highorder bits have different values. Zeros are shifted into low-order bit positions. Here are
some normalization illustrations for 16-bit words:
NORMALIZE
SPECIAL
FUNCTIONS
Initial
Normalized
0000001011000111
0101100011100000
1110101101000101
1010110100010100
0110101101011010
0110101101011010
0000000000000000
Cannot be normalized
1111111111111111
1000000000000000
Each normalize instruction is executed in one microcycle. During this microcycle one upshift occurs if the two highorder bits of the most significant slice S ALU operand are both O. or both 1. No operation occurs if the two high-order
bits differ. In order to complete the normalization process for a multibit word that has many leading 0 or 1 bits. you
must re-execute the normalize instruction the required number of times to shift out leading similar bits. If. for example.
there are five leading 0 bits. followed by a 1 bit. you will have to execute a normalize instruction four times before the
data is normalized. On the fifth execution of the normalize instruction the data will be left unaltered.
Your logic must identify the point at which data has been normalized; the normalize instruction outputs appropriate
status signals to identify normalization - as we will describe shortly,
If binary data is being interpreted as a signed binary number. then a positive number. after normalization. will have a 0
in the high-order bit and a 1 in the adjacent bit:
01XXX----
After normalization a negative number will have a 1 in the high-order bit and a 0 in the adjacent bit.
10XXX----
The single length normalization instruction generates a data word out of the Q registers of
parallel 2903 slices. Thus. you would generate an 8-bit data word out of two parallel slices as
follows:
7
6
o
2
4
~BitNo.
2903
SINGLE
LENGTH
NORMALIZATION
I I I I I I I
~~
MSS
LSS
Q
Q
Register
Register
Four 2903 slices generate a 16-bit data word as follows:
15
14
13
12
11
10
9
8
6
5
4
0 ~ Bit No.
3
Ll.llllUJl!LWIJ
MSS
IS
IS
Q
Q
Q
Q
Register
Register
Register
Register
LSS
MSS means Most Significant Slice. IS means Intermediate Slice. LSS means Least Significant Slice.
The double length normalization instruction generates a data word out of the Q register and
the local RAM location addressed by B. Two 2903 slices would generate a 16-bit word as
follows:
15
14
13
12
11
10
9
8
7
6
5
4
3
o
2
LSS
MSS
Local
LSS
Local
MSS
Q
Q
RAM
RAM
Register
Register
8-67
2903
DOUBLE
LENGTH
NORMALIZATION
~BitNo.
There are some differences between the single and double length normalization instructions resulting from the
fact that the double length normalization instruction must use local RAM, and the ALU, while the single length
normalization instruction needs Q register logic only. We will therefore look at the single length normalization
instruction first.
The single length normalization instruction performs a number of upshifts until the most significant 2903 Q register
has different values in its two high-order bits. Each upshift requires one microcycle. therefore the total execution time
for the normalization instruction is variable. But the C(N+4) and OVR outputs are used to identify the last, and second to the last microcycles of the single length normalization instruction. On the second to the last cycle the OVR
signal is output high; OVR therefore outputs the Exclusive-OR of 02 and 01 at the most significant slice. C(N+4).
likewise. outputs the Exclusive-OR of 03 and 02 at the most significant 2903 slice. This may be illustrated as follows:
o
Register of
Most Significant Slice
.~
103102101 J 00 1
~
C(N+4) = 03
<±> 02
~
OVR
= 02
EE> 01
Thus C(N+4) goes high on the last microcycle of the single length normalization instruction. while OVR goes high on
the previous microcycle.
You cannot normalize a data word that is initially all 0 bits. Since zeros are shifted into the low-order bit position
with each upshift. the normalization operation would never end. The single length normalization instruction therefore
outputs a high Signal on the 0 status line and terminates in a single microcycle. For this to be possible the single length
normalization instruction uses Z status logic to indicate 0 register contents. rather than ALU output. That is to say. Z is
output high when all Q register bits are 0, not when all ALU outputs are O.
You will now understand the special information output via C(N+4). OVR. and Z signals. as shown in Table 22-8 for the
single length normalization instruction.
During each microcycle of a single length normalization instruction the Q register contents are recycled through
Q shifter logic. ALU logic, which would otherwise be unused, adds the contents of CN to the S operand input.
This logic allows you to count the number of microcycles - and therefore upshifts - performed by the single length
normalization instruction. Assuming that WE. OEy. OEB. and 10 are all low and CN is high. then the RAM location addressed by B becomes a microcycle counter. This RAM location becomes the ALU S operand. and the destination for
the ALU output. The ALU output is simply theS operand input incremented by 1. assuming that CN is indeed high.
Single length normalization instruction pin connections are illustrated in Figure 8-20.
You can. if you wish. maintain a microcycle counter in external memory by inputting the ALU S operand from DBO-083
and outputting the ALU result at YO-Y3. This requires that WE and OEB be input high.
If you execute the single length normalization instruction with 10 high. then the 0 register contents also become the
ALU S input. Now on each microcycle the 0 register contents. before they are upshifted. are output by the ALU to YOY3. and/or local RAM. optionally incremented by 1 if CN is input high.
Let us now 4;txamine the double length normalization instruction. The RAM location addressed by B provides the high-order half of the word being normalized. 0103 from the most
significant 2903 slice must therefore be connected to SIOO at the least Significant 2903 slice.
Also, you cannot use ALU logic to count instruction microcycles since ALU logic contributes to the normalization operation. Therefore CN must be input low, and if you wish
to count microcycles you must use external logic or an extra microcycle per cycle.
2903
DOUBLE
LENGTH
NORMALIZATION
The high-order half of the word being normalized can come from internal or external RAM. If it comes from internal
RAM then the RAM location addressed by B must provide the S operand to the ALU. and must receive the ALU output.
But you can also use external RAM to provide the high-order half of the word being normalized; now OBO-OB3 generates the ALU S operand and the ALU output is transmitted to YO-Y3. For this to occur OEB and WE must both be high.
The C(N+4) and OVR statuses identify the last and second to the last microcycles of the double length normalization instruction's execution - just as they do for the single length normalization instruction. The double
length normalization instruction also terminates in a single microcycle when you attempt to normalize a word which is
initially O. At this time the Z status is output high. For this to be possible double length normalization logic tests the
combined contents of the 0 register and ALU output in order to generate a Z status - as indicated in Table 8-8. Double length normalization pin connections are illustrated in Figure 8-21.
8-68
+5V
Z = 00 • 01 ••• ON
Z
03
0103
0100
0103
0100
SIOOt---~
SIOO
SI03
SIOO
SI03
SIOO
CNt---~
03
OVR
N
Z
0100
03¥02
02¥01
Z
0100t----t 0103
SI03
C(N+4)
CN
C(N+4)
+5V
IS
2903
W/MSS
MSS
2903
LSS
CN
IS
2903.
W/MSS
+5V
C(N+4)
LSS
2903
W/MSS
LSS
LSS
WE
WE
CN
WE
WE
-
ALU output = [S] + CN. If WE. OEB. OEy and 10 are low. [B] = [y] = [B] + CN
[0] = [a] upshifted one bit
MSS means Most Significant Slice
IS means Intermediate Slice
LSS means Least Significant Slice
[S] =
[B] =
[y] =
[0] =
S ALU input
Local RAM contents addressed by B
y output
0 register contents
Figure 8-20. 2903 Single Length Normalization Function Pin Connections
Another simple 2903 special function is the Sign/Magnitude Twos Complement. This function converts negative twos complement numbers to this positive form, while leaving positive twos complement numbers alone. This may be illustrated as follows for 16-bit numbers:
2903 SIGN/
MAGNITUDE
TWOS
COMPLEMENT
FUNCTION
Initial
After Sign/Magnitude
Twos Complement
0110010111010010
0110010111010010
Unchanged positive number
1110101111010101
0001010000101011
Twos complement of negative number
The 2903 uses slightly devious logic in order to implement the Sign/Magnitude Twos Complement function. This is the
actual ALU algorithm executed:
ALU output =
ALU output =
[S) means ALU S operand.
S is
[S)
[S)
+ CN
+ CN
if Z status is 0
if Z status is 1
the complement of the S operand.
8-69
+5V
()
Z
.:.
~
= FO • Pi . . . FN • QO • Q1 • ON
-
z
z
z
z-
0103
0100
0103
0100
0103
0100
QI03
0100
F3 . . . SI03
SIOO
SI03
SIOO
SI03
SIOO
SI03
SIOO
F3 ¥F2'" C(N+4)
F2 ¥F'4-
OVR
F34-N
CN
W/MSS
MSS
2903
+5V
IS
2903
r-4.
CN
LSS
WE
WE
•
I~
OEy.
+5V
IS
2903
()
W/MSS 1..........0
+5V
LSS ~
ALU output = [S] + CN. If WE. OEB.
ALU output is upshifted one bit
[0] = [0] upshifted one bit
C(N+4)
CN
C(N+4)
C
W/MSS
LSS
~
WE
C(t-.J+4)
LSS
2903
~
-
~O
---
CN - 0
W/MSS
WE
LSS~
~
CN and 10 are low. [B]
= [y] = 2
[B]
MSS means Most Significant Slice
IS means Intermediate Slice
LSS means Least Significant Slice
[B] = Local RAM contents addressed by B
[S] = S ALU input
[y] = y output
[0] = a register contents
Figure 8-21. 2903 Double Length Normalization Function Pin Connections
During execution of the Sign/Magnitude Twos Complement instruction. the Zero status at the most significant 2903
slice directly outputs the high-order S operand bit - which is the sign bit for a twos complement number. The Zero
status becomes an input to intermediate and least significant slices. which therefore receive the sign bit from the most
significant slice. For a 16-bit number this may be illustrated as follows:
Most
Significant
Slice
15
z=y
14
13
Least
Significant
Slice
Intermediate
Slices
12
11
10
9
8
6
5
4
3
0 4--BitNo.
2
z=y
z=y
z=y
!
!
t
Now you can connect pins of 2903 slices in any way to make use of the Sign/Magnitude Twos Complement
ALU logic, but to use it for its intended purpose, the connections illustrated in Figure 8-22 are required.
8 ... 70
+5V
4)
j=
--
S3
-
-
Z
Z
:~
~
-
......
-
Z
Z
0103
0100
0103
0100
0103
0100
0103
0100 t--
SI03
SIOO
SI03
SIOO
SI03
SIOO
SI03
SIOO I - -
C(N+4)
OVR
N
C(N+4)
CN
W/MSS
MSS
2903
~
LSS
r---<>
WE
C(N+4)
CN
+5V
IS
2903
W/MSS ~~
+5V
LSS
+5V
C)
CN
LSS
2903
W/MSS 1----4
~
LSS
WE
j,.
C(N+4)
CN
IS
2903
()
W/MSS
~
WE
WE
LSS~
j,.
•
ALU output = [S] + CN if Z = 0, or [5) + CN if Z = 1. If WE, OEB, OEy, CN and 10 are low,
[B) remains unaltered if Z = 0, or [B) = [B) + 1 if Z = 1
In either case [y) = ALU output
Neither ALU nor
a shifter function
'OVR = 1 if ALU input is 1000····00, the most negative binary number.
"N = F3 if Z = 0, or F3 E9 S3 if Z = 1
MSS means Most Significant Slice
IS means Intermediate Slice
LSS means Least Significant Slice
[S) = S ALU input.
(S) = complement of S ALU input
[F) = ALU output
[B) = Local RAM contents addressed by B
[0) = a register contents
F3 = High-order ALU output bit from most significant slice
S3 = High-order ALU S operand input bit to most significant slice
Figure 8-22. 2903 Sign/Magnitude Twos Complement Function Pin Connections
By connecting Z to CN positive. twos complement numbers are passed unaltered through the ALU:
[F)
CN
=
=
[S] + CN if Z = 0
Z. therefore [S] = [S]
+0
But a negative twos complement number is complemented and then incremented:
[F) = [5] + CN if Z = 1
CN = Z. therefore [S] = [5]
+ CN
In other words. a twos complement number is twos complemented - which generates a positive number. (If you are
unclear on this twos complement logic refer to Volume 1. Chapter 2.)
8-71
The negative status, N, is output high at the most significant slice if a negative twos complement number was
converted to its positive form. This is the actual logic used by the most significant 2903 slice:
If
Z
= 0,
N
= F3
Z = 0 when S3 = 0, in which case [F)
If
= [S]
Therefore N = F3 = S3 = 0
Z = 1, N = F3 E9 S3
Z = 1 when S3 = 1, in which case [F) = [S]
Therefore N = F3 ~ S3 = S3
S3 = 1
e
+1
The Overflow status indicates the only overflow condition which can occur when a Sign/Magnitude Twos Complement conversion is performed. There is no twos complement positive representation for the most negative twos
complement number which can be represented:
If
[S] = 1000·····0
[F) =0111·····1 + 1, = 1000·····0
If this most negative number is received at the S operand, it is passed through unaltered and the Overflow status from
the most significant slice is output high.
The Sign/Magnitude Twos Complement instruction places no restrictions on where the S operand may come from. Any
of the three options - external memory, local RAM, or the Q register - may provide the S operand to the ALU.
The third and last of the simple 2903 special functions is the Increment. This special function adds 1, plus the Carry In to the S operand. This algorithm may be illustrated as follows:
[F) = [S] + 1 + CN
2903
INCREMENT
FUNCTION
[F) is the ALU output. [S] is the ALU S operand input. and CN is the Carry In. If CN is 0, you increment by 1; if CN is 1,
you increment by 2. This is useful in byte/word machines if the Program Counter is kept in local RAM.
Once again the S operand may come from external or local RAM or from the Q register.
The increment special function makes no special use of status logic.
2903
UNSIGNED
Let us now look at the unsigned multiply special function. The algorithm used by the 2903
MULTIPLY
to perform an unsigned multiply is exactly the same as the algorithm which we described
earlier in this chapter, when showing how to program an unsigned multiply for the 2901. Initially the multiplier must be in the Q register and the multiplicand in the RAM location which provides the ALU R input.
This may be external RAM connected to DAO-DA3, or local RAM addressed by A. The product will be generated in the
RAM location that receives ALU output. and the Q register. The RAM location connected to ALU output may be external RAM connected to YO-Y3, or local RAM addressed by B; it ultimately holds the upper half of the product. The Q
register holds the lower half of the product. The RAM location that finally holds the upper half of the product must initially contain O. Thus we can illustrate initial and final data locations as follows:
Multiplicand
Initial:
Multiplicand
Final:
R
Multiplier (0)
II
R
S
IQ I
Product
upper lower
II
S
I
Q
I
The 2903 unsigned multiply operation will multiply two 16-bit numbers to generate a 32-bit product. If you
wish to multiply larger numbers you must do so in 16-bit increments and add partial products using additional
microcycles.
If we compare the register utilization illustrated above with the unsigned multiply description given for the
2901, the local RAM location addressed by B in the illustration above becomes the window into which the multiplicand is added whenever a 1 bit is shifted out of the multiplier; but 2903 logic tests this bit internally, outputting the
least significant Q register bit from the least significant 2903 slice via the Z status. The Z status becomes an input to
the most significant and intermediate slices, so that these 2903 devices can also tell whether the multiplicand is to be
added into the product window. Thus the unsigned multiply consists of 16 microcycles. In each microcycle the loworder bit of the Q register in the least significant slice is tested. If this bit is 1. the mu Itiplicand is added to the partial
product. If this bit is 0, no addition is performed. Addition, if it occurs, consists of adding the ALU Rand S inputs, which
probably means adding the contents of the RAM location addressed by A to the contents of the RAM location addressed by B. If A and B are the Rand S ALU inputs, respectively, with the sum returned to the RAM location addressed
by B, then WE, OEB. OEy. EA. and 10 must all be O.
8-72
After the low-order bit of the Q register in the least significant slice has been tested. and a conditional addition has
been performed. the product space (local RAM addressed by B. and the Q register) is downshifted one bit position during the same microcycle. The Carry status following the addition is shifted into the high-order bit of the ALU output for
the most significant slice. If no addition is performed. then the Carry will e_qual O. and 0 will be shifted into the highorder ALU output bit of the most Significant 2903 slice. This may be illustrated as follows:
Most Significant Slice
R input
S input
~
~
----I~--I~
C(N+ 1) = C
'F2
C
\
\
F3
\'----~.
\
F2 F1
SIOO = FO
A single microinstruction performs the actual unsigned multiplication; however. preceding instructions must load the
multiplier and multiplicand into their appropriate registers. and must zero the RAM location to be used for the running
partial product.
Necessary pin connections in a 2903 configuration that uses the unsigned multiply function are illustrated in
Figure 8-23.
The use of status by the unsigned multiply function is straightforward - with the exception of the Zero status which
propagates the current low-order multiplier bit to all 2903 slices as we have already described. The Carry In, CN,
must be O. If it is 1 you get the wrong answer when the multiplicand is added to the product window. The Carry Out,
C(N+4}, the Overflow, and the Sign status are all output by the most significant 2903 slice to reflect the result
of each partial product addition. However. these statuses are useless and should be ignored.
The 2903 will also perform twos complement multiplication on two 16-bit signed binary
numbers to generate a 32-bit signed binary resuft. The algorithm for performing twos complement multiplication is essentially the same as the unsigned multiplication algorithm
which we have already described; the same registers are used to hold the multiplier, the
multiplicand, and results.
2903 TWOS
COMPLEMENT
MULTIPLY
FUNCTION
There are two differences between signed and unsigned multiplication; they are:
1)
2)
We must account for the sign bit of the multiplier. which is not a magnitude bit.
Slightly different logic is needed to generate the bit which is shifted into the high-order ALU output from the most
significant 2903 slice following each downshift.
The logic of twos complement multiplication using Signed binary arithmetic is readily deducible from the unsigned
multiplication algorithm which we described for the 2901. together with the discussion of Signed binary arithmetic
given in Chapters 2 and 3 of Volume 1. Moreover. you the user cannot modify twos complement multiply logic in any
way; therefore a detailed understanding of the algorithm is of academic interest only. The algorithms for Signed and
unsigned binary multiplication remain the same until the last microcycle - at which time the sign bit of the
8-73
multiplier is in the low-order bit of the product space. This may be illustrated as follows:
Local RAM
MSS
IS
IS
------..
LSS
..
Ix/x/xlxlxlxlxlxlxlxlxlxlxlxlxlx
V ~---- ------~~
~......----~
window into which
multiplicand is added
generating partial
product
Q Register
MSS
IS
IS
LSS
~~
I X I X I X I X I X I X I X IX I X I X I X I x·1
X
I I xl I
X
S
t
sign bit. last
bit of multiplier,
which has been
downshifted out
of Q register
If the sign bit is 0, then the multiplier is positive and the multiplicand need not be added again to the partial product;
following the next downshift the multiplication is complete. But if the sign bit is 1, then on the last microcycle the
mu ltiplicand must be subtracted from the partial product before the final downshift.
When the Twos Complement Multiply function is executed, following each downshift, the Exclusive-OR of the
Overflow and Sign statuses is moved into the high-order bit position of the most significant 2903 slice. This ensures that a 1 is shifted into the high-order bit position if addition generated a Carry, or if a negative result must have its
sign extended.
8-74
+5V
C)
::
-.
OO,LSS
-
-
.~
-
-
4,.·
Z
Z
Z
Z
0103
0100
0103
0100
0103
0100
0103
SI03
SIOO
SI03
SIOO
SI03
SIOO
SI03
C(N+4)
CN
OVA . . . OVA
F3'" N
~
+5V
LSS r--o
W/MSS
MSS
2903
WE
j
C(N+4)
C(N+4)
CN
IS
2903
+5V
C)
CN
IS
2903
W/MSS ~
LSS
+5V
()
W/MSS
0100 ~OO
FO, LSS
SIOO
C(N+4)
LSS
2903
H
W/MSS
LSS I--
~
CN ~O
WE
WE
WE
h
j~
j~
LSS~
+ CN if Z = 0, or [A) + [S) + CN if Z = 1
[F) and [0) are downshifted one bit position
MSS means Most Significant Slice
IS means Intermediate Slice
LSS means Least Significant Slice
[S) = S ALU input
[A) = A ALU input
[F) = ALU output
(0) = 0 register contents
00 = Low-order bit of 0 register
FO = Low-order bit of F register
ALU output
= [S)
Figure 8-23. 2903 Unsigned Binary Multiply Function Pin Connections
Figure 8-24 illustrates pin connections needed to execute Twos Complement Multiply and Twos Complement
Multiply Last Cycle special functions.
The only non-obvious aspect of Figure 8-24 is the generation of the Carry In (CN) to the least significant 2903 slice.
This Carry In must be 0 until the last microcycle. at which time it must receive the Zero status. We therefore show the
Twos Complement Multiply Last Cycle instruction code uniquely generating an ENABLE signal which conditions an
AND gate that generates the CN input The AND gate passes through the Zero status during the Twos Complement
Multiply Last Cycle instruction's execution. but at other times the AND gate does not pass the Zero status. generating a
o CN input This function is provided in the 2904 logic.
You must execute twos complement multiply instructions in the proper sequence in order to perform twos complement
multiplication using 2903 devices You execute the Twos Complement Multiply special function fifteen times. then you
execute the Twos Complement Multiply Last Cycle special function.
The two microinstructions which perform the twos complement multiply and the last cycle of the twos complement
multiply must of course be preceded by microinstructions that correctly load registers and zero the memory word being
used for the high-order half of the product
8-75
+5V
()
~~
00, L5S
-
....
OV
Z
:~
-
-
Z
Z
Z
0103
0100
0103
0100
0103
0100
0103
0100
-
SI03
SIOO
SI03
SIOO
SI03
SIOO
SI03
SIOO
-
C(N+4)
ll.- OVR
W/MSS
MSS
2903
F3 4 - N
C(N+4)
CN
WE
14r--o
CN
+5V
IS
2903
()
C)
W/MSS ~
W/M5S
~
l5S
lSS
~
C(N+4)
+5V
lS
2903
+5V
lSS
CN
WE
WE
.~
j ~
C(N+4)
lSS
2903
~~
-
W/M5S
WE
ENABLE
AlU output, not last cycle,
AlU output, last cycle, =
= [5]
+ CN if Z = 0, or
[R] + [5] + CN if Z = 1
[F] and [0] are downshifted one bit position
[5] + CN if Z = 0 or
[5] - [R] - 1 + CN if Z = 1
[F] and [0] are downshifted one bit position
ENABLE is high on last cycle only
MSS means Most Significant Slice
IS means Intermediate Slice
l5S means least Significant Slice
[S] = S AlU input
[R] = R AlU input
[F] = AlU output
[OJ =
register contents
00 = low-order bit of a register
FO = low-order bit of F register
a
Figure 8-24. 2903 Signed Binary Multiply Function Pin Connections
8-76
CN
~
---0=
LSS*
We will now examine the 2903 Twos Complement divide special function.
2903 TWOS
You divide a divisor into a dividend. The answer is called a quotient. and there will be a remainder.
This may be illustrated as follows:
COMPLEMENT
DIVIDE
FUNCTION
Quotient
r = Remainder
Divisor ) Dividend
Conceptually the algorithm for performing binary division is very straightforward. As for decimal division. you begin at
the most significant end of the dividend:
1 0 1 ... 1 1) 1 00· .. 1 1 0 1
110 ... 11
But when you perform binary division the problem reduces to comparing the magnitude of the divisor and the current
dividend field:
~
. - . -. X .. - - . - - -
(
..
DIvisor larger. X = 0
101"'11)"-"-'-'-~
Divisor smaller. X = 1 - 1 0 1· .. 1 1
~
When performing a twos complement divide we begin by subtracting the divisor from the high-order end of the dividend:
1 0 1 0 1 1 0 1 »):f'&it"Q;~;;1;(11 0 1 1 1 1 0 1
/10101101
Dividend most significant field
001 001 0 1--Partial remainder
If the divisor is the smaller number. as it is in the illustration above. then the partial remainder is positive: we add the
next dividend bit and subtract again:
10101101)1101001010111101
10101 101
001001011
10101101
This is equivalent to upshifting the combined partial remainder and dividend residue one bit position. and then
subtracting the divisor again:
I'--
1 0 1 0 1 1 0 1) 1 1 01 00 1 0 1 0 1 11 1 0 1
~10101101~
'--P-a-rt-ia-I-re-m---ainder-p 0
JJ?t
10
~
Dividend residue
---~
0010010110111101
001 001 0 1 1 0 1 1 1 1 0 1-Upshift combined partial
---------~11i0!...1LQOJ1_1LQO~1
remainder and dividend
residue
8-77
But what happens if you get a negative result after subtracting the divisor from the current dividend field? The answer
is that you must add the divisor back to the partial remainder before upshifting the dividend one bit position: and then
subtract the divisor again. But in binary logic this is what happens:
(Partial remainder + Divisor) x 2 - Divisor
t
this is the equivalent of an upshift
The sequence of operations is equivalent to:
Partial remainder x 2 - Divisor
Therefore when you subtract the divisor and get a negative result. you simply upshift the concatenated Partial remainder and Dividend residue fields one bit position. then subtract the divisor on the next step.
This is. in essence. the algorithm used by the 2903 to perform binary division. and is called "non-restoring" division. It
is based on the Twos Complement Divide special function. which performs the following net operations:
[F] = [S]
+ [R]
if Z = 0
[F] = [S] - [R] if Z = 1
[Fl [S1. and [R] are the ALU output. S operand and R operand. respectively.
The Zero status is generated by sign compare logic as the complement of the Exclusive-OR of most significant slice
high-order ALU output and R operand input bits:
Z = R3E9F3
Additional required conditions are that lEN be low and one of the Twos Complement Divide special functions be executed.
The sign compare level is output at the most significant 2903 Zero status and it is input to the Zero status of intermediate and least significant 2903 slices. In effect. the sign compare logic compares the sign of the partial remainder with
the sign of the divisor. This generates the following logic sequence:
1)
If [F] and [R] signs are the same. the divisor had a smaller absolute magnitude than the dividend field from which it
was subtracted. Z is therefore O. so on the next microcycle we get:
2)
If [F] and [R] signs differ. the divisor had a larger absolute magnitude than the dividend field from which it was
subtracted. Z is therefore 1. so on the next microcycle we get:
[F] = [S]
+ [R]
[F] = [S] - [R]
The quotient bits are also determined by comparing the sign of the partial remainder with the sign of the divisor. If the
signs differ. the cu rrent quotient digit is 0 because the divisor has the larger absolute value: but if the signs are the
same. the current quotient digit is 1 because the divisor has the smaller absolute value.
Let us now look at the exact 2903 implementation of the binary division. The two steps defined above do not take
into account the first step - at which time we have no partial remainder. or ALU output. 2903 division logic therefore
demands that the absolute magnitude of the divisor be greater than the absolute magnitude of the most significant half of the dividend. To ensure that the divisor does indeed have larger absolute magnitude, the algorithm
illustrated in Figure 8-25 is recommended in Advanced Micro Devices' literature. We will describe this logic.
even though other logic could achieve the same desired result.
In order to compare the absolute magnitude of divisor and dividend. we need to work only with the most significant
half of the dividend. Comparison instructions destroy the data. therefore we begin by moving the divisor and the most
significant half of the dividend to temporary buffers - in all probability additional locations in local RAM.
When moving the divisor to an alternate RAM location we can test the Zero status to see if the divisor is O. If it is. the
division must be aborted.
Next we use the Sign/Magnitude Twos Complement special function (which we have already described) to generate
positive magnitudes for the copies of the divisor. and the most significant half of the dividend: now we can compare
these magnitudes without bothering about sign.
The Sign/Magnitude Twos Complement function. when executed. ijenerates a positive Overflow status if the data input is the most negative binary number allowed - in our case _2 16-. We take advantage of this Overflow status when
operating on the most significant half of the dividend. If the most significant half of the dividend is -216. then the
divisor cannot possibly be larger. so we downshift the entire dividend one bit position and restart. We also check the
8-78
Overflow status when performing the Sign/Magnitude Twos Complement operation on the divisor. If the divisor is
_2 16, then it must be larger than the dividend, which is not -2 16, since the dividend test was made first. We therefore
go straight to the division operation.
If neither the divisor nor the most significant half of the dividend is -2 16 , we upshift both numbers one bit position to
remove the sign bit. then subtract the most significant half of the dividend from the divisor. If the dividend is larger. it
must be downshifted one bit position - and the test repeated. When the divisor is larger, we are ready to start the division.
