1993_AMD_PAL_Device_Handbook 1993 AMD PAL Device Handbook
User Manual: 1993_AMD_PAL_Device_Handbook
Open the PDF directly: View PDF .
Page Count: 764
Download | |
Open PDF In Browser | View PDF |
PAL® Device Data Book and Design Guide 1993 Advanced Micro Devices Family Standard Packages Part Number Technology Features EE CMOS GALIII Device Equivalent tPD ns Icc mA Page 2-58 10 125 115 115 10 55 UNIVERSAL PAL DEVICES 16V8 PALCE16V8H-5 PALCE16V8H-7 20S,J 20P, J PALCE 16V8H-1 0 PALCE16V8Q-10 20P,S,J PALCE16V8H-15 15 PALCE16V8H-25 20P,S,J 25 90 PALCE16V8Q-25 PALCE16V8Z·15 20P, J 20P, J 25 15 55 0.015 25 0.015 25 30 0.015 115 115 Zero-Power 3.3 V Zero-Power 20P,J 20P, J 24P,28J 3.3V High-Drive 10 PALCE20V8H-5 24S,28J GAL Device Equivalent 24P,28J 5 7.5 125 PALCE20V8H -7 10 115 15 24P,S,28J 10 55 PALCE20V8H-15 24P,28J PALCE20V8Q-15 15 15 90 55 PALCE20V8H-25 PALCE20VSQ-25 25 25 55 7.5 220 10 15 180 180 25 180 5 7.5 115 24P,28J TTL Varied Term Distribution PAL22V10-10 PAL22V10-15 AmPAL22V10A PALCE22V10H·5 28J PALCE22V10H-7 24P, S, 28J EECMOS Varied Term Distribution PALCE22V10H-10 2-124 2-79 2-135 2-214 115 PALCE20V8Q·10 PAL22V10-7 2-104 0.015 PALLV16V8-10 PALCE16V8HD-15 PALCE20V8H-10 24V10 90 55 20P,J PALLV16V8Z·25 PALLV16V8Z-30 22V10 15 PALCE16V8Q-15 PALCE16V8Z-25 20V8 5 7.5 90 2-266 2-294 115 10 120 PALCE22V1 OQ·1 0 24P,28J 10 55 PALCE22V10H-15 24P,S,28J 15 90 PALCE22V10Q-15 24P,28J 15 55 PALCE22V1 OH~25 24P, S, 28J 25 90 PALCE22V1oo-25 24P,28J 25 55 PALCE22V10Z·15 PALCE22V10Z-25 24P, S, 28J Zero Power 15 25 0.015 0.015 2-320 PALLV22V10Z·25 24P, S, 28J 3.3 V Zero-Power 25 0.015 2-331 PALCE24V10H-15 28P, J 28-Pin GAL-Type 15 115 2-345 25 115 PALCE24V10H-25 26V12 PALCE26V12H-15 PALCE26V12H-20 2SP,J Advanced 22V10 Macrocell 15 20 105 105 2-360 29M16 PALCE29M16H-25 24P,28J Advanced Macrocell 25 100 2-377 2-432 ASYNCHRONOUS PAL DEVICES 610 PALCE610H-15 24P,28J PALCE610H-25 20RA10 29MA16 PALCE20RA10H-20 PALCE29MA 16H-25 24P,2SJ 24P,28J Bold part numbers indicate preliminary. EECMOS J-K F/Fs, 15 90 Prog. CLK Prog. ClK 25 90 20 100 2-236 Prog. CLK, Advanced Macrocell 25 100 2-399 PAL ® Device Data Book and Design Guide 1993 A D VA NeE D M Ie ROD E V ICE S © 1993 Advanced Micro Devices, Inc. Advanced Micro Devices reserves the right to make changes in its products without notice in order to improve design or performance characteristics. This publication neither states nor implies any warranty of any kind, including but not limited to implied warrants of merchantability or fitness for a particular application. AMD~ assumes no responsibility for the use of any circuitry other than the circuitry in an AMD product. The information in this publication is believed to be accurate in all respects at the time of publication, but is subject to change without notice. AMD assumes no responsibility for any errors or omissions, and disclaims responsibility for any consequences resulting from the use of the information included herein. Additionally, AMD assumes no responsibility for the functioning of undescribed features or parameters. Trademarks AMD, HAL, MACH, PAL, PALASM and SKINNYDIP are registered trademarks of Auto Vec, ProPAL and ZPAL are trademarks and FusionPLD is a service mark of Advanced Micro Devices, Inc. Product names used in this publication are for identification purposes only and may be trademarks of their respective companies. Your fast time-to-market needs can now be met better than ever with PAL ~ devices from Advanced Micro Devices, Inc. This new data book provides you with a truly diverse selection of lowpower and high-performance CMOS solutions in addition to the highest performing bipolar products in the industry. A number of Application Notes have also been included in order to make this data book a valuable design guide as well. For your high-density PLD requirements, please contact an AMD representative for our latest printing of the MACH~ 1 and 2 or MACH 3 and 4 Family Data Books. Thanks for selecting AMD. Remember, our partnership helps you gain and keep the competitive edge. We're not your competition. ~~{k~ Director of Marketing Programmable Logic c III ~AMD Iv TABLE OF CONTENTS Chapter 1 Introduction ................................................... 1-1 Product Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 1-3 Selecting the Correct CMOS PLD .................................. 1-10 Commercial PLDs for Industrial Applications .......................... 1-22 Chapter 2 PAL Device Data Sheets ......................................... 2-1 16R8 PAL1SR8 Family ........................................ 2-3 PAL 1SR8-4 Com'l Series ................................. 2-1S PAL 1SL8-4/5 PAL 1SR8-4/5 PAL 1SRS-4/5 PAL 1SR4-4/5 PAL1SR8-7 Com'l Series ................................. 2-18 PAL 1SL8-7 PAL 1SR8-7 PAL 1SRS-7 PAL 1SR4-7 PAL 1SR8D/2 Com'l Series. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2-20 PAL 1SL8D/2 PAL1SR8D/2 PAL1SRSD/2 PAL1SR4D/2 PAL 1SR8B Com'l Series ................................. 2-22 PAL1SL8B PAL16R8B PAL1SRSB PAl16R4B PAL 16R8B-2 Com'[ Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2-24 PAL 16L8B-2 PAL16R8B-2 PAL 16R6B-2 PAL1SR4B-2 PAL 16R8A Com'l Series ................................. 2-26 PAL1SL8A PAL16R8A PAL 1SRSA PAL16R4A PAL 1SR8B-4 Com'l Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2-28 PAL1SL8B-4 PAL1SR8B-4 PAL1SR6B-4 PAL16R4B-4 PAL16R8-10/12 Mil Series ................................ 2-30 PAL 16L8-10/12 PAL16R8-10/12 PAL16R6-10/12 PAL 16R4-1 0/12 Table of Contents v l1 AMD PAL16R8B Mil Series ............ , ....................... PAL16L8B PAL16R8B PAL16R6B PAL16R4B PAL16R8B-2 Mil Series .................................. PAL16L8B-2 PAL16R8B-2 PAL16R6B-2 PAL16R4B-2 PAL16R8A Mil Series .................................... PAL16L8A PAL16R8A PAL16R6A PAL16R4A PAL 16R8B-4 Mil Series ........ . . . . . . . . . . . . . . . . . . . . . . . . .. PAL16L8B-4 PAL 16R8B-4 PAL 16R6B-4 PAL 16R4B-4 vi 2-32 2-34 2-36 2-38 16V8 PALCE16V8 Family ..................................... 2-48 PALCE16V8H-5 Com'l ................................ 2-58 PALCE16V8H-7 Com'l ................................ 2-60 PALCE16V8H-10 Com'l ............................... 2-62 PALCE16V80-10 Com'l ............................... 2-64 PALCE16V8H-15/25 0-15/25 Com'l ..................... 2-66 PALCE16V8H-15 Mil .................................. 2-68 PALCE16V8H-20/25 Mil ............................... 2-70 PALLV16V8-10 Com'l ................................... 2-79 PALCE16V8Z Family .................................... 2-95 PALCE16V8Z-15 Com'l, INO .......................... 2-104 PALCE16V8Z-25 Com'l, INO .......................... 2-106 PALLV16V8ZFamily ................................... 2-115 PALLV16V8Z-25INO ................................ 2-124 PALLV16V8Z-30 INO ................................ 2-126 PALCE16V8HO-15 Com'l ................................ 2-135 18P8 AmPAL18P8B/AUNL Com'l ............................. 2-155 20R8 PAL20R8 Family ...................................... PAL20R8-5 Com'l Series ................................ PAL20L8-5 PAL20R8-5 PAL20R6-5 PAL20R4-5 PAL20R8-7 Com'l Series ................................ PAL20L8-7 PAL20R8-7 PAL20R6-7 PAL20R4-7 PAL20R8-10/2 Com'l Series ............................. PAL20L8-10/2 PAL20R8-10/2 PAL20R6-10/2 PAL20 R4-1 0/2 Table of Contents 2-165 2-178 2-180 2-182 AMD 20V8 ~ PAL20R8B Com'l Series ................................ PAL20L8B PAL20R8B PAL20R6B PAL20R4B PAL20R8B-2 Com'l Series ............................... PAL20L8B-2 PAL20R8B-2 PAL20R6B-2 PAL20R4B-2 PAL20R8A Com'l Series ................................ PAL20L8A PAL20R8A PAL20R6A PAL20R4A PAL20R8-10/12 Mil Series ............................... PAL20L8-10/12 PAL20R8-10/12 PAL20R6-10/12 PAL20R4-10/12 PAL20R8-15 Mil Series ................................. PAL20L8-15 PAL20R8-15 PAL20R6-15 PAL20R4-15 PAL20R8B-2 Mil Series ................................. PAL20L8B-2 PAL20R8B-2 PAL20R6B-2 PAL20R4B-2 PAL20R8A Mil Series ................................... PAL20L8A PAL20R8A PAL20R6A PAL20R4A 2-184 PALCE20V8 Family .................................... PALCE20V8H-5 Com'l ............................... PALCE20V8H-7 Com'l ............................... PALCE20V8H-10 Com'l .............................. PALCE20V80-10 Com'l .............................. PALCE20V8H-15/25 0-15/25 Com'l .................... PALCE20V8H-15 Mil .................... , ............ PALCE20V8H-20/25 Mil .............................. 2-204 2-214 2-216 2-218 2-220 2-222 2-224 2-226 2-186 2-188 2-190 2-192 2-194 2-196 20RA10 PALCE20RA10H-20 Com'l, INO .......................... 2-236 22P10 AmPAL22P1 OB/AUA Com'l .............................. 2-249 22V10 PAL22V10 Family, AmPAL22V10/A ........................ PAL22V10-7Com'l ................................ :. PAL22V10-10 Com'l ................................. PAL22V10-15 Com'l ................................. AmPAL22V1 OA Com'l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. PAL22V10-12 Mil ................................... PAL22V10-20 Mil ................................... AmPAL22V10/A Mil .................................. Table of Contents 2-258 2-266 2-268 2-270 2-272 2-274 2-276 2-278 vii ~AMD PALCE22V10 Family ................................... PALCE22V10H-5 Com'l .............................. PALCE22V10H-7 Com'l .............................. PALCE22V10H-10 Com'l ............................. PALCE22V10Q-10 Com'l ............................. PALCE22V10H-15/25 0-15/25 Com'l ................... PALCE22V10H-15/20/25/30 Mil ........................ PALCE22V10Z Family .................................. PALCE22V10Z-15IND' ....... ; ....................... PALCE22V10Z-25 Com'l, IND ......................... PALLV22V10Z-25IND .................................. 2-286 2-294 2-296 2-298 2-300 2-302 2-304 2-313 2-320 2-322 2-331 24V10 PALCE24V10H-15/25 Com'l ............................. 2-345 26V12 PALCE26V12H-15/20 Com'l .................. '........... 2-360 29M16 PALCE29M16H-25 Com'l ................................ 2-377 29MA16 PALCE29MA16H-25 Com'l .............................. 2-399 610 viii PALCE610 Family ..................................... 2-424 PALCE610H-15/25 Com'l ............................. 2-432 PALCE610H-20 Mil ..... '............................. 2-434 Chapter 3 MACH Devices ...................... : . . . . . . . . . . . . . . . . . . . . . . . . .. 3-1 MACH 1 and 2 Device Families ..................................... 3-3 MACH110-12/15/20 ...................................... 3-7 MACH120-15/20 ......................................... 3-9 MACH130-15/20 ........................................ 3-11 MACH210A-10, MACH21 0-12115/20, MACH210AQ-15/20 ....... 3-13 MACHLV210A-15/20 ................. '.' ................. 3-15 MACH220-12/15/20 ..................................... 3-17 MACH230-15/20 ........................................ 3-19 MACH215-12/15/20 ..................................... 3-21 MACH 3 and 4 Device Families .................................... 3-23' MACH435-15/20, 0-25 ................................... 3-27 MACH355-15/20 ........................................ 3-29 MACH445-15/20 ........................................ 3-31 MACH465-15/20 ........................................ 3-33 Chapter 4 General Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 4-1 Inside AMD's CMOS PLD Technology ................................ 4-3 Military PAL Devices ..........".................................. 4-43 Military ProPAL Devices ........................................ ; 4-52 Electrical Characteristic Definitions ....................... . . . . . . . . .. 4-54 fMAX Parameters ................................................ 4-57 Physical Dimensions ............................................ 4-58 Table of Contents AMD~ Chapter 5 Design and Testability . .......................................... 5-1 Basic Design with PLDs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-3 PLD Design Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-8 PLD Design Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-16 Combinatorial Logic Design .................................. 5-27 Registered Logic Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-40 State Machine Design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-60 Testability ............................................. '........ 5-73 Ground Bounce ................................................ 5-92 Metastability .......... : ........................... , ............ 5-95 Latchup ...................................................... 5-97 Converting Bipolar PLD Designs to CMOS .......................... 5-100 High-Speed-Board Design Techniques ............................. 5-107 Minimizing Power Consumption with Zero-Power PLDs .... , ........... 5-135 Designing with the PALCE16V8HD .............................. " 5-138 Chapter 6 Appendices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 6-1 Logic Reference Guide ........................................... 6-3 Signal Polarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 6-25 Glossary ......................................... , . . . . . . . . . . .. 6-30 Worldwide Application Support .................................... 6-37 Table of Contents Ix ~ AMD 1 INTRODUCTION Product Overview Selecting the Corre~~ cOMos PLD 0 0 0 0 0 0 0 0 Commercial PLDs for Industrial Ap~ii~~ii~~~ 0 0 0: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 : : : 000000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00000 0 0 0 0 0 0 0 0 0 000 0 0 0000000000000000000000000000000 Introduction 0 1-3 1-10 1-22 1·1 ~AMD 1·2 Introduction PRODUCT OVERVIEW Advanced Micro Devices Advanced Micro Devices offers the industry's widest variety of Programmable Logic Devices (PLDs), implemented in a variety of technologies. In this section, we will briefly discuss the device families, and look at the various architecture, speed, and power options. More specific device information can be found in the individual data sheets. Discussions on some of the special architectural features of many of the devices can also be found in their respective data sheets. There are six basic PLD areas addressed by Advanced Micro Devices' PLDs: • High-speed PAL® devices • Universal PAL devices • Industry-standard PAL devices • Low-power PAL devices • Asynchronous PAL Devices • High-density PLDs The largest application area is that covered by the Programmable ArrayLogic (PAL) devices. There is a wide variety of PAL devices, ranging from simple devices that address general logic design problems to more sophisticated devices that deal with more complex problems. The area of high-density PLDs is addressed by the MACH® devices which provide a PLD with thousands of gates and very high performance. The Macro Array CMOS High-density (MACH) devices are described briefly Chapter 3. Further information on AMD's MACH product line may be found by obtaining the MACH 1 and 2 Family Data Book (14051) and the MACH 3 and 4 Family Data Book (17466). MACH design assistance is available by obtaining the MACH Technical Briefs Manual (15972) and the MACH Devices Applications Handbook (17020). Development Assistance is available through the FusionPLDsM Partners Catalog (15585). Product Overview 1-3 ~ AMD HIGH·SPEED PAL DEVICES AMD offers the fastest PAL devices on the market today. We were the first to introduce the PAL device in 1975 and have been the first to market with volumes on all successive generations. As the market leader in the PLD arena, we fully expect to introduce even faster devices in the future. Currently, we have the bipolar TIL PAL16RSfamily of 20-pin devices in 5-ns speed grade and the PAL20RS family of 24-pin devices, also in 5-ns speed grade, available in volume production. For extra performance we also have the 16RS family at 4.5 ns, when packaged with the high-performance 2S-pin PLCC pinout. These families include, both registered (16R4, 16R6, 16RS, 20R4, 20R6, 20RS) and combinational devices (16LS, 20LS). They are used in a wide variety of applications where performance and space are critical, often replacing SSIIMSllogic circuits. For applications where the absolute fastest devices are not needed, other speed grades are offered at a lower cost and/or lower power consumption. AMD's Electronically Erasable (EE) CMOS process also provides ~igh-speed universal PAL devices. The PALCE16VS, PALCE20VS, and PALCE22V10 have a 5-ns version; most other EE CMOS devices have a 7.5- or 10-ns tPD, while using half or even a quarter of the power required by their bipolar equivalents. Table 1·1 High-Speed PAL Devices Commercial Specifications Functional Description Part Number Pin Count Array Inputs Array Outputs bldlr. dedctd. reg. fdbk. reg. comb. macrocell Prod. Terms per Output Spd/Pwr tPD Options (ns) fMAX· (MHz) lee (mA) PAL16L8 20 6 10 - - 8 - 7 -4 4.5 125 210 PAL16R8 20 - 8 8 8 - - 8 -5 5 117 210 PAL16R6 20 2 8 6 6 2 PAL16R4 20 4 8 4 4 4 - 7~ PALCE16V8 20 O~ 8-10 8-0 - - 8 7~ H-5 5 142.8 125 PAL20L8 24 6 14 - 8 - 7 -5 5 117 210 PAL20R8 24 - 12 8 8 - - 8 PAL20R6 24 2 12 6 6 2 - 7~ - 7~ 7~ PAL20R4 24 4 12 4 4 4 PALCE20V8 24 O~ 12-14 8-0 - - 8 7~ H-5 5 142.8 125 PALCE24V10 28 0-10 14-16 10-0 - - 10 7~ H-15 15 45.5 115 PALCE22V10 24 0-10 12 10-0 - - 10 8-16 H-5 5 142.8 115 PAL22V10 24 0-10 12 10-0 - 10 8-16 -7 7.5 91 220 PALCE610 24 0-16 4 16-0 - - 16 8 H-15 15 45.5 90 ·fMAX is defined as 1/(ts + teo) for the external feedback. 1-4 Product Overview AMD~ UNIVERSAL PAL DEVICES Have your design needs included a non-standard mix of inputs and outputs or choosing a variable number of combinatorial/registered/latched inputs and/or outputs for the given application? How about stocking only one or two PLOs to reduce your inventory costs? The solution to your problem is AMO's family of universal PAL devices. We pioneered . the concept of user-programmable output logic macrocells with the PAL22V10. With this macrocell, you can configure an output for combinatorial or registered operation and active-low or active-high polarity. This is what makes the PAL22V10 universal, for it can substitute for virtually all of the standard 24-pin PAL devices on the market. The PALCE26V12 is a 28-pin version which provides more inputs and outputs for those designs that don't quite fit into a PAL22V10. But we did not stop there. A second new feature pioneered with the PAL22V10 is variable product term distribution; the 10 outputs on this device are arranged in pairs with 16, 14, 12, 10 or eight product terms on each output. With up to 16 product terms, the PAL22V10 can implement far more complex logic functions than can be supported by other 24-pin devices. No wonder the PAL22V10 is the most popular PAL device on the market today. And now both faster (7.5 ns, 91 MHz) and reprogrammable low-power CMOS (7.5 ns at 130 rnA; 15 ns at 55 rnA) versions are available from AMO. The PALCE16V8 and PALCE20V8 are EE CMOS universal devices that have the additional capability of directly taking the designs of standard 20- and 24-pin PAL devices, respectively. They provide a cost-effective means of reducing inventory, lowering power, and reducing risk. The PALCE24V10 extends this architecture to 28 pins. The PALCE610 adds to the basic macrocell by providing 16 1/0 macrocells that can be configured with 0, T, J-K, or S-R flip-flops. The PALCE29M16 further enhances the macrocell concept. Its macrocell can be an input or an output macrocell that can be configured three ways: combinatorial, latched or registered. Sixteen of these macrocells are available in a 24-pin 300-mil package. And eight of the macrocells can be buried, allowing the connecting pins to be used as dedicated inputs. The PALCE29M16 also offers variable product term distribution. For those applications where registers and latches are not needed, the AmPAL 18P8 (20 pins) and AmPAL22P1 0 (24 pins) are ideal. These PAL devices with programmable polarity can flexibly replace almost all standard 20- and 24-pin combinatorial PAL devices. As a result, they significantly reduce your inventory. They are available in several speed-power grades to meet most application requirements. Product Overview 1·5 ~ AMD Table 1·2 Universal PAL Devices Commercial Specifications Functional Description Array Inputs Array Outputs Spd/Pwr tpo Options (ns) fMAX· (MHz) Icc (mA) H-5 H-7 0-10 H-10 0-15 H-15 0-25 H-25 . 5 7.5 10 10 15 15 25 25 142.8 100 66.7 66.7 45.5 45.5 37 37 125 115 55 115 55 90 55 90 7-8 H-5 H-7 0-10 H-10 0-15 H-15 0-25 H-25 5 7.5 10 10 15 15 25 25 142.8 100 66.7 66.7 45.5 45.5 37 37 125 115 55 115 55 90 55 90 10 7-8 H-15 H-25 15 25 45.5 37 90 90 - 10 8-16 H-5 H-7 0-10 H-10 0-15 H-15 Z-15 0-25 H-25 5 7.5 10 10 15 15 -15 25 25 142.8 100 83.3 83.3 50 50 50 33.3 33.3 115 115 55 120 55 90 0.015 55 90 - - 10 8-16 -7 -10 -15 A 7.5 10 15 25 91 71 50 28.5 220 180 180 180 - - 10 4 H-20 20 37 90 12-0 - 12 8-16 H-15 H-20 15 20 50 40 105 105 4 16-0 - - 16 8 H-15 H-25 15 25 45.5 37 90 90 8-16 5 16-8 16 8-16 H-25 25 28.5 100 10 - - 8 8 - 8 B A AL L 15 25 25 35 B A AL 15 25 25 Prod. Terms per Output Pin Count bldlr. dedctd. reg. fdbk. PALCE16V8 20 0-8 8-10 8-0 - - 8 7-8 PALCE20V8 24 0-8 12-14 8-0 - - 8 PALCE24V10 28 0-10 14-16 10-0 - - PALCE22V10 24 0-10 12 10-0 - PAL22V10 24 0-10 12 10-0 PALCE20RA10 24 10 10 PALCE26V12 28 0-12 14 PALCE610 24 0-16 PALCE29M16 24 AmPAL18P8 20 Part Number AmPAL22P10 24 10 12 reg. comb. macrocell - - - - 10 - ·fMAX is defined as 1/(ts + teo) for the external feedback. 1·6 Product Overview 8 - - - - 180 180 90 90 180 180 90 AMD~ INDUSTRY-STANDARD PAL DEVICES As we have increased speed on the TTL PAL devices, we have also reduced the power consumption on the slower devices by as much as 75 percent. As a result, both the industry-standard 20-pin and 24-pin PAL device families are available in a variety of speed and power grades. This allows the designer to select the optimum performance at the lowest possible cost and power consumption. These 20-pin and 24-pin devices are used in applications where the advantages of reduced package count, such as higher reliability and lower power consumption, improve the overall price-performance of the end-product. Often these benefits are realized by replacing Schottky, ALS, LS and some CMOS SSI/MSIIogic circuits with these PAL devices. Table 1·3 Standard PAL Devices Commercial Specifications Functional Description Part Number Pin Count Array Inputs bidir. dedctd. Array Outputs reg. comb. macrocell reg. fdbk. - PAl16L8 20 6 10 - PAL16R8 20 - 8 8 8 PAl16R6 20 8 6 6 2 Prod. Terms per Output Spd/Pwr tPD Options (ns) fMAX· (MHz) Icc (mA) 8 - 7 8 15 - - - 8 8-2 25 25 90 2 - 7-8 A 25 25 180 180 PAl16R4 20 4 8 4 - 7-8 8-4 35 16 55 PAL20L8 24 6 14 - - 8 - 7 8 15 - 210 PAL20R8 24 - 12 8 8 - - 8 8-2 25 25 105 PAL20R6 24 2 12 6 6 2 - 7-8 A 25 25 210 PAL20R4 24 4 12 4 4 4 - 7-8 4 4 ·fMAX is defined as 1/(ts + teo) for the external feedback. Product Overview 1-7 ~ AMD LOW-POWER PAL DEVICES AMD is the only major supplier of programmable logic devices to offer a broad line of low-power CMOS devices. And we are the only PLD supplier with such a comprehensive CMOS programmable logic line. There are two basic types of CMOS PAL devices: those that dissipate essentially no power when in a quiescent state, and faster devices which draw nominal current even when quiescent. Devices are thus classified as "zero-power" orl'quarter-power." Zero-power PAL devices are particularly suited for products that are portable or battery operated. In a standby mode they consume less than 151lA. Quarter-power CMOS devices can cut system power consumption 50 percent by replacing equivalent CMOS PAL devices. The PALLV16V8 and PALLV22V10 are devices designated to operate at 3.3 V for battery-operated applications. Table 1-4 Low-Power PAL Devices Functional Description Part Number PALCE16V8 Array Inputs Commercial Specifications Array Outputs Pin Count bldlr. dedctd. reg. fdbk. 20 0-8 8-10 8-0 reg. comb. macrocell - - 8 Prod. Terms per Output 7-8 PALLV16V8 PALCE22Vl0 24 0-10 12 - 10-0 - 10 PALLV22Vl0 ·fMAX is defined as 1/(ts + teo) for the external feedback. 1·8 Product Overview 8-16 Spd/Pwr (PD Options (ns) fMAX· (MHz) Icc (mA) 0-10 0-15 Z-15 0-25 Z-25 10 15 15 25 25 66.7 45.5 45.5 37 33.3 55 55 0.015 55 0.015 -10 Z-25 Z-30 10 25 30 66.7 33.3 22 90 0.015 0.015 0-15 Z-15 0-25 Z-25 15 15 25 25 50 50 33.3 33.3 55 0.015 55 0.015 Z-25 25 33.3 0.015 AMD~ ASYNCHRONOUS PAL DEVICES Currently AMD makes three devices that support asynchronous and bus interface applications. The PALCE20RA 10 is optimized for asynchronous applications. It contains ten D-type flip-flops, driven by a PAL array. Each flip-flop has individually programmable Clock, Reset and Preset product terms. With such features, this device is well suited to replacing glue logic in your system. The PALCE29MA16 combines some of the advantages of the PALCE29M16 with the advantages of the PALCE20RA 10. It has one dedicated Clock/Latch Enable input as well as product terms for each of the 16 macrocells to allow individual clocking, asynchronous Reset and asynchronous Preset. It also features variable product term distribution. To top it off, the PALCE29MA16 is electrically reprogrammable in a plastiC 300-mil package. The PALCE610 is a general purpose PLD. It has 16 independently-configurable macrocells. Each macrocell can be configured as either combinatorial or registered. The registers can be D, T, J-K or S-R type flip-flops. The device has 4 dedicated input pins and 2 clock pins. Asynchronous clocking is available since each clock pin controls 8 of the 16 macro cells. Table 1·5 Asynchronous PAL Devices Commercial Specifications Functional Description Part Number Pin Count Array Inputs bldir. dedctd. Array Outputs reg. fdbk. reg. comb. macrocell - PALCE20RA10 24 10 10 PALCE29MA16 24 8-16 5 16-8 PALCE610 24 0-16 4 16-0 Prod. Terms per Output Spd/Pwr Options tPD (ns) f MAX• (MHz) Icc (mA) - - 10 4 H-20 20 50 90 - - 16 4-12 H-25 25 28.5 100 - 16 8 H-15 H-25 15 25 45.5 37 90 90 ·fMAX is defined as 1/(ts + teo) for the external feedback. Product Overview 1·9 Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs Application Note by Steve Holte, Applications Engineer INTRODUCTION The purpose of this application note is to provide a survey of AMD's CMOS PLDs (Programmable Logic Devices). This includes both PAL (Programmable Array Logic)devicesandthemoregeneralrealmofPLDstowhich PAL devices be long. With the proliferation of parts the selection ofthe best PLDforyour application may seemdifficult.lfyouareanewPLDuser,thisoverviewwillguideyou through the wide variety of different device architectures, speed, and power grades. This tutorial should increase yourunderstandingofthebasiccharacteristicfeaturesthat make a device appropriate for a given application. Figure 1 shows a "CMOS PAL Selection Route Map." This can be used as a convenient model of the discussion throughout this paper. It can also be used as a reference guide when you are selecting a PLD for a particular application. The Benefits of AM D's CMOS PLDs Before addressing individual products, it is important to understand why CMOS technology is used in PLDs. There are two universal benefits of AM D's CMOS PLDs: electrical erasability and low power. Electrical Erasability Because PLDs are programmable, electrical erasability is probably the most important advantage that CMOS technology can bring. AMD's electrically-erasable CMOS has benefits that make it superior to both UVerasable CMOS and bipolar technologies. The most important advantage is the ability to erase the device electrically in a matter of seconds as opposed to hours for UV-based technologies, and not at all in the case of bipolar. The chief benefit to the user is a very high quality device. This is realized through the ability to erase and reprogram the device many times at various test points in the factory. In fact, the quality is so good that programming and post-programming functional rejects are virtually non-existent. 