If you scale the divisor or the dividend, then the quotient must be scaled in compensation. Divisor, dividend and quotient scaling logic is entirely your responsibility.
Combining the data preparation and division programs, the sequence of 2903 special functions shown in Table
8-9 is recommended in v~ndor literature to perform binary division. Table 8-9 shows a 16-bit divisor divided into
a 32-bit dividend to generate a 16-bit quotient and a 16-bit remainder.
The instruction sequence preceding the actual division instructions implement Figure 8-26 logic. These instructions
need no special discussion. But we do need to clarify the manner in which status signals output by the 2903 are
handled. The 2903 outputs status and data at the same time. For timing details refer to the 2903 microcycle description given earlier in this chapter. Some 2903 functions require status output by one 2903 slice to be input to other
2903 slices within the same microcycles; for an example of this look at the way Z is used by the Twos Complement
Divide special function. Status is output early enough in the microcycle for an output to become an input to another
2903 slice within the same microcycle. But external logic will not have time to process any 2903 status outputs in the
process of generating 2903 inputs for the same microcycle. Status output in one microcycle must be processed by external logic during the next microcycle. In Table 8-9 the comments associated with each microinstruction identify
relevant status, if any, which is generated during the microinstruction's execution. Comments make clear the fact that
the generated status must be tested during the next microcycle's execution time. Status output by the 2903 is usually
tested by microprogram address generation logic. Later in this chapter, when we describe microprogram address
generation devices. the consequences of testing status while executing the next microinstruction will become self-evident.
The three divide instructions use 2903 local RAM and Q registers as follows:
Divisor register
Dividend Most
Significant
Half register
Divisor buffer
Dividend Most
Significant
Half buffer
Dividend Least
Significant
Half register
.,
Q register
Remainder
Quotient
I
The divisor and dividend require initial memory locations identified as registers in the illustration above. The divisor
and the most significant half of the dividend also require temporary buffers. The contents of these buffers are destroyed
in the process of comparing the divisor and dividend magnitudes.
8-79
Move the divisor and
most significant half
of the dividend to
temporarY buffers
YES
Perform twos complement sign/magnitude
convention on divisor
and most significant
half of dividend in
temporary buffers
Exit
NO
Downshift
dividend one
bit position
YES
Subtract most
significant half of
dividend from divisor
Move least significant
half of dividend to
Q register
YES
Downshift
dividend one
bit position
Start division
Figure 8-25. 2903 Binary Division Data Preparation Algorithm
8-80
Table 8-9. A Possible 2903 Twos Complement Binary Division Microprogram
Microinstruction
No
1
18 - 15
14-11
10
4
6
0
EA
CN
AO - A3
BO - B3
0
0
Divisor register
Divisor buffer
(RO)
(R3)
Comment
Copy divisor to temporary buffer.
2
4
6
0
0
0
Dividend (MS)
register (R 1)
Dividend (MS)
buffer (R2)
Copy dividend most significant half to temporary buffer.
3
5
0
0
X
.0
X
Dividend (MS)
buffer (R2)
Convert dividend (MS) from twos complement to sign/magnitude version.
Test OVR externally while next microinstruction is being executed. If OVR is 1,
branch to subroutine that downshifts dividend.
4
5
0
0
X
0
X
Divisor buffer
Convert divisor (MS) from twos complement to sign/magnitude version.
Test OVR externally. If OVR is 1, branch to microinstruction 9.
(R3)
5
9
4
0
X
0
X
Dividend (MS)
buffer (R2)
Shift out sign bit of dividend (MS) half in temporary buffer.
6
9
4
0
X
0
X
Divisor buffer
Shift out sign bit of divisor in temporary buffer.
(R3)
F
7
2
0
0
1
Dividend (MS)
buffer (R2)
Divisor buffer
(R3)
Subtract sign bit stripped divisor from sign bit stripped dividend (MS) half.
If Carry = 1 (dividend larger) branch to subroutine that downshifts dividend
or upshifts divisor.
8
6
6
0
0
0
Dividend (LS)
register (R4)
X
9
A·
0
0
0
0
Divisor register
Dividend (MS)
register (Rl)
Double length normalize dividend in MS register and Q register.
Dividend· (MS)
register (R 1)
Execute twos complement divide instruction fourteen times.
Dividend (MS)
register (R 1)
Twos complement divide final instruction.
(RO)
C·
10
0
0
0
Z
Divisor register
(RO)
11
E
0
0
0
Z
Divisor register
(RO)
Copy dividend least significant half to Q register.
·CN is connected to Z status while these two special functions are being executed.
Before the actual division begins, the least significant half of the dividend is moved to the 0 register. The quotient is
ultimately returned in the 0 register and the remainder in the Dividend Most Significant Half register. Taking a simple
case. if local RAM is used to implement Divisor and Dividend registers and buffers. then we can illustrate local RAM
and 0 registers utilization as follows:
Initial:
RO - Divisor
R1 - Dividend. most significant half
R2 - Copy of R1. dividend most significant half
R3 - Copy of divisor
R4 - Dividend. least significant half
o - Dividend. least significant half
Final:
R1 - Remainder
o - Ouotient
The 0 register. which initially holds the least significant half of the dividend. ultimately holds the quotient. As the dividend is upshifted out of the Q register and into the Dividend Most Significant Half register. quotient bits get shifted into
the 0 register via 00.
8-81
If you look again at Table 8-9, you will see that the actual division operation executes three functions:
1)
The Double Length Normalize function. which serves as the first divide function.
2)
The Twos Complement Divide function; this function is executed N-2 times. where N is the number of divisor and
quotient bits.
3)
A final Twos Complement Divide Correction function completes the division.
Necessary pin connections for the Double Length Normalize function are given in Figure 8-21. Figure 8-26
shows necessary pin connections for the Twos Complement Divide and Twos Complement Divide Correction
functions.
Zero status logic is used to transmit sign compare information from the most significant 2903 slice to intermediate and least significant slices. The level transmitted is the complement of the Exclusive-OR of the most significant
bits of the ALU output and R operand input. This may be illustrated as follows:
R3 R2 R1
53 52 51
RO
50
ALU
F3
F2
F1
FO
Z = R3 E& F3
This Z status logic works only when an A or C special function code is input via 18-15, and iEN is simultaneously
low. The Z status also becomes the CN input to the least significant slice in order to neutralize CN within the ALU
functions performed. This may be illustrated as follows:
[F]
so [F]
[F]
so [F]
+ [R] + CN if Z = 0
+ [R] if CN = Z
[S] - [R] - 1 + CN if Z =
[S]
[S]
1
[S] - [R] if CN =Z
The Q register and ALU register are connected so that an upshift causes the high-order bit of the Q register to be input
to the low-order ALU bit. The high-order ALU bit is lost. and the next quotient digit is shifted into the least significant
bit of the Q register. This may be illustrated as follows:
Lost bit
ALU register
Q register
The level actually output at SI03 is also R3
e F3. This becomes the next bit shifted into the quotient.
The final Twos Complement Divide Correction function forces a 1 into the low-order quotient bit. leaving the remainder
adjusted accordingly.
8-82
Q3, MSS
+5V
~)
:~
-- -
-
F3 + R3
••
-
,~
z
Z
Z
Z
....
0103
0100
QI03
QIOO
0103
0100
QI03
QIOO
""'""-
SI03
SIOO
SI03
SIOO
SI03
SIOO
SI03
SIOO
-
C(N+4)
-
C(N+4)
CN
OVR
-N
W/MSS
MSS
2903 _
LSS
WE
~
f-<>
+5V
j
C(N+4)
CN
CN+5V
IS
C)
2903
W/MSS ~
+5V
IS
2903
()
W/MSS
LSS
~~
LSS
~
WE
WE
j
.~
C(N+4)
LSS
2903
kJ::
CN
W/MSS
~
WE
LSSQ
ENABLE
ALU output = [S) + [R) + CN if Z = 0, or [S) - [R) + CN - 1 if Z = 1
[0) is upshifted on all microcycles
[F) is upshifted on twos complement divide, but not on twos
complement divide last microcycle.
ENABLE is high on last cycle only
MSS means Most Significant Slice
IS means Intermediate Slice
LSS means Least Significant Slice
[S) = S ALU input
[R) = R ALU input
[F) = ALU output
[Q) = a register contents
03, R3 and F3 are most significant bits of
a register, R ALU input and ALU output
Figure 8-26. 2903 Signed Binary Twos Complement Divide Pin Connections
Merely understanding the pin connections and functions shown in Figure 8-25 is quite straightforward. Understanding how binary division is performed using these pin connections, and the three binary division functions, is not selfevident. Let us therefore take a very simple example and analyze divide logic in conjunction with this example. Consider the following simple division:
18 16
~
= 3 remainder 3
We have a 4-bit divisor and an 8-bit dividend which generate a 4-bit quotient and a 4-bit remainder. We must therefore
execute the Double Length Normalize function, followed by two Twos Complement Divide functions, and a Twos Complement Divide Correction function.
8-83
For Step 1 we execute the Double Length Normalize function. This upshifts the dividend and generates the sign of the
quotient at SI03 of the most significant slice. The quotient sign bit gets shifted into the low-order 0 register bit. Logic
may be illustrated as follows:
Step 1
[R]
[F]
a 1 1 1 aaa 1
Initial
Final
[a]
[5]
5103
z
1aaa
11II1I1
a a a 1 a 1 1 1 a a 1 1 a a a 0-0
SI03 is the Exclusive-OR of the most significant slice ALU output and R operand input bits. This may be illustrated as
follows:
Step 1
[F]
[R]
[5]
[a]
a 1 1 1 aaa 1
Initial
5103
z
1aaa
This SI03 logic says that when the divisor and the dividend have the same sign, the quotient is positive; the quotient is
negative when the divisor and the dividend have opposite signs. What is not self-evident is the fact that we have
multiplied the dividend by two before starting to work with the divisor. In consequence, we must finally upshift the
quotient and the remainder to generate answers that stand numerical comparison.
Moving on to Step 2, we execute the Twos Complement Divide function for the first time. The Zero status is 1:
Step 1
[F]
[R]
Initial
[a]
[5]
5103
z
011100011000
Therefore, we subtract the divisor from the high-order four dividend bits. This may be illustrated as follows:
Step 2
[F]
7
[R]
In;t;a~/
(
0011
100 1
Final
1 1aa
a111
[5]
00
I
[a]
,/o;OjOlO
1aaa
--'----7
aaaa-
5103
z
o
0
a
This step is very logical. It is equivalent to initially subtracting the divisor from the dividend in any decimal division:
25)237642
25
-2
8-84
In our binary example the divisor is larger than the dividend. even though the dividend has been upshifted; therefore
the next quotient bit shifted into the Q register is O. 0 is indeed output by the most significant slice at 5103:
Step 1
[Fi
[R]
IS]
[a]
SI03
Initial
0001
01 1 1
0011
0000
0
Final
!;~r:::~~o
Z
0
ED
In Step 3 the Twos Complement Divide function is executed a second time. The Zero status is now 0:
Step 2
Initial
[F)
[R]
000 1 0 111
IS]
[a]
SI03
00 11
0000
0
z
Final
Therefore. during Step 3 we add the divisor to the high-order four bits of the shifted dividend. This addition. and the
subsequent upshift. may be illustrated as follows:
Step 3
[R]
IS]
[a]
SI03
Z
0
0
111101111110 0000-0
0
[F)
,nl"ac::;?100/ 11
Final
~"--7
1
Adding the divisor to the upshifted dividend is also self-evident. We got a negative answer during Step 2. therefore (as
described earlier) we must now compensate by adding the divisor to the upshifted dividend. The dividend is still
smaller than the divisor. so once again 5103 outputs 0 at the most significant 2903 slice:
Step 3
Initial
[F]
[R]
[S]
[a]
SI03
z
1 10 0
0 111
1000
0000
o
o
Final
and zero gets shifted into the Q register to become the next quotient bit.
8-85
Finally. in Step 4 we execute a Twos Complement Divide Correction function. Once again we test the Z status. which is
O. therefore we add the divisor to the high-order four bits of the upshifted dividend. Together with the final shift this
may be illustrated as follows:
Step 4
[R]
[F]
c;
"'"8'''''7'''
Final
010'1
0 111
[0]
5103
z
0000
0
o
[5]
0
III
0 1 0 1 0001
-v-~
t
Forced input
During the final shift a 1 is forced into the quotient to become the quotient low-order bit. The four high-order dividend
.
bits do not change. Thus our final answer is:
Quotient = 0001
Remainder = 0101
In order to test the numeric accuracy of our answer we must upshift one bit position:
Quotient = 0010
Remainder = 1010
Thus. the answer is 2 with a remainder of A 16 -
which is not 3 with a remainder of 3. but it is correct.
You r external logic (2904) must upshift thequotient and the remainder. if your algorithm demands it. and must adjust
the quotient and the remainder if your algorithm requires the remainder to be less than the divisor.
8-86
THE 2902 CARRY LOOK-AHEAD DEVICE
This device serves just one function: when performing binary addition or subtraction using cascaded 2901 or
2903 systems, it creates parallel carry inputs for 4-bit slices beyond the least significant slice. Carry Look·
Ahead logic has been described in detail in Volume 1, Chapter 4. We will therefore provide a simple summary of
the 2902 device in this chapter, stating its logic functions, but omitting Carry Look-Ahead theory.
The 2902 is packaged as a 16-pin DIP. All signals are TTL-level compatible and a single +5V power supply is required.
The 2902A is a faster version of the 2902.
2902 PINS AND SIGNALS
Figure. 8-27 illustrates pins and signal assignments for the 2902 Carry Look-Ahead device. Figure 22-28 shows
a 2902 device connected to four parallel 2901 devices. If you replace the 2901 devices with 2903 devices, connections between the 2902 and the 2901 or 2903 devices do not change.
0
y
OR
Sync/Enable
Logic
".-
v
(}
External
Controls
A-
v
4
4_
A
~
..
~
Microprogram
PROM/ROM
-
I I
U
Microinstruction
Register
II
~)
"11,/
'-,
Microprogram
Addresses
V'
To 2901 or 2903
Slices
"II",,"
~
Logic
Sequence
and Enable
Controls
0-
Status
Output
from
2901/2903
Figure 8-31. Function of Microprogram Sequencer Logic in a 2901 or 2903 Based System
8-93
Macroinstruction object codes have been described in considerable detail in Volume 1. Chapter 7. This discussion
emphasizes the fact that macroinstruction object codes are selected to optimize Central Processing Unit operations.
without regard to microprograms. or how microprograms may be stored in a memory device. This being the case. there
is no chance that the op-code portion of any macroinstruction will have a bit pattern that addresses the correct
microinstruction. or initial microinstruction that must be executed in response to the macroinstruction's execution. Instead. a mapping read-only memory or a programmable logic array is used as an address translator. The mapping ROM
or PLA treats the op-code portion of the macroinstruction as an input. In the ROM. the actual microprogram starting address is stored at the location addressed by the op-code bit pattern. Conceptually. this may be illustrated as follows:
Assembly Language
Object Code
I
Microprogram ROM
Microinstructions
to be executed
Op Code
I
Data
This actual bit pattern
is treated as a memory
address in mapping ROM
The contents of the
addressed memory
word is the address
of the first microinstruction
in microprogram ROM
The size of the mapping ROM and the width of the address which it outputs depend on the size of the microprogramthat is to say. the length of the microprogram in terms of the number of microinstructions. This may be illustrated as
follows:
Microinstruction
bit width has no
effect on mapping
ROM size
r~----~~~~--~--"
P---------------~~~
.~
Number of
microinstructions
determines Size
of mapping ROM
••
--------_.. ---If. for example. 256 or fewer microinstructions constitute the entire microprogram. then an 8-bit address can be output
by the mapping ROM. irrespective of whether the microinstructions are 16 bits wide. 64 bits wide or have any other bit
width. But. if the total length of the microprogram were 1024 microinstructions. then a 1a-bit address wou Id have to be
output by the mapping ROM.
8-94
If every macroinstruction resulted in the execution of a single microinstruction. then there would be no need for
290912911 Microprogram Sequencers. The mapping ROM could output a single address to the microprogram ROM.
The contents of the addressed microprogram ROM would be output to the Microinstruction register - to become the
microinstruction that enables operations requ ired by the macroinstruction.
But it is most unlikely that the Central Processing Unit's assembly language will consist of instructions that are all primitive enough to be implemented via a single microinstruction. In particular. as Central Processing Units become more
complex. an ever larger number of microinstructions may have to be executed in response to a single macroinstruction
execution; and the sequence in which these microinstructions are stored may also become more complicated. The
2909 and 2911 Microprogram Sequencers provide the logic which takes you from the initial microinstruction through
the microprogram.
There is one very important conceptual aspect of the 2909 and 2911 Microprogram Sequencers which must be clearly
understood. These Microprogram Sequencer devices are. like the 2901 and 2903. cascadable 4-bit devices. But there is
no relationship between the number of 2909/2911 devices which are cascaded. 2901 or 2903 devices are cascaded to
give you the required CPU word width. Two 2901/2903 devices generate an 8-bit word; four 2901/2903 devices
generate a 16-bit word; eight 290112903 devices generate a 32-bit word. etc. 2909 or 2911 devices are cascaded to
address the required length of microprogram memory. The number of 2909 or 2911 devices cascaded together is in no
way influenced by the width of the microinstruction. or the width of the CPU word. For example. if the microprogram
has 256 or fewer microinstructions. two 2909 or 2911 devices cascaded together are sufficient. This holds true
whatever the microinstruction width may be. and whatever the CPU word width may be. Thus the number of parallel
2901 or 2903 slices has no bearing whatsoever on the number of parallel 2909 or 2911 devices. You cannot even
generalize by stating that there will be a tendency to require more 2909/2911 devices as the number of parallel
290112903 devices increases. Rather. the width of the microinstruction will increase with the number of 290112903
devices and. as we have already stated. the width of a microinstruction has no bearing on the length of the
microprogram. or the number of parallel Microprogram Sequencer devices that will be needed.
2909/2911 MICROPROGRAM SEQUENCER PINS AND SIGNALS
Pins and signal assignments for the 2909 and 2911 Microprogram Sequencers are illustrated in Figure 8-32.
These signals are most easily understood in conjunction with the functional logic illustrations for the two
devices which are given in Figures 8-33 and 8-34.
Central to the logic of 2909aild 2911 Microprogram Sequencers is the Output Multiplexer,
which receives four inputs. SO and S 1 are two control inputs that select an output as
follows:
SO
S1
o
0
1
1
1
0
1
o
2909/2911
OUTPUT
SELECT
Output Multiplexer Source
Microprogram Counter
Address register
Stack
Direct inputs (via 00-03)
We will for the moment ignore the Microprogram Counter and Stack. two data storage areas whose functions will be
described shortly.
00-03 are four data input lines. Data input via these four lines can be selectedVby the Output
Multiplexer and output immediately (if SO and S1 are both high).
2909/2911
IMMEDIATE
DATA INPUT
Data input via RO-R3 is held in the Address register. Timing for Address register access may be
illustrated as follows:
CP
2909/2911
ADDRESS
REGISTER
-----~
RO - R3
Address
ADDR
Register _ _ _ _ _ _ _""'_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
8-95
-
RE
R3
R2
R1
RO
OR3
03
OR2
02
OR1
01
ORO
DO
GND
-
-----
----
----..--..-
--
1
2
3
4
5
6
7
8
9
10
11
12
13
14
2909
Pin Name
RO - R3
DO - 03
RIDO - R/D3
YO - Y3
ORO - OR3
ZERO
RE
OE
FE
PUP
CN
C(N+4)
SO,S1
CP
VCC,GND
28
27
26
25
24
23
22
21
20
19
18
17
16
15
..
--:
-
- -
--
-----
...
-.-
-
CP
VCC (+5V).
CP
(+5V) VCC
PUP
RE
FE
D/R3
D/R2
D/R1
DIRO
C(N+4)
CN
OE
Y3
Y2
Y1
YO
51
SO
ZERO
GND
ZERO
SO
Description
Address register input
Data input
Combined Address register and data input
Address output
Address output mask
Zero Address output control
Address register input enable
Address output enable
Stack enable
Stack pushlpop select
Program Counter Carry in
Program Counter Carry out
Output select
System clock
Power, Ground
1
2
3
4
5
6
7
20
19
18
17
2911
8
9
10
16
15
14
13
12
11
Type
Input
Input
Input
Output, Tristate
Input
Input
Input
Input
Input
Input
Input
Output
Input
Input
Figure 8-32. 2909 and 2911 Microprogram Sequencer Pins and Signal Assignments
8-96
PUP
FE
C(N+4)
CN
OE
Y3
Y2
Y1
YO
51
RE----------------~
RO-R3~
________
~~1
Address
Register
Microprogram
Counter
00-03 __________,
50
----~
Output Multiplexer
51~
OR3----------+------+------4---~
OR2----------+------+----~
----------+---~
ORO - - - - - - -.....
OR1
Incrementer
ZERO ----------4H----...+-----~~--_
YO
Y1
Y2
Y3
C(N+4)
Figure 8-33. 2909 Microprogram Sequencer Functional Logic
8-97
RE ------------------~
.....I----FE
Stack
Po.inter
.....1----PUp
4x
4-Bit
Stack
Microprogram
Counter
RIDO - R/D3 '--________.....
SO
---1"
Output Multiplexer
S1
ZERO ----------~----~~----~~--~
....I---CN
YO
Y1
Y2
Y3
C(N+4)
Figure 8-34. 2911 Microprogram Sequencer Functional Logic
8-98
As illustrated above. the Address register enable signal RE must be low before data can be written into the Address register via RO-R3.lf RE is low. then data is written into the Address register when the clock signal CP makes its
high-to-Iow transition. But RE has no effect on Address register output. Whether RE is low or high. the Address register
contents are transmitted to the Output Multiplexer.
The 2911 shares D and R inputs. Data input at RIDO-R/03 will be written into the Address register if RE is low. and
will be output via the Output Multiplexer if SO and 51 are both high.
2909/2911
DATA
OUTPUT
Multiplexer output lines VO-V3 have their own enable signal OE. If this signal is high, VO-V3
outputs are floated. This allows you to disconnect Microprogram Sequencer devices from the
microprogram ROM. something you may do when switching to an external tester.
2909/2911
The 2911 has one set of conditioning logic on the Y outputs. The 2909 has two sets of conditionOUTPUT
ing logic on the Y outputs. Both devices have a ZERO input which, when low, unilaterally
ZERO
forces the four lines VO, V1, V2, and V3 to output O. Frequently you will use the ZERO line as a
CONTROL
restart -with an initialization microinstruction sequence origin at microinst~uction number in
the microprogram ROM. The 2909, but not the 2911, has four mask signals, ORO, OR1, OR2,
2909
and OR3, which can individually force VO, V1, V2, and V3, respectively, low. Typically you
OUTPUT
will use the mask signals to implement conditional logic. For example. we have already seen how
MASK
the Overflow status (OVR), output by the most significant 2901 and 2903 slice. signals an overflow or "exceptional" condition. By tying the Overflow status to ORO. you can implement
microinstruction pairs. The Output Multiplexer outputs a 0 low-order microinstruction address via YO. which ORO can
override and convert to 1. For an 8-bit microprogram address this may be illustrated as follows:
a
1-*-,...----
, . . . - - - - - - - Output by high-order 2909
Output by low-order 2909
XXXXYYYO
~
______
+L.___ ORO to low-order 2909 can change
~I
X X X X Y Y YO . .
M;c",;nstruct;on exec..ed ;1 OVR
this address bit to 1
~0
• X X X X Y Y Y 1 . . . . - - Microinstruction executed if OVR = 1
Of course. having a mask line associated with every microinstruction address output line lets you generate more complex conditional logic schemes than the simple illustration above.
There are two internal locations within the 2909 and the 2911 which can hold addresses. These are the
Microprogram Counter and the Stack.
2909/2911
Let us first look at the Microprogram Counter. This location is equivalent to the typical Central MICROPROGRAM
Processing Unit Program Counter.
COUNTER
When SO and S1 are both low. Microprogram Counter contents are read by the Output Multiplexer. and are output via
YO-Y3.
New data is written into the Microprogram Counter whenever data is input from the Output Multiplexer. whether or not
the Microprogram Counter was selected as the Output Multiplexer input. Data written back to the Microprogram
Counter is taken from the YO-Y3 path following the OR and AND gates. Therefore. if you use either of the output conditioning gates. you will also modify the Microprogram Counter contents. This. of course. is no different to a Central ProceSSing Unit's Program Counter. which is also modified by a restart or jump instruction.
Data being written back to the Microprogram Counter passes through an Incrementer. The Incrementer adds the CN level to data which is on its way to the Microprogram Counter. Thus if
CN is low. the Incrementer passes data through unmodified; but the data is incremented if CN is
high. An Incrementer overflow generates a high C(N+4) output.
2909/2911
INCREMENTER
Let us look at the various ways in which you may use Microprogram Counter logic.
In the simplest case, you may wish to sequentially access a number of microinstructions.
You can begin the sequence by inputting the first microinstruction address to the Address register
via RO-R3. or as immediate data via 00-03. Remember. 2909 and 2911 devices are cascadable;
therefore we are not limited to 4-bit addresses. The initial address. when output by the Output
Multiplexer. also gets written to the Microprogram Counter. Assuming that CN is high. the address
8-99
2909/2911··
SEQUENTIAL
ADDRESSES
written into the
Microprogram Counter will be one more than the starting address input via RO-R3 or 00-03. For a number of subsequent microcycles. you will continue to select the Microprogram Counter. leaving CN high. so that Microprogram
Counter contents are incremented on each microcycle. Selecting addresses arbitrarily. this sequence may be illustrated
as follows:
00-03
SO .§.l
1
1
0
0
0
0
0
0
etc.
30
XX
XX
XX
CN
YO-Y3
30
31
32
33
Microprogram
Counter Contents
31
32
33
34
etc.
XX represents "don't care" inputs.
There are some non-obvious problems that can occur when you generate sequential microinstruction addresses using
Program Counter logic as illustrated above.
The next very simple microprogram counter sequence involves the re-execution of a single
microinstruction - as you may do while performing a normalize or twos complement divide
operation using the 2903 special functions. If the Output Multiplexer selects the Microprogram
Counter contents while CN is input low. then the Microprogram Counter contents will not change
on succeeding microcycles - and the same microinstruction will be executed repeatedly.
SINGLE
INSTRUCTION
RE-EXECUTION
You can skip a microinstruction with an even address within an otherwise consecutive instruction sequence. To do this you keep CN high. so that the Microprogram Counter increments
on each microcycle. but you input a high ORO mask bit in order to skip an instruction. This may be
illustrated as follows:
INSTRUCTION
SKIP
SO
o
Sl
CN
-o - - 0
o o
o
o
o
ORO
2909/2911
Microprogram
Counter Contents
34
35
37
38
39
etc.
0
1
0
0
o
o
o
2909/2911
etc.
ORO forces low-order
address bit to 1.
You can also use mask bits to jump between microprogram pages. For example. within a 256microinstruction program you can jump in sixteen microinstruction increments by inputting a
high signal at the ORO mask bit of the high-order 2909 slice. This may be illustrated as follows:
Both 2909's
Most Significant 2909
Address
34
35
46
47
Least Significant 2909
SO
S1
ORO
Microprogram
Counter
CN
ORO
0
0
0
0
0
0
0
1
0
0010
0010
0011
0011
1
1
1
1
a
a
0
0
0
a
0
Microprogram
Counter
0100
0101
01 10
01 1 1
etc.
By applying high inputs to other mask bits you can span almost any number of microinstructions in a Single jump.
The 2909/2911 Stack has four locations.
A Stack Pointer identifies the currently selected Stack location. If the Output Multiplexer
receives SO low and S1 high. then it reads the contents of the currently selected Stack location
and outputs this data via YO-Y3. This output address. like all other output addresses. passes through the Incrementer
and is loaded into the Program Counter.
8-100
You modify the Stack Pointer address using the FE and PUP signals. FE must be low in order to modify the Stack
Pointer. FE does not have to be low in order to output data from the Stack to the Output Multiplexer.
When FE and PUP are both low. the Stack Pointer decrements on the low-to-high transition of CPo Timing may be illustrated as follows:
CP
PUP
Decrement
Sfack
Pointer
If FE is low and PUP is high. the Stack Pointer is incremented. then the contents of the Microprogram Counter are
loaded into the newly addressed Stack register. Timing may be illustrated as follows:
CP
PUP
contents to
Stack
Increment
Stack
Pointer
The address in the Program Counter which gets written into the newly addressed Stack register will be the address
which was output by the Output Multiplexer - incremented by one. assuming that CN is high. This enables traditional
subroutine call and return logic. as we will see soon.
The Stack Pointer is a roll-over counter. That is to say. it will decrement from 0 to 3:
and it will increment from 3 to 0:
This is normally not advantageous.
Let us look at some of the address sequences which can be generated using the Stack.
8-101
Consider first a typical subroutine caU- in the classical assembly language sense. The
2909/2911
microinstruction which causes the subroutine call increments the Stack Pointer. Assuming that
SUBROUTINE
CN is high, the address of the microinstruction which causes the subroutine call is incremented by
CALL
one, written into the Microprogram Counter, and thence to the newly addressed Stack register, to
the address of the microinstruction to which execution will return at the end of the subroutine. This sequence may be
illustrated as follows:
,,
,
,
,,
,
I
,,
,
, ,II'
With reference to the illustration above, the microinstruction which causes the subroutine call is arbitrarily assumed to
reside in microprogram read-only memory location A216. Therefore the subroutine call begins with the Microprogram
Sequencer outputting address A216, as illustrated above by
CD
For the microinstruction in location A216, only those bits that affect the Microprogram Sequencer are of interest to us.
These bits cause the Microprogram Sequencer to receive high inputs at CN and PUP, with a low input at FE. This
microinstruction will also provide the address for the next microinstruction, arbitrarily assumed to be 2E16 at @ in
the illustration above. But this address will be output on the next microcycle. On the current microcycle, CN causes the
The new incremented address A316
currenfaddress (A216) to be incremented to A3...l..6. This is shown above at
is written to the Microprogram Cou nter; see C§) above. Since FE is low and PUP is high, the Stack Pointer is increabove) and the current Microprogram Counter contents, which is A316, is saved in the Stack.
mented (
® .
®
On the next microcycle the address 2E16 will be output. initiating the subroutine's execution. When the subroutine
completes execution, it has the return address A316 stored at the Stack register identified by the Stack Pointer.
8-102
The final instruction of the subroutine must execute a Stack PUP and cause the Output Multiplexer to select the Stack
as its input. This requires S1 to be high while SO. FE. and PUP are all low. The Output Multiplexer will read A316 from
the Stack and output this address next. A316 is incremented to A416 and returned as the new Microprogram Counter
contents. The Stack Pointer decrements. Thus a classical Stack-Oriented Return-from-Subroutine has been executed.
A subroutine can consist of a single microinstruction. If you look again at the subroutine call illustration given earlier.
the first instruction of the subroutine. which in our illustration will be the instruction stored at location 2E16. has
nothing said about its Microprogram Sequencer bits. If this instruction causes a Return-from-Subroutine. then you have
created a single-microinstruction subroutine.
Using the Stack you can nest subroutines to a depth of four. In most microprograms. nesting
to a depth of fou r is perfectly adequate.
A computed multidestination jump is easily implemented using a 2909 or 2911 Microprogram
Sequencer. A 16-way jump to individual instructions can be achieved by inputting data via OROOR4 to the least significant 290912911. while generating more significant portions of the address
from some other location. such as the Address register.
2909/2911
SUBROUTINE
NESTING
2909/2911
MULTIPLE
JUMP
2909/2911 SYNC/ENABLE LOGIC
There are innumerable ways in which the Sync/Enable logic portion of Figure 8-31 could be designed. At its
most elementary level, 2909/2911 control signals will be generated (possibly from a read-only memory) based
on Microinstruction register outputs which become a read-only memory address. This may be illustrated as
follows:
Enable/Select
FE
PUP
RE
S1
SO
ZERO
eN
Microinstruction
Register
Il
ORO
OR1
OR2
OR3
ROM Address
----
--
--
--
--
I
Sync/Enable
ROM
5J
A 12-bit wide read-only memory would be required in the illustration above to generate eleven input signals needed by
the 2909/2911. plus an Enable/Select signal for the R input 2-IN MUX. (The clock signal CP could not be generated by
a ROM.) It would be possible to generate 4096 different combinations of 12 signals. Very few of these possibilities will
ever be encountered. In all probability. a maximum of 32 different signal combinations may be seen. in which case a 32
x 12-bit read-only memory will suffice. with a 5-bit address provided by the Microinstruction register. Each
microinstruction stored in the microprogram ROM will then contain a 5-bit address field; the address field selects one
of the 32 signal combinations that define the next step of Microprogram Sequencer control inputs. Thus we are able to
achieve address continuity within a microprogram. In many applications. the 29811A provides this function.
But in Figure 8-31 three sets of inputs to the Sync/Enable logic section are shown.
The "logic sequence and enable" control emanating from the Microinstruction register becomes the 5-bit address
which we have already described.
8-103
Table 8-11. The 2903 Twos Complement Binary Division Microprogram Subroutine from Table 8-9,
with 2911 Addressing Microinstruction Fields Added
Microinstruction
No
1
18-1&
14-11
10
4
6
0
EA
CN
AO-A3
80 - 83
0
0
Divisor register
Divisor buffer
(RO)
(R3)
Comment
Copy divisor to temporary buffer.
2
4
6
0
0
0
Dividend (MS)
register (Rl)
Dividend (MS)
buffer (R2)
Copy dividend most significant half to temporary buffer.
3
5
0
0
X
0
X
Dividend (MS)
buffer (R2)
Convert dividend (MS) from twos complement to sign/magnitude version.
Test OVR externally while next microinstruction is being executed. If OVR is 1,
branch to subroutine that downshifts dividend.
4
5
0
0
X
0
X
Divisor buffer
Convert divisor (MS) from twos complement to sign/magnitude version.
Test OVR externally. If OVR is 1. branch to microinstruction 9.
(R3)
&
9
4
0
X
0
X
Dividend (MS)
buffer (R2)
Shift out sign bit of dividend (MS) half in temporary buffer.
6
9
4
0
X
0
X
Divisor buffer
Shift out sign bit of divisor in temporary buffer.
(R3)
7
F
2
0
0
1
Dividend (MS)
buffer (R2)
Divisor buffer
(R3)
Subtract sign bit stripped divisor from sign bit stripped dividend (MS) half.
If Carry = 1 (dividend larger) branch to subroutine that downshifts dividend
or upshifts divisor.
S
6
6
0
0
0
Dividend (LS)
register (R4)
X
9
A"
0
0
0
0
Divisor register
Dividend (MS)
register (Rl)
Double length normalize dividend in MS register and Q register.
Dividend (MS)
register (Rl)
Execute twos complement divide instruction fourteen times.
Dividend (MS)
register (Rl)
Twos complement divide final instruction.
(RO)
10
C"
0
0
0
Z
Divisor register
(RO)
11
E
0
0
0
Z
Divisor register
(RO)
Copy dividend least significant half to Q register.
'CN is connected to Z status while these two special functions are being executed.
8-104
Table 8-11. The 2903 Twos Complement Binary Division Microprogram Subroutine from Table 8-9,
with 2911 Addressing Microinstruction Fields Added (Continued)
Microinstruction bits covering 2903 and two 291 1 input
(Bit positions are arbitrary and have no signlficancel
:s
11
i
1
2
3
2903 Bits
18 17 16 16 14 13 12 11 10
o
o
0
0
0
0
0
o
o
1
0 0 1 1 0
0 0 1 1 0
0 0 0
0
o
o
----...-.5
1 0 1
0 0 0
0
6
4
1 o 0 1 o 1 o 0
-....-'-...-.-
9
'7
0
X X X X 0 0 1 0
-....-'
A2
X X X X X X X X
4
1 1 1 1
o
0 1 0
X
0
X X X X 0 0 1
0
0
X
0
0
0
8
9
2
o
o
-...---...-.6
6
1 1
0 1 1 0
X
0
o
0 0 1
0
0
1
0
0
1
0
0
0
0
0
0
10
0
0
0
1 1 000000 0
0
X
0
C
'1
1 1 1
o
0 0 0 0
0
1
X X X X X X X X
0 1
o
0
0 X X X X
o0 0 o0 0 0 1
-....-'-...--
X X X X X X X X
T T T T T T T T
~
Scale divisor or
Dividend Subroutine
X X X X X X X X
Rl
0
0
0
0
X
1 0
Z 1 1 1 X
1 1 1
---
1 0 0
1 1 1 X
1
1 1 1
0
1 0 0
1 1 1 X
1 1 1 1 1 0
1
-----
Select next sequential instruction. Load
abort address into Address register.
Select next sequential instruction, or abort
address if Z = 1 from microinstruction 1.
Select next sequential instruction.
OVA = O. Select next sequential instruction.
OVA = 1. Call "scale dividend" subroutine.
OVA = O. Select next sequential address.
OVR = 1. Branch to microinstraction 8.
1 1 1 X
Select next sequential instruction.
1 1 1 X
Select next sequential instruction.
1 0 0
1 1 1 X
1 1 1
1 1 0
C(N+41 = O. Select next sequential instruction.
C(N+41 = 1. Call "scale divisor or dividend"
subroutine.
1 0 0
1 0 0
-------
1 0 0
1
1 1 1 X
Select next sequential instruction.
1 0/11
X
Execute this instruction 14 times, then select
next sequential instruction.
1 1 0
0
Return from subroutine.
0
1
-...-----AO
1 1 1
0
X X X X X X X X 1 0 0
Rl
0 000000
1 1 1 X
0
0
0 000000 1
RO
0 0 0
0
R4
RO
~~
0
R3
A2
---
CNFE
B B B B B B B B 1 0 0 1 1 1 X
~
Microinstruction
1 1 1 1 1 1 X
8 Address
--.,-~
0
----....-'
E
0 0 1
~
0
----....-'
o
S S S S S S S S
~
--.,--....-'
0 0 00 0
0
A2
X X X X 0 0 1 1
-....-'
A3
0
-....-'-....-'
A
--.,-
0
0
1 0 1
X X X X 0 0 1 0
iI:l
1
Scale Dividend
Subroutine
R3
0
----....-'
F
1
-....-'
0 1 0 1 00 0
1
X
0
o
-...-.---.,9
6
0
0
A A A A A A AA
~
Abort Address
R3
0
0
0
1 0 1
RO
X X X X X X X X
0
0
o
o
--.,--...-.-
1
000000 1
--.,---.,-
2911 Addressing Operations
D7D6D6D4D3D2D1 DO RES1 SO
0 o 0 1 o 0 1 0
-....-'-....-'
Al
R2
0
--.,--....-'
6
0
4
5
4
EA CNA3A2A1AOB3B2B1 BO
-....-'--.,0
6
4
1
2911 Bits
X X X X X X X X
Rl
1 1 0
2
X = "Don't Care" bits
8-105
Status outputs from the 2901 or 2903 devices have already been encountered; we have. for example. discussed the
way in which the Overflow status may generate one of the OR mask lines. The most effective way of handling status
outputs from 2901 or 2903 slices is to use the test input of 29811 A. This might be illustrated as follows:
.....
.--
.-..
-.....
t +t ,
13
MUX
-..
12
11
10
29811A
TEST
...
...
-......
.-..
COUNTER LOAD
COUNTER ENABLE
MAP ENABLE
PIPELINE ENABLE
-....
--.. FE
PUP
51
SO
+
OUTPUT ENABLE
2909/2911 ADDRESSING EXAMPLE
We will now complete our discussion of 2909 and 2911 operations by looking at a specific example - the twos
complement binary division microprogram summarized in Table 8-9. Table 8-9 shows only those
microinstruction bits required by the 2903 device. In Table 8-11 we add microinstruction bits for two 2911
Microprogram Sequencers. We assume that the two 2911 Microprogram Sequencers address the 2903 twos complement binary division microprogram as a subroutine. within a 256-microinstruction microprogram memory.
In the lower half of Table 8-11. RAM locations used by the twos complement binary division microprogram are identified using register designations "RO" through "R4". These designations represent RAM locations with addresses 0
through 4. respectively.
Additional microinstruction bits added in the lower half of Table 8-11 provide the two 2911 Microprogram Sequencers with their data and control inputs. Eight data input bits are needed. four for each of the two 2911 Microprogram
Sequencers. But since both 2911 Microprogram Sequencers will receive exactly the same control inputs. one set of
control Signals is sufficient. Table 8-11 shows the six control signals RE. S1. SO. ZERO. CN. FE. and PUP being generated by individual microinstruction bits. But earlierwe generated these control codes out of a read-only memory. In the
discussion which follows we will compare these two methods of generating control inputs.
Microinstruction 1 in Table 8-11 copies the divisor into a temporary register. If the divisor is not O. the next sequential microinstruction must be executed. If the divisor is O. microprogram execution must be aborted. since you
cannot divide by O. The most significant 2903 slice outputs Z high if the divisor is O. Z is output tow if the divisor is not
O. The Z status is output at the end of the microinstruction 1 microcycle. too late to influence 2911 Microprogram Sequencer logic. Microinstruction 1. therefore. provides the 2911 with data needed to evaluate the Z status while
microinstruction 2 is executing. SO and S 1 are both low while CN is high; therefore.while microinstruction 1 is executing. the 2911 Output Multiplexer selects the Microprogram Counter as its source. then increments the Microprogram
Counter contents. Thus. the next sequential microinstruction. microinstruction 2. is selected. Simultaneously.
microinstruction 1 inputs the abort address to the 2911 Address register. The abort address is input at DO-D7 while RE
is low.
If logic associated with execution of microinstruction 2 conflicted in any way with testing the Z status. then we would
have to insert a dummy microinstruction in front of microinstruction 2. whose sole function would be to test the Z
status.
But microinstruction 2 has no such conflicting logic. so we can use its execution time to test the ZERO status generated
by microinstruction 1. If this ZERO status is 1. then the abort address. input to the Address register by microinstruction
1. will be selected; execution of microinstruction 2 becomes redundant insofar as 2903 logic events are concerned.
but. providing execution of microinstruction 2 is inconsequential. no harm is done.
The 2911 bits provided by microinstruction 2 can be quite simple. S1 is 0 and SO is connected to the ZERO status; if
the ZERO status is O. the Output Multiplexer will select the Microprogram Counter - and thus the next sequential instruction. But if the ZERO status is 1. the Address register will be selected by the Output Multiplexer - and thus an
abort will occur. In order to connect SO to a ZERO status we would probably include an additional enable bit. not shown
in microinstruction 2. This enable bit. when high. will link SO to the ZERO status. but when low will cause SO to be
8-106
I:F_-so
derived directly from its microinstruction bit. This may be illustrated as follows:
SO Bit _ _
ENABLE
-
ZERO---Addressing logic for microinstructions 3 and 4 is similar to that which we have just described for microinstructions 1 and 2.
Microinstruction 3 converts the most significant half of the dividend from twos complement to sign/magnitude form. If
the dividend has the largest possible twos complement value. then the Overflow status is setand the dividend must be
downshifted. If the dividend has any other twos complement value. the Overflow status is not set and the next sequential instruction must be executed. But as we found when examining microinstruction 1. the Overflow status is generated at the end of the microinstruction 3 microcycle. therefore the 2911 bits shown in Table 8-11 for microinstruction
3 simply select the next sequential microinstruction - microinstruction 4. But while microinstruction 4 is being executed. 2911 logic tests the Overflow status generated by microinstruction 3. Now we can delay testing microinstruction 3 Overflow status until microinstruction 4 has been executed because execution of microinstruction 4 does not
conflict with Overflow = 1. If it did. we would have to insert a dummy microinstruction between 3 and 4 which did
nothing at the 2903. but gave the 2911 an additional microcycle time within which to test the Overflow status generated by microinstruction 3. and determine subsequent addressing based on Overflow status level. Since this additional
dummy instruction is not needed. microinstruction 4 provides the address for the "Scale Oividend" subroutine via data
bits 00-07. Two sets of control inputs are generated. If the Overflow status left over from microinstruction 3 is 0 then
the control inputs to the 2911 simply select the next sequential microinstruction - microinstruction 5. But if the Overflow status left over from microinstruction 3 is 1 then control inputs cause the 2911 to push Microprogram Counter
contents onto the Stack. then have the Output Multiplexer choose as its source the address input at 00-07. Thus the
scale dividend subroutine is called.
If you look at the 2911 control bits of microinstruction 4 in Table 8-11. you will see that three bits. SO. S1. and FE.
must change. depending on the level of the Overflow status left over from the execution of microinStruction 3. PUP
need not change. since its level is not significant when OVR is low. We could control all three bit levels using an enable
bit. as described for the ZERO status in microinstruction 2. But it is probably simpler and cheaper to use a read-only
memory device. as suggested earlier in this chapter. Consider the following possibility:
Control
Bits
~
D7
D6' D5 D4
D3 D2 D1
DO A4 A3 A2 A 1 AO
Microinstruction bits for 2911
ZERO-----------------+-+--+-+--+---------}
OVR-------------------------r--r--r-~~~------------
C(N+4)---------------------------~~~~~~~--------------
;i~ess
Instead of having seven separate 2911 control bits in every microinstruction. we now have five control bits. Three
status bits (ZERO. OVR. and C(N+41. generated by the 2903) provide the three low-order bits for an 8-bit read-only
memory address. The five microinstruction control bits. together with the three status bits generated by the 2903,
create an 8-bit read-only memory address. The addressed read-only memory location outputs the seven signal levels
required by the 2911. For microinstructions 1 through 4, read-only memory might arbitrarily be mapped as shown in
Figure 8-35.
Microinstruction 5 has 2911 addressing bits that are similar to microinstruction 4. Microinstruction 4 not only
tests the Overflow status left over from microinstruction 3. but itself generates an Overflow status which must be
tested while microinstruction 5 is being executed. In the event that the 2903 Overflow status generated by
microinstruction 4 is low. the next sequential microinstruction. microinstruction 6. is selected. But if the Overflow
status is high. then a branch to microinstruction 8 occurs. Since this is a simple branch. no push is required; therefore
FE remains high in both cases. See Figure 8-35.
8-107
Address Bits from
Microinstruction
Status from
2903
Sync/Enable
ROM Map
~
A4 A3 A2 A 1 AO
1
2
2
3
4
4
5
5
6
7
8
8
9
10
10
11
~ 0~
(j
IffiN
X
I-----------,~(J
o
0
0
000
oo
01
X
X
X
X
X
0 L-_ _ _ _ _ _ _---.
000
o 0 1
000
000
o
1
X
X
1
o
0
0
000
001
o
o
o
o
o
o
o
0
0
0
0
0
0
0
o
0
X
X
X ~---10XOX
1
0
X
1
X
11
1
1
__
+_.
_
XOX
X
1
X
o
o
0
X
0
X
010
X
X ~------.
X
1
X
X
o
1
0
X ~------.
X ~--X 1------,
X
X
1
1
1
0
0
1
X
X
X
X
X
X
0 r--1 I-X f-
o
~
X
--
I----~_+.
---
----
--
-
10
11
12
13
14
15
1
a a
1
o 1
o 0
o 1
o 0
~
/
2B
2C
,
2D
~
2E
2F
31
32
33
L-_ _ _ _ _ _ _~--~
34
35
36
L-_ _ _ _ _ _ _ _~~~
37
a
38
0
L-_ _ _ _ _ _ _ _ _
J
~.
- t
3F
1
1
o
o
o
0
1
1
0
0
o
o
1
a
0
1
1
1
1
1
1
o 0
o 0
1
1
1
1
1
1
o
I
0
I
I
I
I
I
I
I
II
I
I
1
1
1
1
1
1
1
1
o
o
o
1
1
1
o
1
1
1
1
1
1
o
1
1
1
1
1
1
1
1
o
o
o
o
o
o
o
o
o
o
o
o
o
1
1
o
1
1
o
o
o
o
1
1
1
0
0
0
0
1
1
1
1
1
0
1
1
1
1
0
0
0
0
1
0
1
0
0
1
1
1
0
1
1
0
1
1
0
0
I•
II
1
1
1
II
II
II
1
0
II
i i i i i I i i
0
1
1
0
Figure 8-35. 2911 Sync/Enable ROM for the 2903 Twos Complement Binary
Division Microprogram Subroutine
8-108
I
I
1
1
1
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
1
o o
o o
a o
o o
o o
0
0
,•
1
1
o 0
o
o
o
o
o
o
o
o
30
0
1
1
1
1
1
o
I
o
o 0
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
o
o
o
...I----------~:--
1
1
1
1
1
o
2A
-
1
o
o
o
~
-------~_=_~
1
o
o
o
o
o
---------~--.\~
--
I
o
o
o
o
o
o
o
o
o
o
o
20
I
o
1
o
16
17
18
19
I
I
I
I
1
1
1
0
o
E
F
I
I
I
I
o 0
o 0
o 1
o 0
o 1
27
28
29
~~
....
A
B
C
D
-
--
o
o
o
o
8
9
1E
1F
~
eN FE
000
1
1
1 0
00001110
7
--
I~
RE S1 SO
o
1
1A
1B
1C
1D
---
L-_ _ _ _ _ _
o
1
1
0
0
Microinstructions 6 and 7 both have very simple 2911 addressing logic. Each microinstruction chooses the next
sequential microinstruction. But the Carry status generated by the 2903 following execution of microinstruction 7 must
be tested by address generation logic while microinstruction 8 is being executed. The fact that microinstruction 5 can
branch directly to microinstruction 8 causes no problem since microinstruction 5 will always generate a low 2903 Carry
status: thereforhe Carry status testing logic which we are about to describe does no harm.
If the Carry status generated by the 2903 during execution of microinstruction 7 is low. then no further divisor or dividend scaling is needed. and the next sequential microinstruction is selected. But if the 2903 generates a high Carry
status while executing microinstruction 7. then a subroutine must be called to scale the divisor or dividend. Therefore
microinstruction 8 provides the address of the scaling subroutine at the data inputs DO-D7; it generates the
same two sets of 2911 control inputs which we described for microinstruction 4. However. the level of C(N+4)
determines which set of 2911 control inputs are selected by microinstruction 8. See Figure 8-35.
We now come to the last three microinstructions (9, 10, and 11) which together perform the actual twos complement binary division. Microinstruction 9 must be executed once. followed by fourteen executions of microinstruction 10. and one execution of 11. Fourteen executions of microinstruction 10 are required because we are dividing a
16-bit divisor into a 32-bit dividend. From our previous discussion of the twos complement binary division algorithm.
recall that microinstruction 10 must be executed N times. where the divisor has N bits and the dividend has 2N bits.
Our 2911 addressing logic implements this multiple execution requirement by keeping the CN input low for thirteen
executions of microinstruction 10. But this requires an external counter. Here is one possibility:
A4
A 3 - -....._
x, 4 Counter.
advanced
ENABLE
A2
A1--_--
by ENABLE
AO
-.r-
ZERO--------~
There are innumerable ways (including the use of a 2910) in which the counting logic required to augment
microinstruction 10 may be implemented. The one we have shown assigns one set of eight read-only memory addresses to a preloaded counter. This preloaded counter provides the level of the low-order address bit (zero) which must
be input to the 2911 select ROM. The preloaded counter decrements once every time ENABLE goes high: it outputs
COUNT low until it decrements to O. Upon decrementing to 0 the counter outputs COUNT high.
Apart from the external counter required by microinstruction 10. microinstructions 9 and 10 generate simple next sequential address controls for the 2911. Microinstruction 11 causes the 2911 to pop its Stack - on the assumption that
the entire twos complement binary division microprogram is itself a microprogram subroutine. Thus. if a read-only
memory is used to generate 2911 control inputs Figure 8-35 illustrates the final read-only memory map and addressing bits required for the microinstruction sequence given in Table 8-11.
8-109
THE 2910 MICROPROGRAM SEQUENCER
This device is an enhancement of the 2909/2911 Microprogram Sequencers which we have just described.
Here is a summary of 2910 enhancements, as compared to the 2909 and 2911:
1)
The 2910 is a 12-bit device capable of addressing up to 4096 microinstructions. The 2909 and 2911 are 4-bit
devices. Therefore the 2910 is equivalent to three parallel 290912911 devices.
2)
The 2910 has sixteen address generation sequences, selected individually via four instruction code inputs. Many of
these sixteen address generation sequences have alternate forms that depend on the level of a condition code. The
2909/2'911 has no such address generation sequences; rather. individual control signal levels (such as output
selects SO and S 1. and output masks ORO-OR3) provide more primitive control of addressing logic.
The 2910 has three enable signals. One of the three is output low on each microcycle. You can use these enable
signals to selectively strobe data out of different sources to meet the specific needs of any microcycle. The 2909
and 2911 have no such enable outputs.
The Address register of the 290912911 becomes an Address register or a down counter within the 2910. As a
down counter. the 2910 Address register controls loop iteration.
The 2910 has a five-level Stack. as compared to the 2909/2911 four-level Stack. Also, the 2910 has slightly
different Stack Pointer logic.
3)
4)
5)
The fact that the 2910 is an enhancement of the 2909/2911 does not necessarily mean that the 2910 is always
the part of choice. The 2910 is more expensive. In most cases, a small microprogram consisting of 256 or fewer
microinstructions is more economically served by two 2909 or 2911 devices. rather than a single 2910 device.
The 2910 is packaged as a 40-pin DIP or a 42-pin flat package. The device is manufactured using bipolar technology. A
single +5V power supply is required. All signals are TTL-level compatible.
2910 MICROPROGRAM SEQUENCER PINS AND SIGNALS
For an overview of Microprogram Sequencer logic, and how it is used within a 2900-based system. refer to our earlier
discussion of this subject given for the 290912911 devices. The discussion which follows assumes that you understand
how a Microprogram Sequencer fits into a 2900-based configuration.
2910 pins and signal assignments are illustrated in Figure 8-35. Figure 8-37 illustrates 2910 functional logic.
We will describe signals in conjunction with functional logic.
The Output Multiplexer is central to logic of the 2910 Microprogram Sequencer. An instruction code input via 10-13 determines which of the four possible inputs will be
selected by the Output Multiplexer. In contrast. the 2909 and 2911 have two control inputs.
SO and S 1. that determine the source which the Output Mu Itiplexer selects.
2910
MICROPROGRAM
COUNTER
00-011 are twelve data input lines Which, like the 2911, can input data to the Address
register, or to the Output Multiplexer. If RLO is low, then data input at 00-011 is loaded
into the Address register on the low-to-high transition of clock signal CPo Data input via 00011 is selected by the Output Multiplexer when an appropriate instruction is input via 10-13. Timing for an Address
register access may be illustrated as follows:
CP
DO - D11
Address-----------v-------~'--------------Register _ _ _ _ _ _ _ _ _ _ _,,_ _ _ _ _Data
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
The 2910 outputs the microprogram address via VO-V11. These address output lines have
their own enable signal OE. If this signal is high. YO-Y11 outputs are floated. This allows you to
disconnect Microprogram Sequencer devices from the microprogram ROM. something you will
likely do when switching to a test program.
8-110
2910
ADDRESS
OUTPUT
Y4
04
Y5
05
VECT
Pi::
MAP
13
12
VCC
11
10
CCEN
CC
RLD
FULL
06
Y6
07
Y7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
2910
DIP
Pin Name
DO - 011
YO - Yll
VECT
MAP
PI:
RLO
OE
CI
10 -13
CC
CCEN
FULL
CP
VCC.GNO
Figure 8-36.
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
03
Y3
02
Y2
01
Yl
DO
YO
CI
CP
GNO
OE
Yll
011
Yl0
010
Y9
09
Y8
08
03
Y4
04
Y5
05
VECT
PL
MAP
13
12
VCC
11
10
CCEN
CC
RLO
FULL
06
Y6
07
Y7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
2910
FLAT
PACKAGE
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
Description
Type
Combined Address register and data input
Address output
Vector address enable
Map address enable
Pipeline address enable
Address register input enable
Output enable
Carry in
Instruction code
Condition code
Condition code enable
Stack full error indicator
System clock
Power. Ground
Input
Output. Tristate
Output
Output
Output
Input
Input
Input
Input
Input
Input
Output
Input
2910 Microprogram Sequencer Pins and Signal Assignments
8-111
Y3
GNO
02
Y2
01
Yl
DO
YO
CI
CP
GNO
OE
Yll
011
Yl0
010
Y9
09
GNO
Y8
08
~
~
Address Register/
Down Counter
Stack
Pointer
Zero
l----
i
5by -~
_-------T"""""--I~ 12-Bit_1'r-Stack
~
_ _ _ _ _ _ _ ___
_--------I.--f-
,
DO-Dll&~~
Microprogram
Counter
Output Multiplexer
~
~
~
r
J'\.