1-10 ~ Advanced Micro Devices A second major benefit of electrical erasability is the ease of reprogramming the device by the user. This saves time when prototyping, and allows for recoverY of large volumes of devices that may need to be reprogrammed for a variety of reasons, such as midproduction bug fixes. Erasure takes place automatically by the device programmer, and is completely transparent to the user. Low Power The most well-known attribute of CMOS is low power consumption. All PALCE (Programmable Array Logic CMOS Electrically-Erasable) devices are offered in halfpower versions; they require at most halfthe supply current of their first-generation bipolar counterparts. Some devices are also offered in quarter-power versions. This is achieved by taking the latest process technology and designing to favor even lower power over the fastest speed possible. CMOS uses less current than bipolar because most of the current flow only takes place while the transistors are actually switching. With bipolar, current flows through the transistors all the time. AMD's half- and quarter-power CMOS devices take advantage of this as much as possible. However, in order to achieve high speed it is necessary to operate some transistors on the Chip in the linear region. Because this circuitry is essentially always switching, the power consumption does not go to zero as it would in a conventional CMOS device. Lower power requirements are ideal for applications which have tight power budgets, such as mobile telecommunications. Smaller power supplies also reduce cost and lowers heat dissipation. This results in smaller cooling fans, or perhaps no fan at all. It also allows the system designer to pack everything even tighter since less empty space is needed for air circulation. This can make the circuit board fit in a smaller package, again reducing cost. Publication# 17402 Rev. A Issue Date: February 1993 Amendment/O AMD~ Global Clocks Individual Macrocell Clocks & Resets/Presets & Resets/Presets Synchronous More Output Drive _-......1'-----., 16V8HD 15 ns 16V8 5f7 .5/10/15/25 ns More Inputs Lower Power More PTs & I/Os 22V10Z 15/25 ns 20V8 5f7 .5/10/15/25 ns 16V8Z 15/25 ns Lower Powe More PTs & I/Os 22V10 5f7 .511 0/15/25 ns More PTs, I/Os & Clocks More Inputs More PTs & I/Os 24V10 15/25 ns 26V12 15/20 ns More Flexibility More I/Os Density 610 15/25 ns MACW U Family 10/12/15/20 ns & Speed More PTs & I/Os 29M16 25 ns 17402A-l More Flexibility Means Choice of: JK, SR, T & D Flipflops, liP & DIP MacroceJ/s, Latch or Reg. MacroceJ/s III Asynchronous Clocking Figure 1. CMOS PAL Selection Route Map Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs 1-11 ~ AMD Zero-Power The Selection Process Some devices are also offered in zero-standby power versions. Instead of a/ways operating certain transistors in the linear region to achieve high speed, this circuitry can shut down and the device goes into standby mode. Standby mode is defined as anytime the inputs do not switch for an extended period of time (typically 50 ns). When this happens the current consumption will almost go to zero (Icc < 151lA). The outputs will maintain the current state held while the device is in standby mode. When selecting a CMOS PLD, start by determining both the functional and size requirements for your application. When any input switches, the internal circuitry is fully enabled and the power consumption returns to normal. This feature results in considerable power savings for operation at low to medium frequencies. Zero-standby devices are desirable for a number of reasons. In portable and field-installed equipment that rely on batteries, battery life is extended. In solar powered systems, fewer solar cells are required. Functional Requirements The functional requirements of a given application are what' determine which device architecture should be used. The functional criteria include such issues as the clocking scheme, macrocell flexibility and output drive. The clocking scheme can be synchronous, where qll registers within a device use the same clock signal, or asynchronous, where each register can be clocked individually using any logic signal or combination of logic Signals available to the device. These two alternatives are illustrated in Figure 2. elK a. Registers with Synchronous Clocking 17402A-2 b. Registers with Asynchronous Clocking Figure 2. Basic Clocking Schemes Macrocell flexibility refers to the ability to configure the output in various ways. A macrocell (Figure 3) takes the basic sum-of-products logic and adds functionality through features like storage elements, optional path Inputs AND OR controls, polarity, and feedback. This concept will be further illustrated as each device macrocell is explained throughout the selection process. Macrocell Outputs 17402A-3 Figure 3. PLD Block Diagram with the Macrocell Included 1-12 Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs AMD~ Size Requirements The combination of number of inputs and outputs required determines the device size that should be used. The number of product terms required to implement a particular design also factor into this decision. The best approach to selecting the appropriate device is to begin with the simplest and smallest devices and upgrade as necessary to accommodate your application. Combinatorial and Synchronous Applications Starting from the top of the flow chart in Figure 1 and taking the path for synchronous designs leads one to those devices best suited for simple state machines, encoders, decoders, muxes, and similar logic applications. Forthese applications, D-type registered or combinatorial (non-registered) logic is needed. The first choice is the PALCE16V8, PALCE20V8, or PALCE24V10, depending on the number of inputs or outputs needed. The macrocells (Figure 4) can be configured to use combinatorial or D-type registered outputs in any combination. The D-type register operates very simply; data presented at the D input of the register will be clocked into the register on the rising edge of the clock signal. The output can be configured as active low- or active high-output depending on the requirements of the downstream devices and the efficiency of the logic. To ~==~--------------------------------------~11 OE 11 Vee Adjacent Macrocell 101------.... 00 01 (lOX )------4~ D QI------~ 01-----------.. From Adjacent SG1 Pin 17402A-4 Figure 4. PALCE16V8 Macrocell Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs 1-13 ~ AMD More Product Terms and I/O's If the application requires the features of the PALCE16V8 macrocell, but requires more programma- ' ble gale functions, move further down the flowchart. The next device, the PALCE22V10 (Figure 5) has varied product term distribution. The number of product terms varies among outputs, with up to 16 product terms on some outputs. In addition, it has global synchronous preset and asynchronous reset product terms. These are connected to all macrocells configured as registers, facilitating easy power-up and system reset. This versatility contributes significantly to the 22V10 being the world's most popular PAL architecture. When combined with the PALCE16V8 family, these two device families will likely handle about 80% of PLD applications. The PALCE26V12, a 28-pin version of the 22V10 increases versatility by adding more inputs and outputs, and by adding another global clock. Any macrocell can use one of the two clocks. This allows'the logic to be partitioned giving greater design flexibility. In addition, registered outputs can be configured as bidirectional pins on the 26V12. Since historically most applications in bipolar technology had been done in PAL16R8, PAL20R8, and PAL22V10 families, these types of applications can easily have their power lowered with half-, quarter-, and zero-power versions of the PALCE16V8, PALCE20V8, or PALCE22V10. •• • lIOn Figure 5. PALCE22V10 Macrocell Diagram Table 1. Summary of the PALCE16V8 and PALCE22V10 Family Architectures Macrocell Type Device Dedicated Inputs I/O's Clocks Product Terms PALCE24V10 16V8 14 10 1 7-8 7-8 7-8 PALCE22V10 22V10 12 10 1 8-16 PALCE26V12 22V10 14 12 2 8-16 PALCE16V8 16V8 8-10 8 1 PALCE20V8 16V8 10-12 8 1 1-14 Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs AMD~ high-output drive capability. This compares to 24 rnA low-output and 3.2 rnA high-output drive for other PALCE16V8 devices. The pALCE16V8HD also has some unique macrocell features. Because this device is designed to drive a bus, it can be configured with opendrain outputs. Open-drain (open-collector) configuration is sometimes used in bus applications because it provides controlled VOH, termination, and wire-NOR capability. Because the PALCE16V8HD is designed to take inputs directly from a noisy bus, all inputs have 200 mV input threshold hysteresis to improve noise immunity. The inputs can be configured as direct or latched, making additional buffering devices unnecessary. Additionally, the macrocell can be configured as either a 0- or T -type register. The T -type register is more efficient for counter applications because fewer product terms are consumed as hold states. Table 2. Summary of the Speed and Power Requirements for the PALCE16V8, PALCE20V8, and PALCE22V10 Families of Devices Device and Speed Grade Icc PALCE16V8H-5 125 mA static PALCE16V8H-7/-10 115 mA at 25 MHz PALCE 16V8H-15/-25 90 mA at 15 MHz PALCE 16V8Q-15/-25 55 mA at 15 MHz PALCE16V8Z-25 15 IlA in standby mode PALCE20V8H-5 125 mA static PALCE20V8H-7/-10 115 mA at 25 MHz PALCE20V8H-15/-25 90 mA at 15 MHz PALCE20V8Q-15/-25 55 mA at 15 MHz PALCE22V10H-5 140 mA at 25 MHz PALCE22V10H-7 140 mA at 25 MHz PALCE22V1 OH-1 0 120 mA at 25 MHz PALCE22V10H-15125 90 mA static PALCE22V1oo-25 55 mA static PALCE22V1 OZ-25 15 IlA in standby mode Complex Functions For those applications that require 0, T, J-K, or S-R register capability, the PALCE610 (Figure 6) has this flexibility. This device has 16 macrocell outputs and four dedicated inputs. The J-K, S-R, and T registers allow easy implementation of counters and larger state machines. Bus Applications For applications that require bus interaction, the PALCE16V8HD features 64 rnA low-output and 16 rnA 1/016 1/015 1/014 1/013 1/012 1/011 1/010 1109 CLK1 17402A-6 Figure 6. PALCE610 Block Diagram Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs 1-15 ~ AMD More Flexibility Moving down to the bottom of the flow chart reveals a more flexible device than the PALCE610. Applications that make use of embedded, or buried, registers can take advantage of the PALCE29M16 (Figure 7). Buried register operation is very useful when a state machine uses state bits that do not need to be brought outside the chip. This allows the pin associated with the macrocell to be available as an input. Eight of the 16 macrocells have dual feedback capability. This means that these macrocells have two independent feedback paths: one from the register and one from the 1/0 pin. The other eight macrocells have single feedback, where both paths are available but, only one or the other can be used. Since almost every pin is an 1/0 pin this device has 29 available inputs to the programmable AND array. 17402A-7 Figure 7. PALCE29M16 Block Diagram 1-16 Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs AMD~ The macrocells (Figure 8) can be configured as latches or registers. Latch operation allows the flip-flop in the macrocell to become transparent while the latch is enabled. When the latch is disabled, the flip-flop will hold the current state. This kind of operation is useful in sample and hold applications. Also, the register or latch may be used with the macrocelJ input pin for synchronizing signals. The active level of the latch enable, as well as the clock edge (rising or falling) are both programmable. Like the PALCE22V1 0, there is varied product term distribution among the macrocells. Also, preload capability is available using a global product term to define the preload condition. Preload capability aIJows arbitrary states to be loaded directly into the register, making it unnecessary to cycle through long, complex vector sequences to get the device in a particular state. This is normally only performed by the device programmer when it performs functional tests. The preload product term allows preload to be engaged by hand, without the need for supervoltages (voltages above Vee needed to engage preload on most PLDs). ~~~~?~----------------------------------~~ OE PTs FOR BANKS { OF 4 MACROCELLS COMMON ASYNCHRONOUS PRESET Po t---------t.,;>o-...----tC>II/OX I1CLKlLE - ........- t COMMON ASYNCHRONOUS RESET TOANO ARRAY :===~ ______________ ~RX TOANO ARRAY 17402A-8 Figure 8. PALCE29M16 Macrocell Diagram Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs 1-17 ~ AMD Asynchronous Applications Starting at the top of the flow chart again, but this time taking the path for asynchronous applications, you will find PALCE20RA10. The PALCE20RA10 is the simplest of the asynchronous devices. Each macrocell (Figure 9) is clocked individually using one product term per macrocell. Also, reset, preset, and output enable are controlled individually with one product term each. There is also a global output enable pin which is combined with the product term enable to determine if the output is enabled or disabled. A dedicated global preload pin allows all registered macroce lis to be preloaded simultaneously using normal TTL levels. The PALCE610 has the distinction of bridging the gap between synchronous and asynchronous registerclocking. The PALCE61 0 macrocells can be clocked via individual product terms for each macrocell, or the macrocells can be clocked in banks of eight via two dedicated clocks. This is done by using a clock/output enable mux (Figure 10). If the macrocell is configured as PL combinatorial or as a synchronous register, output enable/disable is controlled by a product term. If asynchronous register mode is desired, the same product term is used as a clock and the macrocell is always enabled. As mentioned above, the PALCE610 can act as a synchronous or asynchronous PAL device. As shown, a special function product term can be steered to control either the output enable orthe macrocell clock. In the latter mode, each register can be individually clocked. If your application requires the basic features of the PALCE29M 16, except with individual macrocell control, the PALCE29MA16 (Figure 11) should be considered. Four product terms in each macrocell are dedicated to control clocking, output enable, asynchronous reset, and asynchronous preset. These functions are controlled either globally or in blocks on the PALCE29M16. A common clock pin and output enable pin are still maintained, but the user has a choice of using eitherthe common control pin or individual macrocell control via the control product term. OE Output 17402A-9 Figure 9. PALCE20RA10 Macrocell Diagram 1-18 Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs AMD~ Vee ---------~ CLK-----i 17402A-l0 o Register Figure 10. PALCE610 Macrocell (Configured as aD-Register) with the Output Enable/Clock Mux e~~~~ _____________________________________ ~ INDIVIDUAL OE INDIVIDUAL ASYNCHRONOUS PRESET Po t---------iI~.....---1<-,;>II/OFx COMMON CLKlLE (PIN) INDIVIDUAL CLK/IE" INDIVIDUAL ASYNCHRONOUS RESET TO AND ARRAY ::==~-- __ -----....J RFx TO AND ARRAY 17402A-ll Figure 11. PALCE29MA16 Macrocell Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs 1-19 ~ AMD MACH Devices At the bottom of the flow chart, both the synchronous and asynchronous design paths converge into the MACH Family. MACH devices extend AMD's PLD offerin~s into t.he rea!m of what is referred to as mid-density. Mid-density deVices allow multiple smaller PLD designs to be consolidated into one device. Mid-density covers replacement of just a couple of smaller PAL devices, all th.e way up to designs that would traditionally be done With small gate arrays. These devices span from 900 to 3600 gates, with 32 to 64 macrocells, and are offered in 44- to 84-pin packages. MACH devices use PAL blocks that are interconnected using a switch matrix. The members of the families are differentiated by the number of pins, macrocells, clocks, and the amount of interconnect. The MACH 1 family has output macrocells; the MACH 2 family has output and buned macrocells. All signals, whether registered or combinatorial can be buried. The basic macrocell, common to both families resembles the PALCE22V10 macroc~1I with the additional choice of using D- or T-type registers. The MACH210 is illustrated in Figure 12. 1/00-1/07 2 44 x 68 AND Logic Array and Logic Allocator Switch Matrix 44 x 68 44 x 68 AND Logic Array and Logic Allocator AND Logic Array and Logic Allocator OE OE 2 1/024-1/031 1/016-1/023 CLKo/12, CLK1/1s 17402A-12 Figure 12. MACH210 Block Diagram 1-20 Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs AMD~ Both synchronous and asynchronous versions are available. The asynchronous 'MACH215 macrocell looks more like a PALCE20RA10 macrocell, rather than PALCE22V10 type macrocells. The synchronous devices are better suited to structured designs; the asynchronous MACH215 is better suited for random logic collection. All MACH devices have the advantage of fast (10, 12, 15, and 20 ns), predictable timing, which is au nique advantage when compared to other mid-density PLDs. SUMMARY Selecting the appropriate PLD for a given application involves matching your requirements with various device capabilities. Following the guidelines in this article along with the "CMOS PAL Selection Route Map" makes it easier. AMD's wide array of electrically-erasable CMOS PLDs, combined with strong third party support through our FusionPLDsM relationships, means an excellent selection of devices, software, and programming hardware. This gives you a virtual toolbox of solutions for your system logiC requirements, along with the strong technical support you expect. Selecting the Correct CMOS PLD-An Overview of Advanced Micro Devices' CMOS PLDs 1·21 COMMERCIAL PLDs FOR INDUSTRIAL APPLICATIONS l1 Advanced Micro Devices Customers have expressed an interest in using AMD programmable logic devices over the industrial temperature range. To serve your need, we recommend the following: 1. Use any standard commercial device from the following family datasheets: PALCE16V8 PALCE20V8 PALCE22V10 PALCE610 2. Slow down all commercial timing parameters by 20%. 3. Add 20 mA to the commercial Icc. All s~andard AMD warranties will apply to products used as noted above at Vee of +5 V ±10% over the temperature range of -40°C to 85°C. This approach will allow you to use broadly available commercial devices to fill your industrial temperature needs at no price premium. We are able to assure the performance to specification of these products since they are characterized and monitored over the full military temperature range (-55°C to 125°C). 1·22 Commercial PLDs For Industrial Applications 2 PAL DEVICE DATA SHEETS PAL 16R8 Family ......................................................... 2-3 PALCE16V8 Family ..................................................... 2-48 PALLV16V8-10 .......................................................... 2-79 PALCE16V8Z Family .................................................... 2-95 PALLV16V8Z Family ................................................... 2-115 AmPAL 18P8B/AUAlL ................................................... 2-155 PAL20R8 Family ....................................................... 2-165 PALCE20V8 Family .................................................... 2-204 PALCE20RA10H-20 .................................................... 2-236 AmPAL22P10B/AUA ................................................... 2-249 PAL22V10 Family/AmPAL22V10A ......................................... 2-258 PALCE22V10 Family ................................................... 2-286 PALCE22V10Z Family .................................................. 2-313 PALLV22V10Z-25 ...................................................... 2-331 PALCE24V10H-15/25 ................................................... 2-345 PALCE26V12-15/20 .................................................... 2-360 PALCE29M16H-25 ..................................................... 2-377 PALCE29MA16H-25 .................................................... 2-399 PALCE610 Family ..................................................... 2-424 PAL Device Data Sheets 2·1 ~AMD 2·2 PAL Device Data Sheets _ COM'L: -4/5f7/B/B-2/A, 012 ~ MIL: -10/12/B/B-2/A1B-4 Advanced Micro Devices PAL 16R8 Family 20-Pin TTL Programmable Array Logic DISTINCTIVE CHARACTERISTICS • • As fast as 4.5 ns maximum propagation delay Popular 20-pln architectures: 16L8, 16R8, 16R6,16R4 • Programmable replacement for high-speed TTL logic • • Register preload for testability Power-up reset for Initialization • Extensive third-party software and programmer support through FusionPLD partners • • 2o-Pln DIP and PLCC packages save space 2B-Pln PLCC-4 package provides ultra-clean high-speed signals GENERAL DESCRIPTION The PAL16R8 Family (PAL16L8, PAL16R8, PAL16R6, PAL 16R4) includes the PAL 16R8-5/4 Series which provides the highest speed in the 20-pin TTL PAL device family, making the series ideal for high-performance applications. The PAL 16R8 Family is provided with standard 20-pin DIP and PLCC pinouts and a 28-pin PLCC pinout. The 28-pin PLCC pinout contains seven extra ground pins interleaved between the outputs to reduce noise and increase speed. The devices provide user-programmable logic for replacing conventional SSI/MSI gates and flip-flops at a reduced chip count. The family allows the systems engineer to implement the design on-chip, by opening fuse links to configure AND and OR gates within the device, according to the desired logic function. Complex interconnections between gates, which previously required time-consuming layout, are lifted from the PC board and placed on silicon, where they can be easily modified during prototyping or production. The PAL device implements the familiar Boolean logic transfer function, the sum of products. The PAL device is a programmable AND array driving a fixed OR array. The AN D array is programmed to create custom product terms, while the OR array sums selected terms at the outputs. In addition, the PAL device provides the following options: - Variable input/output pin ratio - Programmable three-state outputs - Registers with feedback Product terms with all connections opened assume the logical HIGH state; product terms connected to both true and complement of any single input assume the logical LOW state. Registers consist of O-type flip-flops that are loaded on the LOW-to-HIGH transition of the clock. Unused input pins should be tied to Vee or GND. The entire PAL device family is supported by the FusionPLD partners. The PAL family is programmed on conventional PAL device programmers with appropriate personality and socket adapter modules. Once the PAL device is programmed and verified, an additional connection may be opened to prevent pattern readout. This feature secures proprietary circuits. PRODUCT SELECTOR GUIDE. Device Dedicated Inputs Outputs PAL16La 10 6 comb. 7 7 a 2 comb. a 7 4 reg. 4 comb. a 7 2 comb. PAL16Ra a a reg. PAL16R6 a 6 reg. PAL16R4 a Publication# 16492 Rev. B Amendment/O Issue Date: June 1993 Product Termsl Output Feedback Enable I/O - prog. prog. reg. pin reg. I/O pin prog. reg. I/O pin prog. 2-3 ~AMD BLOCK DIAGRAMS PAL16L8 INPUTS PROGRAMMABLE AND ARRAY (32 x 64) 16492B-1 PAL16R8 elK INPUTS PROGRAMMABLE AND ARRAY (32 x 64) 16492B-2 2·4. PAL16R8 Family AMD~ BLOCK DIAGRAMS PAL16R6 ClK INPUTS PROGRAMMABLE AND ARRAY (32 x 64) 16492B-3 ClK PAL16R4 PROGRAMMABLE AND ARRAY (32 x 64) 16492B-4 PAL16R8 Family 2-5 ~AMD CONNECTION DIAGRAMS Top View 20-Pin PLCC DIP (NOTE 1) w Vee 11 12 (NOTE 10) W I- 01- ~> ~ o (,) (,) (NOTE 9) 13 (NOTE 8) 14 (NOTE 7) 18 (NOTE 9) 15 (NOTE 6) 17 (NOTE 8) 16 (NOTE 5) 16 (NOTE 7) 17 18 (NOTE 4) 15 (NOTE 6) (NOTE 3) 14 (NOTE 5) GND (NOTE 2) 9 10 11 12 13 164928-5 164928-6 28-Pin PLCC 18 GND (NOTE 2) (NOTE 3) GND (NOTE 4) GND PIN DESIGNATIONS 11 ClK GND I I/O (NOTE 1) Vee 21 (NOTE 10) GND (NOTE 9) o GND Vee Clock Ground Input Input/Output Output Output Enable Supply Voltage OE Note: Pin 1 is marked for orientation. 164928-7 2-6 Note 16L8 16R8 16R6 16R4 1 10 elK elK elK 2 19 OE OE OE 3 01 01 1/01 1/01 4 1/02 02 02 1/02 5 1/03 03 03 03 6 1/04 04 04 04 7 1105 05 05 05 8 1/06 06 06 06 9 1/07 07 07 1/07 10 Os Os I/0s lias PAL16R8 Family AMO~ ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL T 16 -r- R 8 -I'"" FAMILY TYPE PAL = Programmable Array Logic NUMBER OF ARRAY INPUTS OUTPUTTYPE _ _ _ _ _ _ _ _-1 -5 " "r- P C t OPTIONAL PROCESSING Blank = Standard Processing OPERATING CONDITIONS C = Commercial (O°C to +7S°C) PACKAGE TYPE P = 20-Pin Plastic DIP (PD 020) J = 20-Pin Plastic Leaded Chip Carrier (PL 020) 28-Pin Plastic Leaded Chip Carrier for -4 (PL 028) D = 20-Pin Ceramic DIP (CD 020) R = Registered L = Active-Low Combinatorial NUMBER OF OUTPUTS _ _ _ _ _ _----J SPEED -4 = 4.S ns tPD -S =S ns tPD -7 = 7.S ns tPD D= 10 nstpD VERSION Blank = First Revision 12 = Second Revision Valid Combinations PAL16L8 PAL16R8 -SPC, -SJC, -4JC PAL 16R6 Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of . specific valid combinations and to check on newly released combinations. PAL16R4 PAL16L8-7 PAL16R8-7 PC, JC, DC PAL16R6-7 PAL16R4-7 PAL16L8D/2 PAL16R8D/2 PC,JC PAL16R6D/2 PAL16R4D/2 PAL16RS-4/S/7, 0/2 (COm'l) 2-7 ~AMD ORDERING INFORMATION Commercial Products (MMI Marking Only) AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL _____ T...J. 16 -r- R 8 B -2 C N -r- I'" -I'" • _ FAMILY TYPE PAL = Programmable Array Logic NUMBER OF ARRAY INPUTS OUTPUT TYPE _ _ _ _ _ _ _ _-...J ~ R = Registered L = Active-Low Combinatorial NUMBER OF OUTPUTS - - - - - - - ' OPTIONAL PROCESSING Blank = Standard Processing PACKAGE TYPE N = 20-Pin Plastic DIP (PO 020) NL = 20-Pin Plastic Leaded Chip Carrier (PL 020) J = 20-Pin Ceramic DIP (CD 020) OPERATING CONDITIONS C = Commercial (O°C to +75°C) SPEED B = Very High Speed (15 ns-35 ns tpo) A = High Speed (25 ns-35 ns tpo) -------------.....1 POWER Blank = Full Power (155 mA-180 mA Icc) -2 = Half Power (80 mA-gO mA Icc) -4 = Quarter Power (55 mA Icc) Valid Combinations PAL16L8 PAL16R8 PAL16R6 B. B-2. A. B-4 CN. CNL. CJ Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. Note: Marked with MMllogo. PAL16R4 2-8 PAL 16R8/B/B-2/A/B-4 (Com'l) AMD~ ORDERING INFORMATION APL Products AMD programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STD-883 requirements. The order number (Valid Combination) is formed by a combination of: PAL FAMILY TYPE-PAL = Programmable Array Logic T 16 -r- R -r- 8 r- R -10 IB' -r- .r- NUMBER OF _ _ _ _ _ _ _- - 1 A L LEAD FINISH A = Hot Solder Dip ARRAY INPUTS OUTPUT TYPE _ _ _ _ _ _ _ _- - - l PACKAGE TYPE R = 20-Pin Ceramic DIP (CD 020) 2 = 20-Pin Ceramic Leadless Chip Carrier (CL 020) R = Registered L = Active-Low Combinatorial NUMBER OF OUTPUTS - - - - - - - - - - ' SPEED-----------------~ -10 = 10 ns tPD -12 = 12nstpD DEVICE CLASS IB = MIL-STD-883C Class B Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. Valid Combinations PAL16L8 PAL16R8 -10, -12 IBRA,/B2A PAL16R6 PAL16R4 Military Burn-In Group A Tests Military burn-in is in accordance with the current revision of MIL-STD-883, Test Methods 1015, Conditions A through E. Test conditions are selected at AMD's option. PAL 16R8·1 0/12 (Mil) Group A tests consist of Subgroups 1, 2, 3, 7, 8, 9, 10, 11. 2·9 ~AMD ORDERING INFORMATION APL Products (MMI Marking Only) AMD programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STD-883 requirements. The order number (Valid Combination) is formed by a combination of: ___T PAL 16 R 8 B -2 M ---~ FAMILY TYPE PAL = Programmable Array Logic NUMBER OF ARRAY INPUTS OUTPUT TYPE _ _ _ _ _ _ _----1 J 1883B l PACKAGE TYPE (Per 09-000) J = 20-Pin Ceramic DIP R = Registered L = Active-Low Combinatorial NUMBER OF OUTPUTS OPTIONAL PROCESSING /8838 = MIL-STD-883, Class 8 (CD 020) W = 20-Pin Ceramic Flatpack (CFL020) L = 20-Pin Ceramic Leadless Chip Carrier (CL 020) ---------1 SPEED 8 = Very High Speed (20 ns-50 ns tpo) A = High Speed (30 ns-50 ns tpo) OPERATING CONDITIONS M = Military POWER 81ank = Full Power (180 mA Icc) -2 = Half Power (90 mA Icc) -4 = Quarter Power (55 mA Icc) Valid Combinations PAL16L8 PAL16R8 PAL16R6 8, 8-2, A,8-4 MJ/8838, MW/8838, MU8838 Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. Note: Marked with MMllogo. PAL16R4 Group A Tests Military Burn-In Military burn-in is in accordance with the current revision of MIL-STD-883, Test Methods 1015, Conditions A through E. Test conditions are selected at AM D's option. 2-10 PAL 16R8B/B-2/A/B-4 (Mil) Group A tests consist of Subgroups 1, 2, 3, 7, 8, 9, 10, 11. AMD~ FUNCTIONAL DESCRIPTION Standard 20-Pin PAL Family Registers with Feedback The standard bipolar 20-pin PAL family devices have common electrical characteristics and programming procedures. Fourdifferent devices are available, including both registered and combinatorial devices. All parts are produced with a fuse link at each input to the AND gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Utilizing an easily-implemented programming algorithm, these products can be rapidly programmed to any customized pattern. Extra test words are preprogrammed during manufacturing to ensure extremely high field programming yields, and provide extra test paths to achieve excellent parametric correlation. Pinouts The PAL 16R8 Family is available in the standard 20-pin 01 P and PLCC pinouts and the PAL 16R8-4 Series is available in the new 28-pin PLCC pinout. The 28-pin PLCC pinout gives the designer the cleanest possible signal with only 4.5 ns delay. The PAL 16R8-4 pinout has been designed to minimize the noise that can be generated by high-speed signals. Because of its inherently shorter leads, the PLCC package is the best package for use in high-speed designs. The short leads and multiple ground signals reduce the effective lead inductance, minimizing ground bounce. Placing the ground pins between the outputs optimizes the ground bounce protection, and also isolates the outputs from each other, eliminating cross-talk. This pinout can reduce the effective propagation delay by as much as 20% from a standard DIP pinout. Design files for PAL 16R8-4 Series devices are written as if the device had a standard 20-pin DIP pinout for most design software packages. Variable Input/Output Pin Ratio The registered devices have eight dedicated input lines, and each combinatorial output is an 1/0 pin. The PAL 16L8 has ten dedicated input lines and six of the eight combinatorial outputs are 1/0 pins. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vee or GNO. Programmable Three-State Outputs Each output has a three-state output buffer with threestate control. On combinatorial outputs, a product term controls the buffer, allowing enable and disable to be a function of any product of device inputs or output feedback. The combinatorial output provides a bidirectional 1/0 pin and may be configured as a dedicated input if the output buffer is always disabled. On registered outputs, an input pin controls the enabling of the three-state outputs. Registered outputs are provided for data storage and synchronization. Registers are composed of D-type flip-flops that are loaded on the LOW-to-HIGH transition of the clock input. Register Preload The register on the PAL 16R8 Family can be preloaded from the output pins to facilitate functional testing of complex state machine deSigns. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. Power-Up Reset All flip-flops power-up to a logic LOW for predictable system initialization. Outputs of the PAL 16R8 Family will be HIGH due to the active-low outputs. The Vcc rise must be monotonic and the reset delay time is 1000 ns maximum. Security Fuse After programming and verification, a PAL 16R8 Family design can be secured by programming the security fuse. Once programmed, this fuse defeats read back of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. When the security fuse is programmed, the array will read as if every fuse is programmed. Quality and Testability The PAL 16R8 Family offers a very high level of built-in quality. Extra programmable fuses provide a means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Technology The PAL16R8-5, -7 and 0/2 are fabricated with AMO's oxide isolated bipolar process. The array connections are formed with highly reliable PtSi fuses. The PAL 16R8B, B-2, A and B-4 series are fabricated with AMO's advanced trench-isolated bipolar process. The array connections are formed with proven TiW fuses for reliable operation. These processes reduce parasitic capacitances and minimum geometries to provide higher performance. PAL16R8 Family 2-11 ~AMD LOGIC DIAGRAM DIP and 20-Pin PLCC (28-Pin PLCC) Pinouts 16L8(-4) ~Vcc 10 1 (24) 0 3 4 7 8 11 12 1516 1920 2324 2728 (23) 31 0 1 ..... \ 19 (22) J 7 ~ m--t:c (25) 8 ~rO J 8 ~ L -= 18 (20) 15 3 (26) <. .> 16 1 \ I ~ 1 ..... . ~ 1 t )--vo- 39 5 > .~.rO - (13) J ..... I ... ~ <.t--' 48 l ~ I> \ 55 8 ~ (4) 56 ~ 14 (12) rD GND -= (11) 13 (10) ~rO GND -= (9) J -:l> ..... ... \ I ~ GND~ ..... GND ... 63 (6) (15) 40 47 18 9 (5) -= (14) .J ~ \ 7 (3) 16 1/°5 (16) ~rD GND 32 b 6 (2) - (17) .-J 31 0(1) 17 (18) rD GND <. 24 \ 5 (28) GND ill - (19) ~ 23 4 (27) GND (21) 3 4 7 8 11 12 1516 1920 2324 2728 12 (8) 11 19 (7) 31 -= 164928-8 2·12 PAL16R8 Family AMD~ LOGIC DIAGRAM DIP and 20-Pin PLCC (28-Pin PLCC) Pinouts 16R8 (-4) 2 ~Vee (23 ) elK 1~ (24) ..... 0 3 4 7 8 11 12 1516 1920 23 24 2728 31 0 "'" Q 7 ~ ~ L./ I-' 15 3 (26) .> 23 ~ ..> ~ [.J' 31 5 (28) Vee > L./ (1) 39 ..... >~ 6 > (2) 40 < ~ "'" t...../ 47 ....> ... (20) ~ M ~ h ~ h ~ ~ ~ H ~ ~ 48 .J 55 8 (4) ~ h 32 0 7 (3) (21 ) 24 ... > 56 ~ ~ 9 (5) ~ GND~ (6) o 3 4 7 8 11 12 15 16 19 20 23 24 27 28 -= GND GND ill -= (19) (18) GND ill -= (17) (16) GND ill -= (15) (14) GND ill -= (13) (12) GND ill -= (11) (10) GND rD -= (9) ~H Q 63 08 (22 ) -= '- 16 ~ 4 (27) ~~ ill (8) ~ (7) 31 164928-9 PAL16R8 Family 2·13 ~AMD LOGIC DIAGRAM DIP and 20-Pin PLCC (28-Pin PLCC) Pinouts 16R6 (-4) elK 20 ~Vee (23) too... 1 (24) ..... 0 3 4 7 8 11 12 1516 19 20 23 24 2728 31 0 J 6_ 7 ~ (20). GND rO .". (19) > 16 >-~ ~ 23 4 (27) ~ -. 24 31 5 (28) ~ ~ (1) 39 .... > 40 ........ f-- ./ 47 7 (3) ..... > ""'" t.../ .of > 56 t~..... ~ 63 9 (5) ... > GND~ (6) ~ ~. ~ ~ ~ H, ~ H 48 55 8 (4) ~ ~ 32 Vee D 6 (2) GND rO -= (21) a ~~ ...., 15 3 (26) 19 (22) :J \ o 34 78 1112 1516 192023242728 j (18) GND rD .". (17) (16) GND rD -= (15) (14) GND rD .". (13) (12) GND rD .". (11) (10) GND rO -= (9) 12 (8) L(28) 32 I-- ~ Vee D L-/ (1) (14) 39 6 (2) .. ~ GND ":" (13) I"'" 40 GND ":" (15) '- '~ ........ J (12) 47 7 (3) b <~ ~ 56 ~l ~ .rv 63 18 9 (5) '= J t:t»] 55 8 (4) GND rD (11) ~ 48 o 3 4 7 8 11 12 1516 1920 23 24 2728 GND rD (9) '= 1 -+-+H------f-HH- VT Output 164928-16 Input to Output Disable/Enable 164928-17 OE to Output Disable/Enable Notes: 1. VT= 1.5 V 2. Input pulse amplitude 0 V to 3.0 V 3. Input rise and fall times 2 ns-3 ns typical. . 2-40 PAL16R8 Family " AMD~ KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L /77// May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State WAVEFORM XXXXXX ]}) EK KSOOOO10-PAL SWITCHING TEST CIRCUIT 5V Output O - -.....- - -.....-~J Test Point 164928-18 Military Commercial Specification tPD, teo CL Sl Closed tpzx, tEA Z~ Z~ tpxz, tER H H: Opeh L: Closed ~Z: L~Z: Open Closed 50 pF 5 pF R1 R2 R1 R2 All but All but All but All but 8-4: 8-4: 8-4: 8-4: 200n 390n 390n 750n 8-4: 8-4: 8-4: 8-4: 800n 1.56 kn 800n 1.56 kn PAL16R8 Family Measured Output Value 1.5 V H ~Z: VOH-0.5 V L ~ Z: VOL + 0.5 V 2-41 ~AMD MEASURED SWITCHING CHARACTERISTICS for the PAL 16R8-5 Vee = 4.75 V, TA = 75°C (Note 1) 5.0 -5 4.5 tPD, ns 4.0 3.5 3.0 -+---+--+---+---t---+--i---I 2 3 5 4 6 7 8 Number of Outputs Switching 164928-12 tpo vs. Number of Outputs Switching 10 8 tPD, ns 6 -5 4 2 o 50 100 150 200 250 CL, pF tpo vs. Load Capacitance Vee = 5.25 V, TA 164928-13 = 25°C Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified ' where tpo may be affected. 2-42 PAL16R8-5 AMD~ CURRENT VS. VOLTAGE(I-V) CHARACTERISTICS for the PAL16R8-4/5 Vee = 5.0 V, TA = 25°C IOL, rnA 15 --+--+---t---t--tf---t--i'- VOL, V 164928-14 Output, LOW VOH, V 164928-15 Output, HIGH h, JlA 20 2 3 VI, V -100 -150 -200 164928-16 Input PAL16R8-5 2-43 ~AMD MEASURED SWITCHING CHARACTERISTICS for the PAL16R8-7 =4.75 V, TA = 75°C (Note 1) Vee . 7.5 7 tPD, ns 6.5 6-f--+--f.-~--+--+--+---I 2 3 4 5 6 7 8 NUM8ER OF OUTPUTS SWITCHING 164928-12 tPD VS. Number of Outputs Switching tPD, ns CL, pF 164928-13 tPD VS. Load Capacitance Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where tPD may be affected. 2-44 PAL16R8-7 AMD~ CURRENT VS. VOLTAGE (I-V) CHARACTERISTICS for the PAL16R8-7 Vee = 5.0 V, TA = 25°C IOl.mA 15 --+--+----+-+--+-+-+--1- VOL. V 164928-14 Output, LOW IOH. mA 20 VOH. V 164928-15 Output, HIGH h.Jl.A 20 2 3 VI. V -80 164928-16 Input PAL16R8-7 2-45 ~AMD INPUT/OUTPUT EQUIVALENT SCHEMATICS Typical Input Vee Input ProgramNerify Circuitry 164928-29 Typical Output ---------4~--_o Vee 40n NOM ~~----.--------.-() Input, Output ProgramNerifyl 1/0 Test Circuitry Pins Preload Circuitry 164928-30 2-46 PAL16R8 Family AMD jt1 POWER-UP RESET The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. The output state will be HIGH due to the inverting output buffer. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vee Parameter Symbol can rise to its steady state, two conditions are required to ensure a valid power-up reset. These conditions are: • The Vee rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Description Max Unit tPR Power-Up Reset Time 1000 ns ts .Input or Feedback Setup Time tWL Clock Width LOW See Switching Characteristics 4V~-------------------------------------- Vee Power Registered Active-Low Output Clock 9-t~ 164928-31 Power-Up Reset Waveform PAL16R8 Family 2-47 _ COM'L: H-5/7/10/15/25, Q-10/15/25 ~ MIL: H-15/20/25 PALCE16V8 Family EE CMOS 20-Pin Universal Programmable Array Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • • • • Pin, function and fuse-map compatible with all 20-pin GAL devices Electrically erasable CMOS technology provides reconflgurable logic and full testability High-speed CMOS technology - 5 ns propagation delay for "-5" version - 7.5 ns propagation delay for "-7" version Direct plug-In replacement for the PAL16R8 series and most of the PAL 1OH8 series Outputs programmable as registered or combinatorial In any combination • • • • • • • • Programmable output polarity Programmable enable/disable control Preloadable output registers for testability Automatic register reset on power up Cost-effective 20-pln plastic DIP, PLCC, and SOIC packages Extensive third-party software and programmer support through FusionPLD partners Fully tested for 100% programming and functional yields and high reliability 5 ns version utilizes a split leadframe for Improved performance GENERAL DESCRIPTION The PALCE16V8 is an advanced PAL device built with low-power, high-speed, electrically-erasable CMOS technology. It is functionally compatible with all 20-pin GAL devices. The macrocells provide a universal device architecture. The PALCE16V8 will directly replace the PAL16R8 and PAL 1OH8 series devices, with the exception ofthe PAL16C1. The PALCE16V8 utilizes the familiar sum-of-products (AND/OR) architecture that allows users to implement complex logic functions easily and efficiently. Multiple levels of combinatorial logic can always be reduced to sum-of-products form, taking advantage of the, very wide input gates available in PAL devices. The equations are programmed into the device through floatinggate cells in the AND logic array that can be erased electrically. The fixed OR array allows up to eight data product terms per output for logic functions. The sum of these products 2-48 feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial with an activehigh or active-low output. The output configuration is , determined by two global bits and one local bit controlling four multiplexers in each macrocell. AMD's FusionPLD program allows PALCE16V8 designs to be implemented using a wide variety of popular industry-standard deSign tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a deSigner can use a tool that is already installed and familiar. Please refer to the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. Publication# 16493 Rev. B Issue Date: June 1993 AmendmentlO l1 AMD BLOCK DIAGRAM 11-18 Programmable AND Array 32 x 64 164938-1 CONNECTION DIAGRAMS Top View DIP/SOIC PLCC/LCC 0 Vee CLKl/o II ~ 1/07 12 II0s 13 1/05 14 1/04 15 1/03 Is 1/02 17 1/01 18 1/00 ...£-I I/0s 1/05 15 1/04 Is 1/03 17 1102 ..!F 10 11 CI z (!) PIN DESIGNATIONS OE l"- ~ 18 9 164938-2 Vee () () > 14 Note: Pin 1 is marked for orientation ClK GND I 1/0 () 13 OE/lg GND -= -oJ 12 13 g I~ g 0> 0 164938-3 Clock Ground Input InpuVOutput Output Enable Supply Voltage PALCE16V8 Family 2-49 ~ AMD ORDERING INFORMATION Commercial Products AMO programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL CE _ _ _ _ _ _T....J -r- 16 V 8 H ·5 P CIS -"-.r-"r-'r- _ FAMILY TYPE - PAL = Programmable Array Logic TECHNOLOGY ___________________-....J CE .. CMOS Electrically Erasable [ OPTIONAL PROCESSING Blank ARRAY INPUTS Standard Processing PROGRAMMING DESIGNATOR Blank NUMBEROF----------------------------~ = = Initial Algorithm 14 First Revision 15 .. Second Revision (Same Algorithm as 14) OUTPUTTYPE----------------------------~ V - Versatile NUMBER OF FLIP·FLOPS -------------~ POWER-----------------------~ H .. Half Power (90 - 125 mA Icc) Q .. Quarter Power (55 mA Icc) OPERATING CONDITIONS C = Commercial (O°C to +75°C) SPEED--------------------------------~ PACKAGE TYPE -5 -7 -10 -15 -25 P = 20·Pin Plastic DIP (PO 020) J = 20·Pin Plastic Leaded Chip Carrier (PL 020) S = 20·Pin Plastic Gull,Wing Small Outline Package (SO 020) .. .. .. .. .. 5 ns tPD 7.5 ns tPD 10 ns tPD 15 ns tPD 25 ns tPD Valid Combinations Valid Combinations PALCE16VSH-5 JC PAlCE16V8H-7 PC,JC PALCE 16VSH-1 0 PC,JC, SC PALCE 16VSQ-1 0 PC,JC, SC PALCE16VSH-15 PC,JC, SC 15 14 15 PALCE 16VSQ-15 PC,JC Blank, PALCE16VSH-25 PC, JC, SC 14 PALCE16VSQ-25 PC,JC 2·50 Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMO sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE16V8H·517/10/15/25, Q~10/15/25 (Com'l) AMD~ ORDERING INFORMATION APL Products (Military) AMD programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STD-883 requirements. The order number (Valid Combination) is formed by a combination of: PAL CE 16 V 8 H -15 E5 IB R A I-..Ju -- FAMILY TYPE _ _ _ _ PAL = Programmable Array Logic 1L TECHNOLOGY - - CE = CMOS Electrically Erasable NUMBER OF _ _ _ _ _ _ _ _ _- - J LEAD FINISH A = Hot Solder Dip PACKAGE TYPE R = 20-Pin Ceramic DIP (CD 020) 2 = 20-Pin Ceramic Leadless Chip Carrier (CL 020) ARRAY INPUTS OUTPUT TYPE - - - - - - - - - - - - ' V = Versatile DEVICE CLASS = Class B NUMBER OF FLIP-FLOPS _ _ _ _ _ _ _- - J !8 POWER ---------------------------~ H = Half Power (130 mA Icc) PROGRAMMING DESIGNATOR Blank = Initial Release E4 = First Revision (Different Algorithm from Blank) SPEED -15 = 15 ns tPD -20 = 20 ns tPD -25 = 25 ns tPD Valid Combinations PALCE16V8H-15 E4 PALCE16V8H-20 Blank, E4 PALCE16V8H-25 IBRA IB2A Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. Group A Tests Group A tests consist of Subgroups 1,2,3, ~a ~ 1~ 11. Military Burn-In Military burn-in is in accordance with the current revision of MIL-STO-883, Test Method 1015, Conditions A through conditions are selected at AMO's option. PALCE16V8H-15/20/25 (Mil) E. Test 2-51 ~ AMD FUNCTIONAL DESCRIPTION The PAlCE16V8 is a universal PAL device. It has eight independently configurable macrocells (MCo-MC7). Each macrocell can be configured as registered output, combinatorial output, combinatorial 1/0 or dedicated input. The programming matrix implements a programmable AND logic array, which drives a fixed OR logic array. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Pins 1 and 11 serve either as array inputs or as clock (elK) and output enable (OE), respectively, for all flip-flops. Unused input pins should be tied directly to Vee or GND. Product terms with all bits unprogrammed (disconnected) assume the logical HIGH state and product terms with both true and complement of any input signal connected assume a logical lOW state. The programmable functions on the PAlCE16V8 are automatically configured from the user's design specifi- cation, which can be in a number of formats. The design specification is processed by development software to verify the design and create a programming file. This file, once downloaded to a programmer, configures the device according to the user's desired function. The user is given two design options with the PAlCE16V8. First, it can be programmed as a standard PAL device from the PAl16R8 and PAl10H8 series. The PAL programmer manufacturer will supply device codes for the standard PAL device architectures to be used with the PAlCE16V8. The programmer will program the PAlCE16V8 in the corresponding architecture. This allows the user to use existing standard PAL device JEDEC files without making any changes to them. Alternatively, the device can be programmed as a PAlCE16V8. Here the user must use the PAlCE16V8 device code. This option allows full utilization of the macrocell. To Adjacent Macrocell I/OX *SG1 */n macrocells MCo and MCl, SG1 is rep/aced by SGO on the feedback multiplexer. PAlCE16V8 Macrocell 2-52 PALCE16V8 Family From Adjacent Pin 164938-4 AMD Configuration Options Each macrocell can be configured as one of the following: registered output, combinatorial output, combinatorial 110, or dedicated input. In the registered output configuration, the output buffer is enabled by the OE pin. In the combinatorial configuration, the buffer is either controlled by a product term or always enabled. In the dedicated input configuration, it is always disabled. With the exception of MCo and MC7, a macrocell configured as a dedicated input derives the input signal from an adjacent 110. MCo derives its input from pin 11 (OE) and MC7 from pin 1 (ClK). The macrocell configurations are controlled by the configuration control word. It contains 2 global bits (SGO and SG1) and 16 local bits (SlOo through Sl07 and Sl1 0 through Sl17). SGO determines whether registers will be allowed. SG1 determines whether the PAlCE16V8 will emulate a PAL 16R8 family or a PAL 1OH8 family device. Within each macrocell, SlOx, in conjunction with SG1, selects the configuration of the macroce II , and Sl1x sets the output as either active low or active high for the individual macrocell. The configuration bits work by acting as control inputs forthe multiplexers in the macrocell. There are four mUltiplexers: a product term input, an enable select, an output select, and a feedback select multiplexer. SG1 and SLOx are the control signals for all four multiplexers. In MCo and MC7, SGO replaces SG1 on the feedback multiplexer. This accommodates ClK being the adjacent pin for MC7 and OE the adjacent pin for MCo. ~ use the feedback path of MC7 and pin 11 will use the feedback path of MCo. Combinatorial 1/0 in a Non-Registered Device The control bit settings are SGO =1, SG1 = 1, and SlOx = 1. Only seven product terms are available to the OR gate. The eighth product term is used to enable the output buffer. The signal at the I/O pin is fed back to the AND array via the feedback multiplexer. This allows the pin to be used as an input. Because ClK and OE are not used in a non-registered device, pins 1 and 11 are available as inputs. Pin 1 will use the feedback pathof MC7 and pin 11 will use the feedback path of MCo. Combinatorial 1/0 in a Registered Device The control bit settings are SGO = 0, SG1 = 1 and SlOx = 1. Only seven product terms are available to the OR gate. The eighth product term is used as the output enable. The feedback signal is the corresponding I/O signal. Dedicated Input Configuration The control bit settings are SGO = 1, SG 1 =0 and SlOx = 1. The output buffer is disabled. Except forMCo and MC7 the feedback signal is an adjacent 110. For MCo and MC7 the feedback Signals are pins 1 and 11. These configurations are summarized in Table 1 and illustrated in Figure 2. Table 1. Macrocell Configuration Registered Output Configuration The control bit settings are SGO =0, SG1 = 1 and SlOx = O. There is only one registered configuration. All eight product terms are available as inputs to the OR gate. Data polarity is determined by Sl1 x. The flip-flop is loaded on the lOW-to-HIGH transition of ClK. The feedback path is from Q on the register. The output buffer is enabled by OE. Combinatorial Configurations The PAlCE16V8 has three combinatorial output configurations: dedicated output in a non-registered device, 110 in a non-registered device and 110 in a registered device. Dedicated Output in a Non-Registered Device The control bit settings are SGO = 1, SG 1 = 0 and SlOx = O. All eight product terms are available to the OR gate. Although the macrocell is a dedicated output, the feedback is used, with the exception of pins 15 and 16. Pins 15 and 16 do not use feedback in this mode. Because ClK and OE are not used in a non-registered device, pins 1 and 11 are available as input Signals. Pin 1 will SGO SG1 SLOx Cell Configuration Devices Emulated Device Uses Registers 0 Registered Output PAL 16RB, 16R6, 0 1 16R4 1 Combinatorial 1/0 PAL16R6,16R4 0 1 Device Uses No Registers PAL1OHB, 12H6, 0 Combinatorial 1 0 14H4, 16H2, 10LB, Output 12L6, 14L4, 16L2 PAL12H6,14H4, 1 0 1 Input 16H2, 12L6, 14L4, 16L2 1 1 1 Combinatorial I/O PAL16LB Programmable Output Polarity The polarity of each macrocell can be active-high or active-low, either to match output Signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is through a programmable bit Sl1 x which controls an exclusive-OR gate at the output of the AND! OR logiC. The output is active high if Sl1 x is 1 and active low if SL1x is O. PALCE16V8 Family 2·53 ~ AMD OE--------------, OE----------------------~ >---~ D Y)--~ Q D Q elK Registered Active High Registered Active Low Combinatorial 1/0 Active High Combinatorial 1/0 Active Low Combinatorial Output Active High Combinatorial Output Active Low Notes: 1. Feedback is not available on pins 15 and 16 in the combinatorial output mode. Adjacent 1/0 pin 2. This configuration is not available on pins 15 and 16. Note 2 Dedicated Input Figure 2. Macrocell Configurations 2-54 PALCE16V8 Family 164938-5 AMD ~ Power-Up Reset Programming and Erasing All flip-flops power up to a logic LOW for predictable system initialization. Outputs of the PALCE16V8 will depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be HIGH. If combinatorial is selected, the output will be a function of the logic. The PALCE16V8 can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erase operation is required. . Register Preload The register on the PALCE16V8 can be preloaded from the output pins to facilitate functional testing of complex state machine· designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. Security Bit A security bit is provided on the PALCE16V8 as a deterrent to unauthorized copying of the array configuration patterns. Once programmed, this bit defeats readback and verification of the programmed pattern by a device programmer, securing proprietary designs from competitors. The bit can only be erased in conjunction with the array during an erase cycle. Quality and Testability The PALCE16V8 offers a very high level of built-in quality. The erasability of the device provides a direct means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. , Technology The high-speed PALCE16V8 is fabricated with AMD's advanced electrically erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with TTL devices. This technology provides strong input clamp diodes, output slew-rate control, and a grounded substrate for clean switching. Electronic Signature Word An electronic Signature word is provided in the PALCE16V8 device. It consists of 64 bits of programmable memory that can contain user-defined data. The signature data is always available to the user independent of the security bit. PALCE16V8 Family 2-55 ~AMD LOGIC DIAGRAM 4 ClKlI 11 2 15 6 19 0 ~ 4 7 8 1 1 ~ th 10 ~ 0 ic ~ II & ~, \SlO, >p 7 ~ U ~~~l~ iJ." ..... o 0 ~ ""UlIII V0 7 .~ 11 ~ OSlO, ~~ 1O I 8 _~ .n ;c ~r VCC.,f=t0C & >:£:D- ~~ ~, \SlO, ox 15 , if _ -lim JW 11 3 ox SGI ~~ 10 ~ 16 . 4 t10 VCC;=tg~ 4 ~ J ",+SlO. SGI - SGI ~~ '0 ~D- . ... .:rr. SlO. -n1- 10 VCC.;:=t0O 4 1 st ~SlO. w----t ~ 11 O""W 31 ~ --_.- ~ ox .A 24 .:n:. SlO, >t,D- ~ 23 4 IZiivCC 1- VCCFO ox Slh ~ ~ ox .... ..... ..... -mL" SGI31: SlO. 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 ClK OE 164938-6. 2-56 PALCE16V8 Family AMD ~ LOGIC DIAGRAM (continued) o 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 a.K OE I Sl0 ll1. ~ , >-r~ 39 m - IT • ~ 32 1- 1O vcc.;;t0o ~il ~.. ': " = ~- ... ox SGI.:rt:. ~~ 1Ti vee;:t~~ ~ 40 47 • ~ ~ >r~ ~II \SlO, II 001 SlO, 1IT 1O vee.;:t0o ~il l ~ J l. Sl0 SGI ..... ~ , >-r> 55 Ill-1. -mI ~ ox .A 48 SlO, . ~ 0: ~ "~ -mI mb ox SGI ~il 1Ti ~ ~ .. " 0: . 1P ~ 56 .:rL. SlO, 1O vee.;:too l II ~Slo. ll.2I'" 63 ~ rn--t -mJ 1100 mb ox m1.:rL.SlOg 50 k.Q I I I I I . II Programmlng ESD Protection and Clamping Ll2n.!~!l. -= ____ _ Typical Input Vee Vee > 50 k.Q Provides ESD Protection and Clamping Preload Circuitry Typical Output Feedback Input 164938-15 * Device Rev Letter Filter Only Filter and Pullups PALCE16VSH-10 E,F I PALCE16VSH-15 D, E,F,G H, I PALCE16VSQ-15 D,G H PALCE16VSH-25 D,G H PALCE16VSQ-25 D,G H 2-76 Topside Marking: AMD CMOS PLD's are marked on the top of the package in the following manner: PALCEXXXX Date Code (3 numbers) Lot 10 (4 characters)- -(Rev. Letter) The Lot /D and Rev Letter are separated by two spaces. PALCE16V8 Family AMD~ POWER-UP RESET The PALCE16V8 has been designed with the capability to reset during system power-up. Following power-up, all flip-flops will be reset to LOW. The output state will be HIGH independent of the logic polarity. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset Parameter Symbol and the wide range of ways Vec can rise to its steady state, two conditions are required to insure a valid power-up reset. These conditions are: • • The Vee rise must be monotonic. Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Descriptions Power-Up Reset Time tPR Min ts Input or Feedback Setup Time tWL Clock Width LOW Power Registered Output Clock Unit ns See Switching Characteristics Vee 4Vf. L. Max 1000 tPR -.l ZI ~ tWL 164938-17 Power-Up Reset Waveform PALCE16V8 Family 2-77 ~ AMD TYPICAL THERMAL CHARACTERISTICS 14 Devices (PALCE16V8H-10/4) Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Sjc Sja Sjma Typ. Parameter Description Thermal Impedance, Junction to Case Thermal Impedance, Junction to Ambient Thermal Impedance, Junction to Ambient with Air Flow 200 Ifpm air 400 Ifpm air 600 Ifpm air 800 Ifpm air PDIP PLCC Unit 25 71 61 55 51 47 22 64 55 51 47 45 °CIW °CIW °CIW °CIW °CIW °CIW 15 Devices (PALCE16V8H-7/5) Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Typ. Parameter Description Sjc Thermal Impedance, Junction to Case Sja Thermal Impedance, Junction to Ambient Sjma Thermal Impedance, Junction to Ambient with Air Flow 200 Ifpm air 400 Ifpm air 600 Ifpm air 800 Ifpm air PDIP PLCC Unit 29 70 64 58 53 23 61 53 47 44 °CIW X X °CIW °CIW °CIW °CIW °CIW Plastic ale Considerations The data listed for plastic S~ are for reference only and are not recommended for use in calculating junction temperatures. The heat-flow paths in plastic-encapsulated devices are complex, making the Sjc measurement relative to a specific location on the package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom center of the package. Furthermore, S~ tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a similar environment. 2·78 PALCE16V8 Family ~ PALLV16V8-10 Advanced Micro Devices Low-Voltage 20-Pin EE CMOS Universal Programmable Array Logic DISTINCTIVE CHARACTERISTICS • • • • • • Low-voltage operation, 3.3 V JEDEC compatible - Vee = +3.0 V to +3.6 V Pin, function and fuse-map compatible with all 20-pin GAL devices Electrically-erasable CMOS technology provides reconfigurable logic and full testability Direct plug-in replacement for the PAL16R6 series and most of the PAL 1OH6 series Designed to interface with both 3.3 V and 5 V logic • • • • • • • Programmable output polarity Programmable enable/disable control Preloadable output registers for testability Automatic register reset on power up Cost-effective 20-pin plastic DIP, PLCC, and SOIC packages Extensive third-party software and programmer support through FusionPLD partners Fully tested for 100% programming and functional yields and high reliability Outputs programmable as registered or combinatorial in any combination GENERAL DESCRIPTION The PALLV16V8 is an advanced PAL device built with low-voltage, high-speed, electrically-erasable CMOS technology. It is functionally compatible with all 20-pin GAL devices. The macrocells provide a universal device architecture. The PALLV16V8 will directly replace the PAL 16R8 and PAL 1OH8 series devices, with the exception of the PAL 16C1. The PALLV16V8 utilizes the familiar sum-of-products (AND/OR) architecture that allows users to implement complex logic functions easily and efficiently. Multiple levels of combinatorial logic can always be reduced to sum-of-products form, taking advantage of the very wide input gates available in PAL devices. The equations are programmed into the device through floating-gate cells in the AN D logic array that can be erased electrically. The fixed OR array allows up to eight data product terms per output for logic functions. The sum of these Publication# 1n13 Rev. A Issue Date: June 1993 Amendment/O products feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial with an active-high or active-low output. The output configuration is determined by two global bits and one local bit controlling four multiplexers in each macrocell. AMD's FusionPLD program allows PALLV16V8 designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accu rate, quality support. By ensuring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a designer can use a tool that is already installed and familiar. Please refer to the Software Reference Guide to PLD Compilers for certified development systems and the Programmer Reference Guide for approved programmers. This document contains information on a product under development at Advanced Micro Devices, Inc. The information Is ~~~!~~~~~~,~IP you evaluate this product. AMD reserves the right to change or discontinue work on this proposed product 2-79 ~ PRELIMINARY AMD BLOCK DIAGRAM h -Is Programmable AND Array 32x 64 1100 OE/lg 1/01 1/02 110a 1/04 liDs 1/05 1/07 17713A-1 CONNECTION DIAGRAMS (Top View) DIP/SOIC PLCC ~ elK/I a Vee 11 1/07 12 lIDs 13 1105 14 1104 L. VOs 15 1103 15 1/04 Is 110 2 16 V0 3 17 1101 17 1102 Is OE 2·80 2 1 20 19 • 18 0 Z CI> I~ ~ ~ Note: Pin 1 is marked for orientation. Clock Ground Input Input/Output Output Enable Supply Voltage PALLV16V8-10 lIDs 10 11 12 13 (!) PIN DESIGNATIONS Vee 3 ~ 17713A-2 ClK GND I I/O ~ 9 '01:/1 9 .... ~ 0 b 1100 GND U -= ...J ..!:l 17713A-3 AMD~ PRELIMINARY ORDERING INFORMATION Commerical Products AMD programmable logic products for industrial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL J LV 16 V --- 8 -10 FAMILY TYPE PAL = Programmable Array Logic TECHNOLOGY LV = Low-Voltage P C L NUMBER OF ARRAY INPUTS PACKAGE TYPE P = 20-Pin Plastic DIP (PD 020) J '" 20-Pin Plastic Leaded Chip Carrier (PL 020) S '" 20-Pin Plastic Gull-Wing Small Outline Package (SO 020) OUTPUT TYPE _ _ _ _ _ _ _ _ _--1 V = Versatile NUMBER OF FLIP-FLOPS OPERATING CONDITIONS C = Commercial (O°C to +7S°C) --------1 SPEED -10 = 10 ns tPD Valid Combinations Valid Combinations PALLV16V8-10 I PC, JC, SC Valid Combinations list configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALLV16V8-10 (Com'l) 2-81 ~ AMD PRELIMINARY FUNCTIONAL DESCRIPTION The PALLV16VS is a low-voltage, EE CMOS version of the PALCE16VS. The PALLV16VS isa universal PAL device. It has eight independently configurable macrocells (MCo-MC7). Each macrocell can be configured as registered output, combinatorial output, combinatorial 110 or dedicated input. The programming matrix implements a programmable AN D logic array, which drives a fixed OR logic array. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Pins 1 and 11 serve either as array inputs or as clock (CLK) and output enable (OE), respectively, for all flip-flops. Unused input pins should be tied directly to Vcc or GND. Product terms with all bits unprogrammed (disconnected) assume the logical HIGH state and product terms with both true and complement of any input signal connected assume a logical LOW state. The programmable functions on the PALLV16VS are automatically configured from the user's design specification, which can be in a number of formats. The design specification is processed by development software to verify the design and create a programming file. This file, once downloaded to a programmer, configures the device according to the user's desired function. The user is given two design options with the PALLV16VS. First, it can be programmed as a standard PAL device from the PAL16RS and PAL10HS series. The PAL programmer manufacturer will supply device codes for the standard PAL device architectures to be used with the PALLV16VS. The programmer will program the PALLV16VS in the corresponding architecture. This allows the user to use existing standard PAL device JEDEC files without making any changes to them. Alternatively, the device can be programmed as a PALLV16V8. Here the user must use the PALLV16V8 device code. This option allows full utilization of the macrocell. To Adjacent Macrocell I/Ox >-----41~ 0 QI-----4 From Adjacent Pin -In macrocells MCo and MC7, SG1 is replaced by SGO on the feedback multiplexer. Figure 1. PALLV16V8 Macrocell 2-82 PALLV16V8-10 17713A-4 AMD~ PRELIMINARY Configuration Options Each macrocell can be configured as one of the following: registered output, combinatorial output, combinatorial liD, or dedicated input. In the registered output configuration, the output buffer is enabled by the DE pin. In the combinatorial configuration, the buffer is either controlled by a product term or always enabled. In the dedicated input configuration, it is always disabled. With the ~xception of MCo and MCl, a macrocell configured as a dedicated input derives the input signal from an adjacent liD. MCo derives its input from pin 11 (DE) and MCl from pin 1 (ClK). The macrocell configurations are controlled by the configuration control word. It contains 2 global bits (SGO and SG1) and 16 local bits (SlOo through SlOl and Sl1 0 through Sl17). SGO determines whether registers will be allowed. SG1 determines whether the PAllV16V8 will emulate a PAL 16R8 family or a PAL 1OH8 family device. Within each macrocell, SlOx, in conjunction with SG1, selects the configuration of the macrocell, and Sl1 x sets the output as either active low or active high for the individual macrocell. The configuration bits work by acting as control inputs forthe multiplexers in the macrocell. There are four mUltiplexers: a product term input, an enable select, an output select, and a feedback select multiplexer. SG1 and SlOx are the control signals for all four multiplexers. In MCo and MCl, SGO replaces SG1 on the feedback multiplexer. This accommodates ClK being the adjacent pin for MCl and DE the adjacent pin for MCo. Registered Output Configuration The control bit settings are SGO = 0, SG1 = 1 and SlOx =O. There is only one registered configuration. All eight product terms are available as inputs to the OR gate. Data polarity is determined by Sl1 x. The flip-flop is loaded on the lOW-to-HIGH transition of ClK. The feedback path is from Q on the register. The output buffer is enabled by DE. will use the feedback path of Me? and pin 11 will use the feedback path of MCo. Combinatorial 1/0 In a Non-Registered Device The control bit settings are SGO = 1, SG1 = 1, and SlOx =1. Only seven product terms are available to the OR gate. The eighth product term is used to enable the output buffer. The signal at the liD pin is fed back to the AND array via the feedback multiplexer. This allows the pin to be used as an input. Because ClK and DE are not used in a non-registered device, pins 1 and 11 are available as inputs. Pin 1 will use the feedback path of MC7and pin 11 will use the feedback path of MCo. Combinatorial 1/0 in a Registered Device The control bit settings are SGO = 0, SG1 = 1 and SlOx = 1. Only seven product terms are available to the OR gate. The eighth product term is used as the output enable. The feedback signal is the corresponding liD signal. Dedicated Input Configuration The control bit settings are SGO = 1, SG1 = 0 and SlOx =1. The output buffer is disabled. Except for MCo and MCl the feedback signal is an adjacent liD. For MCo and MCl the feedback signals are pins 1 and 11. These configurations are summarized in Table 1 and illustrated in Figure 2. Table 1. Macrocell Configuration SGO SG1 SLOx Cell Configuration Devices Emulated Device Uses Registers 0 1 0 0 1 1 Registered Output Combinatorial I/O Device Uses No Registers 1 0 0 Combinatorial Output 1 0 1 Input 1 1 1 Combinatorial I/O Combinatorial Configurations The PAllV16V8 has three combinatorial output configurations: dedicated output in a non-registered device, I/O in a non-registered device and liD in a registered device. Dedicated Output In a Non-Registered Device The control bit settings are SGO = 1, SG1 = 0 and SlOx =O. All eight product terms are available to the OR gate. Although the macrocell is a dedicated output, the feedback is used, with the exception of MC3 and MC4. MC3 and MC4 do not use feedback in this mode. Because ClK and DE are not used in a non-registered device, pins 1 and 11 are available as input signals. Pin 1 PAL 16R8, 16R6, 16R4 PAL 16R6, 16R4 PAL 1OH8, 12H6, "14H4. 16H2. 10L8. 12L6, 14L4. 16L2 PAL 12H6, 14H4, 16H2. 12L6, 14L4, 16L2 PAL16L8 Programmable Output Polarity The polarity of each macrocell can be active-high or active-low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is through a programmable bit Sl1 x which controls an exclusive-OR gate at the output of the ANDI OR logiC. The output is active high if Sl1 x is 1 and active low if Sl1x is O. PALLV16V8-10 2-83 ~ PRELIMINARY AMD OE-----------------------, OE---------------------, XJ-----t D Registered Active Low Registered Active High Combinatorial 110 Active Low Combinatorial 1/0 Active High Combinatorial Output Active Low Combinatorial Output Active High Notes: 1. Feedback is not available on pins 15 and 16 in the combinatorial output mode. Adjacent 1/0 pin Note 2 2. The dedicated-input configuration is not available on pins 15 and 16. Dedicated Input 17713A-5 Figure 2. Macrocell Configurations 2-84 Q PALLV16V8-10 PRELIMINARY \ Benefits of Lower Operating Voltage AMD~ Security Bit The PALLV16V8 has an operating voltage range of 3.0 V to 3.6 V. Low voltage allows for lower operating power consumption, longer battery life, and/or smaller batteries for notebook applications. Because power is proportional to the square of the voltage, reduction of the supply voltage from 5.0 V to 3.3 V significantly reduces power consumption. This directly translates to longer battery life for portable applications. Lower power consumption can also be used to reduce the size and weight of the battery. Thus, 3.3-V designs facilitate a reduction in the form factor. A lower operating voltage results in a reduction of I/O voltage swings. This reduces noise generation and provides a less hostile environment for board design. Lower operating voltage also reduces electromagnetic radiation noise and makes obtaining FCC approval easier. Power-Up Reset All flip-flops powerupto a logic LOWforpredictable system initialization. Outputs of the PALLV16V8 will depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be HIGH. If combinatorial is selected, the output will be a function of the logic. Register Preload The register on the PALLV16V8 can be pre loaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. The preload function is not disabled by the security bit. This allows functional testing after the security bit is programmed. A security bit is provided on the PALLV16V8 as a deterrent to unauthorized copying of the array configuration patterns. Once programmed, this bit defeats readback of the programmed pattern by a device programmer, securing proprietary designs from competitors. However, programming and verification are also defeated by the security bit. The bit can only be erased in conjunction with the array during an erase cycle. Electronic Signature Word An electronic signature word is provided in the PALLV16V8 device. It consists of 64 bits of programmable memory that can contain user-defined data. The signature data is always available to the user independent of the security bit. Programming and Erasing The PALLV16V8 can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its unprogrammed state. Erasure is automatically performed by the programming hardware. No special erase operation is required. Quality and Testability The PALLV16V8 offers a very high level of built-in quality. The erasability if the device provides a direct means of verifying performance of all the AC and DC . parameters. In addition, this verifies complete programmability and functionality of the device to yield. the highest programming yields and post-programming function yields in the industry. Technology The high-speed PALLV16V8 is fabricated with AMD's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. This technology provides strong input-clamp diodes, output slew-rate control, and a grounded substrate for clean switching. PALLV16V8-10 2-85 ~ PRELIMINARY AMD LOGIC DIAGRAM 4 11 12 15 6 19 3 4 7 8 31 elK/IO 1 ~i=l 10 L 0 nr>-tD- Ft~=t~* , J SGI \SlO, - o 0 7 11 ~ ox SGO ~i=l 10 l 8 , J SlO, -n1- 10 vcc;:t0o ~ >-tD- ~. \SlO, SGI ~ox 15 su, ... . 3 ~II 10 .~ " Sll, ... I2QIv cc 10 vee o-joo v~ >0- r--lIID mrJ " lox _ _ SGI ~i=l 10 l 16 tIh . ~. ~ 10 vee;t~~ , J \SlO. SGI - >--tD- 23 .:rr. 51.0, r"ox [ill! 1105 ~ " Sll • .A 4 " --- SGII~ SlO. ~ 10 L J mfi 10 vcc.;=l.0o ~i=l , . >-tD- ~::1}d 24 SGI 31 ... 5 \SlO. - ~I illL " Sll • ox SGI SlO. 17713A-6 3 4 2-86 7 8 11 12 15 16 19 20 23 24 27 28 31 PALLV16V8-10 ClK OE AMD~ PRELIMINARY LOGIC DIAGRAM (continued) 3 4 7 8 11 12 15 16 1920 23 24 2728 31 CLK OE SGl rtth ~. 56~ 63~ VCC~m.+11 R~pl,l'"\&' >tD- ~ J ~~';)O-"""""'L.i11'J.-..r.1""I/OO • • • SLl, 18 SLO, iTIb o 11 OX!--- 9 '---------<(f- --£I:1) 001 9 o 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 GND~ 17713A-6 (concluded) PALLV16V8-10 2-87 ~ PRELIMINARY AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Commercial (C) Devices Ambient Temperature with Power Applied ............. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air ............ O°C to +75°C Supply Voltage with Respect to Ground ............. -0.5 V to + 7.0 V Supply Voltage (Vee) with Respect to Ground ........... +3.0 V to +3.6 V DC Input Voltage ............ -0.5 V to Vee + 0.5 V DC Output or 110 Pin Voltage ............... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Latehup Current (TA = -40°C to 85°C) ................... 100 rnA, Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERICAL operating ranges unless otherwise specified PRELIMINARY , Parameter Symbol VOH VOL Parameter Description Output HIGH Voltage Output LOW Voltage Min Test Conditions = VIH or VIL = Min VIN = VIH or VIL Vee = Min VIN IOH Vee IOH IOL IOL = -2 rnA = -100 IlA = 2 rnA = 100 ~A Max Unit 2.4 V Vee-O.2 V V 0.4 V 0.1 V 2.0 V VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) 0.8 V IIH Input HIGH Leakage Current 10 ilL Input LOW Leakage Current = Vee, Vee = Max (Note 2) VIN = 0 V, Vee = Max (Note 2) VOUT = Vee, Vee = Max VIN = VIH or VIL (Note 2) VOUT = 0 V, Vee = Max VIN = VIH or VIL (Note 2) VOUT = 0.5 V, Vee = Max (Note 3) Outputs Open (lOUT = 0 rnA) Vee = Max, f = 15 MHz 10 IlA IlA IlA -100 IlA -150 rnA 55 rnA IOZH Off-State Output Leakage Current HIGH IOZl Off-State Output Leakage Current LOW Ise Output Short-Circuit Current Icc Supply Current VIN -100 -30 Notes: 1. These are absolute values with respect to device ground and all overshoots due to system or tester noise are included. 2. 110 pin leakage is the worst case of IJL and IOZL (or IJH and IOZH). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-88 PALLV16V8-10 (Com'l) AMD~ PRELIMINARY CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Description Test Condition Input Capacitance VIN= 2.0 V Output Capacitance VOUT= 2.0 V Typ Unit Vcc = 5.0 V, TA = 25°C, 5 pF f = 1 MHz 8 pF I I Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) PRELIMINARY Parameter Symbol Min Parameter Description tPD Input or Feedback to Combinatorial Output ts Setup Time from Input or Feedback to Clock tH Hold Time tco Clock to Output tWL Clock Width tWH Max Unit 10 ns 7.5 0 ns 7.5 ns LOW 6 ns HIGH 6 ns External Feedback I 1/(ts+tco) 66.7 MHz 71.4 MHz fMAX Maximum Frequency (Note 3) tpzx OE to Output Enable 10 ns tpxz OE to Output Disable 10 ns Internal Feedback (fCNT) No Feedback I 1/(tS+tH) 83.3 MHz tEA Input to Output Enable Using Product Term Control 10 ns tEA Input to Output Disable Using Product Term Control 10 ns Notes: , 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested,' but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALLV16V8-10 (Com'l) 2-89 ~ PRELIMINARY AMD SWITCHING WAVEFORMS Input or Feedback _ _ ___ Input or Feedback Combinatorial Output VT ~VT -~""'---VT 17713A-7 Clock - - - - - - - - . . . ; . Registered Output _ _ _ _ _ _ _ _..... 17713A-8 Combinatorial Output Registered Output Input tER Clock VT Output 17713A-10 17713A-9 Input to Output Disable/Enable Clock Width tpxz tpzx VT Output 17713A-11 OE to Output Disable/Enable Notes: 1. Vr = 1.5 V for Input Signals and 1.65 V for Output Signals. 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns -5 ns typical. 2-90 tEA PALLV16V8-10 AMD~ PRELIMINARY KEY TO SWITCHING WAVEFORMS WAVEFORM INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L IllZZ May Change from L to H Will be Changing from L to H 'XXXXXX Don't Care, Any Change Permitted Changing, State Unknown ]}) Does Not Apply Center Line is HighImpedance "Off" State CK KSOOOO10-PAL SWITCHING TEST CIRCUIT 3.3 V Output D--""---""---1r.l Test Point 17713A-12 Specification 51 52 CL R1 R2 Measured Output Value tPD, teo Closed Closed 1.65 V tpzx, tEA Z~ Z~ 1.65 V Z~ tpxz, tER H H: Open L: Closed ~Z: L~Z: Open Closed H: Closed L: Open Z ~ H ~Z: 30 pF 1.6K Closed L~Z:Open 5 pF PALLV16V8-10 1.6K H ~ Z: VOH - 0.5 V L ~ Z: VOL + 0.5 V 2-91 ~ PRELIMINARY AMD ENDURANCE CHARACTERISTICS The PALLV16V8 is manufactured using AM D's advanced Electrically Erasable process. This technology uses an EE cell to replace the fuse link used in bipolar Symbol parts. As a result, device can be erased and reprogrammed - a feature which allows 100% testing at the factory. Parameter Test Conditions tOR Min Pattern Data Retention Time Max Storage Temperature Max Operating Temperature 20 Years N Min Reprogramming Cycles Normal Programming Conditions 100 Cycles Min Unit 10 Years ROBUSTNESS FEATURES The PALLV16V8 has some unique features that make it extremely robust, especially when operating in highspeed design environments. Pull-up resistors on inputs and I/O pins cause unconnected pins to default to a known state. Input clamping circuitry limits negative overshoot, eliminating the possibility of false clocking caused by subsequent ringing. A special noise filter makes the programming circuitry completely insensitive to any positive overshoot that has a pulse width of less than about 100 ns. INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee Vee >50kn I I I I ESD Protection and Clamping I . II Programmmg L'2n~~!!: =- ____ _ Typical Input Vee Vee >50kn Provides ESD Protection and Clamping Preload Circuitry =Typical Output 2-92 PALLV16V8-10 Feedback Input 17713A-14 AMD~ PRELIMINARY POWER-UP RESET The PALLV16V8 has been designed with the capability to reset during system power-up. Following power-up, all flip-flops will be reset to LOW. The output state will be HIGH independent of the logic polarity. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vee can rise to its steady state, two conditions are required to insure a valid power-up reset. These conditions are: • The Vee rise must be monotonic. • Following reset, the clock input must not be driven from LOWto HIGH until all applicable input and feedback setup times are met. Parameter Symbol Parameter Descrlptlo'ns Min tPR Power-Up Reset Time ts Input or Feedback Setup Time twL Max Unit 1000 ns See Switching Characteristics Clock Width LOW t Vee 4V Power Registered Output Clock ~ tPR zL{ ~~ PALLV16V8·10 j 17713A-15 2·93 ~ AMD 9 2-94 GL'.DeVices _ COM'L:-15/25 PALCE16V8Z FAMILY Advanced Micro Devices Zero-Power 20-Pin EE CMOS Universal Programmable Array Logic DISTINCTIVE CHARACTERISTICS • • • • • • Zero-Power CMOS technology - 15 J,JA Standby Current - 15 ns propagation delay for "-15" version - 25 ns propagation delay for "-25" version Unused product term disable for reduced power consumption Available In Industrial operating range - Tc ~ IND: -15/25 =-40°C to +85°C - Vcc = +4.5 V to +5.5 V HC- and HCT-Compatible inputs and outputs Pin, function and fuse-map compatible with all 20-pin GAL devices Electrically-erasable CMOS technology provides reconfigurable logic and full testability • • • • • • • • • Direct plug-In replacement for the PAL16R8 series and most of the PAL 10H8 series Outputs programmable as registered or comblnato rial In any combination Programmable output polarity Programmable enable/disable control Preloadable output registers for testability Automatic register reset on power up Cost-effective 20-pin plastic DIP and PLCC packages Extensive third-party software and programmer support through FusionPLD partners Fully tested for 100% programming and functlonal yields and high reliability GENERAL DESCRIPTION The PALCE16V8Z is an advanced PAL device built with zero-power, high-speed, electrically-erasable CMOS technology. It is functionally compatible with all 20-pin GAL devices. The macrocellsprovide a universal device architecture. The PALCE16V8Z will directly replace the PAL 16R8 and PAL1OH8 series devices, with the exception of the PAL 16C1. The fixed OR array allows up to eight data product terms peroutputfor logic functions. The sum ofthese products feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial with an active- high or active-low output. The output configuration is determined by two global bits and one local bit controlling four multiplexers in each macrocell. The PALCE16V8Z provides zero standby power and high speed. At 15 J,JA maximum standby current, the PALCE16V8Z allows battery powered operation for an extended period. AMD's FusionPLD program allows PALCE16V8Z designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a designer can use a tool that is already installed and familiar. Please refer to the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. The PALCE16V8Z utilizes the familiar sum-of-products (AND/OR) architecture that allows users to implement complex logic functions easily and efficiently. Multiple levels of combinatorial logic can always be reduced to sum-of-products form, taking advantage of the very wide input gates available in PAL devices. The equations are programmed into the device through floating-gate cells in the AN D logic array that can be erased electrically. Publication# 13061 Rev. C Issue Date: June 1993 Amendment/O 2-95 ~AMD BLOCK DIAGRAM 11-18 Programmable AND Array 32 x 64 OEll9 1/00 1/01 1/02 1/03 1/04 1106 1105 1/07 13061C-l CONNECTION DIAGRAMS Top View DIP PLCC 12 elK/I 0 Vee I, 1/07 12 1/0 6 13 1/05 14 1/04 15 1/0 3 16 1/02 17 11O, 18 GND 18 ClK/lo Vee 1/0 7 GND OE/Igi/Oo lID, 1/00 PIN DESIGNATIONS L>E!l g ClK 13061C-2 GND Note: 1/0 Pin 1 is marked for orientation OE Vee 2-96 I, PALCE16V8Z Family Clock Ground Input InputlOutput Output Enable Supply Voltage AMO l1 ORDERING INFORMATION Industrial Products AMD programmable logic products for industrial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: _---'T PAL CE 16 V 8 Z -15 P -~-~ FAMILY TYPE PAL = Programmable Array Logic TECHNOLOGY CE = CMOS Electrically Erasable NUMBER OF ARRAY INPUTS OUTPUT TYPE V = Versatile OPERATING CONDITIONS I = Industrial (-40°C to +85°C) = Commercial (0°0 to +75°C) -----------1 o PACKAGE TYPE P = 20-Pin Plastic DIP (PO 020) J = 20-Pin Plastic Leaded Chip Carrier (PL 020) NUMBER OF FLIP-FLOPS POWER Z = Zero Power (15 j.1A Icc Standby) SPEED -15 = 15 ns tPD -25 = 25 ns tPD Valid Combinations Valid Combinations PALCE16V8Z-15 PALCE16V8Z-25 I I PI,JI, PC, JC Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE16V8Z-15/25 (Com'I, INO) 2-97 ~AMD FUNCTIONAL DESCRIPTION The PALCE16V8Z is the zero-power .version of the PALCE16V8. It has all the architectural features of the PALCE16V8. In addition, the PALCE16V8Z has zero standby power and unused product term disable. The PALCE16V8Z is a universal PAL device. It has macrocells eight independently configurable (MCo-MC7). Each macrocell can be configured as registered output, combinatorial output, combinatorial lID or dedicated input. The programming matrix implements a programmable AND logic array, which drives a fixed OR logic array. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Pins 1 and 11 serve either as array inputs or as clock (CLK) and output enable (DE), respectively, for all flip-flops. Unused input pins should be tied directly to Vec or GND. Product terms with all bits unprogrammed (disconnected) assume the logical HIGH state and product terms with both true and complement of any input signal connected assume a logical LOW state. The programmable functions on the PALCE16V8Z are automatically configured from the user's design specification, which can be in a number of formats. The design specification is processed by development software to verify the design and create a programming file. This file, once downloaded to a programmer, configures the device according to the user's desired function. The user is given two design options with the PALCE16V8Z. First, it can be programmed as a standard PAL device from the PAL 16R8 and PAL1OH8 series. The PAL programmer manufacturer will supply device codes for the standard PAL device architectures to be used with the PALCE16V8Z. The programmer will program the PALCE16V8Z in the corresponding architecture. This allows the userto use existing standard PAL device JEDEC files without making any changes to them. Alternatively, the device can be programmed as a PALCE16V8Z. Here the user must use the PALCE16V8Z device code. This option allows full utilization of the macrocell. To Adjacent Macrocell ~~~----------------------------------------~1 OE 10~--"'" 11 1 Vcc :' >---~a---l D 00 01 IIOX Q~----l From Adjacent Pin ~/n macrocells MCo and MCl, SG1 is replaced by SGO on the feedback multiplexer. Figure 1. PALCE16V8Z Macrocell 2-98 PALCE16V8Z Family 13061C-4 AMD~ Configuration Options Each macrocell can be configured as one of the following: registered output, combinatorial output, combinatorial liD, or dedicated input. In the registered output configuration, the output buffer is enabled by the DE pin. In the combinatorial configuration, the buffer is either controlled by a product term or always enabled. In the dedicated input configuration, it is always disabled. With the exception of MCo and MC7, a macrocell configured as a dedicated input derives the input signal from an adjacent I/O. MCo derives its input from pin 11 (DE) and MC7 from pin 1 (ClK). The macrocell configurations are controlled by the configuration control word. It contains 2 global bits (SGO andSG1) and 16 local bits (SlOothrough Sl07and Sl10 through Sl 17). SGO determines whether registers will be allowed. SG1 determines whetherthe PAlCE16V8Z will emulate a PAL 16R8 family or a PAL 1OH8 family device. Within each macrocell, SlOx, in conjunction with SG1, selects the configuration of the macrocell, and Sl1 x sets the output as either active low or active high for the individual macrocell. The configuration bits work by acting as control inputs forthe multiplexers in the macrocell. There are four mUltiplexers: a product term input, an enable select, an output select, and a feedback select multiplexer. SG1 and SlOx are the control signals for all four multiplexers. In MCo and MC7, SGO replaces SG1 on the feedback multiplexer. This accommodates ClK being the adjacent pin for MC7 and DE the adjacent pin for MCo. Combinatorial 1/0 In a Non-Registered Device The control bit settings are SGO =1, SG1 = 1, and SlOx = 1. Only seven product terms are available to the OR gate. The eighth product term is used to enable the output buffer. The signal at the liD pin is fed back to the AND array via the feedback multiplexer. This allows the pin to be used as an input. Because ClK and DE are not used in a non-registered device, pins 1 and 11 are available as inputs. Pin 1 will use the feedback path of MC7 and pin 11 will use the feedback path of MCo. Combinatorial 1/0 in a Registered Device The control bit settings are SGO =0, SG1 = 1 and SlOx = 1. Only seven product terms are available to the OR gate. The eighth product term is used as the output enable. The feedback signal is the corresponding 1/0 signal. Dedicated Input Configuration The control bit settings are SGO =1, SG1 = 0 and SlOx = 1. The output buffer is disabled. Except for MCo and MC7 the feedback signal is an adjacent liD. For MCo and MC7 the feedback signals are pins 1 and 11. These configurations are summarized in Table 1 and illustrated in . Figure 2. Table 1. Macrocell Configuration SGO SG1 SLOx Cell Configuration Devices Emulated Device Uses Registers Registered Output Configuration The control bit settings are SGO = 0, SG1 = 1 and SlOx = O. There is only one registered configuration. All eight product terms are available as inputs to the OR gate. Data polarity is determined by Sl1 x. The flip-flop is loaded on the lOW-to-HIGH transition of elK. The feedback path is from Q on the register. The output buffer is enabled by DE. Combinatorial Configurations The PAlCE16V8Z has three combinatorial output configurations: dedicated output in a non-registered device, liD in a non-registered device and liD in a registered device. Dedicated Output In a Non-Registered Device The control bit settings are SGO = 1, SG1 = 0 and SlOx = O. All eight product terms are available to the OR gate. Although the macrocell is a dedicated output, the feedback is used, with the exception of MC3 and MC4. MC3 and MC4 do not use feedback in this mode. Because ClK and DE are not used in a non-registered device, pins 1 and 11 are available as input signals. Pin 1 will use the feedback path of MC7 and pin 11 will use the feedback path of MCo. 0 1 0 Registered Output 0 1 1 Combinatorial I/O 1 0 0 Combinatorial Output 1 0 1 Input 1 1 1 Combinatorial I/O PAL 16R8, 16R6, 16R4 PAL16R6,16R4 Device Uses No Registers PAL 1OH8, 12H6, 14H4, 16H2, 10L8, 12L6, 14L4, 16L2 PAL 12H6, 14H4, 16H2, 12L6, 14L4, 16L2 PAL16L8 Programmable Output Polarity The polarity of each macrocell can be active-high or active-low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is through a programmable bit Sl1 x which controls an exclusive-OR gate at the output of the ANDI OR logic. The output is active high if Sl1 x is 1 and active low if Sl1x is O. PALCE16V8Z Family 2-99 ~ AMD OE - - - - - - - - - - - - - , OE - - - - - - - - - - - - . . elK Registered Active Low Registered Active High Combinatorial 1/0 Active Low Combinatorial 110 Active High Combinatorial Output Active Low Combinatorial Output Active High Notes: 1. Feedback is not available on pins 15 and 16 in the combinatorial output mode. 2. The dedicated-input configuration is not available on pins 15 and 16. Adjacent 1/0 pin Note 2 Dedicated Input 13061C-5 Figure 2. Macrocell Configurations 2·100 PALCE16V8Z Family AMD~ Zero-Standby Power Mode Security Bit The PALCE16V8Z features a zero-standby power mode. When none of the inputs switch for an extended period (typically 50 nS)"the PALCE16V8Z will go into standby mode, shutting down most of its internal circuitry. The current will go to almost zero (Icc < 15 JlA). The outputs will maintain the states held before the device went into the standby mode. A security bit is provided on the PALCE16V8Z as a deterrent to unauthorized copying of the array configuration patterns. Once programmed, this bit defeats readback of the programmed pattern by a device programmer, securing proprietary designs from competitors. However, programming and verification are also defeated by the security bit. The bit can only be erased in conjunction with the array during an erase cycle. When any input switches, the internal circuitry is fully enabled and power consumption returns to normal. This feature results in considerable power savings for operation at low to medium frequencies. This savings is illustrated in the Icc vs. frequency graph. Product-Term Disable On a programmed PALCE16V8Z, any product terms that are not used are disabled. Power is cut off from these product terms so that they do not draw current. As shown in the Icc vs frequency graph, product-term disabling results in considerable power savings. This savings is greater at the higher frequencies. Further hints on minimizing power consumption can be found in the Application Note, "Minimizing Power Consumption with Zero-Power PLDs". Power-Up Reset All flip-flops powerupto a logic LOWforpredictable system initialization. Outputs of the PALCE16V8Z will depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be HIGH. If combinatorial is selected, the output will be a function of the logic. Electronic Signature Word An electronic signature' word is provided in the PALCE16V8Z device. It consists of 64 bits of programmable memory that can contain user-defined data. The signature data is always available to the user independent of the security bit. Programming and Erasing The PALCE16V8Z can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its unprogrammed state. Erasure is automatically performed by the programming hardware. No special erase operation is required. Quality and Testability The PALCE16V8Z offers a very high level of built-in quality. The erasability if the device provides a direct means of verifying performance of all the AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to yield the highest programming yields and post-programming function yields in the industry. Technology Register Preload The register on the PALCE16V8Z can be preloaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. The high-speed PALCE 16V8Z is fabricated with AM D's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with HC and HCT devices. This technology provides strong input-clamp diodes, output slew-rate control, and a grounded substrate for clean switching. The preload function is not disabled by the security bit. This allows functional testing after the security bit is programmed. PALCE16V8Z Family 2-101 ~ AMD LOGIC DIAGRAM 4 8 11 12 15 6 19 !3 4 27 8 31 elK/IO 1 ~ii 10 & L J\ 0 t SGI - SlO IT 7 ~D- 7 Sl17 0: FTI::~th» ~ ~ 10 8 3 SlO7 rn vcc;:t.oi 10 - & I J\SlO, SGI r---@l ~D- ~::~»- . ~W 51.1, ~ ~ +~ 10 23 ~ 10 vcc.;=h~ & >-tD- ~::~ I -- -- SGI +ii 10 - " ~ SlO. t 10 • \SlO, SGI - <:: V05 vcc~~~ L J .. llLJ 11 .....-v 24 ...,.". ~~ SlI • .A 5 SlO, \Slo. SGI - 16 31 *" 1111 SGI 4 ---UiJ V0 7 11 OX < .... - 15 V ~ .... m--{k: I2ii1v cc vcc~H ~DSlI, ~ ~illL ...... WiI ~ 11 ox SGI SlO, 13061C-6 3 4 2·102 7 8 11 12 -15 16 19 20 23 24 27 28 31 PALCE16V8Z Family ClJ( OE AMD~ LOGIC DIAGRAM (continued) 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 CLK OE Wb .r ~-----------j~========~11r--~; VCCo-Jg~ 10 16 .rn 7 IS 9 '----------<<}-l.------.wl 001 9 3 4 7 8 11 12 15 16 19 20 2324 27 2S 31 GND~ 130S1C-S (concluded) PALCE16V8Z Family 2-103 ~ AMD PRELIMINARY ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Ambient Temperature with Power Applied ............. -55°C to + 125°C Supply Voltage with Respect to Ground ............ -D.5 V to + 7.0 V DC Input Voltage. . . . . . . . . .. -D.5 V to Vee + 0.5 V DC Output or I/O Pin Vottage ............... -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Latehup Current (TA = O°C to 75°C) .................. 100 mA Commercial (C) Devices Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Ambient Temperature (TA) Operating in Free Air ............ O°C to +75°C Supply Voltage (Vee) with Respect to Ground .......... +4.75 V to +5.25 V Industrial (I) Devices Operating Case Temperature (Te) ............ -40°C to +85°C Supply Voltage (Vee) with Respect to Ground ............ +4.5 V to +5.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges unless otherwise specified PRELIMINARY Parameter Symbol VOH VOL Min Parameter Description Test Conditions Output HIGH Voltage VIN = VIH or VIL Vee = Min 10H = 6 mA IoH = 20 JlA VIN = VIH or VIL Vee = Min IoL Output LOW Voltage 10l = 24mA = 6mA 10l = 20 JlA \ VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Notes 1 and 2) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Notes 1 and 2) IIH Input HIGH Leakage Current ilL Input LOW Leakage Current = Vee, Vec = Max (Note 3) = 0 V, Vee = Max (Note 3) VOUT = Vee, Vee = Max VIN = VIH or VIL (Note 3) VOUT = 0 V, Vee = Max VIN = VIH or VIL (Note 3) Vee = Max (Note 4) VOUT = 0.5 V f = 0 MHz Outputs Open (lOUT = 0 mA) f = 25 MHz Vee = Max IOZH Off-State Output Leakage Current HIGH IOZl Off-State Output Leakage Current LOW Isc Output Short-Circuit Current Icc Supply Current Max Unit 3.84 Vee -0.1 V V V 0.5 0.33 0.1 2.0 V V V V ___O,~ __ V_ VIN 10 VIN -10 JlA JlA 10 JlA -10 JlA -150 mA 15 75 mA -30 JlA Notes: 1. These are absolute values with respect to device ground and all overshoots due to system or tester noise are included. 2. Represents the worst case of HC and HCT standards, allowing compatibility with either. 3. 110 pin leakage is the worst case of hL and lozL (or I/H and lozH ). 4. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-104 PALCE16V8Z-15 (Com'l, IND) AMO~ PRELIMINARY CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Condition Input Capacitance VIN= 2.0 V Output Capacitance VOUT= 2.0 V I I Typ Unit Vee = 5.0 V, TA = 25°C, 5 pF f = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges (Note 2) PRELIMINARY Parameter Symbol Parameter Description tPD Min Input or Feedback to Combinatorial Output Max Unit 15 ns ts Setup Time from Input or Feedback to Clock 10 ns tH Hold Time 0 ns teo tWL Clock to Output Clock Width tWH 10 ns LOW 8 HIGH 8 ns 50 MHz Internal Feedback (teNT) 58.8 MHz No Feedback 62.5 External Feedback 11/(ts+teo) ns fMAx Maximum Frequency (Note 3) tpzx OE to Output Enable 15 ns tpxz OE to Output Disable 15 ns tEA Input to Output Enable Using Product Term Control 15 ns tER Input to Output Disable Using Product Term Control 15 ns 1 1/(tWH+tWL) MHz Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALCE16V8Z-15 (Com'l, INO) 2-105 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Ambient Temperature with Power Applied ............. -55°C to + 125°C Supply Voltage with Respect to Ground ............ -0.5 V to + 7.0 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vec + 0.5 V DC Output or 110 Pin Voltage ............... -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Latchup Current (TA = O°C to 75°C) .................. 100 rnA Commercial (C) Devices Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Ambient Temperature (TA) Operating in Free Air ............ O°C to +75°C Supply Voltage (Vee) with Respect to Ground .......... +4.75 V to +5.25 V Industrial (I) Devices Operating Case Temperature (Te) ............ -40°C to +85°C Supply Voltage (Vee) with Respect to Ground ............ +4.5 V to +5.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges unless otherwise specified Parameter Symbol VOH Parameter Description Test Conditions Output HIGH Voltage VIN = VIH or Vil Vee = Min Min IoH = 6 rnA IOH = 20 JlA IOL = 24 rnA IOL = 6 rnA VOL Output LOW Voltage VIH Input HIGH Voltage Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Notes 1 and 2) ilH Input HIGH Leakage Current ill VIN = VIH or VIL Vee = Min IOL = 20 JlA Guaranteed Input Logical HIGH Voltage for all Inputs (Notes 1 and 2) Max Unit 3.84 Vee-O.l V V V 0.5 0.33 0.1 2.0 V V V V 0.9 V VIN = Vee, Vee = Max (Note 3) 10 Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 3) -10 J.LA J.LA IOZH Off-State Output Leakage Current HIGH VOUT = Vee, Vee = Max VIN = VIH or VIL (Note 3) 10 JlA lozl Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or Vil (Note 3) -10 JlA Isc Output Short-Circuit Current VOUT=0.5 V -150 rnA Icc Supply Current Outputs Open (lOUT = 0 rnA) Vee = Max 15 90 rnA Vee = Max (Note 4) f = 0 MHz f = 25 MHz -30 J.LA Notes: 1. These are absolute values with respect to device ground and all overshoots due to system or tester noise are included. 2. Represents the worst case of HC and HCT standards, allowing compatibility with either. 3. VO pin leakage is the worst case of liL and IOZL (or I/H and lozH ). 4. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2·106 PALCE16V8Z·25 (Com'l, INO) AMO~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Condition Input Capacitance VIN= 2.0 V Output Capacitance VOUT= 2.0 V I 1 ,Typ Unit Vee = 5.0 V, TA = 25°C, 5 pF f = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges (Note 2) Parameter Symbol Parameter Description Min tPD Input or Feedback to Combinatorial Output (Note 3) ts Setup Time from Input or Feedback to Clock tH Hold Time teo tWL fMAx 25 20 Clock Width 8 ns ns 8 ns 33.3 MHz Internal Feedback (teNT) 50 MHz No Feedback 50 HIGH Maximum Frequency (Note 4) ns ns 10 LOW Unit ns 0 Clock to Output tWH Max External Feedback 11/(ts+teo) 1 1/(tS+tH) MHz tpzx OE to Output Enable 25 ns tpxz OE to Output Disable 25 ns tEA Input to Output Enable Using Product Term Control 25 ns tER Input to Output Disable Using Product Term Control 25 ns Notes: 2. See Switching Test Circuit for test conditions. 3. This parameter is tested in Standby Mode. When the device is not in Standby Mode, the tPD will typically be 2 ns faster. 4. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALCE16V8Z·25 (Com'l, INO) 2·107 ~ AMD SWITCHING WAVEFORMS Input or Input or Feedback Combinatorial Output VT Feedback _ _ _---' € . VT -~-VT Clock - - - - - - - Registered Output _ _ _ _ _ _ _----' 13061C-7 13061C-8 Combinatorial Output Registered Output Input tER Clock VT Output 13061C-10 13061C-9 , Clock Width Input to Output Disable/Enable tpxz tpzx VT Output 13061C-11 OE to Output Disable/Enable Notes: 1. Vr = 1.5 V for Input Signals and 2.5 V for Output Signals. 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns - 5 ns typical. 2-108 tEA PALCE16V8Z Family AMD~ KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L IIIII May Change from L to H Will be Changing from L to H X'/YYY:i JJ) g" 12 1/06 13 1/0 5 14 1/0 4 14 1/05 15 1/0 3 Is 1/04 16 1/0 2 16 1/03 17 1/01 h 1/02 Ie lIDo 1 '20 19 • b 9 10 11 ~ Cl ~1I9 Z C) PIN DESIGNATIONS OE -= 1/0 7 174228-2 Vee ~ ~ -I 11 GND ClK GND I I/O ---- --------._--------_._ .. a elK/I 0 1/06 12 13 C1> I~ ~ Note: Pin 1 is marked for orientation. Clock Ground Input Input/Output Output Enable Supply Voltage PALLV16V8Z Family ~ 174228-3 AMO~ ORDERING INFORMATION Industrial Products AMD programmable logic products for industrial applications are available with several ordering options. The order number. (Valid Combination) is formed by a combination of: _----IT ---PAL LV 16 V 8 Z -25 P FAMILY TYPE PAL = Programmable Array Logic TECHNOLOGY LV = Low-Voltage NUMBER OF ARRAY INPUTS OPERATING CONDITIONS I = Industrial (-40°C to +85°C) OUTPUT TYPE - - - - - - - - - - - - - ' V = Versatile PACKAGE TYPE P = 20-Pin Plastic DIP (PD 020) J = 20-Pin Plastic Leaded Chip Carrier (PL 020) S = 20-Pin Plastic Gull-Wing Small Outline Package (SO 020) NUMBER OF FLIP-FLOPS - - - - - - - - - ' POWER Z = Zero Power (15 ~ Icc Standby) SPEED ----------------~ -25 = 25 ns tPD -30 = 30 ns tPD Valid Combinations PALLV16V8Z-25 PALLV16V8Z-30 I PI JI SI I " Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALLV16V8Z-25/30 (INO) 2-117 ~ AMD FUNCTIONAL DESCRIPTION The PALLV16V8Z is a low-voltage, EE CMOS version of the PALCE16V8. In addition, the PALLV16V8Z has zero standby power and unused product term disable. The PALLV16V8Z is a universal PAL device. It has eight independently configurable macrocells (MCo-MC7). Each macrocell can be configured as registered output, combinatorial output, combinatorial 1/0 or dedicated input. The programming matrix implements a programmable AND logic array, which drives a fixed OR logic array. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Pins 1 and 11 serve either as array inputs or as clock (CLK) and output enable (OE), respectively, for all flip-flops. Unused input pins should be tied directly to Vcc or GND. Product terms with all bits unprogrammed (disconnected) assume the logical HIGH state and product terms with both true and complement of any input signal connected assume a logical LOW state. The programmable functions on the PALLV16V8Z are automatically,configured from the user's design specification, which can be in a number of formats. The design specification is processed by development software to verify the design and create a programming file. This file, once downloaded to a programmer, configures the device according to the user's desired function. The user is given two design options with the PALLV16V8Z. First, it can be programmed as a standard PAL device from the PAL16R8 and PAL 1OH8 series. The PAL programmer manufacturer will supply device codes for the standard PAL device architectures to be used with the PALLV16V8Z. The programmer will program the PALLV16V8Z in the corresponding architecture. This allows the userto use existing standard PAL device JEDEC files without making any changes to them. Alternatively, the device can be programmed as a PALLV16V8Z. Here the user must use the PALLV16V8Z device code. This option allows full utilization of the macrocell. To Adjacent Macrocell 11 I/Ox }-----tI.....-t D Qt----i Qt---~I-+-.. 1--____ *In macrocells MCo and MC7, SG1 is replaced by SGO on the feedback multiplexer. Figure 1. PALLV16V8Z Macrocell 2-118 PALLV16V8Z Family From Adjacent Pin 174228-4 AMD~ Configuration Options Each macrocell can be configured as one of the following: registered output, combinatorial output, combinatorial I/O; or dedicated input. In the registered output configuration, the output buffer is enabled by the DE pin. In the combinatorial configuration, the buffer is either controlled by a product term or always enabled. In the dedicated input configuration, it is always disabled. With the exception of MCo and MC7, a macrocell configured as a dedicated input derives the input signal from an adjacent I/O. MCo derives its input from pin 11 (DE) and MC7 from pin 1 (ClK). The macrocell configurations are controlled by the configuration control word. It contains 2 global bits (SGO and SG1) and 16 local bits (SlOothrough Sl07and Sl10 through Sl17). SGO determines whether registers will be allowed. SG1 determines whether the PAllV16V8Z will emulate a PAL 16R8 family or a PAL 1OH8 family device. Within each macrocell, SlOx, in conjunction with SG1, selects the configuration of the macrocell, and Sl1x sets the output as either active low or active high for the individual macrocell. The configuration bits work by acting as control inputs forthe multiplexers in the macrocell. There are four mUltiplexers: a product term input, an enable select, an output select, and a feedback select multiplexer. SG1 and SlOx are the control signals for all four multiplexers. In MCo and MC7, SGO replaces SG1 on the feedback mUltiplexer. This accommodates ClK being the adjacent pin for MC7 and DE the adjacent pin for MCo. Registered Output Configuration will use the feedback path of MC7 and pin 11 will use the feedback path of MCo. Combinatorial I/O In a Non-Registered Device The control bit settings are SGO = 1, SG1 = 1, and SlOx = 1. Only seven product terms are available to the OR gate. The eighth product term is used to enable the output buffer. The signal at the va pin is fed back to the AND array via the feedback muttiplexer. This allows the pin to be used as an input. Because ClK and DE are not used in a non-registered device, pins 1 and 11 are available as inputs. Pin 1 will use the feedback path of MC7 and pin 11 will use the feedback path of MCo. Combinatorial I/O in a Registered Device The control bit settings are SGO = 0, SG1 = 1 and SlOx =1. Only seven product terms are available to the OR gate. The eighth product term is used as the output enable. The feedback signal is the corresponding I/O signal. Dedicated Input Configuration The control bit settings are SGO = 1, SG1 = 0 and SlOx =1. The output buffer is disabled. Except for MCo and MC7 the feedback signal is an adjacent I/O. For MCo and MC7 the feedback signals are pins 1 and 11. These configurations are summarized in Table 1 and illustrated in Figure 2. Table 1. Macrocell Configuration SGO SG1 SLOx Cell Configuration Devices Emulated The control bit settings are SGO = 0, SG1 = 1 and SlOx =O. There is only one registered configuration. All eight product terms are available as inputs to the OR gate. Data polarity is determined by Sl1 x. The flip-flop is loaded on the lOW-to-HIGH transition of ClK. The feedback path is from Q on the register. The output buffer is enabled by DE. 0 1 0 a 1 1 1 a a 1 0 1 1 1 1 Combinatorial Configurations The PAllV16V8Z has three combinatorial output configurations: dedicated output in a non-registered device, I/O in a non-registered device· and I/O in a registered device. Dedicated Output In a Non-Registered Device The control bit settings are SGO = 1, SG1 = 0 and SlOx =O. All eight product terms are available to the OR gate. Although the macrocell is a dedicated output, the feedback is used, with the exception of MC3 and MC4. MC3 and MC4 do not use feedback in this mode. Because ClK and DE are not used in a non-registered device, pins 1 and 11 are available as input signals. Pin 1 Device Uses Registers Registered Output PAL 16R8, 16R6, 16R4 PAl16R6,16R4 Combinatorial 1/0 Device Uses No Registers Combinatorial PAL 1aH8, 12H6, Output 14H4, 16H2, 10LB, 12L6, 14L4, 16L2 Input PAl12H6,14H4, 16H2, 12L6, 14L4, 16L2 PAL16L8 Combinatorial 1/0 Programmable Output Polarity The polarity of each macrocell can be active-high or active-low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is through a programmable bit Sl1 x which controls an exclusive-OR gate at the output of the AND/ OR logic. The output is active high if Sl1 x is 1 and active low if SL1x is O. PALLV16V8Z Family 2-119 ~ AMD OE - - - - - - - - - - - - , OE - - - - - - - - - - -...... )-----10 D----i 0 Q Q elK Registered Active Low Registered Active High Combinatorial 1/0 Active Low Combinatorial 1/0 Active High Combinatorial Output Active Low Combinatorial Output Active High Notes: 1. Feedback is not available on pins 15 and 16 in the combinatorial output mode. 2. The dedicated-input configuration is not available on pins 15 and 16. Figure 2. Macrocell Configurations 2-120 PALLV16V8Z Family Adjacent 1/0 pin Note 2 Dedicated Input 174228-5 AMD~ Benefits of Lower Operating Voltage The PALLV16V8Z has an operating voltage range of 3.0 V to 3.6 V. Low voltage allows for lower operating power consumption, longer battery life, and/or smaller batteries for notebook applications. Because power is proportional to the square of the voltage, reduction of the supply voltage from 5.0 V to 3.3 V significantly reduces power consumption. This directly translates to longer battery life for portable applications. Lower power consumption can also be used to reduce the size and weight of the battery. Thus, 3.3-V designs facilitate a reduction in the form factor. A lower operating voltage results in a reduction of I/O voltage swings. This reduces noise generation and provides a less hostile environment for board design. Lower operating voltage also reduces electromagnetic radiation noise and makes obtaining FCC approval easier. Zero-Standby Power Mode The PALLV16V8Z features a zero-standby power mode. When none of the inputs switch for an extended period (typically 50 ns), the PALLV16V8Z will go into standby mode, shutting down most of its internal circuitry. The. current will go to almost zero (Icc < 15 JlA). The outputs will maintain the states held before the device went into the standby mode. When any input switches, the internal circuitry is fully enabled and power consumption returns to normal. This feature results in considerable power savings for operation at low to medium frequencies. This savings is illustrated in the Icc vs. frequency graph. Product-Term Disable On a programmed PALLV16V8Z, any product terms that are not used are disabled. Power is cut off from these product terms so that they do not draw current. As shown in the Icc vs frequency graph, product-term disabling results in considerable power savings. This savings is greater at the higher frequencies. Further hints on minimizing power consumption can be found in the Application Note "Minimizing Power Consumption with Zero-Power PLDs." Power-Up Reset All flip-flops power up to a logic LOW for predictable system initialization. Outputs of the PALL V16V8Z will depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be HIGH. If combinatorial is selected, the output will be a function of the logic. complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. The preload function is not disabled by the security bit. This allows functional testing after the security bit is programmed. Security Bit A security bit is provided on the PALLV16V8Z as a deterrent to unauthorized copying of the array configuration patterns. Once programmed, this bit defeats readback of the programmed pattern by a device programmer, securing proprietary designs from competitors. However, programming and verification are also defeated by the security bit. The bit can only be erased in conjunction with the array during an erase cycle. Electronic Signature Word An electronic signature word is provided in the PALLV16V8Z device. It consists of 64 bits of programmable memory that can contain user-defined data. The signature data is always available to the user independent of the security bit. Programming and Erasing The PALLV16V8Z can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its unprogrammed state. Erasure is automatically performed by the programming hardware. No special erase operation is required. Quality and Testability The PALLV16V8Z offers a very high level of built-in quality. The erasability if the device provides a direct means of verifying performance of all the AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to yield the highest programming yields and post-programming function yields in the industry. Technology The high-speed PALLV16V8Z is fabricated with AMD's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. This technology provides strong input-clamp diodes, output slew-rate control, and a grounded substrate for clean switching. Register Preload The register on the PALLV16V8Z can be preloaded from the output pins to facilitate functional testing of PALLV16VBZ Family 2-121 ~ AMD LOGIC DIAGRAM 4· 7 8 11 12 15 16 19 20 324 27 8 31 elK/IO 1 r-r~F.l~l------------t;=======V=~='~~~ ~~. u~ 13 4 174228-6 3 4 2·122 7 8 11 12 15 16 19 20 23 24 27 28 31 PALLV16V8Z Family ClK OE AMD~ LOGIC DIAGRAM (continued) 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 eLl( OE SOl $lO, ~--------< and 28-pin PLCC packages • • Outputs conflgurable as 0 or T flip-flops Outputs programmable as registered or combinatorial in any combination • Extensive third-party software and programmer support through FusionPLD partners • • Automatic register reset on power-up Fully tested for 100% programming and functional yields and high reliability • High output-current drive capability (64 mA 10L) • Programmable Totem-Pole or Open-Drain Outputs • • GENERAL DESCRIPTION The PALCE16V8HD is the first CMOS PAL device to combine high-current drive capability with a PAL architecture. The PALCE16V8HD can sink up to 64 rnA for bus applications. It also has an advanced PAL architecture using a programmable macrocell to help provide a universal solution. The PALCE16V8HD utilizes the familiar sum-ofproducts (AND/OR) architecture that allows users to implement complex logic functions easily and efficiently. Multiple levels of combinatorial logic. can always be reduced to sum-of-products form, taking advantage of the very wide input gates available in PAL devices. The equations are programmed into the device through floating-gate cells in the AN D logic array that can. be erased electrically. The fixed OR array allows up to eight data product terms per output for logic functions. The su m of these products feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial with an activehigh or active-low output. The output configuration is Publication# 15559 Rev.D Issue Date: June 1993 Amendment/O determined by two global bits and one local bit controlling four multiplexers in each macrocell. The PALCE16V8HD has some additional features that make it an ideal choice for bus applications. These include input hysteresis of 200 mV, clean output-switChing Signals, programmable totem-pole or open-drain output configurations, programmable direct or latched inputs, and programmable D- or T -type output registers. AMD's FusionPLD program allows PALCE16V8HD designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accu rate, quality support. By ensu ring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a deSigner can use a tool that is already installed and familiar. Please refer to the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. 