Incrementer
'"
-
CI
++++++
OE------------------~••~
.-
---:::::
~
Control
Logic
-
v'
YO - Y11
Figure 8-37. 2910 Microprogram Sequencer Functional Logic
8-112
I0
I1
I2
I3
CC
CCEN
Note that the 2910 has no zero input control on the address output lines. nor does it have any OR mask inputs. The
2909 and 2911 have such logic on the address output path. Instead. the 2910 uses its instruction codes to generate
equ iva lent addressing logic.
Like the 2909 and the 2911. the 2910 has two internal locations within which addresses can be held. These are
the Microprogram Counter and the Stack.
The 2910 Microprogram Counter logic is functionally identical to that of the 290912911
Microprogram Counter. Any address output by the output multiplexer also passes through
the incrementer, and is then stored in the Microprogram Counter. If CI is input high, then
the address output by the multiplexer is incremented by 1 before being written to the
Microprogram Counter. If CI is input low. then the address is written into the Microprogram
Counter without being incremented.
2910
MICROPROGRAM
COUNTER
2910
INCREMENT
The 2910 incrementer generates no Carry Out. Thus if you connect two 2910 Microprogram Sequencers in parallel. you
cannot increment the resultant 24-bit Microprogram Counter contents across the low- and high-order halves of the address. But. given current microprogramming technology. this is unlikely to pose any problem. Few single
microprograms have more than 4096 microinstructions. and you can use multiple 2910 Microprogram Sequencers to
address large microprograms in discrete 4096-microinstruction blocks.
The 2910 Microprogram Sequencer has a five-level Stack. The Stack is addressed by a Stack
Pointer which initially addresses Stack Register O. Appropriate instruction codes. input via 10-13.
generate Stack pushes or pops. A Stack push writes the Microprogram Counter contents to the
currently selected Stack location and then increments the Stack Pointer. The Stack Pointer always points to the last
word written. A Stack pop decrements the Stack Pointer. You must select the Stack as the output mu Itiplexer source in
order to load the current top-of-stack plus one via the incrementer into the Microprogram Counter.
If you execute more than five pushes. the Stack Pointer continues to address the last Stack location (number 4); the
Microprogram Counter contents are written into this location. overwriting prior data. This will inevitably cause an error.
The 2910 Stack Pointer does not wrap around. and will not increment from 4 to O. Stack full is indicated by the FULL
signal. When either four words have been pushed and a fifth push is selected via 10-13 or five words have been pushed.
FULL goes low. Thus it can be tested to see if the Stack is full.
If you try to pop the Stack when the Stack Pointer is addressing location O. then the Stack Pointer continues to select
location O. Since no data is written into the Stack following a pop. no prior information will be destroyed: however. in
all probability you have a microprogramming error. No error signal is output at this time.
2910
The various addressing operations which the 2910 Microprogram Sequencer can perform
INSTRUCTION
are identified using an instruction code input at 10-13. CCEN (condition code enable) and CC
(condition code) are two additional control inputs that in some cases modify the addressing
CODES
operations which the 2910 Microprogram Sequencer will perform. These six 2910 signals10-13. CC. and CCEN - together replace the 290912911 signals SO. S 1. ZERO. ORO-OR3. FE. and PUP.
....
-~~2910 responses to 10-13 inputs are summarized in Table 8-12. As shown in this table. many
2910
of the responses depend on the level of the CC and CCEN inputs. Vendor literature describes CC
CONDITION
as a "condition code" and CCEN as an enable: however. in effect. CCEN must be low while CC
CODES
is high to select a "fail" condition, while any other combination of CC and CCEN input levels
select a "pass" condition. Your design logic can use CC as a pass/fail selector. with CCEN as an override. This logic
may be illustrated as follows:
CC _ _ _ _ _ _ _ _
~------- Condition Code
CCEN
--L./
In response to each instruction code input at 10-13.2910 logic performs operations which we will describe individually.
Also. 2910 logic outputs low one of the three Signals PI.. MAP. or VECT. You can use these three signals as you see fit.
however their intended purpose is to enable one of three possible inputs to 00-011. Figure 8-38 functionally illustrates timing for PL. MAP. and VECT as part of the 2910 microcycle response.
VECT low will normally select a restart interrupt or other special address.
MAP will normally enable the mapping ROM out of which a microprogram starting address is generated in response to
a macroinstruction op-code.
VECT and MAP are the exceptional enable outputs. PL is the enable signal which is usually output low. This signal will
likely enable the Microinstruction register. or a connected Oata register. allowing the microinstruction to determine the
next data input to appear at 00-011. With reference to Figure 8-31. Figure 8-~39 identifies how VECT. MAP and PL
will likely be used.
8-113
.. 14
One
Microcycle
1-I
DO - 011.
10 -13
Control
Inputs
YO - Y11
pC MAP
or VECT
I
I
~
\
t\
I)
I
\
I
}
II
.!
I
'-
~I
I
I
I
CP~
One
Microcycle
"I
"
I
~
I
t
t
I
I
Figure 8-38. 2910 Microcycle Eve:1t Sequences
2910 MICROPROGRAM SEQUENCER INSTRUCTIONS
We will now examine 2910 Microprogram Sequencer instructions in detail. First, we will describe each instruction individually, then we will look at the instructions functionally, showing when instructions are likely to be
used, and in what combinations.
Instruction 0 (JZ) is a Jump-to-Zero instruction. This instruction forces the address to be output at YO-Y11;
therefore the Microprogram Counter will subsequently hold the address 0 or 1. depending on the level of the CI input
The JZ instruction also resets the Stack Pointer to O. effectively clearing the Stack. Since this instruction restarts execution with the microinstruction stored at microprogram memory location O. JZ is frequently used as a power-up. reset. or
Restart instruction code.
Instruction 1 (CJS) is a Conditional Jump-to-Subroutine instruction. A "pass" condition pushes the current
Microprogram Counter contents onto the Stack and takes the next microinstruction address from the 00-011 inputs. A
"fail" condition causes the next sequential microinstruction to be executed. This may be illustrated as follows:
M- Address HM" must be input
at DO - 011. Address "N+ 1"
is saved on the Stack.
8-114
Macroinstruction
D
Macroinstruction
Register
--------JJ
Mapping
PROM/ROM
Restart
Address
Register
--
OE
"'
II..
~~ OE
)
2910
Microprogram
Sequencer
~10
v
J
Microprogram
PROM/ROM
J
L----t.~~M
"fail"
"pass"
+
N
N+ 1
Address
"M"
must be
input at DO
- 011.
The return address
"N+ 1" is pushed
onto the Stack
Instruction code 6 is a Conditional Jump Vector (CJV). It is significant principally because it is the only instruction
code which generates a low VECT output. This low output usually enables a special Address register, out of which a
direct memory access or interrupt address is fetched.
If CC is in the "pass" condition, then the output multiplexer takes as its source DO-D11. If CC is in the "fail" condition,
the next sequential instruction is executed.
Instruction code 'is a Conditional Jump (JRP) which differs from the Conditional Jump-to-Subroutine (instruction
code 5) only in that no push occurs. If the condition code "passes", then the microinstruction execution sequence
jumps to the address which is input at DO-D11; otherwise, the microinstruction sequence jumps to the address held in
the Address register. This may be illustrated as follows:
l
I-------+
"faW'
Address uP" must P - - - - - - - N
have been loaded
previously into
the Address register.
"pass"
l
M
Address "M" must
be input at DO - 0 11 .
.
Instruction code 8 is a Repeat-Loop-Until-Counter-Is-Zero (RFCT) instruction. The purpose of this instruction code
is to re-execute one or more microinstructions some fixed number of times. The microinstruction loop to be re-executed
has its starting address stored in the Stack. The loop count is handled by the Address register. which must be
preloaded with a number that is one less than the required count. For example, if the Address register is loaded with an
initial value of 8, then the microinstruction loop will be executed 9 times. Thus, you can load into the Address register
values ranging between 0 and 4095 to generate counts ranging between 1 and 4096.
8-117
When the 2910 receives an RFCT instruction code, it examines the Address register contents. If the Address register
does not contain 0, then the output multiplexer takes as its source the currently addressed Stack location; the Address
register contents are then decremented. If the Address register contents are 0, then the output multiplexer selects the
Microprogram Counter as its source; also the Stack is popped.
We can illustrate the RFCT instruction code as follows:
M~:=J Address "M" must
Address register
contains O.
Pop Stack.
have been pushed
onto the Stack.
<: - -- N
-N+ 1 ~
Address register does
not contain O. Decrement
Address register.
Now the RFCT instruction, as illustrated above, re-executes a loop if the branch address M precedes N, where N is the
microprogram location. But you can push any address onto the Stack prior to executing an RFCT code. For example,
RFCT could be used to branch some fixed number of times without re-executing a loop. This may be illustrated as
follows:
1
Add"", "",;,... C---- Nt ___ ---- M
contains
O.
Po
St ck.
p a
--N+ 1 ~ Address reg,st~r
.
does not contain
O. Decrement
Address register
and branch· to "M".
Address "M" must
have been pushed
onto the Stack.
Instruction code 9, the Repeat-Register-Until-Counter-Is-Zero (RPCT) instruction, is almost identical to instruction code 8. RPCT again uses the Address register as a counter. If when RPCT is executed the Address register does not
contain 0, then it is decremented and the output multiplexer chooses as its source an immediate address input at 00011. If the Address register contents are 0, then the output multiplexer selects as its source the Microprogram Counter.
The Stack is not used.
If the microinstruction which generates the RPCT code supplies its own address at 00-011, then this microinstruction
gets re-executed the number of times specified by the Address register contents. This may be illustrated as follows:
Address register
contains O.
Pop Stack.
C-- --
Nl---,
--N+1+~--~N+1
Addn>ss "N+'" must
be input at DO - 011.
Address register does
not contain O. Decrement
Address register.
But a microinstruction does not have to supply its own address at 00-011 when specifying the RPCT instruction code.
It can input any address at 00-011. Thus, RPCT can be used to re-execute an instruction loop, or RPCT can be used to
execute any other branch some fixed number of times, as illustrated for the RFCT instruction code.
Instruction code A is a Conditional Return-from-Subroutine (CRTN).lf the condition code "passes", then a Returnfrom-Subroutine occurs. The output multiplexer takes the currently addressed Stack register as its source, then pops
the Stack. If the condition code "fails", then the next microinstruction address is taken from the Microprogram Counter.
This may be illustrated as follows:
"N+1" must be the address
currently in the top of
the Stack.
I .~
r
N+~'M---~
w
N+'~
8-118
'
"f T'
.0
Instruction code B is a Conditional Jump-and-Return instruction (CJPP). If the condition code "passes", then the
Stack is popped, but the output multiplexer takes 00-011 as its source. This is equivalent to a subroutine return where
the return address is not taken from the Stack, even though the Stack is popped: rather, the return address is taken
from 00-011. If the condition code "fails", then normal sequential microinstruction execution occurs with the next
microinstruction address being taken from the Microprogram Counter.
Instruction code C is a Load Counter instruction (LDCC). When this instruction code is executed, microprogram execution continues sequentially, with the output multiplexer taking the Microprogram Counter as its source: however,
data input at 00-011 is loaded into the Address register.
Instruction code 0 is a Conditional End-of-Loop (LOOP). As long as the condition code "fails", the output mUltiplexer selects the currently addressed Stack register as its source. When the condition code "passes", the Microprogram
Counter is selected as the output multiplexer source and the Stack is popped. This may be illustrated as follows:
M
"pass'<:'~: ~
~=:J
l
Address "M" must
have been pushed
onto the Stack.
"faW'
The LOOP instruction code is equivalent to the RFCT instruction code, with the condition code, rather than the Address
register, determining the number of loop iterations.
The LOOP instruction code does not have to be used to re-execute a loop. In the illustration above a loop is re-executed
only because the address M is shown preceding N, the microinstruction address where the loop code is generated. If
the address M does not precede N, then the LOOP instruction code becomes a simple conditional jump, where the
jump address is held at the top of the Stack.
Instruction code E is a Continue instruction (CONT). This is the normal default instruction code. It causes the next
sequential microinstruction to be addressed. The output multiplexer selects the Microprogram Counter as its source.
No other operations occur.
Instruction code F is a three-way branch (TWB); it uses the condition code and the Address register. Whether the
condition code "passes" or "fails", the Address register contents are decremented to O. If the condition code "passes",
then the Stack is popped and the Microprogram Counter is selected as the output multiplexer source. If the condition
code "fails", then the Stack is selected by the output multiplexer while the Address register is decrementing, but when
the Address register has decremented to 0, the Stack is popped and the output multiple~er selects as its source data input at DO-011. These options may be illustrated as follows:
On "pass" decrement
Address register to
M~:=:J
- - -- N
"pass" 0, then
pop <·"'N+1
Stack.
Address register does not contain O.
Decrement address register.
Address "M" must have been
pushed onto the Stack.
"fail"
I~'
"fail"
P
Address register contains O.
Address "P" must be
input at DO - 011.
The TWB instruction code is, in effect. a conditional loop execute. While the condition code is failing, the microinstruction sequence between M and N is re-executed a number of times, as defined by the Address register: microprogram
execution then branches to an address input at 00-011. But any single loop iteration can be bypassed by a "pass" condition code: moreover, when the Address register has decremented to 0, microprogram execution can continue sequentially, or it can branch to the address input at 00-011.
8-119
2910 MICROPROGRAM SEQUENCER ADDRESSING SEQUENCES
Let us now examine some of the more common microprogram address sequences which you are likely to encounter, and how these address sequences will be generated using a 2910 Microprogram Sequencer.
A simple sequence of consecutive microinstructions represents the simplest case:
Address
Sequence
2910 Instruction
Code Sequence
N
N+1
N+2
N+3
etc
CONT
CONT
CO NT
CONT
etc
You generate sequential microinstruction addresses. as requ ired by the sequence above. using instruction code E. with
the CI input high.
You will likely initialize your microprogram in one of two ways:
1)
2)
2910
MICROPROGRAM
Following a restart or special condition. use instruction code 0 (JZ); this forces a 0 output
INITIALIZATION
at Y. 0 or 1 can be written to the Microprogram Counter - depending on the level of the
CI input.
At the end of an instruction fetch. use instruction code 2 (JMAP) to take the microprogram starting address from a
mapping ROM. We described this instruction code earlier.
The 2910 Microprogram Sequencer gives you many ways of jumping within a 2910
microprogram. Any instruction code that causes the output multiplexer to select 00-011 or the MICROPROGRAM
Address register as its source can be used to generate a microprogram jump. Instruction code 3 ...J_U
__M_P
______~
(CJP) jumps to the address input at 00-011 if the condition code passes. So does instruction code
6 (CJV). but instruction code 6 (CJV) outputs VECT low - and the VECT enable signal is commonly used to select a
special OMA or interrupt address. Instruction code 7 (JRP) is a dual. unconditional jump; if the condition code
"passes". the jump address is taken from 00-011. while the jump address is taken from the Address register if the condition code "fails".
A microprogram subroutine is called using instruction code 5 (JSRP). This instruction code
pushes the Microprogram Counter contents onto the Stack; the subroutine starting address is
taken from 00-011 if the condition code "passes". and from the Address register if the condition
code "fails". The Address register must have been preloaded with an appropriate starting address
if the condition code "fails". Generally. instruction code C (LOCT) is used to load the Address
register.
2910
MICROPROGRAM
JUMP-TOSUBROUTINE
A normal Return-from-Subroutine will use instruction code A (CRTN) with the condition code held "passing".
There are two types of loop you may encounter in a microprogram. You may wish to re-execute a single
microinstruction, or a sequence of microinstructions some number of times.
The simplest way of re-executing a single microinstruction some fixed number of times is to load the Address register
with a number one less than the required count. then issue instruction code 9 (RPCT). This may be illustrated as
follows:
Address
Sequence
N
N+ 1 (M times)
N+2
2910 Instruction
Code Sequence
LOCT with M-1 to 00-D11
RPCT with N+1 to 00-011
CONT
You can alternatively use instruction code 8 (RFCT) to re-execute a single microinstruction. but the microinstruction
address must be preloaded onto the Stack.
8-120
... More frequently. instruction code 8 (RFCT) will be used at the end of a multi-microinstruction loop. This may be illustrated as follows:
2910 Instruction
Code Sequence
Address
Sequence
Re-execute
loop N+1
to N+M+1
P times
[
N
N+1
N+2
N+M
N+M+1
N+M+2
PUSH with P-1 to DO-D11 and CC "passing"
CO NT
CO NT
CO NT
RFCT
CO NT
As illustrated above. the microinstruction preceding the loop must input instruction code 4 (PUSH) with the condition
code passing. This simultaneously loads the count into the Address register while pushing the next sequential address
onto the Stack. Subsequently instruction code 8 (RFCT) decrements the Address register. selecting the address which
PUSH saved on the Stack. until the Address register has decremented to O.
2910 ADDRESSING EXAMPLE
As we did for the 2911 Microprogram Sequencer, we will now look again at the twos complement binary division microprogram given in Table 8-9, adding address generation microinstruction bits needed by the 2910
Microprogram Sequencer. See Table 8-13.
We are going to treat the twos complement binary division microprogram as a subroutine. We assume. therefore. that
microinstruction 1 is executed following a Jump-to-Subroutine instruction code input to the 2910 Microprogram Sequencer.
Microinstruction 1 inputs a CONT instruction code to 10-13 of the 2910 Microprogram Sequencer. This causes the
next sequential microinstruction to be executed. Since the 2910 does not sample its DO-D11 inputs. these bits are irrelevant. CI is input high since the 2910 Microprogram Counter contents must increment. RLD and CCEN are high
since no data is to be written into the Address register. and the condition code is not used.
Although microinstruction 1 itself transmits the simplest possible addressing logic to the 2910. the ZERO status output
by the 2903 while microinstruction 1 is executed contributes to microinstruction 2 address generation logic. The binary division subroutine must be aborted if the divisor is O. This abort condition is indicated by a ZERO status output
following execution of microinstruction 1. This ZERO status is output too late during microinstruction 1's microcycle to
be considered by 2910 addressing logic during execution of microinstruction 1. Therefore, in the event that the
subroutine must be aborted. microinstruction 2 gets executed gratuitously. but causes an abort exit from the
subroutine after its gratuitous execution. The CJP instruction code is input to the 2910 Microprogram Sequencer
by microinstruction 2 in order to achieve this end. This instruction code causes the next sequential microinstruction
to be addressed if the condition code "fails", while the addressing input at DO-D11 is selected if the condition code
"passes". The address of the microinstruction to be selected following an abort is therefore input via bits DO-D11. In
order to enable the condition code. CCEN is low. CC must be connected to the complement of Z while microinstruction
2 is being executed. This allows the ZERO status output by the execution of microinstruction 1 to generate the CC input during microinstruction 2 - and thus generate an abort. if needed. We do not show the logic which causes CC to
be connected to the complement of the 2903 Z status. There are many ways in which such "one time" connections can
be made. Possibly the simplest technique is to add microinstruction bits which enable a specific NOR gate linking the
2910 CC input with the 2903 Z output. This may be illustrated as follows:
CC----~----~L_~
z----------~L_~
8-121
Table 8-13. The 2903 Twos Complement Binary Division Microprogram Subroutine from Table 8-9,
with 2910 Addressing Microinstruction Fields Added
Microinstruction
EA
CN
AO -A3
80 - 83
a
a
a
Divisor register
(Ra)
Divisor buffer
(R3)
Copy divisor to temporary buffer.
6
a
a
a
Dividend (MS)
register (R 1 )
Dividend (MS)
buffer (R2)
Copy dividend most significant half to temporary buffer.
5
a
a
x
a
x
Dividend (MS)
buffer (R2)
Convert dividend (MS) from twos complement to sign/magnitude version.
Test aVR externally w/lile next microinstruction is being executed. If OVR is 1,
branch to subroutine that downshifts dividend.
4
5
a
a
x
a
x
Divisor buffer
(R3)
Convert divisor (MS) from twos complement to sign/magnitude version.
Test aVR externally. If OVR is 1, branch to microinstruction 9.
5
9
4
a
x
a
x
Dividend (MS)
buffer (R2)
Shift out sign bit of dividend (MS) half in temporary buffer.
6
9
4
a
x
a
x
Divisor buffer
(R3)
Shift out sign bit of divisor in temporary buffer.
7
F
2
a
a
1
Dividend (MS)
buffer (R2)
Divisor buffer
(R3)
Subtract sign bit stripped divisor from sign bit stripped dividend (MS) half.
If Carry = 1 (dividend larger) branch to subroutine that downshifts dividend
or upshifts divisor.
8
6
6
a
a
a
Dividend (LS)
register (R4)
X
9
A*
a
a
a
a
Divisor register
(Ra)
Dividend (MS)
register (Rll
Double length normalize dividend in MS register and Q register.
10
c·
a
a
a
z
Divisor register
(Ra)
Dividend (MS)
register (R 1l
Execute twos complement divide instruction fourteen times.
11
E
a
0
0
Z
Divisor register
(Ra)
Dividend (MS)
register (Rll
Twos complement divide final instruction.
No
18 -15
14 -11
10
1
4
6
2
4
3
Comment
Copy dividend least significant half to Q register.
'CN is connected to Z status while these two special functions are being executed.
8-122
Table 8-13. The 2903 Twos Complement Binary Oivision Microprogram Subroutine from Table 8-9.
with 2910 Addressing Microinstruction Fields Added (Continued)
Mlcrol••tnlcdon bl .. covering ZII03 ond ZII10 Inpu"
IBlt poold.... ore ..... ltr1Iry ond hove no .1•• 111.....1
Z810 BI..
21103BI..
-- _'-...--'----- -1817181111413 IZ 11 10
o
1 0 0 0 1 1 0 0
Ell eN A3A2Al AOB3 BZ Bl
0
0
0 0
6000
4
0 0 0 0
RO
80 13 IZ 11 10Dll Dl0 DII DB D7 D8 DII D4 D3 DZ Dl DO
1 1
R3
0100011000000010010
~-eo
0
0
-;;;-
1 0 1 0 0 0 0 0
X
0
X X X X 0
o
0
1 0 1
o0
5
0 0 0
X
0
1
0 1
o
9
o0
1
0
--..,-.'
R3
X
0
--..,-.'
9
o0
1
o
1
o0
11
X
X
X
X
X
1 0
0011A
1 1 lOX
_
~~
o
1
1
1
A
A
A
A
A
A
A
A
A
A
:;,.,
X
X
X
X
X
X
X
X
X
X
X
0 1 1 .......
B
-
5
B
B
S
5
5
5
5
5
5
5
5
'V
~
SC81e dividend subroutine
3 = CJP
B .!....B
B
B
B
B
B
V
~
Microinstruction 8 address
1
1
o
1
1
CC is generated by 2903 Z otatus ICC = ZI. Select next
sequential microinstruction unless RO from microinstruction
1 is O.. Then abort.
1
1
1
Select next sequential instruction.
1
1
1
0
1
1
0
1
1
0
1
1
0
1
1
Then branch to microinstruction 8.
1
1
5elect next sequential instruction.
X
X
X
X
X
X
X
X
X
1
1
1
1
0
1
0 0
X
X
X
X
X
X
X
X
X
X
X
1
1
1
0
1
1
1
1
1 1 0 0
0
0
0
0
0
0 0 0 0
0
0
0
0
0
1 0 0 0
0
1 1
1 1 lOX
CC is genarated by OVA st.tus. ICC = OVRI.
Select next sequential instruction unless
OVR
= 1. Then call uscals dividend" subroutine.
CC is generated by OVA status. ICC = OVRI.
Select next sequential instruction unless OVA = 1.
X
4
Select .ext seque.tial instruction.
0
X
6
A
X
X X X X 0 0 1 1 1 1 lOX
'-...--' - - . . , - . '
R3
E = CONT
5elect next sequential instruction.
'-...--' - - . . , - . ' - - . . , - . '
0
R2
-- -6
X
0
2
1 1
X
X
---..,-.'
o
0
R2
1 1 1 1 0 0 1 0 0
F
X
111
1 = CJ5
--..,-.'
1 0 0 1 () 1 0 0 0
10
X X X X 0
4
X
E = CONT
1 1
X X X X 0 0
0
0
----o
O' 1 0
R2
---..,-.'
X
-;;-~ ~'-""-----""V'
Address for abort
0
5
1 1 lOX
E=CONT
1 1 0 0 0 0 0 0 0
--..,-.'.-.-.'
COO
0 X
0
1 1 1 0 0 0 0 0 0
0
- -E- . . , 0 .' 0
0
X
0
R3
E = CONT
1 0 0 X X X X
~-.!.
1 = CJ5
R4
0 0 0 0 0 0 0 1
. - . - . ' '-...--'
Rl
RO
~~
T
-
T
0
0
X
1
-
1 0 1 0
X
T
T
T
T
T
V
0...2... 0
0......2.. 1
1
0
'V
0 0 0 0 0 0 0
'-...--'
Rl
T
~
Scale divisor or
dividend subroutine
Count
X
RO
T
4 = PU5H
0 0 0 0 0 0 0 1 1 0 0 0 X
--..,-.' ' - . . . - - ' RO
Rl
8 = RFCT
-
T
X
X
X
X
X
X
X
X
.J...
= 13
X
X
A = CRTN
X
X
X
X
X
X
X
X
X
X
0
1
1
0
1
1
1
1
0
1
1
0
1
1
1
1
1
1
1
1
1
1
1
1
CC is generated by CIN+41. ICC = CIN+4H.
Select next sequential instruction unless
C(N+4) = 1. Then call "scale divisor or dividend" subroutine.
Load count into address register, push
next sequential instruction address onto
Stack and select next sequential instruction.
Re-execute instruction addraosed by 5tack until count
decrements to O.
Ratum from subroutine.
x = "Oon't Care" bits
While it may appear wasteful to dedicate a single microinstruction bit to enabling such a connection. perhaps only
once in an entire microprogram. in fact the economics of microprogramming often favor wasting a single bit in this
way. The alternative is to have additional logic which generates a particular input, from one of a variety of outputs. depending on complex combinations of circumstances. To evaluate this reasonin~ look at Table 8-13. where CC may be
connected at different times to a Zero. Overflow or C(N+4) status. You cannot guarantee that the CC generation logic
will be a known function of any bit field within the microinstruction. since the same bit field may recur somewhere else
in the microprogram without having the same CC logic requirements. Your only alternative to dedicating a
microinstruction bit to each CC generation possibility would be to decode the microinstruction address itself; and that
would not be simple.
Microinstruction 3 is another example of a microinstruction that generates a status output which contributes to
subsequent addressing logic - but not until the next microinstruction. Microinstruction 3 itself provides the 2910
with a CaNT instruction code. which together with a high CI causes the next sequential microinstruction address to be
output by the 2910. The data bits 00-011 are not used; RLO and CCEN are both high (i.e .. in the "off" state). At the
2903. microinstruction 3 performs a twos complement to sign/magnitude conversion on the contents of RAM location
R2. If this operation generates a high Overflow status. then R2. which contains the high-order half of the dividend. must
be scaled by shifting down one bit position. The subroutine should be called. if needed. after microinstruction 3 has
been executed. and before microinstruction 4 is executed. But microinstruction 3 outputs OVR too late in its microcycle
for this status to contribute to the next microinstruction address. Therefore microinstruction 4 provides the 2910 with
addressing logic that tests the OVR status from microinstruction 3's execution. CJS. the Conditional Jump-toSubroutine. is specified by inputting 1 as the 2910 instruction code. The address of microinstruction 1's subroutine is
held in data field bits 00-011. CCEN is output low so that CC can be tested. CC must be connected to the complemen t
of the 2903 OVR status output. If CC is low - and OVR is high - then the subroutine addressed by 00-011 will be
called. Otherwise. since CI is high. the 2910 will output the next sequential microinstruction address - which is the
address of microinstruction 5. RLO is output high since the address at 00-011 must not be written into the Address
8-123
register. As we did in instruction 2. again in microinstruction 4 an additional microinstruction bit will likely connect CC-
to-OVR via NOR gates. And once again we will choose to waste this CC-to-OVR connector bit for the vast majority of
microinstructions where such a connection does not apply.