2-135 ~AMD BLOCK DIAGRAM GNOVee OE/lg 18 17 Is 1/02 1/03 GND lEll3 14 15 11 elK/lo 99 1/00 1/01 66 Vee 1/04 110s 1/05 1/07 6 Vee 155590-1 CONNECTION DIAGRAMS Top View SKINNVDIP PLCC 0 ClK/lo Vee h 1/07 12 lE/13 14 ~ I/Os GND 1/05 15 1/04 Is Vee • o ,... !D ~~ ~ 28 27 26 25 GND 24 1/°5 15 NC 23 1/°4 NC Vee 1/°3 22 GND 1/03 Is OE/lg 1102 17 21 20 1/01 18 19 GND 12 13 14 15 16 17 18 1/00 Q) ili 10 0 0 0 z z ~ C} o ~ N ~ ~ ~ 155590-3 PIN DESIGNATIONS 2-136 2 0 z 14 155590-2 OE 3 0 lE/13 Note: Pin 1 is marked for orientation. Vee 4 .....I 18 GND = = = = = = = = - 17 Vee ClK lE GND I I/O NC -!.'I Clock lateh Enable Ground Input Input/Output No Connect Output Enable Supply Voltage PALCE16V8HD-15 AMD~ ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: T PAL CE 16 V 8 HD -15 P C -r-- ~ FAMILY TYPE ---' PAL= Programmable Array Logic PROGRAMMING DESIGNATOR Blank = Initial Release TECHNOLOGY CE = CMOS Electrically Erasable OPERATING CONDITIONS C = Commercial (OOC to +75°C) NUMBER OF ARRAY INPUTS PACKAGE TYPE P = 24-Pin Plastic SKINNYDIP (PD 3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) OUTPUT TYPE - - - - - - - - -...... V = Versatile NUMBER OF FLIP-FLOPS - - - - - - - ' DRIVE HD = High Output Drive SPEED -15 = 15 ns tPD Valid Combinations PALCE16V8HD-15 Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE16V8HD-15 (Com'l) 2-137 ~ AMD FUNCTIONAL DESCRIPTION The PAlCE16V8HD is a universal PAL device with eight independently configurable macrocells (MCo-MC7). Each macrocell can be configured as a registered output, combinatorial output, combinatorial 1/0 or dedicated input. The programming matrix implements a programmable AND logic array, which drives a fixed OR logic array. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Pins 1 and 10 serve either as array inputs or as clock (ClK) and output enable (DE), respectively, for all flip-flops. All inputs to the array can be individually programmed as either direct or transparent-latch inputs. lE/b is the latch enable pin. The inputs to the array also have a minimum of 200 mV of hysteresis. Unused input pins should be tied directly to Vee or GND. Product terms with all bits unprogrammed (disconnected) assume the logical HIGH state and product terms with both true and complement of any input signal connected assume a logical lOW state. The programmable functions on the PAlCE16V8HD are automatically configured from the user's design specification, which can be in a number of formats. The design specification is processed by development software to verify the design and create a programming file. This file, once downloaded to a programmer, configures the device according to the user's desired function. 1 - - - - - . To Adjacent 11 Macrocell SlOx SG1 >-----......-IDfT elK (lOX QI--~ Q~--------~ Sl5x From 'SG1 "In macrocells MCo and MCl, SGt is replaced by SGO on the feedback multiplexer. Adjacent Pin 15559D-4 Figure 1. PAlCE16V8HD 1/0 Macrocell 15559D-5 Figure 2. PALCE16V8HD Input Macrocell 2-138 PALCE16V8HD-15 AMD~ Device Configuration The configuration of the PAlCE16V8HD is controlled by the configuration control word. It contains 2 global bits (SGO and SG1) and 48 local bits (SlOa through Sl07, SUa through Sl h, Sl21 through Sl2a, Sl3a through Sl37, SL40 through SL47 and Sl5a through Sl57). SGO determines whether registers will be allowed. SG1 and the individual SlOx bits select the output macrocell configuration as registered output, combinatorial input, combinatorial output, or combinatorial 1/0. Sl3x sets the feedback path to the array as either direct or latched. SL4x sets the output buffer as either a totem pole or an open drain. Sl5x sets the register as either a D or T type flip-flop. At each input pin, Sl2x sets the input as direct or latched. Input Pin Configuration Options Each input pin can be configured as either a direct input or a transparent latch. The input-pin configuration is set by the local fuse Sl2x. When Sl2x is unprogrammed, the input is direct. When Sl2x is programmed, the input is through a corresponding transparent latch. The latch is enabled via lEl13. The latches hold data when lEll3 is low. They are transparent when lElI3 is HIGH. I/O Macrocell Configuration Options Each 1/0 macrocell can be configured as one of the following: registered output, combinatorial output, combinatorialllO, or dedicated input. In the registered output configuration, the output buffer is enabled by theOE pin. In the combinatorial configuration, the buffer is either controlled by a product term or always enabled. In the dedicated input configuration, it is always disabled. With the exception of MCa and MC7, a macrocell configured as a dedicated input derives the input signal from an adjacent 1/0. MCa derives its input from pin 10 (OE) and MC7 from pin 1 (ClK). These configurations are summarized in Table. 1 and illustrated in Figure 3. In the registered configurations all eight product terms are available as inputs to the OR gate. The flip-flop is loaded on the lOW-to-HIGH transition of ClK. The output buffer is enabled by OE. Feedback to the array can be either direct or latched. Direct feedback is from Q of the register to the productterm array. latched feedback is from Q of the register through a transparent latch to the product-term array. lEll3 is the latch-enable signal. Combinatorial Configurations The PAlCE16V8HD has three combinatorial output configurations: dedicated output in a non-registered device, 1/0 in a non-registered device and 1/0 in a registered device. Dedicated Output in a Non-Registered Device In this configuration, the output buffer is always enabled; therefore all eight product terms are available to the OR gate. The feedback to the array is from an adjacent 1/0 pin. 1/03 and 1/04 do not have connections to adjacent macrocells; therefore, MC3 and MC4 do not have feedback to the array in this mode. Because ClK and OE are not used in a non-registered device, pins 1 and 10 are available as input signals. Pin 1 will use the feedback path of MC7 and pin 10 will use the feedback path of MCa. Combinatorial I/O In a Non-Registered Device Only seven product terms are available to the OR gate in this configuration. The eighth product term is used to enable the output buffer. The signal at the 1/0 pin is fed back to the AND array via the feedback multiplexer. This allows the pin to be used as an input. Because ClK and OE are not used in a non-registered device, pins 1 and 10 are available as inputs. Pin 1 will use the feedback path of MC7 and pin 10 will use the feedback path of MCa. Combinatorial I/O in a Registered Device The feedback path in each macrocell can be programmed as either direct or latched. The feedback configuration is set by the local fuse Sl3x. When SL3x is unprogrammed, the corresponding feedback path is direct to the array. When Sl3x is programmed, the corresponding feedback path is through a corresponding transparent latch. In this configuration only seven product terms are available to the OR gate. The eighth product term is used as the output enable. The feedback signal is the corresponding 1/0 signal. The latch is enabled via lElI3. The latches hold data when lElI3 is lOW. They are transparent when lElI3 is HIGH. The output buffer is disabled in this configuration. Except for MCa and MC7 the feedback signal is an adjacent 1/0. For MCa and MC7 the feedback signals are pins 1 and 10. Registered Output Configurations There are two registered configurations: D-type and Ttype. Tl:le type is selected by Sl5x. Dedicated Input Configuration Pins 16 (19) and 19 (23) do not have connections to adjacent macrocells. The dedicated-input configuration is not available on these pins. PALCE16V8HD-15 2-139 ~ AMD Table 1. Macrocell Configuration SGO SG1 SLOx SL5x Cell Configuration Device Uses Registers 0 1 0 0 T-Type Registered Output 0 1 0 1 D-Type Registered Output 1 0 0 X Combinatorial Output 1 0 1 X Dedicated Input 1 1 1 X Combinatorial I/O Device Uses No Registers Programmable Output Polarity The polarity of each macrocell can be active-high or active-low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. 2-140 Selection is through a programmable bit SL 1x which controls an exclusive-OR gate at the output of the AND/ OR logiC. The output is active high if SL 1x is 1 and active low if SL1x is O. Output Buffer Configurations The output buffer can be configured as either a totempole output or an open-drain output. This configuration is set by SL4x . The buffer is a totem-pole output when SL4x is unprogrammed and an open-drain output when SL4x is programmed. In the totem-pole configuration, the output voltage levels are the standard VOH and VOL levels. In the open-drain configuration, VOL is the standard value. However, VOH will depend on the termination circuitry. PALCE16V8HD-15 AMD~ OE-----------------------, }------t OE -----------------------, Off Q D - - - - - t Off Q elK Registered Active High Registered Active Low Combinatorial 1/0 Active Low Combinatorial 1/0 Active High lE/13 lE/13 Combinatorial 110 Active Low with Latched Feedback Combinatorial VO Active High with Latched Feedback Note: 1. All output and /10 configurations are valid as either totem-pole outputs or opencollector outputs. 15559D-6 Figure 3. Macrocell Configurations PALCE16V8HD-15 2-141 ~ AMD Combinatorial Output Active High Combinatorial Output Active Low To Logic Array Combinatorial Output Active Low with Latched Feedback Input~.JJ D or I/O Pin LE Combinatorial Output Active High with Latched Feedback Qr To Logic Array ~ <:J ~ L....-_---I Adjacent I/O pin Note 3 Latched Input Dedicated Input 155590-7 Notes: 1. All output and 110 configurations are valid as either totem-pole outputs or opencol/ector outputs. 2. Feedback is not available on pins 16 and 19 in the combinatorial output mode. 3. The dedicated-input configuration is not available on pins 16 and 19. Figure 4. Macroceil Configurations 2-142 PALCE16V8HD-15 AMD~ Power-Up Reset Electronic Signature Word All flip-flops power up to a logic LOW for predictable system initialization. Outputs of the PALCE16V8HD will depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be HIGH. If combinatorial is selected, the output will be a function of the logic. An electronic signature word is provided in the PALCE16V8HD device. It consists of 64 bits of programmable memory that can contain user-defined data. The signature data is always available to the user independent of the security bit. Register Preload The PALCE16V8HD can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erase operation is required. The register on the PALCE16V8HD can be pre loaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. Security Bit A security bit is provided on the PALCE16V8HD as a deterrent to unauthorized copying of the array configuration patterns. Once programmed, this bit defeats readback of the programmed pattern by a device programmer, securing proprietary designs from competitors. However, programming and verification are also defeated by the security bit. The bit can only be erased in conjunction with the array during an erase cycle. Programming and Erasing Quality and Testability The PAL16V8HD offers a very high level of built-in quality. The erasability of the device provides a direct means of verifying performance of all the AC and DC parameters. In addition, this helps verify complete programmability and functionality of the device to yield the highest programming yields and post-programming function yields in the industry. Technology The high-speed PALCE16V8HD is fabricated with AMD's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with TTL devices. This technology provides strong input-clamp diodes, output slew-rate control, and a grounded substrate for clean switching. PALCE16V8HD-15 2-143 ~AMD LOGIC DIAGRAM SKINNYDIP/Flatpack (PLCC) Pinouts o 3 • 7 • 11 12 15 16 19 20 23 2. 27 28 31 C~. 1~-----------~~~++-HH+-+HH-++H~~~++~H++-HH+---------------------~----------------, (2) ~Vee .rtn, (28) ITL+ .r€TI GND (25) .r-tch tl!d (2.) rtch tl!d _& -@) VO. (23) !ill Vee (21) .r@l GND (20) o 3 • 7 8 11 12 15 16 19 20 23 2. 27 28 31 LE OE CLK 155590-8 2-144 PALCE16V8HD-15 AMD~ LOGIC DIAGRAM (continued) o 3 4 7 a 11 12 15 18 19 20 23 24 27 28 eLK LE OE 31 ri=! . 4011111111~~ '~;t>-H-+-+-IIro;T=1 I~: 1l}G :::::stl -1SLO, 0-~17 ~ a (9) ~1 ~ SL~ 47 ~ 0 q.SL4, (18) SGl SLO, SGl SLO, LE # SL3, SL2, -f"n. ~ i @~P" : a (10) ":>0---+---/15 VO, ~~==:::::: I sll, a 0 LE ~DIT ~ 0 LE : SL2. -r-n. ~ i 17 @~1 (11) 0 o LE a • SL2. '--------<<}~.10~--------------rHH_++rr~++_H~_+HH_+rh~++~~+_----------------------------------------~ (121 Vcc[ill o 3 4 7 8 11 12 15 18 19 20 23 24 27 28 31 (13) 155590-8 (concluded) GlDIill-J.. (14) • PALCE16V8HD-15 2-145 ~AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature Commercial (C) Devices Ambient Temperature with Power Applied ............ -55°C to + 125°C Temperature (TA) Operating in Free Air ....................... O°C to +75°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ............ +4.75 V to +5.25 V DC Input Voltage. . . . . . . . .. -0.5 V to Vcc + 0.5 V DC Output or lID Pin Voltage ............... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage . . . . . . . . . . . . . . .. 2001 V Latchup Current (TA = O°C to 75°C) ..................... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage Totem-pole Configuration IOH =-16 mA Vee == Min VIN = VIH or Vil VOL Output LOW Voltage IOl = 64 mA Vee = Min VIN = VIH or Vil VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) Min Hysteresis (Notes 2 and 3) Vee = Min hH Input HIGH Leakage Current VIN = 5.25 V, Vee = Max (Note 4) III Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 4) IOZH Off-State Output Leakage Current HIGH Your = 5.25 V, Vee = Max VIN = VIH or Vil (Note 4) IOZl Off-State Output Leakage Current LOW Your = 0 V, Vee = Max VIN = VIH or VIL (Note 4) Isc Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 5) Icc Supply Current Outputs Open (lOUT = 0 mA) Vee = Max, f = 25 MHz VHYS Max Unit 2.4 V 0.5 2.0 V 0.8 200 -30 V V mV 10 ).LA -10 ).LA 10 JlA -10 JlA -150 mA 115 mA Notes: 1. These are absolute values with respect to device ground and all overshoots due to system or tester noise are included. 2. Hysteresis is the difference between the positive going input threshold voltage and the negative going input threshold voltage. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. 4. 110 pin leakage is the worst case of liL and IOZL (or liH and IOZH). 5. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT == 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-146 PALCE16V8HD-15 (Com'l) AMD~ CAPACITANCE Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN= 2.0 V Output Capacitance VOUT= 2.0 V I I Vee = 5.0 V, TA = 25°C, f= 1 MHz Typ Unit 5 pF 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol Parameter Description Min tPD Input, I/O, or Feedback to Combinatorial Output Max Unit 15 ns ts Setup Time from Input, I/O, or Feedback to Clock 10 ns tH Register Data Hold Time 0 ns teo Clock to Output twL tWH Clock Width fMAX Maximum Frequency (Note 3) 10 lLOW I HIGH External Feedback 11/(ts + teo) Internal Feedback (feNT) 11/(twL + tWH) No Feedback ns 6 6 ns ns 50 MHz 66 MHz 83.3 MHz ns tSIL Input Latch Setup Time 4 tHIL Input Latch Hold Time 6 tlGO Input Latch Enable to Combinatorial Output ns 15 ns Input Latch Enable Width HIGH 15 tlGS Input Latch Enable to Output Register Setup Time 10 tPDL Input, I/O, or Feedback to Output Through Transparent Input Latch tSLR Setup Time from Input, I/O, or Feedback Through Transparent Input Latch to Output Register tHLR Hold Time from Input, I/O or Feedback Through Input Latch to Output Register tpzx DE to Output Enable 15 ns tpxz DE to Output Disable 15 ns tWIGH ns ns ns 15 10 ns 0 ns tEA Input, I/O, or Feedback to Output Enable 15 ns tER Input, I/O, or Feedback to Output Disable 15 ns tEAL Input, I/O, or Feedback to Output Enable Through Transparent Latch (Note 3) 15 ns tERL Input, I/O, or Feedback to Output Disable Through Transparent Latch (Note 3) 15 ns Notes: 2. See Switching Test Circuit, page 15, for test conditions. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALCE16V8HD-15 (Com'l) 2-147 ~ AMD SWITCHING WAVEFORMS Input, ~( ,~ V I/O,or _ _ _--'/\_ _ _ _ _--'Jr\. T Feedback ~ Input, VO, or Feedback Combinatorial Output -~-VT Clock ts tH ~ VT ________________ '~2OCiktco VT Registered Output _ _ _ _ _ _ _ _---' 155590-10 155590-9 Registered Output Combinatorial Output Latched Input Latched Input LE tSLR --f'I1I-~""'LE Clock----- Combinatorial Output _ _--t"-' Registered Output _ _ _ _..L.JI'-_ _ _ _ _ _-'-II..M'I'155590-12 155590-11 Latched Input with Registered Output Latched Input with Combinatorial Output Input Latched Latch Enable Clock VTLatched 155590-14 155590-13 Input Latch Enable Width Clock Width Notes: 1. VT= 1.5V 2. Input pulse amplitude 0 V to 3.0 V 3. Input rise and fall times 2 ns - 5 ns typical. 2-148 PALCE16V8HD-15 AMD~ SWITCHING WAVEFORMS Input,---~ 1/0, or Feedback - - - - - ' tER >-+-+H-----f-+-4H- VT Output tpxz tEA Output tpzx >-+-+H-.....;..;;;.;.;...~~-+-4H- _ _ _ _--L..L..I.:....II VT ------L~"_1 155590-16 155590-15 Input to Output Disable/Enable DE to Output Disable/Enable Latched _ _ _~ Input, 1/0, or Feedback - - - - - ' tERL tEAL Output 155590-17 Input to Output Disable/Enable Through Transparent Latch 4 r' 4-1-r'_ _ _"'T""'"I 3 3 Vo Output Voltage 2 Vo Output Voltage 2 VTH I 1 11 _I I 1 2 3 4 VI Input Voltage Notes: 0 I 0 I I I I 1 2 3 4 VI Input Voltage 155590-18 155590-19 1. Vr= 1.5 V 2. Input pulse amplitude 0 V to 3.0 V 3. Input rise and fall times 2 ns - 5 ns typical. PALCE16V8HD-15 2-149 ~ AMD KEY 10 SWITCHING WAVEFORMS WAVEFORM INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L /77// May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State xxxxxx KSOOOO10-PAL SWITCHING TEST CIRCUIT SV Output 0 - -. . .- - -......-"'") Test Point 15559D-20 Commercial Specification. S1 tpo, tpOl, teo Closed tpzx, tEA, tEAL Z~ tpxz, tER, tERl Cl R1 Measured Output Value 1.S V H: Open Z ~ L: Closed SO pF H ~Z: Open S pF 1.S V 80n 160n H ~ Z: VOH - O.S V L ~Z: Closed 2-150 R2 L ~ Z: VOL + O.S V PALCE16V8HD-15 AMD~ ENDURANCE CHARACTERISTICS The PALCE16V8HD is manufactured using AMD's advanced Electrically Erasable process. This technology uses an EE cell to replace the fuse link used in bipolar parts. As a result, the device can be erased and reprogrammed-a feature which allows 100% testing at the factory. Endurance Characteristics Symbol Parameter Min Unit Max Storage Temperature 10 Years Test Conditions tOR Min Pattern Data Retention Time Max Operating Temperature (Military) 20 Years N Min Reprogramming Cycles Normal Programming Conditions 100 Cycles INPUT/OUTPUT EQUIVALENT SCHEMATICS Vcc ESD ProgramlVerify Protection Circuitry Typical Input Vee Preload Circuitry Typical Output PALCE16V8HD-15 Feedback Input 15559D·21 2-151 ~ AMD MEASURED SWITCHING CHARACTERISTICS for the PALCE16V8HD-15 (Note 1) 12 11 tPD. ns 10 9 8 o 50 100 150 200 250 300 CL. pF tPD vs. Load Capacitance Vee =5.25 V, TA =25°C 155590-22 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where tPD may be affected. 2-152 PALCE16V8HD-15 AMD~ POWER-UP RESET The PALCE16V8HD has been designed with the capability to reset during system power-up. Following powerup, all flip-flops will be reset to LOW. The output state will be HIGH independent of the logic polarity. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset Parameter Symbol and the wide range of ways Vec can rise to its steady state, two conditions are required to insure a valid power-up reset. These conditions are: • • The Vcc rise must be monotonic. Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Descriptions Min tPR Power-Up Reset Time ts Input or Feedback Setup Time Max Unit 1000 ns See Switching Characteristics Clock Width LOW tWL 4 Power Registered Output Clock v-jL----------------f~ ... L.I-__ tPR Vcc ~ II '\ tWL 155590-23 Power-Up Reset Waveform PALCE16V8HD-15 2-153 ~AMD TYPICAL THERMAL CHARACTERISTICS Measured at 25°C ambient. These parameters are not tested. PALCE16V8HD-15 Parameter Symbol Typ. PLCC Unit 9jc Thermal impedance, junction to case 22 17 9ja Thermal impedance, junction to ambient 70 55 200 Ifpm air 65 47 400 Ifpm air 60 42 600 Jfpm air 56 38 800 Ifpm air 53 36 °elW °elW °elW °elW °elW °elW 9jma Parameter Description Thermal impedance, junction to ambient with air flow SKINNYDIP Plastic alc Considerations The data listed for plastic 9jc are for reference only and are not recommended for use in calculating junction temperatures. The heat-flow paths in plastic-encapsulated devices are complex, making the 9jc measurement relative to a specific location on the . package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom centerof the package. Furthermore, 9jc tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a similar environment. 2-154 PALCE16V8HD-15 ~ _COM'L Advanced Micro Devices AmPAL 18P88/AL/A/L 20-Pin Combinatorial TTL Prorammable Array Logic DISTINCTIVE CHARACTERISTICS • • • • • As fast as 15 ns maximum propagation delay Universal combinatorial architecture Programmable output polarity Programmable replacement for high-speed TIL logic • Extensive third-party software and programmer support through FusionPLD partners 20-pin DIP and 20-pin PLCC packages save space GENERAL DESCRIPTION The Am PAL 18P8 utilizes Advanced Micro Devices' advanced oxide-isolated bipolar process and fuse-link technology. The devices provide user-programmable logic for replacing conventional SSI/MSI gates and flipflops at a reduced chip count. The AmPAL 18P8 allows the systems engineer to implement the design on-chip, by opening fuse links to configure AND and OR gates within the device, according to the desired logic function. Complex interconnections between gates, which previously required timeconsuming layout, are lifted from the PC board and placed on silicon, where they can be easily modified during prototyping or production. The PAL device implements the familiar Boolean logic transfer function, the sum of products. The PAL device is a programmable AND array driving a fixed OR array. The AND array is programmed to create custom product terms, while the OR array sums selected terms at the outputs. In addition, the PAL device provides the following options: - Variable inpuVoutput pin ratio - Programmable three-state outputs Product terms with all fuses opened assume the logical HIGH state; product terms connected to both true and complement of any single input assu me the logical LOW state. Unused input pins should be tied to Vcc or GND. The entire PAL device family is supported by the FusionPLD partners. The PAL family is programmed on conventional PAL device programmers with appropriate personality and socket adapter modules. See the Programmer Reference Guide for approved programmers. Once the PAL device is 'programmed and verified an additional fuse may be opened to prevent pattern readout. This feature secures proprietary circuits. BLOCK DIAGRAM AmPAL18P8 Inputs 1/0 1/0 Publication# 05799 Rev. G AmendmentiO Issue Date: January 1992 1/0 1/0 1/0 1/0 1/0 1/0 OS799G-1 2-155 ~ AMD PRODUCT SELECTOR GUIDE tpo Icc IOL Family ns (Max) mA{Max) mA{Mln) Very High-Speed ("B") Versions 15 180, 24 High-Speed ("Ai Versions 25 180 24 High-Speed, Half-Power ("AL") Versions 25 90 24 Half-Power ("Li Versions 35 90 24 CONNECTION DIAGRAMS Top View DIP PLCC ~ ~ vee 1/0 3 2 1/0 1/0 1/0 110 1/0 1/0 1/0 9 10 11 12 13 (!) 