While microinstruction 4 provides the 2910 with addressing inputs that test the Overflow status generated by
microinstruction 3, the Overflow status itself is mQdified by the 2903 while microinstruction 4 is being executed. This presents no timing problems. since the 2910 has stopped sampling its CC input by the time microinstruction 4 modifies the Overflow status. See Figure 8-~8.
Microinstruction 5 must provide the 2910 with addressing inputs that account for the Overflow status generated by microinstruction 4, just as microinstruction 4 had to provide the 2910 with addressing inputs that took into
account the Overflow status generated by microinstruction 3. Microinstruction 5 causes the 2910 to execute a conditional jump. rather than a conditional subroutine call. since a high Overflow status generated by the 2903 while executing microinstruction 4 requires microinstructions 5. 6. and 7 to be bypassed. We do not bypass microinstruction 5.
since it is during this microinstruction's execution that the 2910 will test the Overflow status from microinstruction
4 - and determine if a jump is required. This would present a problem if execution of microinstruction 5 upset the
2903 logic sequence; but it does not. If executed unnecessarily no harm is done. But remember. there may be circumstances under which you may have to insert a dummy microinstruction that generates no operation at a 2903. but
gives the 2910 time to test 2903 status from the previous microinstruction.
In our present example microinstruction 5 inputs a CJP instruction code at 10-13 of the 2910 and the address of
microinstruction 8 at DO-D11. CCEN is low so that 2910 condition code logic will sample CC. which remains connected
to the 2903 OVR status. as it was during execution of microinStruction 4. RLD is high so that the address input at DOD11 does not get written into the Address register. CI is high since the next sequential microinstruction must be
selected if OVR is low - and the condition code test "fails".
If the condition code test "fails" at the 2910 during execution of microinstruction 5. then microinstructions 6 and 7
are next executed sequentially. Each of these microinstructions provides the 2910 with a simple CONT input at 10-13.
no data inputs. and CCEN and RLD disabled with CI high. so that the Microprogram Counter will increment
Microinstruction 7 subtracts modified versions of the divisor from the dividend. If the Carry Out generated by the 2903
at C(N+4) is 1. then the divisor must be upshifted (or the dividend must be downshifted) in order to guarantee that the
divisor ultimately has the larger absolute value C(N+4) is generated by the 2903 at the end of the microinstruction 7
microcycle - too late for the 2910 to take this status output into account until the next microcycle. during which
microinstruction 8 is executed. In Table $~13 we show microinstruction 8 providing a Conditional-Jump-toSubroutine input to 10-13, with a subroutine address input to 00-011. This 2910 addressing logic is identical to
that which we have already described for microinstruction 4. except that the condition code CC will now be connected
to the 2903 C(N+4) output - again via a special enabling microinstruction bit The fact that microinstruction 5 might
have caused the 2910 to generate a branch to microinstruction 8 is not a problem. since C(N+4) will be low following
execution of microinstruction 5: therefore if microinstruction 8 is executed next. the condition code must fail
Microinstruction 9 is executed after microinstruction 8 - possibly with a scaling subroutine executed in between
Microinstruction 9 prepares the 2910 for execution of microinstruction 10. A PUSH Instruction code is Input at 1014. with a count input at DO-D11. RLD is low so that the count gets written into the Address register. The 2910 then
outputs the address of microinstruction 10 since CI is high - so the Microprogram Counter gets incremented
MicroinstrlJction 10 outputs an RFCT instruction code to the 2910 via 10-13:This Instruction code causes the 2910
to output the address held at the top of the Stack until the Address register decrements to zero The push performed by
the 2910 while microinstruction 9 was executing loaded microinstruction 10's address onto the Stack. therefore
microinstruction 10 gets re-executed 13 times - for a total of 14 executions
After the Address register has decremented 13 times to O. the RFCT instruction code causes the 2910 to output the
next sequential instruction - that of microinstruction 11. This is the terminating microinstruction for the twos complement binary division microprogram subroutine Therefore a Return-from-Subroutine code. CRTN. IS transmitted to
the 2910 via 10-13 In order to force a pass condition. this being a Conditional Subroutine Return Instruction. code CCEN
IS output high
8-124
THE 2930 AND 2932 PROGRAM CONTROL UNITS
These two parts were designed to provide assembly language instructions with their memory address generation logic. The internal architecture of these two devices approximates II 2901 whose local RAM has been converted into a Stack, while the Q register functions as a local data register.
In reality. the 2930 series devices are hybrid parts that may substitute for 2909/10/11 Microprogram Sequencers, or they may be used separately to implement assembly language instructions' memory addressing logic.
The 2930 series devices are probably more effective as microprogram sequencers. In this role you need only make
sure that timing requirements of the 2930 series devices are compatible with the 2901 or 2903 central logic you are
using. But as assembly language memory address generators 2930 series devices leave a great deal to be desired. They
cannot cope with indirect memory addreSSing. since by its very definition indirect memory addreSSing requires intermediate access of external memory. Also. most minicomputer and microcomputer assembly languages depend on an
external memory stack which cannot be implemented within the small. local stack provided by 2930 series devices.
Therefore. 2930 series devices do little more than add indexes to base addresses. But the value of 2930 series devices
increases dramatically when you are not building a general purpose central processing unit. If you are building dedicated CPU-based logic that uses both microcode and higher level instructions. then you can probably avoid indirect addressing. and you can work with the limited 2930 series stack.
The 2930 series devices are both 4-bit slices. The 2930 is the most advanced of the two devices; it is fully
cascadable and packaged as a 28-pin DIP. The 2932 device is packaged as a 20-pin DIP and is also cascadable.
These two devices differ only in their internal instruction logic.
Both 2930 devices are manufactured using bipolar technology: they use a single +5V power supply and have TTL-level
compatible signals.
2930/32 DEVICE PINS AND SIGNALS
We will describe pins and signal assignments for these two devices together, and in conjunction with their
functional logic. Device pins and signals are illustrated in Figure 8-40. Functional logic is illustrated in Figure
8-41.
The Adder is central to 2930 series operations. The Adder accepts one or two operands as inputs. and generates a
single output. The Adder can perform three different operations:
1)
2)
3)
It can add (with carry) the two operand inputs. CN determines the level of the carry during addition.
It can increment a single operand by adding CN to the operand.
It can output a single operand unaltered.
2930 SERIES
Rand S in Figure 8-41 are the two Adder operands. The R operand can have one of three
sources. The S operand can have one of four sources. An instruction code selects the source for
INSTRUCTION
the Rand S operands. together with Adder operations. The 2930 has a 5-bit instruction code,
CODES
input at 10-14. The 2932 has a 4-bit instruction code, input at 10-13. These instruction codes do
more than control logic around the Adder and its inputs. they also control the Stack Accumu lator and Program
Counter. as summarized in Table 8-14.. This table is keyed to the 2930. The set of 2932 instruction codes is a subset
of the 2930 instruction code set. and is so identified.
The 2930 allows you to enable or disable the Adder output using the output enable signal OE. The 2932 has no
such output enable signal: Adder output from this device is always enabled. But when OE is input high to the 2930 the
YO-Y3 outputs are floated.
Data may be held in three different places within 2930 series devices: these three places are the Accumulator. the Program Counter. and the Stack.
The Accumulator is a single 4-bit location that can receive data input via 00-03, or it can
2930 SERIES
receive output from the Adder. The instruction code determines which of the two inputs. if
ACCUMULATOR
either. will be written into the Accumulator. The 2930. but not the 2932. has a separate Accumulator enable control signal RE. This enable control signal is subordinate to the instruction code. The level of RE is
unimportant when the instruction code specifies that data will be written into the Accumulator. For other instruction
codes. a low RE input causes immediate data at 00-03 to be written into the Accumulator.
8-125
10
EMPTY
CI
OE
C(I+4)
CN
YO
Y1
Y2
Y3
G
---
-----
C(N+4) .GND
.
.-.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
28
27
26
25
24
23
22
21
20
19
18
17
16
15
2930
-...-----
---......
----
----
DO - 03
YO - Y3
RE
OE
10 - 14/10 - 13
lEN
CC
CI
C(I+4)
CN
C(N+4)
G
FULL
EMPTY
CP
VCC. GND
10
FULL
CI
C(I+4)
CN
YO
Y1
Y2
Y3
GND
-- ..-- ------
1
2
3
4
5
6
7
8
9
10
20
19
18
17
2932
16
15
14
13
12
11
P
-... -
Pin Name
P
VCC (+5V)
11
12
13
14
CC
lEN
RE
DO
01
02
03
CP
Description
Data input
Address output
Register input enable
Address output enable
Instruction input
Instruction enable
Instruction condition code
Program Counter carry in
Program Counter carry out
Adder carry in
Adder carry out
Adder carry generate
Adder carry propagate
Stack full indicator
Stack empty indicator
Device clock
Power. Ground
Type
Input
Output
Input
Input
Input
Input
Input
Input
Output
Input
Output
Output
Output
Output
Output
Input
Figure 8-40. 2930/32 Program Control Units Signals and Pin Assignments
8-126
-I
-----
--- .
-- --
VCC(+5V)
11
12
13
DO
01
02
03
C(N+4)
CP
h
DO - 03
-
1
1
-----..
.
17 x 4
Bit
Stack
>
-
-----.. FULL
-
~ lii1v:
Accumulator
0
0
. ,,l,
J
("<
,
~
'Ii
R
~ ~
J
S
3-IN MUX
4-IN MUX
A
"
~
.1
..
CN
~
.
~
'1\
R
S
Program
Counter
Adder
"
),~
('
7
-
~
:::
~
7
Incrementer
--
j
II
Instruction
Decoder
YO - Y3
These signals are present in the 2930,
but not the 2932.
Figure 8-41. Functional Logic for 2930 Series Program Control Units
8-127
~
CI
C(i+4)
Table 8-14. 2930 Series Program Control Unit Instruction Codes Summary
Instruction
Output
atYO-V3
Code
No.
New PC
Contents
Accumulator
Contents
(1)
2932(4)
Instructions
Stack
Operation
No.
1413121110
Comment
131211 10
0
4
8
o
o
None
9
1 0.,0 1
None
None
A
1 0 1 0
CI
CI
CI
CI
[D)
[D)
[D)
[D)
[D)
[PC)
+ CI
[D)
None
[PC)
+ CI
[D)
[PC)
[PC)
+ CI
+ CI
[D)
[PC)·
[PC)
+ CI
[PC)
[PC)
+ CI
[OJ"
None
F
1 1 1 1
[PC)
[PC)
[PC)
[PC)
+ CI
+ CI
[D)
~D)
PUSH [PC)
PUSH [0)
6
2
o
o
01 1 0 1
o1 1 10
01 1 1 1
[S)
[PC)
[PC)
[PC) + CI
[PC) + CI
[PC)
[D)
[D)
[D)
POP
POP
None
3
00 1 1
10
11
12
13
1
1
1
1
[R)
[D)
0
[D)
[D)
[D)
[D)
[0]
None
None
None
None
8
5
o
14
1 0 1 0 0
[PC)
+ [D) + CN
[0]
None
i
15
1 0 1 0 1
[PC)
+ [R] + CN
C
1 1
1
1
1
1
0
1 1
~
16
17
18
19
lA
1 1 0 1 0
[PC)
+ [0] + CN
18
1 1 0 1 1
[PC]
+ [R] + CN
lC
10
1 1 1 0 0
1 1 1 0 1
IE
IF
1 1 1 1 0
1 1 1 1 1
[R + CI
[D) + CI
CI
[R) + [D)
+ CN + CI
[PC) + [D)
+ CN + CI
[PC] + [R]
+ CN + CI
[R) + CI
[D) + CI
CI
[R] + [D)
+ CN + CI
[PC] + [0]
+ CN + CI
[PC) + [R]
+ CN + CI
[S] + CI
[S) + [0]
+ CN + CI
[PC)
[PC)
i
~
1
";:
1
;:,
§
i
1
J
0
1
2
3
4
0000
0000
000 1
000 1
00 1 0
5
001 0 1
[PC)
+ [D) + CN
6
00 1 1 0
[PC)
+ (RI + CN
7
8
00 1 1 1
o 1 00 0
[S)
+ [D) + CN
9
o
1
[R)
+ (D) + CN
A
o
o
o
1
0
8
C
1 o 1 1
1 1 0 0
0
E
F
o0
o1
000
000
0 0 1
00 1
0
0
1
1
1 1
1 1
00
00
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
[PC)
[R)
[D)
[R) + [D) + CN
[PC)
[R)
[R)
[S)
+
+
+
[R)
[0]
0
[0]
[S]
[0]
+ CN
+ CN
+ CN
[PC)
Off 121
CI
[PC) +
[PC) +
[PC) +
[PC) +
[R)
+ [D) + CN·
[0]
[0]
[D)
[0]
[D)
O-[SP)
None
None
0 0 0
1 o 0
1 0 0 0
None
None
None
None
PUSH
PUSH
PUSH
PUSH
[PC)
[PC)
[PC)
[PC)
1 1 0
0 1 0
1 0 1 1
1 0 1
o
o
0
1
[0]
PUSH [PC)
[0]
PUSH [PC)
E
1 1 1 0
[0]
[0]
POP
None
7
o
1 1 1
[D)
[D)
None
None
1
o
0 0 1
Device resat
Output p.rogram Counter contents
Output Accumulator contents
Output immediete data
Output sum of Accumulator contents and
immediate data
Output sum of Program Counter and
immediata data
Output sum of Program Counter and
Accumulator
Output sum of Stack and immediate data
Output Program Counter contents and
unconditionally loed Accumulator
Output sum of Accumulator and immediate
data. and unconditionally loed Accumulator
Output Program Counter contents. Unconditionally loed immediate data into Accumulator
Output and push Program Counter contents
Output Program Counter contents and push
immediate data
Pop and output Stack
Output Program Counter and pop Stack
Output and hold Program Counter
Jump to address in Accumulator
Jump to address given by immediate data
Jump to "0"
Jump to address given by sum of Accumulator
and immeidate data
Jump to address given by sum of Program Counter
contents and immediate data
Jump to address given by sum of Program Counter
and Accumulator contents
Jump to subroutine addrassed by Accumulator
Jump to subroutine addrassed by immediate data
Jump to subroutine origined at 0
Jump to subroutine addressed by sum of
Accumulator contents and immediate data
Jump to subroutine addressed by sum of
Program Counter and immediate data
Jump to subroutine addressed by sum of
Program Counter and Accumulator contents
Retum from subroutine
Retum from subroutine to retum eddress
plus immediate data
Output and hold Program Counter
High impedance output. Hold Program Counter
(1 )The Accumulator is loaded only when Ai: is input low. Exceptions are the three instructions marked •. which cause the Accumulator to be loaded unconditionally.
(2lThe YO - Y3 outputs are in the high impedance state.
(3)Conditional instructions execute as described only when Cc is input low. Otherwise instruction 1 is executed. These instructions are unconditional for the 2932 which has no
Cc input.
(4)The 2932 has no R[ input. therefore only the three instructions marked· load the Accumulator.
8-128
The Program Counter normally receives the Adder output as its new input. The Adder output
2930 SERIES
mayor may not be incremented, depending on the level of the CI input. When the Adder is
PROGRAM
COUNTER
not outputting the prior Program Counter contents. loading this different output into the Program
Counter constitutes a program jump. Many instruction codes bypass this jump logic by recycling
the Program Counter contents back through the incrementer while some other output is generated by the Adder. For
example. one instruction code causes the Adder to output data input at 00-03. But the Program Counter contents are
held unaltered. or incremented. depending on the level of the CI input.
2930 SERIES
Both 2930 series devices have a 17-level Stack. Program Counter contents or data input at
STACK, PUSH,
00-03 can be pushed onto the Stack. The 2930 series push operation is a little unusual. The
POP
Stack Pointer is incremented. then the selected data is written into the newly addressed Stack
location. Thus. following a push. the Stack Pointer addresses the data most recently pushed onto
the Stack. It is more traditional for the Stack Pointer to address the first free Stack location. When you pop a 2930
series Stack, you decrement the Stack Pointer contents, and that is all. Thus. in order to read data off the top of
the Stack. you select an instruction code which reads data from the Stack. then you pop the Stack. The more traditional
Stack architecture requires that you pop the Stack. then read data from the top of the Stack.
The 2930 series Stack Pointer is not wrap-around. If the number of pushes exceeds the
number of pops by more than seventeen. the Stack Pointer continues to address the topmost
Stack location. If you attempt to pop the Stack when the Stack Pointer is addressing location O.
then it continues to address location O.
2930 SERIES
STACK POINTER
When the Stack Pointer is addressing either of the two top Stack locations. the FULL signal is output low. When you
execute a push while the Stack Pointer is addressing the top Stack location. data is written into this top location. overwriting whatever was there before.
The 2930. but not the 2932. has a Stack Empty indicator. This signal. EMPTY. is output low following a reset. or after
the lowest level Stack location has been popped.
2930 series devices are cascadable. However. the 2930 has more cascading logic than the 2932. Unlike the 2901 and
the 2903. the 2930 series devices do not have Shifters along data paths. therefore the Accumulator. Stack. and Program Counter do not need parallel interconnect signals. Parallel interconnect signals are needed only to cascade the
Program Counter as it increments. or the Adder following an increment or addition. For the Program Counter there is a
carry input. CI. and a carry output. C(I+4). The delay between CI being input and C(I+4) being output is very short.
There is plenty of time for this ripple carry to propagate through four slices. for a 16-bit address. within one microcycle.
For timing details see the data sheets at the end of this chapter.
2930 SERIES
Both 2930 series Adders have a carry input, Ct,!, and a carry output, C(N+4). Only the 2930
CARRY LOGIC
has is" and Goutputs, the carry propagate and generate. Therefore only the 2930 allows you to
use a 2902 carry look-ahead generator. When using the 2932 device. you must rely on ripple carry. For a discussion of carry look-ahead and ripple carry. refer to the 2901 and 2902 device descriptions given earlier in
this chapter.
USING 2930 SERIES DEVICES
You must be careful when deciding how to use 2930 device read/write locations. You should not use these locations to implement CPU registers if you are designing any type of general purpose minicomputer or microcomputer.
That is because you will have to perform a sequence of pushes and pops. each requiring a single microcycle. in order to
select an arbitrary location. Moreover. every time you perform a push in order to increment the address. you will
simultaneously write into the newly addressed location - something you may not wish to do. Thus the Stack lets you
have 17 levels of subroutine within the program logic that you use to generate addresses.
You can use the 2930 series Accumulator to implement a CPU register. For example.
many primitive Central Processing Units have a single Index register whose contents can contribute to address generation logic. but not to CPU arithmetic or logic operations. The 2930
series Accumulator would be an ideal location for such an Index register.
2930 SERIES
INDEX REGISTER
If you are designing a special purpose Central Processing Unit using 2930 series devices. you may well be able to use
the 17-level Stack to give you all the CPU registers you require. Knowing in advance the limitations of 2930 series
Stack access. you can design your microprogram around these limitations so that addresses are stored in the proper
serial sequence within the 2930 series Stack.
8-129
DATA SHEETS
This section contains specific electrical and timing data for the following devices:
•
•
•
•
•
Am2901. Am2901A and Am2901 B
Am2902A Carry Look-Ahead Generator
Am2903 Enhanced Microprocessor Slice
Am2910. Am2909. and Am2911 Microprogram Sequencers
Am2930 Program Control Unit
8-01
Am2901/Am2901 A
ALUSOURCE
OPERANOS
MICROCODE
12
11
10
L
L
L
L
H
H
H
H
L
L
H
H
L
L
H
H
L
H
L
H
L
H
L
H
Octal
Code
0
1
2
3
4
5
16
14
13
A
A
a
0
0
0
a
L
L
L
L
H
H
H
H
L
L
H
H
L
L
H
H
L
H
L
H
L
H
L
H
0
0
0
6
7
MICROCODE
S
R
8
8
A
A
0
0
RAM
FUNCTION
a·REG.
FUNCTION
16
0ctI01
Code
Shift
L ..d
Shift
L
L
0
X
NONE
NONE
F .... O
L
H
1
X
NONE
X
NONE
18
17
L
L
Symbol
RPlusS
SMinus R
RMinu,S
RORS
RANDS
RANDS
R EX-OR S
R EX·NOR S
0
1
2
3
4
5
6
7
R+S
S-R
R-S
RVS
RA S
RI\S
R¥S
R'VS
Figure 3. ALU Function Control.
Figure 2. ALU Source Operand Control.
MICROCODE
ALU
Function
0ctI01
Code
Load
V
OUTPUT
RAM
SHIFTER
a
SHIFTER
RAMo
RAM3
Clo
F
X
X
X
X
F
X
X
X
X
03
L
H
L
2
NONE
F .... 8
X
NONE
A
X
X
X
X
L
H
H
3
NONE
F .... 8
X
NONE
F
X
X
X
X
H
L
L
4
DOWN
F/2 .... 8
DOWN
012 .... 0
F
FO
IN3
00
IN3
H
L
H
5
DOWN
F/2 .... 8
X
NONE
F
FO
IN3
00
X
H
H
L
6
UP
2F .... 8
UP
20 .... 0
F
INO
F3
INO
03
H
H
H
7
UP
2F .... 8
X
NONE
F
INO
F3
X
03
X= Don't care. Electrically, the shift pin is a TTL input Internally connected to a three·state output which is In the high·
Impedance state.
8 = Register Addressed by B inputs.
Up is toward MSB, Down is toward LSB.
Figure4. ALU Destination Control.
~210OCTAL
0
1
2
3
4
5
6
7
CI.~
Source
A,a
A,B
o,a
O,B
O,A
D,A
D,a
0,0
14
L 3 ALU .
Function
0
1
2
Cn-L
R PlUI S
Cn-H
Cn-L
SMinul R
Cn - H
Cn- L
R MinulS
Cn-H
A+O
A+8
a
8
A
O+A
0+0
0
A+O+l
A+8+1
0+1
8+1
A+l
O+A+l
0+0+1
0+1
O-A-l
8-A-l
0-1
8-1
A-I
A-D-l
0-0-1
-0-1
O-A
B-A
0
8
A
A-O
0-0
-0
A-O-l
A-8-1
-0-1
-8-1
-A-l
D-A-l
0-0-1
0-1
A-O
A-8
-0
-8
. -A
O-A
0-0
0
3
RORS
AVO
Av8
a
8
A
OVA
OVO
0
4
RANDS
AMi
Ai\8
0
0
0
Di\A
0110
0
5
RANDS
Ai\O
Ai\8
0
8
A
OIlA
0110
0
6
R EX·ORS
A"'O
A"'8
0
8
A
O"'A
O¥O
0
7 REX-NORS
A¥O
A¥ B
Q
ii
A
O"'A
O¥O
0
V = OR; 1\
= AND;
+ * Plus; -
= Minus;
V = EX·OR
Figure 5. Source Operand and ALU Function Matrix.
Data sheets on pages 8-02 through 8-025 Copyright
copyright owner.
©
1978 by Advanced Micro Devices, Inc. Reproduced with permission of
8-D2
Am2901 /2901 A
SOURCE OPERANDS AND ALU FUNCTIONS
Figure 5 results. This matrix fully defines the ALU/source
operand function for each state.
There are eight source operand pairs available to the ALU as
selected by the 10, 11, and 12 instruction inputs. The ALU can
perform eight functions; five logic and three arithmetic. The
13, 14, and 15 instruction input$ control this function selection.
The carry input, Cn , also affects the ALU results when in the
arithmetic mode. The en input has no effect in the logic mode.
When 10 through 15 and en are viewed together, the matrix of
The ALU functions can also be examined on a "task" basis,
i.e., add, subtract, AND, OR, etc. In the arithmetic mode, the
carry will affect the function performed while in the logic
mode, the carry will have no bearing on the ALU output.
Figure 6 defines the various logic operations that the Am2901 A
can perform and Figure 7 shows the arithmetic functions of
the device. Both c:arry-in LOW (en = 0) and carry-in HIGH
(C n = 1) are defined in these operations.
Octal
Group
Function
40
41
45
46
AND
A/\O
A/\8
O/\A
0/\0
30
31
35
36
OR
AVO
AV8
OVA
OVO
60
61
65
66
EX-OR
AVO
AV8
OVA
OVO
1543,1210
70
7 1
75
76
Octal
1543,1210
EX-NOR
72
73
74
77
INVERT
62
63
64
67
AVO
Avo8
9
ADD
8
A
o
PASS
0
8
A
0
1 2
1 3
1 4
PASS
0
8
A
0
2
2
2
1
"ZERO"
0
0
0
0
MASK
A/\O
A/\8
D/\A
0/\0
1
1
1
1
2
2
PASS
4
0 7
Decrement
2 7
32
33
34
37
42
43
44
47
50
51
55
56
A+8
O+A
O+Q
ADD plus
one
8
A
0
0-1
8-1
A-1
0-1
1'5 Compo
3
4
7
-8-1
-A-1
-0-1
1
5
Subtract
(1'sComp)
8-A-1
A-0-1
0-0-1
A-Q-l
A-8-1
O-A-l
O-Q-l
A+0+1
A+8+1
O+A+l
0+0+1
Increment
PASS
8+1
A+l
0+1
0
8
A
0
2'sComp.
(Negate)
-0
-8
-A
-0
Q-A-l
0
Function
Q+l
-Q-l
2
6
0
1
2 5
2 6
Figure 6. ALU Logic Mode Functions.
Cn = 1 (High)
Group
Q
0 2
0 3
D
Function
A+O
0 0
0 1
0 5
0 6
OVA
OVO
Cn = 0 (Low)
Group
Subtract
(2's Comp)
Q-A
8-A
A-O
Q-O
A-O
A-8
O-A
0-0
Figure 7. ALU Arithmetic Mode Functions.
8-03
Am2901
MAXIMUM RATINGS (Above which the useful life may be impaired)
Storage Temperature
Temperature (Ambient) Under Bias
Supply Voltage to Ground Potential
DC Voltage Applied to Outputs for HIGH Output State
DC Input Voltage
DC Output Current, Into Outputs
DC Input Current
-0.5 V to +6.3 V
-0.5 V to +VCC max.
-0.5 V to +5.5 V
30mA
-30 mA to +5.0 mA
OPERATING RANGE
PIN
Ambient Temperature
I Am2901PC. DC
I Am2901DM. FM I
Vcc
I
1 0 ° C to +70°C
1
-55°C to +125°C
4.75 V to 5.25 V
4.50 V to 5.50 V
I
1
STANDARD SCREENING
(Conforms to MI L-STD-883 for Class C Parts)
Step
MIL-STD-883
Method
Pre-Seal Visual Inspection
2010
Leyel
Am2901 PC, DC Am2901DM. FM
Conditions
100%
100%
Stabilization Bake
1008
C
24-hour
150°C
100%
100%
Temperature Cycle
1010
C
-65°C to +150°C
10 cycles
100%
100%
Centrifuge
2001
B
10,OOOG
100% •
100%
Fine Leak
1014
A
5 x 10 -8 atm-cc/cm 3
100% •
100%
Gross Leak
Electrical Test
Subgroups 1 and 7
1014
C2 Fluorocarbon
100% •
5004
See below for
definitions of subgroups
100%
See below for
definitions of subgroups
=5
=7
LTPD = 7
LTPD = 7
LTPD = 7
LTPD = 7
100%
100%
Insert Additional Screening here for Class B Parts
Group A Sample Tests
Subgroup 1
Subgroup 2
Subgroup 3
5005
Subgroup 7
Subgroup 8
Subgroup 9
'Not applicable for Am2901PC
=5
=7
LTPD = 7
LTPD = 7
LTPD = 7
LTPD = 7
LTPD
LTPD
LTPD
LTPD
ADDITIONAL SCREENING FOR CLASS B PARTS
Step
level
MI L·STD·883
Method
Burn·ln
1015
Electrical Test
Subgroup 1
Subgroup 2
Subgroup 3
Subgroup 7
Subgroup 9
5004
Conditions
D
160
~~~:~min.