05799G-2 Note: Pin 1 is marked for orientation. g ~ 05799G-3 PIN DESIGNATIONS 2-156 1/0 1/0 0 Z 1/0 18 1/0 GND Vee • 20 19 1/0 1/0 GND I 1 Ground Input Input/Output Supply Voltage AmPAL 18P8B/AUAlL AMD~ ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: AmPAL 18 P 8 ALP C ____J----I -,... -.- -.. . -.. . _ FAMILY TYPE AmPAL = Programmable Array Logic NUMBER OF ARRAY INPUTS OUTPUT TYPE _ _ _ _ _ _ _ _.....J L OPTIONAL PROCESSING Blank = Standard Processing P = Programmable Polarity OPERATING CONDITIONS C = Commercial (O°C to +7S0 C) NUMBER OF OUTPUTS _ _ _ _ _ _----J SPEED --------------~ B = 15 ns tPD A = 25 ns tPD Blank = 35 ns tPD POWER _ _ _ _ _ _ _ _ _ _ _ _ _----J PACKAGE TYPE P = 20-Pin Plastic DIP (PD 020) J = 20-Pin Plastic Leaded Chip Carrier (PL 020) D = 20-Pin Ceramic DIP (CD 020) L = Low Power (90 mA Icc) Blank = Full Power (180 mA Icc) Valid Combinations AmPAL18P8 PC, JC, DC Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. AmPAL18P88/ALlA/L (Com'l) 2-157 ~ AMD FUNCTIONAL DESCRIPTION All parts are produced with a fuse link at each input to the AND gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Utilizing an easily-implemented programming algorithm, these products can be rapidly programmed to any customized pattern. Information on approved programmers can be found in the Programmer Reference Guide. Extra test words are pre-programmed during manufacturing to ensure extremely high field programming yields, and provide extra test paths to achieve excellent parametric correlation. Variable Input/Output Pin Ratio The AmPAL 18P8 has ten dedicated input lines, and all eight combinatorial outputs are I/O pins. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vee or GND. Programmable Three-State Outputs Each output has a three-state output buffer with threestate control. A product term controls the buffer, allowing enable and disable to be a function of any product of device inputs or output feedback. The combinatorial output provides a bidirectional 110 pin, and may be configured as a dedicated input if the buffer is always disabled. Programmable Polarity The polarity of each output can be active-high or activelow, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean 2-158 expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is through a programmable fuse which controls an exclusive-OR gate at the output of the AND/OR logic. The output is active high if the fuse is 1 (programmed) and active low if the fuse is 0 (intact). Security Fuse After programming and verification, an AmPAL 18P8 design can be secured by programming the security fuse. Once programmed, this fuse defeats readback of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. When the security fuse is programmed, the array will read as if every fuse is programmed. Quality and Testability The AmPAL 18P8 offers a very high level of built-in quality. Extra programmable fuses provide a means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Technology The AmPAL18P8 is fabricated with AMD's advanced oxide-isolated bipolar process. This process reduces parasitic capacitances and minimum geometries to pro~ vide higher performance. The array connections are formed with proven PtSi fuses for reliable operation. AmPAL 18P88/AUA/L AMD~ LOGIC DIAGRAM Inputs (0-35) ell 1 Zl 4a., all01' 12"'4" ,1",St, 2OZ1un 24:r5H17 2eZlla31 3233)435 2- ·· ··· r- , ,. ~~ d:::7 ~~ 7 0--ti= ,. .... · r~6-r ,.,." ,.,. " ~ ~ !:~ < ,.,. ..., .. r- n ..... ~ 2 rb&r rE>t&r rb&r ~b&r t:~ _ . . .R 2< .... ..., .,. .... ..... ~ ..... ~ ~~ 3D ~tj 2- .. .... ~~ ..., ... .... 2 .... ., .. ... . . '; r-~ ;:::~ ct::1 2 14 .... '; r- ~~&r ::tj '; rH 07 §:=J 51 .. ... ...." 51 13 ~ rE>t&r r- ;~ . II 17 II 7D 12 t:tj 7' 9 ..... ,-- 50 8 15 .....~ J-J .7 7 ~ 16 '< ..... 37 6 17 r- ~ 5 ~ > 24 4 18 ~'- 17 3 ~ :> '; < <, 4 ~ Ii 7 e I 1Q 11 1213 '" 15 16111811 2Q 21 22 23 242526 27 28 2130 31 11 323334 35 05799G-4 Am PAL 18P8B/AUAlL 2-159 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Ambient Temperature With Power Applied ............. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air ............ O°C to +75°C Supply Voltage with Respect to Ground ............. -D.5 V to +7.0 V -Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . . . . . .. -0.5 V to +5.5 V DC Input Current .............. -30 rnA to +5 rnA Operating ranges define those limits between which the functionality of the device is guaranteed. DC 1/0 Pin Voltage ............ -D.5 V to Vce Max Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH = -3.2 rnA VIN = VIH or VIL Vee = Min Min VOL Output LOW Voltage IOl= 24 rnA VIN = VIH or VIL Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL ' Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) Max Unit V 2.4 0.5 2.0 V V 0.8 V VI Input Clamp Voltage liN = -18 rnA, Vee = Min -1.2 V 111-1 Input HIGH Current VIN = 2.7 V, Vee = Max (Note 2) 25 ~ III Input LOW Current VIN = 0.4 V, Vee = Max (Note 2) -100 ~ Ii Maximum Input Current VIN = 5.5 V, Vee = Max 1 mA lozH Off-State Output Leakage Current HIGH VOUT= 2.7 V, Vee = Max VIN = VIH or VIL (Note 2) 100 ~ lozl Off-State Output Leakage Current LOW VOUT = 0.4 V, Vee = Max VIN = VIH or Vil (Note 2) -250 ~ Isc Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) -90 mA Icc Supply Current VIN = 0 V, Outputs Open (loUT = 0 rnA) Vee = Max B,A 180 mA AL 90 mA -30 Notes: 1. These are absolute values with respect to device ground and all overshoots due to system andlor tester noise are included. 2. liD pin leakage is the worst case of hL and IOZL (or iJH and loZH). 3. Not more than one output should be shorted at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-160 AmPAL 18P8B/AUA/L (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance YOUTz: Typ 2.0 V Vee = 5.0 V 6 TA = +25°C f .. 1 MHz 9 Unit pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol A AL B Parameter Description Min Max L Max Unit 25 35 ns 15 25 35 ns 15 25 35 ns tPD Input or Feedback to Combinatorial Output 15 tEA Input to Output Enable Using Product Term Control tER Input to Output Disable Using Product Term Control Min Max Min Note: 2. See Switching Test Circuit for test conditions. AmPAL18P88/AL/A/L (Com'l) 2-161 ~ AMD SWITCHING WAVEFORMS Input or Feedback Input _ _ _---I tER Combinatorial Output Output ------'-..1-1.-1 OS799G-S 05799G-6 Input to Output Disable/Enable Combinatorial Output Notes: 1. VT= 1.5 V 2. Input pulse amplitude 0 V to 3.0 V 3. Input rise and fall times 2 ns-5 ns typical. 2-162 tEA AmPAL 18P8B/AUA/L AMD~ KEY TO SWITCHING WAVEFORMS WAVEFORM INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L /77// May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State XXXXXX ]}) EK KSOOOO10-PAL SWITCHING TEST CIRCUIT 5V Ou1put o--.....- - -.....~~) Test Point 05799G-7 Specification tPD tEA tER S1 CL R1 R2 1.5 V Closed Z~ H: Open Z ~ L: Closed H ~Z: Open L ~Z: Measured Output Value Closed 50 pF 1.5 V 2000 5 pF AmPAL 18P8B/AUAlL 3900 H ~ Z: VOH - 0.5 V L ~ Z: VOL + 0.5 V 2-163 ~ AMD INPUT/OUTPUT EQUIVALENT SCHEMATICS ----------------~--~~c Input ProgramlVerify Circuitry 05799G-8 Typical Input -----------------4~----~ ~c 40 n NOM Output Input. 110 Pins ProgramlVerifyl Test Circuitry 05799G-9 Typical Output 2·164 AmPAL 18P88/AUA/L _ COM'L: -517IB/B-2/A,10/2 ~ MIL: -10/12/15/B/A Advanced Micro Devices PAL20R8 Family 24-Pin TTL Programmable Array Logic DISTINCTIVE CHARACTERISTICS • • As fast as 5 ns maximum propagation delay Popular 24-pln architectures: 20LS, 20RS, 20R6,20R4 • • Power-up reset for Initialization Extensive third-party software and programmer support through FusionPLD partners • Programmable replacement for high-speed TTL logic • 24-pln SKINNYDIP and 2S-pln PLCC packages save space GENERAL DESCRIPTION The PAL20RS Family (PAL20LS, PAL20RS, PAL20R6, PAL20R4) includes the PAL20RS-5 Series which is ideal for high-performance applications. The PAL20RS Family is provided in the standard 24-pin DI P and 2S-pin PLCC pinouts. In addition, the PAL device provides the following options: The devices provide user programmable logic for replacing conventional SSI/LSI gates and flip-flops at a reduced chip cost. Product terms with all connections opened assume the logical HIGH state; product terms connected to both true and complement of any single input assume the logical LOW state: Registers consist of D-type flip-flops that are loaded on the LOW-to-HIGH transition of the clock. Unused input pins should be tied to Vcc or GND. The family allows the systems engineer to implement the design on-chip, by opening fuse links to configure AND and OR gates within the device, according to the desired logic function. Complex interconnections between gates, which previously required time-consuming layout, are lifted from the PC board and placed on silicon, where they can be easily modified during prototyping or production. The PAL device implements the familiar Boolean logic transfer function, the sum of products.·The PAL device is a programmable AND array driving a fixed OR array. The AND array is programmed to create custom product terms, while the OR array sums selected terms at the outputs. - Variable inpuVoutput pin ratio Programmable three-state outputs Registers with feedback AM D's FusionPLD program allows PAL20RS Family designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuringthatthirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a designer can use a tool that is already installed and familiar. Please refer to the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. PRODUCT SELECTOR GUIDE Device Dedicated Inputs Outputs Product Terms/Output Feedback Enable PAL20La 14 6 comb. 2 comb. 7 7 a a 7 a 7 I/O prog. prog. PAL20Ra 12 a reg. PAL20R6 12 PAL20R4 12 6 reg. 2 comb. 4 reg. 4 comb. Publication# 16490 Rev. B AmendmentlO Issue Date: June 1993 reg. reg. 1/0 reg. 1/0 pin pin prog. pin prog. 2-165 ~ AMD BLOCK DIAGRAMS PAL20L8 INPUTS PROGRAMMABLE AND ARRAY (40 X 64) 1105 1IC>.3 AS 164908-1 PAL20R8 elK PROGRAMMABLE AND ARRAY (40 X 64) 164908-2 2-166 PAL20R8 Family AMD~ BLOCK DIAGRAMS PAL20R6 CLK INPUTS PROGRAMMABlE AND ARRAY (40 X 64) PAL20R4 ClK INPUTS PROGRAMMABLE AND ARRAY (40 X 64) 05 110 a 1/0 7 164908-4 PAL20R8 Family 2-167 ~AMD CONNECTION DIAGRAMS Top View SKINNVDIP/FLATPACK PLCC/LCC JEDEC: Applies to -5, -7(-12110 mil), (Note 1) -10(-15 mil), B-2 Series Only Vee 11 12 112 13 (Note 9) 14 (NoteS) 15 Is (Note 7) 17 Is (Note 5) 19 110 (Note 3) GND (Note 2) (Note 10) 1 28 27 26 • (Note 6) 25 (NOTE 9) 24 (NOTE 8) 23 (NOTE 7) 22 NC (Note 4) 111 16490B-6 16490B-5 PLCC Applies to B and A Series Only Note: Pin 1 is marked for orientation. (NOTE 9) Note 20L8 20R8 20R6 20R4 14 6 (NOTE 8) 1 10 elK elK elK 15 7 (NOTE 7) NC 8 (NOTE 6) OE OE OE 16 9 3 113 01 1/01 1/01 ? 10 4 1/02 1/03 6 1/04 7 S 1/05 I/0s 05 Os 9 1/07 10 Os 07 Os 02 03 04 05 Os 07 lIDs 1/02 5 01 02 03 04 2 03 04 05 Os 1/07 (NOTE 4) 16490B-7 LCC Applies to B and A Series Only lIDs §" w f- I 1/0 NC o OE Vee 2-168 I- ~~~~J~~ PIN DESIGNATIONS ClK Clock GND W Ground Input Input/Output No Connect Output Output Enable Supply Voltage NC (NOTE 9) (NOTE 8) (NOTE 7) (NOTE 6) (NOTE 5) (NOTE 4) PAl20R8 Family AMD~ ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL T 20 -,- R 8-5 -r- FAMILY TYPE PAL = Programmable Array Logic P C L OPTIONAL PROCESSING Blank = Standard Processing NUMBER OF ARRAY INPUTS OUTPUT TYPE - - - - - - - - - - " R = Registered L = Active-low combinatorial 1...-__ OPERATING CONDITIONS C = Commercial (O°C to +7S0C) NUMBER OF OUTPUTS - - - - - - - - - - ' SPEED -5 = 5 ns tPD -7 = 7.5 ns tPD -10 = 10 ns tPD PACKAGE TYPE P = 24-Pin (300 mil) Plastic SKINNYDIP (PD3024) J = 2S-Pin Plastic Leaded Chip Carrier (PL 02S) D = 24-Pin (300 mil) Ceramic SKINNYDIP (CD3024) VERSION - - - - - - - - - - - - - - - 1 Blank = Revision 1 12 = Revision 2 Valid Combinations PAL20LS-5 PAL20RS-S PAL20R6-S PAL20R4-S PAL20LS-10/2 Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult· the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PC,JC PAL20RS-10/2 PAL20R6-1 012 PAL20R4-1012 PAL20LS-7 PAL20RS-7 PAL20R6-7 PC, JC, DC PAL20R4-7 PAL20R8-S17, -10/2 2-169 ~ AMD ORDERING INFORMATION Commercial Products (MMI Marking Only) AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: . PAL T 20 -~ R 8 B -2 -,---r FAMILY TYPE PAL = Programmable Array Logic C NS 11 R = Registered L = Active-Low Combinatorial NUMBER OF OUTPUTS - - - - - - - - - ' SPEED B = Very High Speed (15 - 25 ns tpo) A = High speed (25 - 35 ns tpo) L...-_ _ _ _ _ ----------------1 POWER Blank Full Power (210 mA Icc) -2 = Half Power (105 mA Icc) CNS, CFN. CJS B.A CNS. CNL, CJS PAL20R8 PAL20R6 OPERATING CONDITIONS C = Commercial (O°C to +75°C) Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. Valid Combinations B-2 Blank = Standard Processing PACKAGE TYPE NS = 24-Pin 300 mil Plastic SKINNYDIP (PD3024) FN = 28-Pin Plastic Leaded Chip Carrier (PL 028), JEDEC pinout NL = 28-Pin Plastic Leaded Chip Carrier (PL 028). non.JEDEC pinout JS = 24-Pin 300 mil Ceramic SKINNYDIP (CD3024) NUMBER OF ARRAY INPUTS OUTPUT TYPE _ _ _ _ _ _ _ _--1 PAL20L8 OPTIONAL PROCESSING PAL20R4 Note: Marked with MMllogo. 2-170 PAL20R8B/AlB-2 AMD~ ORDERING INFORMATION APL Products AMD programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STD-883 requirements. The order number (Valid Combination) is formed by a combination of: 20 PAL FAMILY TYPE PAL = Programmable Array Logic J R 8 -10/B L A L LEAD FINISH A = Hot Solder Dip NUMBER OF ARRAY INPUTS OUTPUT TYPE R = Registered L = Active-low combinatorial PACKAGE TYPE L = 24-Pin (300·mil) Ceramic SKINNYDIP (CD3024) 3 = 28-Pin Ceramic Leadless Chip Carrier (CL 028) . NUMBER OF OUTPUTS SPEED -10 = 10 ns tPD -12 = 12 ns tPD -15 = 15 ns tPD DEVICE CLASS IB = Class B Valid Combinations PAL20L8 PAL20R8 PAL20R6 -10, -12, -15 IBLA,/B3A Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PAL20R4 Group A Tests Group A Tests consist of Subgroups: 1,2,3, 7, 8, 9, 10, 11. Military Burn-In Military burn-in is in accordance with the current revision of MIL-STD-883, Test Methods 1015, Conditions A through Test conditions are selected at AMD's option. PAL20 R8-1 0/12/15 (Mi I) E. 2-171 ~AMD ORDERING INFORMATION APL Products (MMI Marking Only) AMD programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STD-883 requirements. The order number (Valid Combination) is formed bya combination of: PAL T 20 R 8 A ------r-- FAMILY TYPE PAL = Programmable Array Logic NUMBER OF ARRAY INPUTS M JS/883B L PACKAGE TYPE (Per 09-000) JS = 24-Pin 300-Mil Ceramic SKINNYDIP (CD3024) W 24-Pin Ceramic Flatpack "(CFL024) L 28-Pin Ceramic Leadless Chip Carrier (CL 028) OUTPUT TYPE - - - - - - - - - - ' R = Registered L = Active-Low Combinatorial NUMBER OF OUTPUTS - - - - - - - - - - ' SPEED B = Very High Speed (20 ns tPD) A = High speed (30-50 ns tPD) OPTIONAL PROCESSING /883B = MIL-STD-883, Class B ' - - - - - - OPERATING CONDITIONS M = Military POWER Blank = Full Power (210 rnA Icc) Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. Valid Combinations PAL20L8 PAL20R8 PAL20R6 PAL20R4 B,A MJS/883B, MW/883B, MU883B Note: Marked with MMllogo. Group A Tests Group A Tests consist of Subgroups: 1,2,3, 7, 8, 9, 10, 11. Military Burn-In Military burn-in is in accordance with the current revision of MIL-STD-883, Test Methods 1015, Conditions A through E. Test conditions are selected at AMD's option. 2·172 PAL20R8B/A (Mil) AMD~ FUNCTIONAL DESCRIPTION Standard 24-Pin PAL Family Register Preload The standard 24-pin PAL family is comprised of four different devices, including both registered and combinatorial devices. All parts are produced with a fuse link at each input to the AND gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Using any of a number of development packages, these products can be rapidly programmed to any customized pattern. Extra test words are pre-programmed during manufacturing to ensure extremely high field programming yields, and provide extra test paths to achieve excellent parametric correlation. Variable Input/Output Pin Ratio The registered devices have twelve dedicated input lines, and each combinatorial output is an 1/0 pin. The PAL20L8 has fourteen dedicated input lines, and only six of the eight combinatorial outputs are 1/0 pins. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vee or GND. Programmable Three-State Outputs Each output has a three-state output buffer with threestate control. On combinatorial outputs, a product term controls the buffer, allowing enable and disable to be a function of any product of device inputs or output feedback. The combinatorial output provides a bidirectional lID pin, and may be configured as a dedicated input if the buffer is always disabled. On registered outputs, an input pin controls the enabling of the three-state outputs. Registers with Feedback Registered outputs are provided for data storage and synchronization. Registers are composed of D-type flipflops that are loaded on the LOW-to-HIGH transition of the clock input. The register on the PAL20R8 Family can be preloaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. Security Fuse After programming and verification, a PAL20R8 Family , design can be secured by programming the security fuse. Once programmed, this fuse defeats read back of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. When the security fuse is programmed, the array will read as if every fuse is intact. Quality and Testability The PAL20R8 Family offers a very high level of built-in quality. Extra programmable fuses provide a means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Technology The PAL20R8-5, -7 and 1012 are fabricated with AM D's oxide isolated process. The array connections are formed with highly reliable PtSi fuses. The PAL20R8B, B-2, and A series are fabricated with AMD's trench-isolated bipolar process. The array connections are formed with proven TiW fuses. These processes reduce parasitic capacitances and minimum geometries to provide higher performance. Power-Up Reset All flip-flops power-up to a logic LOW for predictable system initialization. Outputs of the PAL20R8 Family will be HIGH due to the active-low outputs. The Vee rise must be monotonic and the reset delay time is 1000 ns maximum. PAL20R8 Family 2-173 ~ AMD LOGIC DIAGRAM DIP (PLCC) Pinouts 20L8 10 1 (2) 0 3 4 7 8 1112 1516 1920 2324 2728 3132 35 36 24 (28) j 39 ~ (3) 8 ~ (27) l ~ ~ 3 (4) " :r .. ...... 22 (26) 16 21 (25) L :1 23 4 > (5) 24 ~ ..1 ..... ~ 31 5 .. > (6) 32 ",""r- " l...... LJ ""1 39 6 (7) .. .. l..... . J 47 ""---",""r- > 48 l -k ...... 55 8 (10) , l L ~ 63 9 .> 64 11 (13) GND ...> (14) .. ...> m l..... ... .-i 71 19 10 (12) vas 18 (21) --- ------ 17 (20) <...------ .> 56 (11) 19 (23) <...------ ~ 40 -' 7 (9) 201 (24) 0 3 4 7 8 11 12 1516 1920 23 24 2728 31 32 3536 16 I/,, 15 3 (4) "'" ./ ~. 31 ... ~ 32 .> 39 40 ---,........ ,......; ,-.. ~ 47 7 (9) ~ H ~ H ~ H 10- -~ 6 (7) ~ h (25) 24 ~ 5 (6) (26) 16 23 4 (5) 23 (27) 10- ... 7- (24) (23) (21) 48 ....... ~ H ~ H ~ (20) . ""- 55 8 (10) > 56 "'" (19) 63 9 (11) > 64 "'" .. t--' 71 19 10 (12) 11 (13) 7- > G NO Ii2J---, (14) ~ .~ ~ o 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 32 35 36 39 ~ (18) 14 111 (17) OE ~ (16) 164908-10 PAL20R8 Family 2-175 ~ AMD LOGIC DIAGRAM DIP (PLCC) Pinouts 20R6 elK 1 (2) ..... V" 0 3 4 7 8 11 12 1516 1920 2324 2728 3132 35 36 (28) 39 ~ J ~ L 3 (4) 22 (26) :J I 15 23 (27) ~ 16 a f1l~ ........ (25) 23 4 (5) > ~ ~ 24 ""'" ~ 31 5 (6) ~ ~ -~ 39 ~ ~ ~ ""'" ~ 47 ... .> 48 ""'" 55 8 .,.... ... ~ 56 --" 63 9 (11) .,..... ..> 64 ,J ~ I-J 71 19 10 (12) 11 (13) ~ ~ ~ H ... .> (14) ~ J (20) (19) 15 (18) 14 111 (17) > GN D I12l---, (21) -~ ~ ~ (10) ~ h. (23) 40 --, 7 (9) ~ H (24) 32 0--. 6 (7) ~ ~ o 3 4 78 111215161920232427283132353639 OE ~ (16) 164908-11 2-176 PAL20R8 Family AMD~ LOGIC DIAGRAM DIP (PLCC) Pinouts eLK 1 (2) 20R4 to.. ..... 0 3 4 7 8 11 12 1516 1920 23 24 2728 ~ Vee 3132 3536 (28) 39 I~ (3) 8 ..: 9::! > 16 ~ 23 4 (5) 22 (26) , 24 ~ 21 (25) >- ~~ a (24) 31 5 (6) .. .> < 1""1-- 32 I-- >-- D-- 39 6 (7) .. ~ ~ - 40 I-- ~ ;....J~ 47 7 (9) ...> L.. ..: 48 ~ ~~ (23) ~~ (21) >-lU ~ (20) 55 8 (10) -> 56 ~ 63 9 (11) .. > 64 ~ ..1. 71 ,J 1 19 10 (12) -> < 11 (13) -> p 55 ... m-r rt, . . ~tlb if ... I (11) .f161 1101 (19j " ax SG rt:.SLOl r: fu ... ox 10 ~ 56 63 19 SL03 r:~~ 47 18 n 0: ~ data-cf-modified-ae57a6826a2821ddd7e1925a->-r>~ ~ fiil. (2ii vcc;:t~~ • ,1 ~L02 ~ SGl 40 .ff81 IaX I - - - (9j' .r: ... ... .... = 1.1 -tt 9.. ~ >p ~ 0: 1'" -db 1a Vcc';oa . .r:.; . • LOO SG1 lox " ~SLOo (i2) .L~ 10 fi'iI- .A <,1'3) 1 1100 (18) IT -SG .f14I ii7i .f131 o (16) 3 4 7 8 11 12 15 16 19 20 2324 27 28 3132 35 36 39 164918-6 (concluded) PALCE20V8 Family 2-213 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Ambient Temperature with Power Applied ............. -55°C to + 125°C Commercial (C) Devices Temperature (TA) Operating in Free Air . . . . . . . . . . . . . . . . . . . .. DoC to 75°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ......... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vcc + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or 1/0 Pin Voltage ............... -0.5 V to Vcc + 0.5 V Static Discharge Voltage ................. 2001 V Latchup Current (TA = DoC to 75°C) . . . . . . . . . . . . . . . . . . . .. 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified ' Parameter Symbol VOH Parameter Description Test Conditions Output HIGH Voltage IOH =-3.2 rnA Min VIN = VIH or Vil Max 2.4 Unit V Vee = Min VOL Output LOW Voltage IOl '" 24 rnA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) hH Input HIGH Leakage Current VIN = 5.25 V, Vee'" Max (Note 2) III Input LOW Leakage Current VIN '" 0 V, Vee '" Max (Note 2) IOZH Off-State Output Leakage Current HIGH VOUT '" 5.25 V, Vee = Max VIN '" VIH or Vil (Note 2) IOZl Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or Vil (Note 2) Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) Ice Supply Current (Static) Outputs Open (lOUT = 0 rnA), VIN = 0 V Vee = Max 0.5 VIN '" VIH or Vil V 2.0 V V -- ·O.S· -30 10 ~ -100 ~ 10 ~ -100 ~ -150 rnA 125 rnA Notes: 1. These are absolute values with respect to device ground all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of hL and IOZL (or hH and IOZH). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2·214 PALCE20V8H-5 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Descriptions Test Conditions Input Capacitance VIN= 2.0 V Output Capacitance VOUT=2.0V I I Typ Unit Vcc = 5.0 V. TA = 25°C. 5 pF f = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol Parameter Description Min (Note 5) Max Unit 5 ns tpD Input or Feedback to Combinatorial Output 1 ts Setup lime from Input or Feedback to Clock 3 tH Hold Time 0 Clock to Output 1 tco tSKEWR Skew Between Registered Outputs (Note 4) twH Clock Width HIGH External Feedback fMAX Maximum Frequency (Note 3) ns 4 ns 1 ns 3 LOW twL ns I 1/(ts+tco) Internal Feedback (tCNT) No Feedback I 1/(tWH+twL) ns 3 ns 142.8 MHz 166 MHz 166 MHz tpzx OE to Output Enable 1 6 ns tpxz OE to Output Disable 1 5 ns tEA Input to Output Enable Using Product Term Control .2 6 ns tER Input to Output Disable Using Product Term Control 2 5 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. 4. Skew testing takes into account pattern and switching direction differences between outputs that have equalloading. 5. Output delay minimums for tpo, teo, tpzx, tpxz, tEA, and tER are defined under best case conditions. Future process improvements may alter these values therefore, minimum values are recommended for simulation purposes only. PALCE20V8H-5 (Com'l) 2-215 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Com~erclal Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 0.5 V DC Output or 1/0 Pin Voltage ............... -0.5 V to Vee + 0.5 V (C) Devices Temperature (TA) Operating in Free Air . . . . . . . . . . . . . .. . . . . .. O°C to 75°C Supply Voltage (Vee) with Respect to Ground ......... +4.75 V to +5.25 V Ambient Temperature with Power Applied ............. -55°C to + 125°C Operating ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Latehup Current (TA = O°C to 75°C) . . . . . . . . . . . . . . . . . . . .. 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Min Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 mA Vee = Min VOL Output LOW Voltage IOL = 24 mA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) VIN = VIH or VIL Input HIGH Leakage Current VIN = 5.25 V, Vee = Max (Note 2) ilL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) IOZH Off-State Output Leakage Current HIGH VOUT = 5.25 V, Vee = Max VIN = VIH or VIL (Note 2) lozl Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or VIL (Note 2) Isc Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) Icc Supply Current (Dynamic) Outputs Open (lOUT = 0 mA) Vee = Max, f = 25 MHz Unit V V 0.5 VIN = VIH or VIL IIH Max 2.4 V 2.0 V 0.8 ---_ .. _- -30 _._-_ .. _---- 10 ~ -100 ~ 10 ~ -100 ~ -150 mA 115 mA Notes: 1. These are absolute values with respect to device ground all overshoots due to system andlor tester noise are included. 2. 110 pin leakage is the worst case of ItL and 10zL (or I/H and 10zH ). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-216 PALCE20V8H-7 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Descriptions Typ Test Conditions Input Capacitance VIN= 2.0V Output Capacitance VOUT=2.0 V I I Unit Vee = 5.0 V, TA = 25°C, 5 pF f = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol tPD Parameter Description Input or Feedback to Combinatorial Output I 8 Outputs Switching I 1 Output Switching Min Max 3 7.5 ns 3 7 ns ts Setup Time from Input or Feedback to Clock 5 tH Hold Time 0 tco Clock to Output 1 tSKEWR twL Skew Between Registered Outputs (Note 4) Clock Width tWH fMAX Maximum Frequency (Note 3) Unit ns ns 5 ns 1 ns LOW 4 ns HIGH 4 ns 100 MHz 125 MHz External Feedback I 1/(ts+tco) I 1/(tWH+tWL) Internal Feedback (fCNT) No Feedback 125 MHz tpzx OE to Output Enable 1 6 ns tpxz OE to Output Disable 1 6 ns tEA Input to Output Enable Using Product Term Control 3 9 ns tER Input to Output Disable Using Product Term Control 3 9 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. 4. Skew testing takes into account pattern and switching direction differences between outputs that have equal loading. 5. Output delay minimums for tPD, tco, tpzx, tpxz, tEA, and tER are defined under best case conditions. Future process improvements may alter these values therefore, minimum values are recommended for simulation purposes only. PALCE20V8H-7 (Com'l) 2-217 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Commercial (C) Devices Temperature (TA) Operating in Free Air ..................... DoC to 75°C Supply Voltage (Vee) with Respect to Ground ......... +4.75 V to +5.25 V Ambient Temperature with Power Applied ............. -55°C to + 125°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 0.5 V DC Output or 110 Pin Voltage ............... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Latehup Current (TA = DoC to 75°C) ..................... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 mA Vee = Min VOL Output LOW Voltage IOl = 24 rnA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Noto 1) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) Min VIN = VIH or Vil Input HIGH Leakage Current VIN = 5.25 V, Vee = Max (Note 2) IlL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) IOZH Off-State Output Leakage Current HIGH Your = 5.25 V, Vee = Max VIN = VIH or Vil (Note 2) lOll Off-State Output Leakage Current LOW Your = 0 V, Vee = Max VIN = VIH or Vil (Note 2) Ise Output Short-Circuit Current Your = 0.5 V, Vee = Max (Note 3) Icc Supply Current (Dynamic) Outputs Open (lour = 0 mAl Vee = Max, f = 25 MHz Unit 2.4 V 0.5 VIN = VIH or Vil hH Max V 2.0 V 0.8 -- ...... 