Am2901DMB. FMB
100%
100%
100%
100%
100%
100%
Return to Group A Tests in Standard Screening
GROUP A SUBGROUPS
ORDERING INFORMATION
(as defined in MIL·STD-883, method 5005)
Subgroup
Package
Type
Temperature
Range
Order
Number
Molded DIP
Hermetic 0 IP
Hermetic 0 IP
Hermetic Flat Pack
Dice
O°C to +70°C
O°C to +70°C
-55°C to +125°C
-55°C to +125°C
O°C to +70°C
AM2901PC
AM2901DC
AM2901DM
AM2901FM
AM2901XC
8-04
10
11
Parameter
Temperature
DC
DC
DC
Function
Function
25°C
Maximum rated temperature
Minimum rated temperature
25°C
Maximum and minimum rated
temperature
25°C
Maximum Rated Temeperature
Minimum Rated Temperature
Switching
Switching
Switching
Am2901
ELECTRICAL CHARACTERISTICS OVER OPERATING RANGE (Unless Otherwise Noted)
(Group A, Subgroups 1, 2 and 3)
Parameters
Description
Test Conditions (Note 1)
Min.
10H = -1.6mA
YO, V" V2, V3
10H
VOH
I CEX
Output HIGH Voltage
Output Leakage Current
for F = 0 Output
VCC = MIN.
VIN = VIH or VIL
VCC = MIN., VOH
VIN = VIH or VIL
z
-1.OmA. Cn +4
Output LOW Voltage
VCC· MIN.,
VIN = VIH or VIL
=-soo"A, F3
2.4
10H = -soo"A
RAMO, 3, 00, 3
2.4
10H = -1.6mA, G
2.4
Volts
=5.5V
250
0.5
10L = S.OmA, OVR, P
0.5
10L - 6.0mA, F3
RAMO. 3. 00. 3
0.5
Guaranteed input logical HIGH
voltage for all inputs
VIL
Input LOW Level
Guaranteed input logical LOW
voltage for,all inputs
VI
Input Clamp Voltage
VCC = MIN .• liN = -ISmA
IlL
IIH
II
Input LOW Current
Input HIGH Current
Input HIGH Current
VCC = MAX.
VIN =0.5V
VCC = MAX.
V IN = 2.7V
2.0
I
10ZH
10ZL
lOS
ICC
Notes: 1.
2.
3'.
4.
Off State (High Impedance)
Output Current
Output Short Circuit Current
(Note 3)
Volts
-1.5
Volts
-0.36
AO. A" A2, A3
-0.36
BO. B,. B2, B3
00,0,,02. 0 3
-0.36
10,1,,12;16,IS
-0.36
13. 14. 15. 17
-0.72
-0.72
RAMO, 3. 00, 3 (Note 4)
-O.S
Cn
-3.6
Clock,OE
20
Ao. A"
20
A2, A3
BO. B,. B2. B3
20
DO. 0,. 02. 03
40
10.ll.12.16. IS
20
13. 14. 15. 17
40
RAMO. 3. 00. 3 (Note 4)
100
Cn
200
1.0
VCC = MAX .• VIlli = 5.5V
VCC = MAX.
RAMO. 3.
00.3
VCC = 5.75V
VO=0.5V
0.7
0.8
Clock.OE
YO. V,.
V2. V3
Volts
Volts
Military
Commercial
I
"A
0.5
10L = 10mA, Cn +4, FmO
Input HIGH Level
VIH
Units
2.4
2.4
10L = 16mA
YO, V" V2, V3,G
VOL
Max.
2.4
10H = -800"A, OVR, P
IOH
Typ.
(Note 2)
Vo =2.4V
50
Vo -0.5V
-50
Vo =2.4V
(Note 4)
100
Vo -0.5V
(Note 4)
-800
YO. V,. V2. V3.G
-15
Cn+4
-15
OVR.P
-15
-40
-40
-40
F3
-15
-40
RAMO. 3. 00. 3
Military
Commercial
-15
mA
,.A
rnA
"A
mA
-40
280
rnA
280
..
For condItIons shown as MIN. or MAX .• use the appropriate value specified under Electricel Characteristics for the epplicable device type.
Typical limits are at Vec = S.OV. 2S·C ambient and maximum loading.
Not more than one output should be shorted et a time. Duration of the short circuit test should not exceed one second.
These are three-state outputs internally connected to TTL inputs. Input characteristics are measured with 167SIn a state such that the
three·state outpu\ is OFF.
Power Supply Current
VCC= MAX.
8-05
185
185
Am2901
GUARANTEED OPERATING CONDITIONS
OVER TEMPERATURE AND VOLTAGE
TABLE I
Tables I, II, and III below define the timing requirements of
the Am2901 in a system. The Am2901 is guaranteed to
function correctly over the operating range when used within
the delay and set·up time constraints of these tables for the
appropriate device type. The tables are divided into three
types of parameters; clock characteristics, combinational delays
from inputs to outputs, and set-up and hold time requirements.
The latter table defines the·time prior to the end of the cycle
(i.e., clock LOW-to-HIGH transition) that each input must be
stable to guarantee that the correct data is written into one of
the internal registers.
CYCLE TIME AND CLOCK CHARACTERISTICS
TIME
Am29010C,PC
Am2901 OM, FM
105ns
120ns
9.5MHz
8.3MHz
Read·Modify-Write Cycle
(time from selection of
A, B registers to end of
cycle)
Maximum Clock Frequency to
Shift Q Register (50% duty
cycle)
The performance of the Am2901 within the limits of these
tables is guaran~ed by the testing defined as "Group A,
Subgroup 9" Electrical Testing. For a copy of the tests and
limits used for subgroup 9, contact Advanced Micro Devices'
Product Marketing.
Minimum Clock LOW Time
30ns
30ns
Minimum Clock HIGH Time
30ns
30ns
Minimum Clock Period
105ns
120ns
TABLE"
MAXIMUM COMBINATIONAL PROPAGATION DELAYS (all in ns, CL oE;; 15pF)
Am2901DC, PC (O°C to +70°C; 5V ±5%)
~
F
rom
Input
Output
Y
F3
Cn +4
G,P
F=O
RL= OVR
RAMO
470
Cn+4
G,P
120
95
90
110
75
65
80
55
90
75
60
40
30
-
90
70
70
75
60
65
65
75
65
70
-
-
45
45
60
-
-
-
50
50
-
-
-
40/25
-
-
-
-
-
-
65
-
-
-
-
-
-
-
-
105
60
125
95
110
120
115
65
110
75
110
70
70
70
100
60
95
50
35
-
-
60
50
-
30
40
60
55
65
65
65
80
65
1345
70
55
60
60
70
1678
55
-
-
-
-
-
-
-
-
-
-
85
100
100
110
95
(Note 6) 115
SET-UP AND HOLD TIMES (all in ns) (Note 1)
Notes
-
-
110
120
85
120
110
65
65
105
65
55
45
60
70
85
70
85
-
105
Am2901 DC, PC (O°C to +70°C, 5V ±5%)
Am2901 DM,FM (-55°C to +125°C, 5'1£ ±10%)
Hold Time
'Set-Up Time
Hold Time
0
120
tpwL + 30
0
tpwL + 15
0
tpwL +15
0
D (arithmetic mode)
D (I = X37) (Note 5)
Cn
100
60
55
0
0
0
110
65
60
0
0
0
1012
85
0
90
0
1345
70
tpwL + 15
0
75
0
0
tpwL + 15
0
30
0
30
0
2,4
3,5
2,4
1678
RAMO, 3,
4
00, 3
-
TABLE III
Set-UpTime
105
tpwL + 30
A,a
Source
B Dest.
-
60
80
OE Enable/Disable 40/25
A bypassing
60
ALU (I = 2xx)
03
80
65
80
85
From Input
F3
Shift
Outputs
F=O
RL= OVR
RAMO 00
470
RAM3
-
85
1012
I
Y
00
RAM3 03
A,a
110
D (arithmetic mode) 100
D (I = X37) (Note 5) 60
Cn
55
Clock
Am2901DM, FM (-55°C to +125°C; 5V ±10%)
Shift
Outputs
Not.: 1. Se. Figure 11 and 12.
2. If the B addr ... i. used a. a source operand, allow for the "A, B source" set-up time; if it is used only for the destination addr•••, use the
uS dest." set-up time.
3. Wh.r. two numbe .. are shown, both must be met.
4. "tpwL" i. the clock LOW time.
5. OVO is the fastest way to load the RAM from the 0 inputs. This function is obtained with I = 337.
6. Using Q register as source operand in arithmetic mode. Clock is not nqrmally in critical speed path when Q is not a source.
8-06
Am2901A
ELECTRICAL CHARACTERISTICS OVER OPERATING RANGE (Unless Otherwise Noted)
(Group A, Subgroups 1,2, and 3) Data in bold face is changed from Am2901
Parameters
Description
Test Conditions (Note 1)
Min.
10H = -1.6mA
VOH
Output HIGH Voltage
VIN.= VIH or VIL
10H = -1.0mA.C n +4
10H = -SOOI'A. OVR. P
2.4
10H - -600IlA. F3
2.4
Output LOW Voltage
2.4
VCC = MIN .• VOH = 5.5V
250
VIN = VIH or VIL
VCC= MIN .•
VOL
Volts
2.4
RAMO. 3. QO. 3
10H - -1.6mA. G
Output Leakage Current
for F = 0 Output
YO. Yl. Y2. Y3
IOL - 20mA (COM'L)
0.5
10L = 16mA (MIL)
0.5
VIN = VIH
G. F =0
10L = 16mA
0.5
or VIL
Cn+4
10L = 10mA
0.5
OVR.P
10L = S.OmA
0.5
10L = 6.0mA
0.5
F3. RAMO. 3.
Units
2.4
10H - -6001'A
ICEX
Max.
2.4
Yo. Yl. Y2. Y3
VCC = MIN.
Typ.
(Note 2)
I'A
Volts
QO.3
Input HIGH Level
VIH
Guaranteed input logical HIGH
2.0
voltage for all inputs (Note 7)
VIL
Input LOW Level
VI
Input Clamp Voltage
voltage for all inputs (Note 7)
Input LOW Current
IlL
Volts
Guaranteed input logical LOW
Input HIGH Current
VCC = MAX .• VIN = 0.5V
AO. AI. A2. A3
-0.36
SO. Bl. B2. B3
-0.36
DO. 01. 02. 03
-0.72
10.ll. 12.16. I S
13. 14. 15. 17
-0.36
VCC = MAX .• VIN = 2.7V
-O.S
-~
Clock.OE
20
AO. AI. A2. A3
20
BO. Bl. B2. B3
20
00. 0 1. 0 2. 0 3
10. 11. 12. 16. IS
13. 14. 15. 17
20
40
RAMO. 3. QO. 3 (Note 4)
100
Cn
200
IOZL
Output Current
40
VCC = MAX .• VIN = 5.5V
1.0
Y2. Y3
Off State (High Impedance)
Vo = 2.4V
50
Vo - 0.5V
-50
Vo - 2.4V
VCC = MAX.
mA
-0.72
00. 3 (Note 4)
YO. Yl.
IOZH
Volts
-0.36
RAMO. 3.
Input HIGH Current
Volts
Clock.OE
Cn
IIH
O.S
-1.5
..
VCC = MIN .• liN = -ISmA
RAMO. 3
(Note 4)
QO.3
Vo = 0.5V
mA
100
-BOO
(Note 4)
lOS
Output Short Circuit Current
VCC = 5.75V. Vo = 0.5V
(Note 3)
YO. Yl. Y2. Y3. G
-30
Cn+4
OVR.P
-30
-85
-30
-85
F3
-30
-85
RAMO. 3. QO. 3
-30
Am2901APC. DC
Power Supply Current
ICC
(Note 6)
VCC= MAX.
(See graph)
-85
-85
TA-25°C
160
250
TA=0°Cto+70°C
160
265
TA-+70°C
160
220
160
280
160
190
TC - _55°C to
Am2901AOM. FM
+125°C
TC-+125°C
Notes: 1.
2.
3.
4.
mA
mA
For conditions shown as MIN. or MAX., use the appropriate value specified under Electrical Characteristics for the applicable device type.
Typical limits are at Vee = 5.0V, 2SoC ambient and maximum loading.
Not more than one output should be shorted at a time. Duration of the short circuit test should not exceed one second.
These are three~state outputs internally connected to TTL inputs. Input characteristics are measured with 1678 in a state such that the threestate output is OFF.
5. "MI L" = Am2901AXM. OM. FM. "COM'L" = Am2901AXC. PC. DC.
6. Worst case ICC is at minimum temperature.
7. These input levels provide zero noise immunity and should only be tested in a static, noise-free environment.
8-07
Am2901A
SWITCHING CHARACTERISTICS
OVER OPERATING RANGE
TABLE IV
CYCLE TIME AND CLOCK CHARACTERISTICS
Tables IV, V, and VI below define the timing characteristics of
the Am2901 A at 25°C over the operating voltage and temperature range_ The tables are divided into three types of parameters; clock characteristics, combinational delays from inputs
to outputs, and set-up and hold time requirements_ The later
table defines the time prior to the end of the cycle (Le_, clock
LOW-to-H IGH transition) that each input must be stable to
guarantee that the correct data is written into one of the
internal registers_
TIME
COMMERCIAL
MILITARY
100
110
15MHz
12MHz
Minimum Clock HIGH Time
30ns
30ns
30ns
30ns
Minimum Clock Period
lOOns
110ns
Read-Modify-Write Cycle
(time from selection of
A, B registers to end of
cycle)
Maximum Clock Frequency to
Shift Q Register (50% duty
cycle) I = 432 or 632
Measurements are made at 1.5V with VI L = OVand VIH = 3.0V.
For three-state disable tests, CL = 5.0pF and measurement is
to O.5V change on output voltage level.
Minimum Clock LOW Time
Commercial = Am2901APC, DC, XC
O°C to +70°C
4.75 to 5.25V
Military =
Am2901ADM, FM, XM
-55°C to +125°C
4.50 to 5.50V
TABLE V
COMBINATIONAL PROPAGATION DELAYS (all in ns, CL = 50pF (except output disable tests))
COMMERCIAL
~
F
Output
rom
Input
F=O
Y
F3
C n +4 G,P RL= OVR
470
o (arithmetic mode)
80
45
80
45
40
75
45
-
65
35
-
D (I = X37) (Note 5)
Cn
40
30
20
50
-
55
30
55
1012
1345
55
55
1678
30
A,B
OE Enable/Disable
A bypassing
ALU (I = 2xx)
Clock
S
(Note 6)
C n +4
G,Y-
85
50
85
50
70
40
100
65
45
-
35
60
25
55
-
-
45
35
60
80
50
-
50
60
55
75
-
03
100
70
-
-
35
70
65
55
80
-
75
70
80
-
-
-
-
35
35
-
-
-
-
-
-
-
-
-
-
65
55
85
75
85
35
95
65
30
45
55
50
70
65
60
50
75
55
50
70
65
75
-
60
60
60
55
-
-
-
-
-
30
30
35
-
35/25
-
-
-
-
-
-
-
40/25
-
-
45
-
-
-
-
-
-
-
50
-
60
60
60
50
75
70
80
30
65
65
2,4
3,5
2,4
-
-
90
60
-
100
tpw L+3O
tpw L+16
0
110
tpw L+3O
tpw L+16
Hold Time
0
1012
0
85
1345
80
0
85
tpw L+ 3O
0
tpw L+ 3O
0
25
0
25
0
4
00, 3
0
0
0
0
Set-Up Time
70
60
55
80
1678
RAMO, 3,
-
MILITARY
Hold Time
0
0
0
0
0
0
D (arithmetic mode)
0(1 = X37) (Note 5)
Cn
-
TABLE VI
COMMERCIAL
Set-Up Time
A,B
Source
B Dest.
F3
00
85
55
Notes
Shift
Outputs
F=O
RL= OVR
RAMO 00
470
RAM3 03
Y
RAMO
RAM3
90
60
SET-UP AND HOLD TIMES (all in ns) (Note 1)
From Input
MILITARY
Shift
Outputs
75
65
60
Notes: 1. See Figure 1"
2_ If the 8 address is used as a source operand, allow for the "A, 8 source" set-up time; if it is used only for the destination address, use the
"8 Dest" set-up time.
3. Where two numbers are shown, both must be met.
4. "tpwL" is the clock LOW time.
5. D V 0 is the fastest way to load the RAM from the D inputs. This function is obtained with 1=337.
6. Using Q register as source operand in arithmetic mode. Clock is not normally in critical speed path when Q is not a source.
8-08
Am2901A
SET-UP AND HOLD TIMES (minimum cycles from each input)
time prior to the clock until the hold time after the clock. The
set-up times allow sufficient time to perform the correct
operation on the correct data so that the correct ALU data
can be written into one of the registers.
Set-up and hold times are defined relative to the clock LOW-toHIGH edge. Inputs must be steady at ali times from the set-up
30
lOll
A.B
80
Figure 11. Minimum Cycle Times from Inputs. Numbers Shown are Minimum Data
Stable Times for Am2901ADC, in ns. See Table III for Detailed Information.
Typical Icc Versus Temperature
300
200
r---,----,-,-,---...,
.............. ,
____
1501---+--+---+"""'-+---l
TYPICAL ICC ~
VCC~15.0IV
100 '------'-:---'--'----'_----'
··55
0 25
70
125
T -
°c
TEMPERATURE _
Figure 12.
8-09
°c
Am2901B
PRELIMINARY DATA
ELECTRICAL CHARACTERISTICS OVER OPERATING RANGE
(Group
A,
Subgroups
1, 2,
and
(Unless Otherwise Noted)
3)
Typ
Test Conditions (Note 1)
Description
Parameters
Min
10H = -1.6mA
(Note 21
Max
Units
2.4
Yo, Yl, Y2, Y3
Output HIGH Voltage
VOH
VCC = MIN.
VIN = VIH or VIL
2.4
IOH = -1.0mA. Cn +4
10H = -8001lA, OVR, P
2.4
10H - -6001lA, F3
2.4
10H = -6001lA
Volts
2.4
RAMO, 3, QO, 3
10H = -1.6mA, G
ICEX
Output Leakage Current
for F = 0 Output
250
~L
VCC = MIN ..
Output LOW Voltage
VOL
2.4
VCC = MIN., VOH = 5.5V
VIN = VIH or VIL
YO, Yl, Y2, Y3
0.5
10L = 16mA (Mill
VIN = VIH
G, F = 0
10L
Cn+4
OVR,P
= 16mA
10L = 10mA
0.5
or VIL
10L = 8.0mA
0.5
10L = 6.0mA
0.5
F3, RAMO, 3,
IlA
0.5
- 20mA ICOM'L1
0.5
Volts
QO,3
VIH
Input HIGH Level
VIL
Input LOW Level
VI
Input Clamp Voltage
Guaranteed input logical HIGH
Volts
2.0
voltage for all inputs (Note 71
Guaranteed input logical LOW
voltage for all
Input LOW Current
IlL
input~
VCC = MAX., VIN = 0.5V
Clock,OE
-0.36
-0.36
BO, Bl, B2, B3
-0.36
DO, Dl, D2, D3
-0.72
10,11,12,16, IS
-0.36
13,14,15,17
-0.72
00, 3
(Note 4)
Input HIGH Current
10ZH
Off State (High Impedance)
10ZL
Output Current
VCC = MAX., VIN = 2.7V
-O.B
20
20
BO, Bl, B2, B3
20
DO, Dl, D2, D3
40
10,11,12,16. IS
20
13,14,15,17
40
RAMO, 3, QO, 3 (Note 4)
100
Cn
200
VCC - MAX., VIN = 5.5V
1.0
YO, Yl,
Vo = 2.4V
50
Y2, Y3
Vo - 0.5V
-50
RAMO, 3
Vo 2.4V
(Note 4)
QO.3
VO= 0.5V
VCC = MAX.
Output Short Circuit Current
VCC = MAX. + 0.5V, Vo = 0.5V
(Note 3)
-SOO
~30
Cn+4
OVR,P
-30
-85
-30
-85
F3
-30
-85
00, 3
TA-25°C
Power Supply Current
(Note 6)
ICC
Am2901 BPC. DC
VCC = MAX.
ISee Fig. 12)
-85
YO, Yl, Y2, Y3, G
RAMO, 3,
TA - O°C to +70°C
TA - +70°C
-30
+125°C
TC - +125°C
Notes: 1.
2.
3.
4.
mA
-85
160
250
265
220
TC = -55°C to
Am2901BDM. FM
mA
100
(Note 4)
lOS
mA
-3.6
AO, Al, A2, A3
II
Volts
AO, Al, A2, A3
Cn
Clock,OE
Input HIGH Current
Volts
-1.5
(Note 7)
VCC = MIN., liN = -lSmA
RAMO, 3,
IIH
O.S
mA
265
19S
For conditions shown as MIN. or M"AX .• USti the appropriate value specified under Electrical Characteristics for the applicable device type.
Typical limits are at Vee = 5.0V, 25 C ambient and maximum loading.
Not more than one output should be shorted at a time. Duration of the short circuit test should not exceed one second.
These are three-state outputs internally connected to TTL inputs_ Input characteristics are measured with 1678 in a state such that the three
state output is 0 F F.
5. "MIL" = Am29018XM, OM, FM. "COM'L = Am2901 axc,·pc, DC.
6. Worst case ICC is at minimum temperature.
7. These input levels provide zero noise immunity and should only be tested in a static, noise-free environment.
8-D10
Am2901B
I. Typical Room Temperature Performance
A. Cycle Time and Clock Characteristics.
The tables below specify the typical performance of the Am2901 B
at 25°C and 5.0V. All data are in ns, with inputs changing between
OV and 3V at 1V/ns and measurements made at 1.5V. For
guaranteed data, see following pages.
Read-Modify-Write Cycle (from selection of A, B registers
to end of cycle.)
Maximum Clock Frequency to shift
I = 432 or 632)
a (50'7r
Minimum Clock LOW Time
duty cycle,
45ns
33MHz
lOns
Minimum Clock HIGH Time
lOns
Minimum Clock Period
SOns
B. Combinational Propagation Delays.
CL
SOpF
=
~
From Input
Y
A, B Address
0
F3
Cn+4
G,P
F=O
OVR
RAMO
RAM3
00
03
38
41
39
33
44
44
50
22
23
24
20
28
29
31
-
Cn
17
19
13
-
22
19
26
1012
30
30
29
27
34
34
38
-
1345
32
32
30
25
32
30
34
-
1678
17
-
-
-
-
-
16
16
A Bypass ALU
(I = 2XX)
22
-
-
-
-
-
-
-
ClockS
29
31
29
23
33
35
40
19
C. Set-up and Hold Times Relative to Clock (CP) Input.
-'
CP:
Input
A, B Source Address
~-- ~_-yI
Set...up Time
Ho!d Time
Set-up Time
Before H -+L
After H -+ L
Before L -+ H
After L -+ H
Ho!d Time
8
o (Note 3j
45 (Note 4)
0
B Destination
Address
4
0
-
35
0
Cn
-
26
0
1012
-
37
0
1345
-
-
38
0
1678
0
RAMO, 3, QO, 3
-
0
Do Not Change
po Not Change
-
0
9
0
D. Output Enable/Disable Times.
Output disable tests performed with CL = SpF and
measured to O.SV change of output voltage level.
Notes: 1. A dash indicates a propagation delay path or set-up time constraint does not exist.
2. Certain signals must be stable during the entire clock LOW time to avoid erroneous operation. This is indicated by the phrase "do not change".
3. Source addresses must be stable prior to the clock H -+ L transition to allow time to access the source data before the latches close. The A
address may then be changed. The B address could be changed if it is not a destination; i.e. if data is not being written back into the RAM. Nonnally
A and B are not changed during the clock LOW time.
4. The set-up time prior to the clock L -+ H transition is to allow time for data to be accessed, passed through the ALU, and retumed to the RAM. It
includes all the time from stable A and B addresses to the clock L -+ H transition, regardless of when .the clock H -+ L transition occurs.
8-D11
Am29018
I. Guaranteed Commercial
Range Performance
A. Cycle Time and Clock Characteristics.
Read-Modify-Write Cycle (from selection of A, B registers
to end of cycte.)
The tables below specify the guaranteed performance of the
Am2901 B over the commercial operating range of O°C to + 70°C,
with Vcc from 4.7SV to S.2SV. All data are in ns, with inputs
switching between OV and 3V at 1 V/ns and measurements made
at 1.SV_ All outputs have maximum DC load.
Maximum Clock Frequency to shift
I = 432 or 632)
This data applies to the following part numbers: Am2901BPC
Am2901BDC
a (50'7.
duty cycle.
69ns
16MHz
Minimum Clock LOW Time
30ns
Minimum Clock HIGH Time
30ns
Minimum Clock Period
69ns
B. Combinational Propagation Delays.
CL = 50pF
~
From Input
V
F3
Cn+4
G,P
F=O
OVR
RAMO
RAM3
00
03
A. B Address
60
61
59
50
70
67
71
-
0
38
36
40
33
48
44
45
-
Cn
30
29
23
-
37
29
38
-
1012
50
47
45
45
56
53
57
-
1345
49
48
44
45
54
49
53
-
1678
28
-
-
-
-
-
27
27
A Bypass ALU
(I = 2XX)
37
-
-
-
-
-
-
-
ClockS
49
48
47
37
58
55
59
29
C. Set-up and Hold Times Relative to Clock (CP) Input.
~K:---- ~
CP:
Input
_------:1
L-
Set-up Time
Before H'" L
Hold Time
After H'" L
Set-up Time
Before L'" H
Hold Time
After L'" H
A. B Source Address
20
o (Note 3)
69 (Note 4)
0
B Destination
Address
9
0
-
-
51
0
Cn
-
-
39
0
1012
-
-
56
0
-
55
0
1345
r----
11
1678
RAMO. 3,
aD.
3
-
0
Do Not Change
Do Not Change
-
0
16
0
D. Output Enable/Disable Times.
Output disable tests performed with C L = SpF and
measured to O.SV change of output voltage level.
Notes: 1. A dash indicates a propagation delay path or set-up time constraint does not exist.
2_ Certain signals must be stable during the entire clock LOW time to avoid erroneous operation. This i" indicated by the phrase "do not change"
3. Source addresses must be stable prior to the clock H ... L transition to allow time to access the source data before the latches close. The A
address may then be changed. The B address could be changed if it is not a destination; i.e. if data is not being written back into the RAM. Normally
A and B are not changed during the clock LOW time.
4. The set-up time prior to the ~Iock L ... H transition is to allow time for data to be accessed, passed through the ALU, and retumed to the RAM. It
includes all the time from stable A and B addresses to the clock L'" H transition, regardless of when the clock H'" L transition occurs.
8-D12
Am29018
II. Am2901 B Guaranteed Military
Range Performance
A. Cycle Time and Clock Characteristics.
Read-Modily-Write Cycle (from selection of A, B registers
to end of cycle.
The tables below specify the guaranteed performance of the
Am2901 B over the military operating range of -55°C to + 125°C,
with V cc from 4.SV to S.SV. All data are in ns, with inputs switching between OV and 3V at 1 V/ns and measurements made at
1.SV. All outputs have maximum DC load.
Maximum Clock Frequency to shift
I ; 432 or 632)
This data applies to the following part numbers: Am2901 BDM
Am2901BFM
a
(50 rl, duty cycle,
88ns
15MHz
Minimum Clock LOW Time
30ns
Minimum Clock HIGH Time
30ns
Minimum Clock Period
88ns
B. Combinational Propagation Delays.
CL
SOpF
=
~
From Input
Y
F3
Cn+4
G,P
QO
F=O
OVR
RAMO
RAM3
Q3
A. B Address
82
84
80
70
90
86
94
-
D
44
38
40
34
50
45
48
-
Cn
34
32
24
-
38
31
39
-
1012
53
50
47
46
59
55
58
-
1345
53
50
46
45
58
50
55
-
1678
29
-
-
-
-
-
27
27
50
-
-
-
-
-
-
-
53
50
49
41
63
58
61
31
A Bypass ALU
(I; 2XX)
ClockS
~-
C. Set-up and Hold Times Relative to Clock (CP) Input.
~~--
CP:
Input
A, B Source Address
--
~r
Set-up Time
Before H -+ L
Hold Time
After H -+ L
Set-up Time
Before L -+ H
Hold Time
After L -+ H
20
o (Note 3)
88 (Note 4)
0
B Destination
Address
9
D
-
-
55
0
Cn
-
-
42
0
1012
-
-
58
0
1345
-
-
58
0
1678
14
RAMO, 3, ao, 3
-
Do Not Change
0
Do Not Change
-
0
18
3
L-_
D. Output Enable/Disable Times.
Output disable tests performed with C L = SpF and
measured to O.SV change of output voltage level.
Notes: 1. A dash indicates a propagation delay path or set-up time constraint does not exist.
2. Certain signals must be stable during the entire clock LOW time to avoid erroneous operation. This is indicated by the phrase "do not change".
3. Source addresses must be stable prior to the clock H -+ L transition to allow time to access the source data before the latches close. The A
address may then be changed. The B address could be changed if it is not a destination; Le. if data is not being written back into the RAM. Normally
A and B are not changed during the clock LOW time.
4. The set-up time prior to the clock L -+ H transition is to allow time for data to be accessed, passed through the ALU. and retumed to the RAM. It
includes all the time from stable A and B addresses to the clock l-+ H transition. regardless of when the clock H -+ L transition occurs.
8-D13
Am2902A
MAXIMUM RATINGS
(Above which the useful life may be impaired)
Storage Temperature
Temperature (Ambient) Under Bias
Supply Voltage to Ground Potential
-O.5V·to +7.0V
DC Voltage Applied to Outputs for HIGH Output State
-0.5V to +VCC max.
DC Input Voltage
-O.5V to +5.5V
DC Output Current, I nto Outputs
30 rnA
DC Input Current
-30 rnA to +5.0 rnA
ELECTRICAL CHARACTERISTICS OVER OPERATING TEMPERATURE RANGE
Am2902Axe
TA
Am2902AXM
TA
Parameters
= oOe to +7oo e
= _55°e to +125°e
Vee = 5.0V ±5% (eOM'L)
Vee = 5.0V ±10% (MIL)
Description
MIN. = 4.75V
MIN. = 4.50V
Test Conditions INote 1)
Vcc = MIN., IOH = -lmA
VIN = VIH or VIL
VOH
Output HIGH Voltage
VOL
Output LOW Voltage
Vcc = MIN., IOL = 20mA
VIN = VIH or VIL
VIH
Input HIGH Level
Guaranteed input logical HIGH
voltage for all inputs
VIL
Input LOW Level
Guaranteed input logical LOW
voltage for all inputs
VI
Input Clamp Voltage
Vee, = MIN., liN
I
1
(Unless Otherwise Noted)
MAX. = 5.25V
MAX. = 5.50V
Typ.
Min.
(Note 2)
MIL
2.5
3.4
COM
2.7
3.4
Max.
Volts
0.5
2.0
Input LOW Current
= -18mA
VCC = MAX., VIN = 0.5V
Input HIGH Current
VCC
= MA;(.,
VIN = 2.7V
II
Input HIGH Current
VCC = MAX., VIN = 5.5V
ISC
Output Short Circuit
(Note 3)
VCC = MAX., VOUT
Power Supply Current
ICC
Notes:
0.8
Volts
-1.2
Volts
-2
P3
-4
P2
-6
-8
PO, Pl, G3
IIH
GO, G2
-14
Gl
-16
Cn
50
P3
100
P2
150
PO, Pl, G3
200
GO,G2
350
Gl
400
= O.OV
-40
mA
-100
mA
VCC= MAX.
MIL
69
99
All Outputs LOW
COM'L
69
109
VCC= MA".
All Ouputs HIGH
MIL
35
COM'L
35
Description
Min.
Typ.
Max.
4.5
7
G"/Pi --+ Cn+i
tpHL
4.5
7
tplH
5
7.5
7
10.5
4.5
6.5
tpHL
6.5
10
tplH
6.5
10
7
10.5
Gi/Pi--+
G
tpHL
tpHL
J.lA
mA
mA
1. For conditions shown as MIN. or MAX., use the appropriate value specified under Electrical Characteristics for the applicable device type.
2. Typical limits are at VCC = 5.0V, 25°C ambient and maximum loading.
3. Not more than one output should be shorted at a time. Duration of the short circuit test should not exceed one second.
Parameters
tplH
mA
1.0
SWITCHING CHARACTERISTICS
(TA = +25°C, Vee = 5.0V)
tplH
Volts
Volts
Cn
IlL
Units
Pi --+ j5
C n --+ Cn+i
8-D14
Units
Test Conditions
ns
ns
I
ns
ns
CL
RL
I
=
=
15pF
2800
Am2903
Am2903
OPERATING RANGE
PIN
Vce
Temperature
TA
= O'C to + 70'e
Te = -55'Cto +125'C
Vee= 5.0V ::!:5o/r
(MIN. = 4.75V. MAX. = 5.25V)
Vee = 5.0V ±10o/r
(MIN. = 4.50V. MAX. = 5.50V)
DC CHARACTERISTICS OVER OPERATING RANGE
Parameters
Description
Output HIGH Voltage
VO H
Output Leakage Current
for Z Output (Note 4)
ICEX
Typ.
Vee = MIN.
VIN = V IH or V IL
10H = =-1.6mA
YO-Y3 • GIN
2.4
10H = =-SOO/LA
OB O_3 . PIOVA
SIOo. S10 3. 0100. 0103.
WAITE. C n +4
2.4
Output LOW Voltage
(Note 2)
Vee = MIN.
VIN = VIH = or V IL
Max.
Units
Vo~s
Vee = MIN .• VO H = 5.5V
V IN = VIH or VIL
250
IOL = 20mA (COM'L)
Yo. Y,. Y2
Y3. Z
VOL
Min.
Test Conditions (Note 1)
0.5
10L = 16mA (MIL)
10L = 12mA (COM'L)
DBa. DB,.
OB2. 0B 3
10L = S.OmA (MIL)
GIN
10L = lSmA
P/OVA
10L = lOmA
/LA
0.5
I
0.5
Volts
0.5
C n + 4 • SIOo
S103.0100
10L = S.OmA
0103. WRITE
0.5
I
Input HIGH Level
. Guaranteed input logical HIGH
voltage for all inputs (Note '6)
VIL
Input LOW Level
Guaranteed input logical LOW
voltage for all inputs (Note 6)
O.S
Volts
VI
Input Clamp Voltage
Vee = MIN .• liN = -lSmA
-1.5
Volts
V IH
Input LOW Current
IlL
VCC = MAX .• V IN = 0.5V
(Note 4)
Volts
2.0
Cn
2.50
Yo. Y,. Y2. Y3
LOS
10 • I,. 12• 13•
14. OAo. OA,.
OA 2. OA3 • SIOo
S103. 0100. 0103.
MSS. DBa. DB,.
OB 2.OB3
0.72
All other inputs
0.36
mA
--'".
Input HIGH Current
IIH
Vce = MAX .. V IN = 2.7V
(Note 4)
Cn
120
Yo. Y,. Y2. Y3
110
10-14. OAO-OA 3
SIO o. S10 3• 0100.
0103. OBo_3•
MSS
40
/LA
90
All other inputs
In!'lut HIGH Current
II
Off State
10ZH
(HIGH Impedance)
Output Current
10ZL
los
Output Short Circuit
Current (Note 3)
ICC
Power Supply Current
(Note 5)
20
Vec = MAX .• V IN = 5.5V
Vec = MAX .•
(Note 4)
1.0
YO-Y3
OBO_3• 0100. 0100.
SIO o • 8103. MS8/1S
Vo = 2.4V
110
Vo = 0.5V
-1130
Vo = 2.4V .
-770
-30
Vo = 0.5V
TA = 25'C
COM'L
Vee = MAX.
MIL
Notes: t.
2.
3.
4.
-85
220
TA = 010 70'C
/LA
90
Vo = 0.5V
Vce = MAX + 0.5V
mA
mA
335
350
TA = 70'C
291
Te = -55 to 125'C
395
Te = 125'C
25S
mA
For conditions shown as MIN. or MAX.. use the appropriate value specified under Electrical Characteristics for the applicable device type.
Typical limits are at vee ~ 5.0V. 25'C ambient and maximum loading.
Not more than one output should be shorted at a time. Duration of the short circuit test should not exceed one second.
YO-3. 060-3. 510 0.3. Q100.3 and WRITE/M5S are three state outputs internally connected to TTL inputs. Z is an open-collector output internally
connected to a TTL input. Input characteristics are measured under conditions such that the outputs are in the OFF state.
5. Worst case Icc is at minimum tenlperature.
6. These input levels provide zero noise immunity and should only be tested in a static. noise-free environment.
8-D15
Am2903
SWITCHING CHARACTERISTICS (Typical Room Temperature Performance)
Tables I, II, and III define the nominal timing characteristics of
the Am2903 at 25°C and 5.0V. The Tables divide the
parameters into three types: pulse characteristics for the clock
and write enable, combinational delays from input to output,
and set-up and hold times relative to the clock and write
pulse.
TABLE I
Write Pulse and Clock Characteristics
Time
Measurements are made at 1.5V with V1L = OV and V1H =
3.0V. For three-state disable tests, CL '= 5.0pF and measurement is to 0.5V change on output voltage level.
Minimum Time CP and WE both LOW
to write
30ns
Minimum Clock LOW Time
30ns
Minimum Clock HIGH Time
50ns
TABLE II
Combinational Propagation Delays, All In ns.
Outputs Fully Loaded. CL = SOpf (except output disable tests)
~t
From Input
SI03
SI03 (Parity)
Y
Cn + 4
G,P
Z
N
OVR
DB
WRITE
QIOo, QI0 3
SIOo
A, B Addresses
(Arith. Mode)
65
60
55
75
64
70
33
-
-
61
69
87
A, B Addresses
(Logic Mode)
56
-
46
67
56
-
33
-
-
55
64
81
DA, DB Inputs
(Logic Mode)
39
-
25
48
38
-
-
-
-
36
47
56
DA, DB Inputs
(Arith. Mode)
39
37
26
52
38
51
-
-
-
36
47
60
EA
44
38
29
54
44
53
-
-
-
42
52
Cn
25
21
-
39
20
38
-
-
--
21
25
10
39
35
24
48
37
48
-
"15
-
41
46
14321
45
43
32
55
44
55
-
"17
-
45
51
18765
25
-
-
37
-
-
-
18
22
24
27
48
lEN
-
-
-
-
-
-
-
10
-
-
-
-
OEB Enable/Disable
-
-
-
-
-
-
7
-
-
-
-
-
OEY Enable/Disable
10
-
-
-
-
-
-
-
-
-
-
-
SIO o, SI03
13
-
-
-
-
-
-
-
-
-
12
18
Clock
58
52
40
72
56
72
24
-
28
55
63
76
'Applies only when leaving special functions.
TABLE III
Set-Up and Hold Times (All in ns)
CAUTION: READ NOTES TO TABLE III.
NA = Not Applicable; no, timing constraint.
HIGH-ta-LOW
Input
With Respect to
to this Signal
LOW-ta-HIGH
~-~
Set-up
Hold
Set-up
Hold
I
Comment
y
Clock
NA
NA
10
0
To store Y in RAM or 0
WE HIGH
Clock
5
Note 2
Note 2
0
To Prevent Writing
WE LOW
Clock
NA
NA
30
0
To Write into RAM
A,S as Sources
Clock
20
0
NA
NA
--
--
See Note 3
B as a Destination
Clock and WE both LOW
0
Note 4
Note 4
0
To Write Data only into
the Correct B Address
0100 ,0103
Clock
NA
NA
10
0
To Shift 0
18765
Clock
30
Note 5
Note 5
0
lEN HIGH
Clock
10
Note 2
Note 2
0
To Prevent Writing
lEN LOW
Clock
NA
NA
10
0
To Write into 0
8-016
Am2910
MAXIMUM RATINGS (Above which the useful life may be impaired)
Storage Temperature
Temperature (Ambient) Under Bias
Supply Voltage to Ground Potential
DC Voltage Applied to Outputs for High Output State
-0.5V to +7.0V
-0.5 V to Vee max.
-0.5V to +5.SV
DC Input Voltage
DC Output Current. Into Outputs
30mA
-30 mA to +5.0 mA
DC Input Current
ELECTRICAL CHARACTERISTICS The Following Conditions Apply Unless Otherwise Specified:
o
MAX. = 5.25V
MIN. = 4.75V
T A = oOe to +70 e
Vee = 5.0V ±5%
MAX. = 5.50V
MIN. = 4.50V
Vee = 5.0V 110%
Te = -55°e to +125°e
Mil
DC CHARACTERISTICS OVER OPERATING RANGE
eOM'l
ParameJers
Description
Test Conditions (Note 1)
Vec = MIN .• IOH = -1.6mA
VIN = VIH or Vil
.-f-Vce = MIN.
YO-l1. IOl
Output HIGH Voltage
vOH
I
Output LOW Voltage
VOL
VIN
Typ.
(Note 2)
Max.
VECT, MAP, FUll, IOl
Units
Volts
2.4
= 12mA
0.5
Volts
= SmA
-'+GUaranteed Input Logical HIGH
voltage for all inputs
Input HIGH Level (Note 4)
VIH
I PL.
= VIH or Vil
Min.
Volts
2.0
- 1--
Vil
Input lOW level (Note 4)
VI
Input elamp Voltage
Guaranteed input logical lOW
voltage for all inputs
--
Input lOW Current
III
Vce
VCC
= MIN.,
= -lSmA
liN
= MAX.,
O.S
= 0.5V
VIN
-1.5
00-11
-0.S7
el. CCEN
-0.54
10-3,~,R05
CC
-0.72
-1.31
CP
-2.14
Input HIGH Current
Vec
= MAX.,
= 2.7V
VIN
II
Input HIGH Current
VCC = MAX., VIN
ISC
Output Short Circuit Current
(Note 3)
Vec=MAX.
Output OFF Current
Vec = MAX.
OE = 2.4V
IOZL
IOZH
--_ ... _-_ .. _---
----
CI, CCEN
30
10-3, OE. RlO
40
CC
50
CP
100
= 5.5V
-30
VOUT
= 0.5V
VOUT - 2.4V
Am291 OPC, DC
Power Supply Current
Vec = MAX
IJ.A
mA
-S5
mA
IJ.A
SO
19S
320
T A - o°c to +70°C
344
TA = +70·C
2S0
Te = -SSoC to
Am29100M, FM
mA
1.0
-SO
TA = 25°e
ICC
Volts
SO
00-11
IIH
Volts
.-
mA
340
+125°C
Te = +12Soe
227
Notes: 1. For conditions shown as MIN. or M AX .. use the appropriate value specified under Electrical Characteristics for the appl icable devic.8 type.
2. TYPical limits are at Vee = 5.0V. 25°e ambient and maximum loading.
3. Not mort! than one output should be shorted at a time. Duration of the short circuit test should not exceed one second.
4. These input levels provide no guaranteed noise immUnity and should only be tested
In
a static·, noise-free environment.
INPUTS
3.0"
I
L
INPUT
I
CLOCK
TO
--OUTPUT
DELAV
TO
OUTPUT
DELAY
See Tables A for ts and th for various
inputs. See Tables B for combinational
delays from clock and other inputs to
outputs. See Figure 5 for timing of a
typical CCU cycle.
-
Figure 2. Switching Waveforms.
8-017
Am2910
SWITCHING CHARACTERIST1CS
The tables below define the Am291 0 switching characteristics. Tables A are set-up and hold times relative to the clock LOW-to-H IGH
transition. Tables B are combinational delays. Tables e are clock requirements. All measurements are made at 1.5V with input levels
at OV or 3V. All values are in ns.
TYPICAL ROOM TEMPERATURE CHARACTERISTICS (T A = 25°C, Vee
Input
ts
th
Di~R
9
3
Di~PC
34
1
10- 13
64
0
CC
46
0
CCEN
49
LI
CI
26
2
RLD
18
2
Input
DO-011
Y
PL, VECr, MAP
Full
14
10- 13
40
CC
21
CCEN
23
29
CP
All other I
29
26
GUARANTEED ROOM TEMPERATURE CHARACTERISTICS lTA
ts
th
Di~R
Y
Pl, VECr, MAP
FiiIf
CCEN
CCEN
= 25°e, Vee = 5.0V, CL = 50pF)
Minimum Clock LOW Time
(Clock periods for other instructions are
determined by external conditions.!
CP(note)
1= 8, 9, 15
CI
RLD
Note: These instructions are conditional on the
counter. Delays from CP to outputs will be
longer if the instruction prior to the clock was
4 or 12 or RLD was LOW.
Minimum Clock Period, 1=14
CC
CC
(Clock periods for other instructions a'e
determined by external conditions'!
Minimum Clock Period, 1=8,9,15
10- 13
10- 13
30
Minimum Clock HIGH Time
DO-Oll
Di~PC
30
Minimum Clock HIGH Time
C. Clock Requirements
B. Combinational Delays
Input
Minimum Clock LOW Time
Minimum Clock Period, 1=14
OE
Input
= 50pF)
Minimum Clock Period, 1=8.9, 15
27
CP(note)
54
1= 8, 9,15
A. Set-up and Hold Times
CL
C. Clock Requirements
B. Combinational Delays
A. Set-up and Hold Times
= 5.0V,
Note: These instructions are conditional 0:1 the
counter. Delays from CP to outputs will b,~
longer if the instruction prior to the clock was
4 or 12 or RLD was LOW.
CP
All other I
OE
GUARANTEED CHARACTERtSTICS OVER COMMERCIAL OPERATING RANGE
Am2910PC, DC (TA = oOe to+70°C, Vec = 4.75V to 5.25V, eL = 50pF)
Input
ts
th
Di~R
15
5
Di~PC
58
3
10- 13
100
0
CC
80
0
CCEN
85
0
CI
45
5
RLD
35
3
C. Clock Requirements
B. Combinational Delays
A. Set·up and Hold Times
Full
Minimum Clock LOW Time
50
-
-
Minimum Clock HIGH Time
30
45
48
-
-
40
-
-
CP(note)
92
1= 8, 9,15
-
-
CP
All other I
-
50
-
-
Input
Y
Do-D11
25
10-13
65
CC
ei!EN
65
OE
Pl, VECT, MAP
-
Minimum Clock Period, 1=8, 9,15
Minimum Clock Period, 1=14
(Clock periods for other instructions are
determined by external conditions.)
Note: These instructions are conditional on the
counter. Delays from CP to outputs will be
longer if the instruction prior to the clock was
4 or 12 or RLD was LOW.
GUARANTEED CHARACTERISTICS OVER MILITARY OPERATING RANGE
Am2910DM, FM (TC = _55°C to +125°C, Vee = 4.5V to 5.5V, CL = 50pF)
Input
Di~R
Di~PC
10. 13
CC
CCEN
CI
RLD
ts
th
C. Clock Requirements
B. Combinational Delays
A. Set·up and Hold Times
Input
Y
Pl, VECT, MAP
Full
Minimum Clock LOW Time
Minimum Clock HIGH Time
00- 0 11
Minimum Clock Period, 1=8,9, 15
10- 13
Minimum Clock Pellod, 1=14
CC
CCEN
(Clock p~riods for other instructions are
determined by external conditions. I
CP(note)
1= 8, 9, 15
CP
All other I
Note: These instructions are conditional on the
counter. Delays from CP to outputs will be
longer if the instruction prior to the clock was
4 or 12 or RLD was LOW.
OE
8-D18
Am2909 • Am2911
for the control of the push/pop stack. Figure 6 shows in detail
the effect of So, S1, FE and PUP on the Am29Q9. These four
signals define what address appears on the Y outputs and what
the state of all the internal registers will be following the clock
LOW-to-HIGH edge. In this illustration, the microprogram
counter is assumed to contain initially some word J, the address register some word K, and the four words in the push/
pop stack contain Ra through Rd.
OPERATION OF THE Am2909/Am2911
Figure 5 lists the select codes for the multiplexer. The two
bits applied from the microword register (and additional com·
binational logic for branching) determine which data source
contains the address for the next microinstruction. The
contents of the selected source will appear on the Y outputs.
Figure 5 also shows the truth table for the output control and
Output Control
Address Selection
OCTAL
S,
So
0
1
L
L
H
H
L
H
L
H
2
3
ORi
SOURCE FOR Y OUTPUTS SYMBOL
Microprogram Counter
Register
Push-Pop stack
Direct inputs
ZERO
OE
Vi
X
H
L
L
L
Z
X
X
J,LPC
REG
STKO
Di
L
H
H
H
L
L
H
Source selected by So 5,
Z = HIgh Impedance
Synchronous Stack Control
H
L
X
= High
= Low
= Don't Care
FE
PUP
PUSH-POP STACK CHANGE
H
L
X
H
L
L
No change
Increment stack pOil'lter, then
push current PC onto 5TKO
Pop stack (decrement stack pointer)
Figure 5.
CYCLE
FE, PUP
S1, So,
REG
J
J+1
K
K
Ra
Rb
Rb
Rc
Rc
Rd
Rd
Ra
-
Pop Stack
End
Loop
J
J+1
K
K
Ra
J
Rb
Ra
Rc
Rb
Rd
Rc
J
-
Push J,lPC
Set-up
Loop
J
J+1
K
K
Ra
Ra
Rb
Rb
Rc
Rc
Rd
Rd
-
Continue
Continue
J
K+1
K
K
Ra
Rb
Rb
Rc
Rc
Rd
Rd
Ra
K
-
Pop Stack;
Use AR for Address
End
Loop
J
K+1
K
K
Ra
J
Rb
Ra
Rc
Rb
Rd
Rc
-
Push J,lPC;
.Jump to Address in AR
JSR AR
J
K+1
K
K
Ra
Ra
Rb
Rb
Rc
Rc
Rd
Rd
K
-
Jump to Address in AR
JMP AR
J
Ra+1
K
K
Ra
Rb
Rb
Rc
Rc
Rd
Rd
Ra
Ra
Jump to Address in STKO;
Pop Stack
RTS
K
K
Ra
J
Rb
Ra
Rc
Rb
Rd
Rc
Ra
-
Jump to Address in STKO;
Push J,lPC
Ra
Jump to Address in STKO
Stack Ref
(Loop)
Pop Stack;
Jump to Address on D
End
Loop
STKO STK1 STK2
STK3 YOUT
PRINCIPLE
USE
J,lPC
COMMENT
N
N+1
o0
N
N+1
000 1
N
N+1
001 X
N
N+1
o
N
N+1
o1
N
N+1
o1
N
N+1
1 000
N
N+1
1 001
J
-
Ra+1
N
N+1
101 X
-
Ra+1
K
K
Ra
Ra
Rb
Rb
Rc
Rc
Rd
Rd
N
N+1
1 1 0 0
J
D+1
K
K
Ra
Rb
Rb
Rc
Rc
Rd
Rd
Ra
-
N
N+1
1 1 0 1
J
D+1
K
K
Ra
J
Rb
Ra
Rc
Rb
Rd
Rc
-
Jump to Address on D;
Push J,lPC
JSR D
N
N+1
1 1 1 X
J
K
K
Ra
Ra
Rb
Rb
Rc
Rc
Rd
Rd
D
-
Jump to Address on D
JMP D
0 0
-
1 0 0
0 1
1 X
-
-
-
J
D+1
J
J
K
-
D
D
x = Don't care, 0 = LOW, , = HIGH, Assume C n = HIGH
Note: STKO is the location addressed by the stack pointer.
Figure 6. Output and Internal Next-Cycle Register States for Am2909/Am2911.
8-019
Am2909 • Am2911
Figure 7 illustrates the execution of a subroutine using the
Am2909. The configuration of Figure 3 is assumed. The
instruction being executed at any given time is the one contained in the microword register (J.tWR). The contents of the
IlWR also controls (indirectly, perhaps) the four signals So, S1,
FE, and PUP. The starting address of the subroutine is applied
to the 0 inputs of the Am2909at the appropriate time.
routine at A". At the time T 2, this instruction is in the IlWR,
and the Am2909 inputs are set-up to execute the jump and
save the return address. The subroutine address A is applied to
the 0 inputs from the IlWR and appears on the Y outputs. The
first instruction of the subroutine, I(A), is accessed and is at
the inputs of the IlWR. On the next clock transition, I(A) is
loaded into the IlWR for execution, and the return addr~ss
J+3 is pushed onto the stack. The return instruction is executed at T 5. Figure 8 is a similar timing chart showing one
subroutine linking to a second, the latter consisting of only
one microinstruction.
In the columns on the left is the sequence of microinstructions
to be executed. At address J+2, the sequence control portion
of the microinstruction contains the comand "Jump to subCONTROL MEMORY
Execute Cycle
Execute
Cycle
Microprogram
Address
Sequencer
Instruction
J-1
TO
T,
T2
T6
T7
J+1
J+2
J+3
J+4
JSR A
Signals
-
I(AI
RTS
T,
T2
rL rL rL
0
T3
T4
Ts
T6
T7
0
Tg
Ta
n- rL rL ru rL ruru
3
L
H
A
0
H
H
H
H
X
X
X
X
2
L
L
X
0
H
X
X
0
H
X
X
0
FE
PUP
D
X
X
X
X
/lPC
5TKO
5TK1
5TK2
5TK3
J+1
J+2
J+3
-
A+2
J+3
A+3
J+3
J+5
-
A+1
J+3
J+4
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Am2909
Output
V
J+1
J+2
A
A+1
A+2
J+3
J+4
J+5
ROM
Output
(V)
I(J+1)
JSR A
I(A)
I(A+1)
RTS
I (J+3)
I (J+4)
I (J+5)
Contents
of/lWR
(Instruction
being
executed)
/lWR
I(J)
I(J+1)
JSR A
I(AI
I(A+1)
RTS
I(J+3)
I (J+4)
S,.50
Am2909
Inputs
(from
/lWR)
Internal
Registers
A
A+1
A+2
To
Clock
en = HIGH
Figure 7. Subroutine Execution.
CONTROL MEMORV
Execute
Cycle
TO
T,
T2
Tg
Microprogram
Address
J-1
J
J+1
J+2
J+3
-
-
T3
T4
T5
T7
TS
A
A+1
A+2
A+3
A+4
T6
Sequencer
Instruction
JSR A
-
JSR B
RTS
-
-
-
B
RTS
-
-
-
-
Execute Cycle
TO
T1
T2
T3
T4
TS
TS
T7
Ta
Tg
0
H
X
X
0
H
X
X
3
L
H
A
0
H
X
X
0
H
X
X
3
L
H
B
2
L
L
X
0
H
X
X
2
L
L
X
0
H
X
X
/lPC
5TKO
5TK1
5TK2
5TK3
J+1
J+2
J+3
-
-
A+1
J+3
A+2
J+3
A+3
J+3
A+4
J+3
A+5
J+3
-
-
B+1
A+3
J+3
J+4
-
-
-
-
-
-
-
Am2909
Output
Clock
Signals
Am2909
Inputs
(from
/lWR)
Internal
Registers
5,.50
FE
PUP
D
n.JIlJIlJrL rururu ruru ru
-
-
-
-
-
-
-
V
J+1
J+2
A
ROM
Output
(V)
I(J+1)
JSR A
Contents
of/lWR
(I nstruction
being
executed)
/lWR
I(J)
I(J+1)
-
-
-
A+1
A+2
B
A+3
A+4
J+3
J+4
I(A)
1(A+1)
JSR B
RTS
I(A+3)
RTS
I(J+3)
I (J+4)
JSRA
I(A)
I(A+1)
JSR B
RTS
I(A+31
RTS
I(J+3)
Figure 8. Two Nested Subroutines. Routine B is Only One Instruction.
8-020
-
-
-
c n = HIGH
Am2909/Am2911
MAXIMUM RATINGS (Above which the useful life may be impaired)
Storage Temperature
Temperature (Ambient) Under Bias
-0.5 V to +7.0 V
-0.5 V to +Vcc max.
-0.5 V to +7.0 V
30mA
-30 mA to +5.0 mA
Supply Voltage to Ground Potential
DC Voltage Applied to Outputs for HIGH Output State
DC Input Voltage
DC Output Current, Into Outputs
DC Input Current
OPERATING RANGE
Ambient Temperature
PIN
Vcc
Am2909/29110C, PC
4.7SV to S.2SV
Am2909/2911 OM, FM
4.S0V to S.SOV
STANDARD SCREENING
(Conforms to MI L-STD·883 for Class C Parts)
Step
Pre·Seal Visual Inspection
Level
Am2909/Am2911PC. DC Am2909/Am2911DM. FM
100%
100%
MIL·STD·883
Method
2010
B
1008
C
24·hour
lS0°C
100%
100%
100%
100% •
100%
Stabi lization Bake
Conditions
100%
Temperature Cycle
1010
C
-6SoC to +lS0°C
10 cycles
Centrifuge
2001
B
10,000 G
A S x 10-8 atm·cc/sec
C2 Fluorocarbon
100% •
100% •
100%
100%
See below for
definitions of subgroups
100%
100%
=S
=7
=7
LTPD = 7
LTPO = S
LTPD = 7
LTPD = 7
LTPD = 7
LTPD = 7
LTPD=7
Fine Leak
1014
Gross Leak
1014
Electrical Test
Subgroups 1 and 7
SOO4
Insert Additional Screening here for Class B Parts
Group A Sample Tests
Subgroup 1
Subgroup 2
Subgroup 3
Subgroup 7
Subgroup 8
LTPD
LTPD
LTPD
See below for
definitions of subgroups
SOOS
LTPD = 7
LTPD = 7
Subgroup 9
• Not applicable for
Am2909PC or
Am2911PC.
ADDITIONAL SCREENING FOR CLASS B PARTS
MI L·STD·883
Method
Step
Burn·ln
1015
Electrical Test
Subgroup 1
Subgroup 2
Subgroup 3
Subgroup 7
Subgroup 9
SOO4
Conditions
D
Level
Am2909/Am2911DMB. FMB
12SoC
160 hours min.
100%
100%
100%
100%
100%
100%
Return to Group A Tests in Standard Screening
GROUP A SUBGROUPS
(as defined in MIL·STD-883, method 5005)
Subgroup
Parameter
DC
DC
DC
Function
Function
10
11
Switching
Switching
Switching
Temperature
2SoC
Maximum rated temperature
Minimum rated temperature
2SoC
Maximum and minimum rated
temperature
2SoC
Maximum Rated Temeperature
Minimum Rated Temperature
8-D21
Am2909 • Am2911
ELECTRICAL CHARACTERISTICS OVER OPERATING RANGE (Unless Otherwise Noted)
Typ.
Paramatars
Test Conditions
Description
Vcc= MIN.,
VOH
VOL
Output HIGH Voltage
VIN = VIH or VIL
Output LOW Voltage
(Note ,)
Min.
MIL
I 10H = -1.0mA
2.4
COM'L
I 10H = -2.6mA
2.4
Max.
(Note 2)
Volts
10L =4.0mA
0.4
VCC = MIN.,
10L =S.OmA
0.45
VIN = VIH or VIL
10L = 12mA
Input HIGH Level
Guaranteed input logical HIGH
voltage for all inputs
VIL
Input LOW Level
Guaranteed input logical LOW
voltage for all inputs
VI
Input Clamp Voltage
Input LOW Current
IlL
VIN =0.4V
VCC= MAX.,
Input HIGH Current
IIH
VIN = 2.7V
MIL
0.7
COM'L
O.S
Push/Pop, OE
-0.72
Others (Note 6)
-0.36
Cn
40
Push/Pop
40
Others (Note 6)
20
0.2
Others (Note 6)
0.1
Output Short Circuit Current
(Note 3)
VCC = MAX.
Power Supply Currr.nt
VCC = MAX. (Note 4)
Output OF F Current
-LOS
Cn, Push/Pop
lOS
10ZH
-1.5
Cn
VIN = 7.0V
Input HIGH Current
ICC
I
I
Volts
VCC= MAX.,
II
10ZL
2.0
VCC = MIN., liN = -lSmA
VCC = MAX.,
Volts
0.5
(Note 5)
VIH
Units
I
Y O -Y 3
-30
-100
1
C n +4
-30
-S5
SO
130
VCC= MAX.,
VOUT=0.4V
-20
0E=2.7V
VOUT=2.7V
20
Volts
Volts
rnA
JJA
rnA
rnA
rnA
JJA
Notes: ,.
2.
3.
4.
For conditions shown as MIN. or MAX .• use the appropriate value specified under Electrical Characteristics for the applicable device tYpe.
Typical limits are at VCC = 5.0V. 25°C ambient and maximum loading.
Not more than one output should be shorted at a time. Duration of the short circuit test should not exceed one second.
Apply GNO to C n , RO. R" R2. R3, ORO. OR" OA 2 • OR3. DO, 0,. 02. and 03' Other inputs open. All outputs open. Measured after a
LOW-to-HIGH clock transition_
5. The 12mA guarantee applies only to YO, Y 1, y 2 and Y 3'
6. For the Am29", 0i and Ri are Internally connected. Loading is doubled (to same values as Push/Pop).
f--IT::L~ I l - - - j
CP
1,-
3,
CLOCK H TO L OCCURS
I~
~
1
CLOCK TO Y; 0' Cn +4
ov
1111
-----t----+I-(TA~~E III)
INPUTSTOYorCn + 4
(TABLE III
(TABLE III
-------------
Y OUT
Cn + 4
I
·1'-----:&m::::
_ _ _ _ _ 1.5V
- ,,\t\\ \\\ \
I
ITAB~~
I
ALL INPUTS
(EXCEPT DEI
I(T~~ 11---1
~\\\\\\\\\~t
I,-----J·ov
________-L
_______________
~~~~~~~~~~~~~__________________________
Figure 12. sWitching Waveforms. See Tables for Specific Values.
8-022
:::
VOL
MPR-098
Am2909 • Am2911
SWITCHING CHARACTERISTICS
OVER OPERATING RANGE
TABLE II
MAXIMUM COMBINATIONAL PROPAGATION DELAYS
(all in ns. CL = 50pF (except output disable tests))
Tables I. II. and III below define the timing characteristics of
the Am2909 and Am2911 over the operating voltage and
temperature range. The tables are divided into three types of
parameters; clock characteristics. combinational delays
from inputs to outputs. and set-up and hold time requirements. The latter table defines the time prior to the end of
the cycle (Le .• clock LOW-to-HIGH transition) that each
input must be stable to guarantee that the correct data is
written into one of the internal registers.
.~
Measurements are made at 1.5V with V IL = OV and V IH =
3.0V. For three-state disable ~ests. CL =5.0pF and measurement is to O.5V change on output voltage level.
TABLE I
CYCLE TIME AND CLOCK CHARACTERISTICS
TIME
COMMERCIAL
MILITARY
Input
Y
Cn+4
Y
Cn+4
~-
17
30
20
32
50. 5 1
30
48
40
50
ORi
17
30
20
32
Cn
ZERO
-
14
-
16
30
48
40
50
OE LOW (enable)
25
-
25
-
DE HIGH (disable)
25
-
25
-
COMMERCIAL
MILITARY
Clock t
5 1 50 = LH
43
55
50
62
Minimum Clock LOW Time
30
35
Clock t
5 150
=
LL
43
55
50
62
Minimum Clock HIGH Time
30
35
Clock t
5 15 0
=
HL
80
95
90
102
Operating Range
Part Numbers
Power Supply
Temperature Range
Commercial
Am2909PC. DC
Am2911 PC. DC
5.0V ± 5%
TA
Military
Am2909DM. FM
Am2911DM
5.0V ± 10%
Tc = -55°C to +125°C
=
O°C to +70°C
TABLE III
GUARANTEED SET-UP AND HOLD TIMES (all in ns) (Note 1)
From Input
Notes
MILITARY
COMMERCIAL
Hold Time
22
5
22
10
5
12
5
26
6
30
7
FE
2!3
5
30
5
Cn
28
5
30
5
30
0
35
3
ORj
30
0
35
3
50. 5 1
45
0
50
0
ZERO
45
0
50
0
RE
2
Ri
PU5H/POP
2
Di
Set-Up Time
Hold Time
Set-Up Time
5
Notes: 1. All times relative to clock LOW-to-HIGH transition.
2. On Am2911. Ri and OJ are internally connected together and labeled OJ. Use Rj set·up and hold times when 0 inputs are used to load register.
Metallization and Pad Layout
25 24
Am2909
Am2911
23
17
22
21 20 19 18
26
27
28
13
DIE SIZE 0.110" X 0.160"
Numbers correspond to DIP pin-out
8-D23
16
Am2930
MAXIMUM RATINGS (Above which the useful life may be impaired)
Storage Temperature
-65 to +150°C
Temperature (Ambient) Under Bias
-55 to +125°C
Supply Voltage to Ground Potential
-0.5 to + 7.0V
DC Voltage Applied to Outputs for High Output State
-0.5V to Vee max.
DClnputVolta~g~e_______________________________________________________________________
-_O_.5__
to__+_5_.5__
V
DC Output Current, Into Outputs
30mA
DC Input Current
-30 to +5.0mA
OPERATING RANGE
Part Number
Temperature
Vee
4.75V to 5.25V
TC
=
4.50V to 5.50V
-55 to +125°C
DC CHARACTERISTICS OVER OPERATING RANGE
Typ
Parameters
VO H
Description
Min
Test Conditions (Note 1)
Vec = MIN.,
VIN = V IL or V IH
Output HIGH Voltage
Yo, Y" Y2, Y 3
G, C n + 4 ,
Ci+4
IOH
=
-1.6rnA
2.4
P, FULL,
EMPTY
IOH
=
-1.2rnA
2.4
(Note 2)
IOl = 20rnA
(COM'L)
V IH
Input HIGH Level (Note 4)
V il
Input LOW Level (Note 4)
VI
Input Clamp Voltage
IlL
Vec = MIN.
VIN = VIL or V IH
Output LOW Voltage
0.5
IOl
=
16rnA (MIL)
0.5
G,C n +4
Ci+4
IOl
=
16rnA
0.5
P, FULL,
EMPTY
IOL
=
12rnA
Input LOW Current
Vec
=
=
MIN., liN
=
MAX., VIN
=
0.5V
I ~
Vee
MAX•• V"
II
Input HIGH Current
Vec
=
IsC
Output Short Circuit Current
Vee
(Note 3)
=
MAX.
Output OFF Current
=
MAX.,
IOZl
Vee
MAX., VIN
=
-.702
-.657
-2.31
I
lee
Notes: 1.
2.
3.
4.
5.
Power Supply Current
(Note 5)
Vee
=
5.0V
Vee
=
MAX.
. --t-
20
40
CC
50
Ci
90
Cn
250
-30
DE
= 2.4V
VOUT
= 0.5V
VOUT
=
1.0
rnA
-85
rnA
50
150
/LA
205
239
170
220
185
For conditions shown as MIN. or MAX., use the appropriate value specified under Electrical Characteristics for the applicable device type.
Typical limits are at Vee = 5.0V, 25'C ambient and maximum loading.
Not more than one output should be shorted at a time. Duration of the short circuit test should not exceed one second.
These input levels provide no gtlaranteed noise immunity and should only be tested in a static-, noise·free environment.
Minimum ICC is at maximum temperature.
8-D24
/LA
-50
2.4V
TA = 25°C
Te = -55 to +125°C
_._._-Te = +125°C
TA = 0 to 70°C
TA = 70°C
rnA
-3.25
I
5.5V
IOZH
Volts
_..._"-_._,.".. f - -
-~-~
Ci
10 - 4• RE, lEN,
CP,5E
~ 2.7V
Volts
-1.5
CC
DO-3
Input HIGH Current
0.8
-.360
~-"~
Cn
IIH
Volts
-18rnA
DO-3
10 -. 4, RE, lEN,
CP,5E
Volts
0.5
2.0
Vec
Units
Volts
YO' Y" Y 2, Y 3
VOL
Max
rnA
Am2930
Am2930 SWITCHING CHARACTERISTICS
Tables A, B, e and' 0 define the timing characteristics of the Am2930. Measurements are made at 1.5V with V IL
V IH = 3.0V. For three-state disable tests, CL = 5.0pF and measurement is to O.5V change on output voltage level.
~
OV and
I. Typical Room Temperature Performance.
Vec = 5.0V, TA = 25°C
TABLE IA
Clock Characteristics.
Minimum Clock LOW Time
Minimum Clock HIGH Time
TABLE IC
Combinational Propagation Delays.
All in ns.
Outputs fully loaded. C L = 50pF.
18ns
20ns
TABLE IB
Output Enable/Disable Times.
All in ns.
C L = 5.0pF for output disable tests.
From To
y
OE
CC
y
(Nole 1)
14 • 0
(Nole 1)
Y
Enable Disable
17
18
39
57
I~
From
Input
CC
Cn
C;
CP
lEN
41
r-
---'
Ci+4
, Y G,P Cn+4 14=L
57
61
61 50
46 32
39
17
25
14
46
52 40
33
37 23
30
0
27
CP:
Output
14 . 0
TABLE 10
Set-up and Hold Times. All in ns.
All relative to clock
LOW-to-HIGH transition.
CiH
14=H Full Empty
69
52
53
29
32
14
40
58
40
43
27
I
Note 1: "Suspend" instruction.
Input
14 . 0
cc
lEN
Cn
C;
o (RE = L,
14 •0 = 0-8 or 10-15)
o (All other conditions)
RE
Set-up Hold
Time Time
68
53
39
28
18
14
44
t3
II. Guaranteed Performance Over Commercial Operating Range.
Vee = 4.75 to 5.25V, TA = 0 to 70 0 e
TABLE IIA
Clock Characteristics.
Minimum Clock lOW Time
Minimum Clock HIGH Time
TABLE IIC
Combinational Propagation Delays.
All in ns.
Outputs fully loaded. CL = 50pF.
31ns
33ns
TABLE liB
Output Enable/Disable Times.
All in ns.
C L = 5.0pF for output disable tests.
From
OE
CC
(Note 1)
14 •0
(Note 1)
To
y
Enable Disable
26
27
y
55
37
y
80
55
~
Output
From
y G,P Cn+4
Input
81 67
77
14 •0
CC
63 45
55
Cn
32
25
C;
61
CP
69 53
0
49 33
40
lEN
TABLE 110
Set-up and Hold Times. All in ns.
All relative to clock
LOW-to-HIGH transition.
~r-
Ci+4
14=L
80
22
43
Ci+4
14=H Full Empty
91
69
72
42
45
22
78
55
55
59
40
Input
14 . 0
CC
lEN
Cn
C;
o (RE = L,
14 . 0 = 0·8 or 10·15)
o (All other conditions)
RE
Note 1: "Suspend" instruction.
Set-up Hold
Time Time
114
75
55
43
32
5
25
66
24
III. Guaranteed Performance Over Military Operating Range.
Vce = 4.5 to 5.5V, Te = -55 to +125°e
TABLE iliA
Clock Characteristics.
Minimum Clock LOW Time
Minimum Clock HIGH Time
TABLE IIIC
Combinational Propagation Delays.
All in ns.
Outputs fully loaded. CL = 50pF.
35ns
35ns
~
TABLE 1110
Set-up and Hold Times. All in ns.
All relative to clock
LOW-Io-HIGH transition.
output
TABLE IIIB
Output Enable/Disable Times.
All in ns.
C L = 5.0pF for output disable tests.
From
OE
CC
(Note I)
14 • 0
(Note 1)
To
y
Enable Disable
31
32
y
60
42
y
85
60
From
Input
14 . 0
CC
Cn
C;
~.
lEN
CiH Ci+4
G,P Cn+ 4 14=L 14=H Full Empty
88
74
82
87
97
78
68 52
47
60
78
37
30
46
23
23
74 58
48
84
66
60
60
- j
45
55 38
65
45
Input
y
Note 1: "Suspend" instruction.
8-025
14 . 0
CC
lEN
Cn
~
o (RE
~ L.
or 10·15)
o (All other conditions)
RE
~0.8
, Set-up Hold
Time Time
124
80
69
52
37
30
72
29
INDEX
8255 PPI devices used with. 1-42. 1-43
6800 support devices not compatible with. 1-44
INS8900/PACE. See PACE/INS8900
CP1600
direct addressing. 2-3
implied addressing. 2-4
I/O port pin characteristics. 2-30
stack addressing. 2-5
MC68000
absolute data addressing. 7-30
address registers. 7-3
autovector interrupt response. 7-27
bus and address error exception processing. 7-25
data registers. 7-2
exception priorities. 7-23
exception vector table. 7-23
externally generated exceptions. 7-23
immediate data addressing. 7-37
implied register addressing. 7-32
internally generated exceptions. 7-22. 7-23
interrupt request exception processing. 7-26
memory interface. 7-9
operating modes. 7-22
program counter relative addressing. 7-32
read timing. 7-13
register direct addressing. 7-30
register indirect address. 7-30
reset exception processing. 7-25
spurious interrupt. 7-27
Stack Pointer. 7-4
Status register. 7-5
wait state. 7-14
write timing. 7-14
MicroNova I/O busA-12
8086
AX register. 5-5
base relative. indexed addressing. 5-15
BCD addition. 5-49
BCD division. 5-51
BCD multiplication. 5-51
BCD subtraction. 5-49
Bus Interface Unit (BIU). 5-30
BX register. 5-5
Code Segment register. 5-8
Control signals. simple and complex. 5-28
CX register. 5-5
data memory base relative addressing. 5-16
Data Segment register. 5-11
Destination Index register. 5-10
direct memory addressing. 5-13
dual bus complexity. 5-28
DX register. 5-5
-8080A register compatibility. 5-5
Execution Unit (EU). 5-30
external memory addressing. 5-24
Extra Segment register. 5-10
hold. in min. and max. mode systems. 5-39
implied memory addressing. 5-14
indirect addressing. 5-22
instruction queue. 5-31
interrupt return. 5-46
interrupt vector table. 5-44
I/O port addressing. 5-21
maskable interrupt. 5-44. 5-45
non-maskable interrupt. 5-44. 5-45
program counter. 5-8
program relative addressing. 5-21
reset. 5-27
Segment registers. 5-7
software interrupts. 5-44. 5-45
Source Index register. 5-10
Stack Pointer register. 5-9. 5-11
Stack Segment register. 5-9
8212. used in INS8900 system
as input port. 1-39. 1-40
as output port. 1-41
8251 US ART. used in INS8900 system. 1-43
8253 Programmable Counter/Timer.
used in INS8900 system. 1-43
8288 Bus Controller
interrupt signals. 5-110
I/O bus mode. 5-109
memory protect. 5-109
write control signals. 5-109
Nova
addressing. 4-6-9
address space. 4-23
busy status. 4-21
done status. 4-21
registers. 4-22
9440
initialization. 4-16
instruction fetch. 4-24
I/O wait states. 4-28
memory read. 4-24
system bus. 4-14
PACE. See a/so PACE/INS8900
clock signals. 1-11
level 0 interrupt problems. 1-24
stack interrupt problems. 1-22
substrate bias voltage. generating. 1-35
TTL-level bus. 1-2
PACE/I NS8900
address latches and decoders. 1-2
bidirectional transceiver element (BTE). 1-2
BTE mode control signals. 1-37
busses. floating. 1-15
CONTIN signal. 1-15
CPU-initiated DMA block data transfers. 1-16
cycle-stealing DMA. 1-17. 1-18
data input cycle. 1-12
data output cycles. 1-13
direct addressing options. 1-24
INS8900. See a/so PACE/INS8900
address/data lines. demultiplexing. 1-38
control signal polarity considerations. 1-39
8251 and 8253 used with. 1-43
xvii
PACE/INS8900 (Continued)
direct indexed addressing. 1-7
OMA block data transfers. 1-16. 1-17
execution speed. 1-1
Extend signal for slow I/O operations. 1-13
Extend used to suspend I/O during DMA operations.
1-17
Halt state. 1-14
interrupts. 1-21-23
logic level. 1-2
machine cycle. 1-12
NHALT signal. 1-15
power supply. 1-1
processor stall. 1-15
registers. saving during interrupts. 1-22
return from interrupt. 1-21
signal differences. 1-10
split base page. 1-6. 1-7
stack interrupts. 1-5
STE clock frequency. 1-35
system timing element (STEl. 1-2
TMS 9900
context switch. 3-5. 3-6
memory addresses. 3-3
direct addressing. 3-6
indexed addressing. 3-6
instruction execution sequences. 3-18
internal operations machine cycle. 3-15
interrupt vector map. 3-27
multiple interrupt hardware considerations. 3-30
program memory addressing. 3-8
TMS 9902
break. 3-91
break logic. 3-86
Control register. 3-86
device initialization. 3-84
error flags. 3-93
internal clock signal. 3-88
interrupts. 3-86. 3-87
receive logic. 3-92
receiver status. 3-87
register addressing. 3-84
reset. 3-86
Status register. 3-87
test mode. 3-86
timer status. 3-87
Transmit/Receive Data Rate register. 3-88
transmit event sequence. 3-90
transmitter status. 3-87
TMS 9903
asynchronous break logic. 3-103
asynchronous receive. 3-110
asynchronous transmit. 3-109
bisync logic. 3-105
clock rate option. 3-106
Control register. 3-100
CRC options. 3-106
device intialization. 3-109
device reset. 3-100
external sync logic. 3-104
HOLC abort. 3-104
initialize CRC. 3-100
initialize transmit/receive. 3-100
interface signal. 3-97
interrupt enable/disable. 3-102
modes. 3-97
monosync logic. 3-105
NRZI select. 3-106
Parameter register. 3-103
parity options. 3-105
Read register addressi ng. 3-100
receive eRe. 3-102
received character size. 3-106
register select. 3-100
SDLC configurations. 3-105
SDLC loop. 3-111
SDLe receive logic. 3-104
serial I/O signals. 3-98
Status register. 3-106
sync strip. 3-105
test mode. 3-102
transmit controls. 3-102
transmit operation. 3-104
Write register addressing. 3-101
TMS 9940
CRU bit utilization. 3-59
hold logic. 3-64
idle logic. 3-64
expansion mode. 3-60
multiprocessor system interface. 3-61
simple CRU I/O mode. 3-59
sync mode. 3-64
TMS 9980 series clock logic. 3-49
2901
ALU logic. 8- 13
carry status. 8-24
data input. 8-33
half-carry status. 8-25
local RAM. 8-7
microcode. sample. 8-25
microinstruction. 8-9
multiply. 8-35
overflow status. 8-24
Q register. 8- 12
RAM and CPU registers. 8-10
rotate operation. 8-25
sample microcode. 8-25
shift operation. 8-25
sign status. 8-24
status logic. 8-24
zero status. 8-24
2903
ALU functions. 8-57
ALU input. 8-48
ALU input options. 8-44
ALU operand options. 8-49
ALU output destinations. 8-61
ALU shifter. 8-61
Arithmetic and Logic Unit (ALUl. 8-49
destination options. 8-59
double length normalization. 8-67. 8-68
increment function. 8-72
local RAM addressing. 8-56
normalize special functions. 8-67
shift logic. 8-60
sign extend logic. 8-63
signal/magnitude twos complement function. 8-69
single length normalization. 8-7
slice significance select. 8-45
status signals. 8-46
three-address microcycle. 8-56
two-address timing. 8-56
twos complement divide function. 8-77
twos complement multiply function. 8-73
unsigned multiply. 8-72
2909 output mask. 8-99
2909/2911
Address. 8-95
xviii
indirect memory addressing. 6-18
instruction fetch machine cycle. 6-23
I/O instructions. 6-36
LDPS instruction. 6-39
M1 and MO instructions. 6-41
memory interface logic. 6-20
memory read machine cycle. 6-23
memory write machine cycle. 6-23
multiply instruction. 6-39
New Program Status Area pointer. 6-8
normal mode. 6-3
primary memory reference instructions. 6-37
principal memory addressing modes. 6-37
Program Counter 6-6
Refresh Counter. 6-28
secondary memory reference instructions. 6-37
shift instructions. 6-41
sixteen-bit registers. 6-9
software traps. 6-32
Special I/O instructions. 6-37
stack. 6-18
stack instructions. 6-41
Stack Pointer. 6-3
status. 6-6
subroutine call. 6-40
system call. 6-40
system mode. 6-3
thirty-two-bit registers. 6-9
wait state. 6-23
Z8001
address representation. 6-3
base address. 6-9
long segmented base relative addressing. 6-17
long segmented direct memory addressing. 6-13
long segmented indexed addressing. 6-15
program relative addressing. 6-18
segmented mode. 6-7
short segmented base relative addressing. 6-16
short segmented indexed addressing. 6-14
Z8002
direct memory addressing. 6-12
indexed addressing. 6-14
program relative addressing. 6-17
short segmented direct memory addressing. 6-13
data output. 8-99
immediate data input. 8-95
incrementer. 8-99
instruction skip. 8-100
jump. 8-100
microprogram counter. 8-99
multiple jump. 8-103
output select. 8-95
output zero control. 8-99
sequential addresses. 8-99
single instruction reexecution. 8-100
stack. 8-100
subroutine call. 8-102
subroutine nesting. 8-103
2910
add ress output. 8- 11 0
condition codes. 8-113
data input. 8-110
increment. 8-113
instruction codes. 8-113
microprogram counter. 8-110. 8-113
microprogram initialization. 8-120
microprogram jump. 8-120
microprogram jump-to-subroutine. 8-120
stack. 8-113
2930 series
accumulator. 8-125
carry logic. 8-129
Index register. 8-129
instruction codes. 8-125
Program Counter. 8-129
Stack Pointer. 8-129
Stack. Push. Pop. 8-129
Z8000
auto-increment. 6-18
auto-decrement. 6-18
base relative addressing. 6-15
block transfer instructions. 6-40
byte registers. 6-9
conditional jump instructions. 6-40
divide instruction. 6-38
implied indexed addressing. 6-15
implied memory addressing. 6-11
xix
OSBORNE/McGraw-Hili Books of Interest
The SOS6 Book
by R. Rector and G. Alexy
SOSO Programming for Logic Design
by Adam Osborne
6S00 Programming for Logic Design
by Adam Osborne
ZSO Programming for Logic Design
by Adam Osborne
SOSOA/SOS5 Assembly Language Programming
by L. Leventhal
6S00 Assembly Language Programming
by L. Leventhal
Z80 Assembly Language Programming
by L. Leventhal
6502 Assembly Language Programming
by L. Leventhal
ZSOOO Assembly Language Programming
by L. Leventhal et al.
Running Wild: The Next Industrial Revolution
by Adam Osborne
PETICBM Personal Computer Guide - Second Edition
by Adam Osborne and Carroll Donahue
PET and the IEEE 4S8 Bus (GPIB)
by E. Fisher and C. W. Jensen
Practical Basic Programs
by L. Poole et al.
Some Common BASIC Programs
by L. Poole and M. Borchers
Payroll with Cost Accounting - CBASIC
by Lon Poole et al.
Accounts Payable and Accounts Receivable - CBASIC
by Lon Poole et al.
General Ledger - CBASIC
by Lon Poole et al.
Some Common Basic Programs - PETICBM
edited by Lon Poole et al.
OSBORNE/McGraw-H ill
Microprocessor Handbook Series
OSBORNE 4 & 8-Bit Microprocessor Handbook
by Adam Osborne and Gerry Kane
OSBORNE 16-Bit Microprocessor Handbook
by Adam Osborne and Gerry Kane
An Introduction to Microcomputers: Volume 3 - Some Real Support Devices
by Gerry Kane and Adam Osborne
8089 I/O Processor Handbook
by Adam Osborne
CRT Controller Handbook
by Gerry Kane
68000 Microprocessor Handbook
by Gerry Kane
Source Exif Data:
File Type : PDF File Type Extension : pdf MIME Type : application/pdf PDF Version : 1.3 Linearized : No XMP Toolkit : Adobe XMP Core 4.2.1-c043 52.372728, 2009/01/18-15:56:37 Create Date : 2012:06:26 13:03:47-08:00 Modify Date : 2012:06:26 15:37:51-07:00 Metadata Date : 2012:06:26 15:37:51-07:00 Producer : Adobe Acrobat 9.51 Paper Capture Plug-in Format : application/pdf Document ID : uuid:7d941447-edfc-425a-ba0e-750962385ef5 Instance ID : uuid:ce1ae8a9-6169-4919-b2fd-c6a7939d1c37 Page Layout : SinglePage Page Mode : UseNone Page Count : 803EXIF Metadata provided by EXIF.tools