10 -30 V --- ~ -10 ~ 10 ~ -10 ~ -150 mA 115 mA Notes: 1. These are absolute values with respect to device ground all overshoots due to system and/or tester noise are included. 2. //0 pin leakage is the worst case of hL and lozL (or hH and lozH ). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-218 PALCE20V8H-10 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN= 2.0 V Output Capacitance VOUT = 2.0 V 1 f = '1 MHz 1 Vee = 5.0 V, TA = 25°C, Typ Unit 5 pF 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol Parameter Description Min (Note 4) Max Unit 3 10 ns tPD Input or Feedback to Combinatorial Output ts Setup Time from Input or Feedback to Clock tH Hold Time 0 Clock to Output 3 teo tWL Clock Width tWH tMAX Maximum Frequency (Note 3) 7.5 ns ns 7.5 ns LOW 6 HIGH 6 ns 66.7 MHz Internal Feedback (teNT) 71.4 MHz No Feedback 83.3 MHz External Feedback 11I(ts+tcO) 11/(twH+twL) ns tpzx OE to Output Enable 2 10 ns tpxz OE to Output Disable 2 10 ns tEA Input to Output Enable Using Product Term Control 3 10 ns tER Input to Output Disable Using Product Term Control 3 10 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. 4. Output delay minimums for tPD, tco, tpzx, tpxz, tEA, and tER are defined under best case conditions. Future process improvements may alter these values therefore, minimum values are recommended for simulation purposes only. PALCE20V8H-10 (Com'l) 2-219 ~ PRELIMINARY AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Temperature (TA) Operating in Free Air . . . . . . . . . . . . . . . . . . . .. O°C to 75°C Supply Voltage (Vee) with Respect to Ground ......... +4.75 V to +5.25 V Ambient Temperature with Power Applied ............. -55°C to + 125°C Supply Voltage with. Respect to Ground :............ -0.5 V to +7.0 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 0.5 V DC Output or I/O Pin Voltage ............... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Latchup Current (TA = O°C to 75°C) ..................... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified PRELIMINARY Parameter Symbol Min Parameter Description Test Conditions VOH Output HIGH yoltage IOH =-3.2 rnA Vee = Min VOL Output LOW Voltage IOL = 24 rnA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) IIH Input HIGH Leakage Current VIN = 5.25 V, Vee = Max (Note 2) IlL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) IOZH Off-State Output Leakage Current HIGH VOUT = 5.25 V, Vee = Max VIN = VIH or VIL (Note 2) IOZL Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or VIL (Note 2) Isc Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) Icc Supply Current (Dynamic) Outputs Open (lOUT = 0 rnA) Vee = Max, f = 15 MHz VIN = VIH or VIL Max Unit 2.4 V 0.5 VIN = VIH or VIL V 2.0 -30 V 0.8 V 10 ~ -10 ~ 10 ~ -10 ~ -150 rnA 55 rnA Notes: 1. These are absolute values with respect to device ground all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of hL and IOZL (or I/H and IOZH ). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-220 PALCE20VSQ-10 (Com'l) AMD~ PRELIMINARY CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN= 2.0 V Output Capacitance VOUT = 2.0 V Typ I Vee = 5.0 V, TA = 25°C, I f = 1 MHz Unit 5 pF 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) PRELIMINARY Parameter Symbol Parameter Description tPD Input or Feedback to Combinatorial Output ts Setup Time from Input or Feedback to Clock tH Hold Time teo tWL Clock Width fMAX Maximum Frequency (Note 3) tpzx OE to Output Enable tpxz tEA tER Max 3 10 Unit ns 7.5 ns 0 ns 3 Clock to Output tWH Min (Note 4) 7.5 ns LOW 6 ns HIGH 6 ns 66.7 MHz Internal Feedback (feNT) 71.4 MHz No Feedback 83.3 External Feedback 11I(ts+teo) 11/(twH+twL) MHz 2 10 ns OE to Output Disable 2 10 ns Input to Output Enable Using Product Term Control 3 10 ns Input to Output Disable Using Product Term Control 3 10 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modifi9d where frequency may be affected. 4. Output delay minimums for tPD, teo, tpzx, tpxz, tEA, and tER are defined under best case conditions. Future process improvements may alter these values therefore, minimum values are recommended for simulation purposes only. PALCE20VaQ-10 (Com'l) 2-221 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Ambient Temperature with Power Applied ............. -55°C to + 125°C Commercial (C) Devices Temperature (TA) Operating in Free Air .................... O°C to +75°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage ........... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or I/O Pin Voltage ............. -0.5 V to Vee + 0.5 V Static Discharge Voltage .......' .......... 2001 V Latchup Current (TA = O°C to +75°C) .................... 100 rnA Stresses above those listed under Absolute Maximum Rat, ings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Min Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 rnA Vee = Min VOL Output LOW Voltage IOl = 24 mA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) IIH Input HIGH Leakage Current III Input LOW Leakage Current = 5.25 V, Vee = Max (Note 2) VIN = 0 V, Vee = Max (Note 2) VOUT = 5.25 V, Vee = Max VIN = VIH or Vil (Note 2) VOUT = 0 V, Vee = Max VIN = VIH or Vil (Note 2) VOUT = 0.5 V, Vee = Max (Note 3) Outputs Open (lOUT = 0 rnA) I Vee = Max, f = 15 MHz I IOZH Off-State Output Leakage Current HIGH IOZl Off-State Output Leakage Current LOW Isc Output Short-Circuit Current Icc Supply Current Unit 2.4 VIN = VIH or Vil VIN Max = VIH or Vil V 0.5 V 2.0 _ ... VIN -30 H Q V 0.8 --. ._---- . - - V..._.. __ 10 ~ -10 ~ 10 ~ -10 J.1A -150 rnA 90 55 rnA Notes: 1. These are absolute values with respect to device ground all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of hL and lozL (or hH and lozH ). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-222 PALCE20V8H-15/25 Q-15/25 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN =2.0 V Typ Output Capacitance VOUT = 2.0 V If = 1 MHz I Vee = 5.0 V, TA = 25°C , Unit 5 pF 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol -15 Parameter Description Min tPD Input or Feedback to Combinatorial Output ts Setup Time from Input or Feedback to Clock 12 tH Hold Time 0 teo fMAX Clock Width Maximum Frequency (Note 3) Min 15 Max 25 15 Unit ns ns 0 10 Clock to Output tWL tWH -25 Max ns 12 ns LOW 8 12 ns HIGH 8 12 ns 45.5 37 MHz External Feedback 11/(ts+tcO) Internal Feedback (fCNT) No Feedback 11/(tWH+tWL) 50 40 MHz 62.5 41.6 MHz tpzx OE to Output Enable 15 20 ns tpxz OE to Output Disable 15 20 ns tEA Input to Output Enable Using Product Term Control 15 25 ns tER Input to Output Disable Using Product Term Control 15 25 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PALCE20V8H-15/25 0-15/25 (Com'l) 2-223 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Military (M) Devices (Note 1) Ambient Temperature with Power Applied ............. -55°C to + 125°C Operating Case Temperature (Te) ........... -55°C to'+125°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ....... +4.5 V to +5.5 V DC Input Voltage ........... -0.5 V to Vee + 1.0 V Note: 1. Military products are tested at Tc = +25"C, + 125"C and -55"C, per MIL-STD-883. DC Output or I/O . Pin Voltage ............... -0.5 V to Vee + 1.0 V Static Discharge Voltage ................. 2001 V Latehup Current (Te =-55°C to +125°C) ................ 100 Operating ranges define those limits between which the functionality of the device is guaranteed. rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Absolute Maximum Ratings are for system design reference; parameters given are not tested. DC CHARACTERISTICS over MILITARV operating ranges unless otherwise specified (Note 2) . Parameter Symbol Parameter Description Test Conditions Min VOH Output HIGH Voltage IOH =-2.0 mA Vee = Min VOL Output LOW Voltage IOl = 12 mA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 3) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 3) 0.8 V hH Input HIGH Leakage Current VIN 10 ~ hL Input LOW Leakage Current = 5.5 V, Vee = Max (Note 4) VIN = 0 V, Vec = Max (Note 4) -10 ~ IOZH Off-State Output Leakage Current HIGH VOUT = 5.5 V, Vee = Max VIN = VIH or Vil (Note 4) 10 IlA IOZl Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or Vil (Note 4) -10 IlA Ise Output Short-Circuit Current VOUT = 0.5 V, Vee (Note 5) -150 mA Icc Supply Current (Dynamic) Outputs Open (lOUT = 0 mAl Vee = Max, f = 25 MHz 130 mA VIN = VIH or Vil VIN 2.4 = VIH or Vil = 5.0 V, TA = 25°C Max V 0.5 2.0 -30 Unit V V Notes: 2. For APL products, Group A, Subgroups 1, 2 and 3 are tested per MIL-STD-833, Method 5005, unless otherwise noted. 3. VIL and V/H are input conditions of output tests and are not themselves directly tested. VIL and V/H are absolute voltages with respect to device ground and include all overshoots due to system andlor tester noise. Do not attemptto test these values without suitable equipment. 4. 110 pin leakage is the worst case of liL and lozL (or hH and IOZH ). 5. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. This parameter is not 100% tested, but is evaluated at initial characterization and at any time the design is modified where Isc may be affected. 2·224 PALCE20V8H·15 (Mil) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN= 2.0 V Output Capacitance VOUT = 2.0 V I Vee = 5.0 V, TA = 25°C, I f = 1 MHz Typ Unit 8 pF 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over MILITARY operating ranges (Note 2) -15 Parameter Symbol Parameter Description Min (Note 5) Max Unit 15 ns tPD Input or Feedback to Combinatorial Output 3 ts Setup Time from Input or Feedback to Clock 12 ns tH Hold Time 0 ns Clock to Output 3 teo tWL Clock Width tWH fMAX tpzx Maximum Frequency (Note 3) 12 ns LOW 10 HIGH 10 ns 41.6 MHz Internal Feedback (teNT) 45.5 MHz No Feedback 50.0 External Feedback /1/(ts+teo) 11/(twH+twL) ns MHz OE to Output Enable 3 15 ns tpxz OE to Output Disable 3 15 ns tEA Input to Output Enable Using Product Term Control (Note 4) 3 15 ns tER Input to Output Disable Using Product Term Control (Note 4) 3 15 ns Notes: 2. See Switching Test Circuit for test conditions. For APL Products, Group A, Subgroups 9, 10, and 11 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. 4. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. 5. Output delay minimums for tpo, teo, tpzx, tpxz, tEA, and tER are defined under best case conditions. Future process improvements may alter these values therefore, minimum values are recommended for simulation purposes only. PALCE20V8H-15 (Mil) 2-225 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Military (M) Devices (Note 1) Ambient Temperature with Power Applied ............. -55°C to + 125°C Operating Case Temperature (Te) ........... -55°C to +125°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ....... +4.5 V to +5.5 V DC Input Voltage . . . . . . . . . .. -0.5 V to Vee + 0.5 V Note: 1. Military products are tested at Tc = +2SOC, +12SOC and -5SOC, per MIL-STD-883. DC Output 0 r I/O Pin Voltage ............... -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Latehup Current (Tc =-55°C to +125°C) ................ 100 rnA Operating ranges define those limits between which the functionality of the device is guaranteed. Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Absolute Maximum Ratings are for system design reference; parameters given are not tested. DC CHARACTERISTICS over MILITARY operating ranges unless otherwise specified (Note 2) Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-2.0 rnA Vee = Min VOl Output LOW Voltage IOL = 12 rnA Min VIN = VIH or VIL Max Unit 2.4 V 0.5 VIN = VIH or VIL V Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 3) Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 3) IIH Input HIGH Leakage Current VIN = 5.5 V, Vee = Max (Note 4) ilL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 4) IOZH Off-State Output Leakage Current HIGH VOUT = 5.5 V. Vee = Max VIN = VIH or VIL (Note 4) IOZl Off-State Output Leakage Vour = 0 V, Vee = Max VIN = VIH or Vil (Note 4) VIL Current LOW Isc Output Short-Circuit Current Vour = 0.5 V, Vee = 5.0 V. TA = 25°C (Note 5) Icc Supply Current (Dynamic) Outputs Open (lOUT = 0 mA) Vee = Max, f = 25 MHz 2.0 V ....... -30 _. . .. 0.8 V 10 ~ -10 ~ 10 ~A -10 ~A -150 mA 130 mA --- Notes: 2. For APL products, Group A, Subgroups 1, 2 and 3 are tested per MIL-STD-833, Method SODS, unless otherwise noted. 3. V/L and V/H are input conditions of output tests and are not themselves directly tested. V/L and V/H are absolute voltages with respect to device ground and include all overshoots due to system andlor tester noise. Do not attempt to test these values without suitable equipment. 4. liD pin leakage is the worst case of ItL and IOZL (or ItH and loZH ). 5. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 Vhas been chosen to avoid test problems caused by tester ground degradation. This parameter is not 100% tested, but is evaluated at initial characterization and at any time the design is modified where Isc may be affected. 2-226 PALCE20V8H-20/25 (Mil) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance VOUT = 2.0 V Typ I Vee = 5.0 V, TA = 25°C, 1t = 1 MHz Unit 8 pF 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over MILITARY operating ranges (Note 2) Parameter Symbol -20 Parameter Description Min -25 Max Min Unit 25 ns Input or Feedback to Combinatorial Output ts Setup Time trom Input or Feedback to Clock 15 20 ns tH Hold Time (Note 5) 0 0 ns teo Clock to Output tWL tWH tMAX 20 Max tPD Clock Width Maximum Frequency (Note 3) 20 15 ns LOW 12 15 HIGH 12 15 ns External Feedback 11 /(ts+teo) 33.3 25 MHz Internal Feedback (teNT) 35.7 26.3 MHz No Feedback 41.7 33.3 11/(tWH+tWl) tpzx OE to Output Enable (Note 3) tpxz OE to Output Disable (Note 3) tEA Input to Output Enable Using Product Term Control (Note 3) tER Input to Output Disable Using Product Term Control (Note 3) 20 ns MHz 20 ns 18 20 ns 20 25 ns 25 ns 18 Notes: 2. See Switching Test Circuit for test conditions. For APL Products, Group A, Subgroups 9, 10, and 11 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. PALCE20V8H-20/25 (Mil) 2-227 ~ AMD SWITCHING WAVEFORMS Input or Feedback _ _ _- - I Input or Feedback Combinatorial Output -~-VT ' - -_ _ _ _---1 ,-V_T_ __ Clock Registered Output _ _ _ _ _ _ ___ 164918-7 164918-8 Registered Output Combinatorial Output Input Clock tER tEA VT Output 164918-9 164918-10 Clock Width Input to Output Disable/Enable tpxz tpzx VT Output 164918-11 OE to Output Disable/Enable Notes: 1. VT= 1.5 V 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns - 5 ns typical. ·2·228 PALCE20V8 Family AMD~ KEY TO SWITCHING WAVEFORMS WAVEFORM \\\\\ /77// 'lIXXXX }1) CK INPUTS OUTPUTS Must be Steady Will be Steady May Change from H to L Will be Changing from H to L May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State KSOOO010-PAL SWITCHING TEST CIRCUIT Output o - - - - i l I t - - - - -..... 164918-12 Switching Test Circuit Military Commercial Specification Sl CL tPD, teo Closed 50 pF tpzx, tEA Z~ 50 pF Z~ tpxz, tER H H: Open L: Closed ~Z: L~Z: Open R1 R1 Measured Output Value R2 1.5 V 2000. 5 pF R2 3900. H-5: 2000. Closed PALCE20V8 Family 1.5 V 3900. 7500. H ~Z: VOH -0.5 V L ~ Z: VOL + 0.5 V 2·229 ~ AMD TYPICAL Icc CHARACTERISTICS Vee 5.0 V, TA 25°C = = 150 125 20VSH-5 100 20VSH-7 ,Icc (mA) 75 20VSH-10 20VSH-15125 50 20VSQ-15125 25 O-+----~----~---+----~----+_--~----_r----~--~~--~ o 10 20 30 Frequency (MHz) 40 50 164918-13 Icc vs. Frequency The selected "typical" pattern utilized 50% of the device resources. Half of the macroce/ls were programmed as registered, and the other half were programmed as combinatorial. Half of the available product terms were used for each macrocell. On any vector, half of the outputs were switching. By utilizing 50% of the device, a midpoint is defined for Icc. From this midpoint, a designer may scale the Icc graphs up or down to estimate the Icc requirements for a particular design. 2-230 PALCE20V8 Family AMD~ ENDURANCE CHARACTERISTICS The PALCE20V8 is manufactured using AMD's advanced electrically erasable process. This technology uses an EE cell to replace the fuse link used in bipolar parts. As a result, the device can be erased and reprogrammed-a feature which allows 100% testing at the factory. Endurance Characteristics Symbol Parameter Min Unit Max Storage Temperature Test Conditions 10 Years 20 Years 100 Cycles tOR Min Pattern Data Retention Time Max Operating Temperature N Min Reprogramming Cycles Normal Programming Conditions INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee ESD ProgramNerify Circuitry Protection Typical Input Vee Preload Circuitry Typical Output PALCE20V8 Family Feedback Input 164918-14 2·231 ~ AMD ROBUSTNESS FEATURES The PALCE20V8X-x/5 have some unique features that make them extremely robust, especially when operating in high-speed design environments. Pull-up resistors on inputs and I/O pins cause unconnected pins to default to a known state. Input clamping circuitry limits negative overshoot, eliminating the possibility of false clocking caused by subsequent ringing. A special noise filter makes the programming circuitry completely insensitive to any positive overshoot that has a pulse width of less than about 100 ns for the /5 versions. Selected /4 devices are also being retrofitted with these robustness features. See the chart below for device listings. INPUT/OUTPUT EQUIVALENT SCHEMATICS FOR /5 VERSION AND SELECTED 14 VERSIONS· Vee Vcc >50kn I I I I I I I ESD Protection and Clamping . = ____ _ Programming LF2n!.~!r Typical Input Vee >50kn Preload Circuitry = Feedback Input Typical Output 164918-15 • Device Rev Letter PALCE20V8H-10 K PALCE20V8H-15 K,J PALCE20V80-15 J PALCE20V8H-25 J PALCE20V80-25 J 2-232 Topside Marking: AMO CMOS PLO's are marked on top of the package in the following manner: PALCEXXXX Oatecode (3 numbers) Lot 10 (4 characters)- -(Rev Letter) The Lot 10 and Rev Letter are separated by two spaces. PALCE20va Family AMD~ POWER-UP RESET The PALCE20V8 has been designed with the capability to reset during system power-up. Following power-up,all flip-flops will be reset to LOW. The output state will be HIGH independent of the logic polarity. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset Parameter Symbol and the wide range of ways Vec can rise to its steady state, .two conditions are required to insure a-valid power-up reset. These conditions are: • The Vcc rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Description Min tPR Power-Up Reset Time ts Input or Feedback Setup Time twL Clock Width LOW Max Unit 1000 ns See Switching Characteristics Power . . . - - - tPR ---~ Registered Output Clock 164918-16 Power-Up Reset Waveforms PALCE20V8 Family 2-233 ~ AMD TYPICAL THERMAL CHARACTERISTICS /4 Devices (PALCE20V8H-10/4) Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Typ. PLCC Unit Sjc Thermal impedance, junction to case 19 19 °eIW Sja Thermal impedance, junction to ambient 73 55 °eIW Sjma Parameter Description Thermal impedance, junction to ambient with air flow SKINNYDIP 200 Ifpm air 61 45 400 Ifpm air 53 41 600 Ifpm air 50 38 800 Ifpm air 47 36 °eIW °eIW °eIW °eIW PLCC Unit /5 Devices (PALCE20V8H-7/5) Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Typ. SKINNYDIP Parameter Description Sjc Thermal impedance, junction to case 18 16 °eIW Sja Thermal impedance, junction to ambient 69 51 °eIW 200 Ifpm air 60 42 400 Ifpm air 54 37 600 Ifpm air 50 36 800 Ifpm air X X °eIW °eIW °eIW °eIW Sjma Thermal impedance, junction to ambient with air flow Plastic S/c Considerations The data listed for plastic Sjc are for reference only and are not recommended for use in calculating junction temperatures. The heat-floW paths in plastic-encapsulated devices are complex, making the Sjc measurement relative to a specific location on the package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom center of the package. Furthermore, Sjc tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a similar environment.-- ----------------------------- 2·234 PALCE20va Family AMD~ 2·235 _COM'L:-20 ~ INO: -20 Advanced Micro Devices PALCE20RA10H·20 24-Pin Asynchronous EE CMOS Programmable Array Logic DISTINCTIVE CHARACTERISTICS • Low power at 90 mA Icc • • As fast as 20 ns maximum propagation delay and 37 MHz fMAx (external) Programmable replacement for high-speed CMOS or TTL logic - • TTL-level register preload for testability Extensive third-party software and programmer support through FusionPLD partners 24-pln SKINNVOIP and 28-pln PLCC packages save space • Individually programmable asynchronous clock, preset, reset, and enable • • Registered or combinatorial outputs • • Programmable polarity GENERAL DESCRIPTION The PALCE20RA10 is an advanced PAL device built with low-power, high-speed, electrically-erasable CMOS technology. The PALCE20RA10 offers asynchronous clocking for each of the ten flip-flops in the device. The ten macrocells feature programmable clock, preset, reset, and enable, and all can operate asynchronously to other macrocells in the same device. The PALCE20RA10 also has flip-flop bypass, allowing any combination of registered and combinatorial outputs. The PALCE20RA10 utilizes the familiar sum-of-products (AND/OR) architecture that allows users to implement complex logic functions easily and efficiently. Multple levels of combinatorial logic can always be reduced to sum-of-products form, taking advantage of the very wide input gates available in PAL devices. The BLOCK DIAGRAM 1/00 2-236 equations are programmed into the device through floating-gate cells in the AND logic array that can be erased electrically. AMD's FusionPLD program allows PALCE20RA10 designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools fo r each device:- The -Fu sion PLD -program -also--greatly reduces design time since a deSigner can use a tool that is already installed and familiar. Please refer to the PLD Software reference guide for certified development systems and the Programmer reference guide for approved programmers. Dedicated Inputs 1/03 1/05 1/06 1/07 I/0a Publication# 15434 Rev.D Issue Date: June 1993 1I0g 154340-1 AmendmentlO AMD ;r1 CONNECTION DIAGRAMS Top View PLCC SKINNYDIP/FLATPACK 0 PL ~ Vee 10 1/0g h -= .,. ~ g I/Oa 12 25 1/07 12 1107 13 24 1/06 13 1/06 14 23 1/05 14 1/05 22 NC 15 1/04 16 1/03 17 1/02 la 1/01 19 GND NC 15 21 1104 16 20 1103 17 19 1/02 1/00 ..!!' ..!2' DE z IW 0 z o Q (!) 15434D-2 0 g g15434D-3 Note: Pin 1 is marked for orientation. PIN DESIGNATIONS GND I 1/0 NC OE PL Vee Ground Input Input/Output No Connect Output Enable Preload Supply Voltage PALCE20RA1OH-20 2-237 ~AMD ORDERING INFORMATION Commercial and Industrial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL CE 20 FAMILY TYPE PAL = Programmable Array Logic T TECHNOLOGY CE = CMOS Electrically Erasable RA 10 H -20 P C L NUMBER OF ARRAY INPUTS OPERATING CONDITIONS I = Industrial (-40°C to +85°C) C = Commercial (OOC to +75°C) PACKAGE TYPE P = 24-Pin 300-mil Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) OUTPUT TYPE RA = Registered Asynchronous NUMBER OF OUTPUTS POWER H = Half Power (Icc = 90 mA) SPEED -20 = 20 ns tPD Valid Combinations PALCE20RA10H-20 2-238 I Valid Combinations PC, JC, PI, JI Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE20F:lA10H-20 (Com'l, INO) Commercial and Industrial Products Output 15434D-5 Figure 1. PALCE20RA10 Macrocell FUNCTIONAL DESCRIPTION The PALCE20RA10 has ten dedicated input lines and ten programmable I/O macrocells. The Registered Asynchronous (RA) macrocell is shown in Figure 1. Pin 1 serves as global register preload and pin 13 serves as global output enable. Programmable output polarity is available to provide user-programmable output polarity for each individual macrocell. Programmable Clock The clock input to each flip-flop comes from the programmable array, allowing any flip-flop to be clocked independently if desired. The programmable functions in the PALCE20RA10 are automatically configured from the user's design specification, which can be in a number of formats. The design specificaiion is processed by development software to verify the design and create a programming file. This file, once downloaded to a programmer, configures the device according to the user's desired function. Programmable Preset and Reset Registered/Active Low CombinatorlaUActive Low In each macrocell, two product lines are dedicated to asynchronous preset and asynchronous reset. If the preset product line is HIGH, the Q output of the register becomes a logic 1 and the output pin will be a logic O.lf the reset product line is HIGH, the Q output of the register becomes a logic 0 and the output pin will be logic 1. The operation of the programmable preset and reset overrides the clock. Combinatorial/Registered Outputs Registered/Active High If both the preset and reset product lines are HIGH, the flip-flop is bypassed and the output becomes combinatorial. Otherwise, the output is from the register. Each output can be configured to be combinatorial or registered. PALCE20RA 1OH-20 Combinatorial/Active High 15434D-6 Figure 2. Macrocell Configurations 2-239 ~ AMD Three-State Outputs Register Preload The devices provide a product term dedicated to local output control. There is also a global output control pin. The output is enabled if both the global output control pin is LOW and the local output control product term is HIGH. If the global output control pin is HIGH, all outputs will be disabled. If the local output control product term is LOW, then that output will be disabled. The register on the PALCE20RA10 can be preloaded . from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transistions from illegal states can be verified by loading illegal states and observing proper recovery. Register preload is controlled by a TTL-level signal, making it a convenient board-level initialization function. Security Bit A security bit is also provided to prevent unauthorized copying of PAL device patterns. Once the bit is programmed, the circuitry enabling verification is permanently disabled, and the array will read as if every bit is programmed. With verification not operating, it is impossible to simply copy the PAL device pattern on a PAL device programmer. The security bit can only be erased in conjunction with the entire pattern. Power-Up Reset All flip-flops power up to a logic LOW for predictable system initialization. Registered outputs of the PALCE20RA10 will be HIGH due to the output inverter. The state of combinatorial outputs will be a function of the logic. Programmable Polarity Quality and Testability The outputs can be programmed either active-LOW or active-HIGH. This is represented by the Exclusive-OR gate shown in the PALCE20RA 10 logic diagram. When the output polarity bit is programmed, the lower input to the Exclusive-OR gate is HIGH, so the output is activeHIGH. Similarly when the output polarity bit is unprogrammed, the output is active-LOW. The programmable output polarity feature allows the user a higher degree of flexibility when writing equations. The PALCE20RA10 offers a very high level of built-in quality. The erasability of the device provides a means of verifying performance of all AC and OC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Programming and Erasing The PALCE20RA10 can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erase operation is required. 2·240 Technology The high-speed PALCE20RA10H-20 is fabricated with AMO's advanced electrically erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be-compatible--with TTL devices. This technology provides strong input clamp diodes, output slew-rate control, and a grounded substrate for clean switching. PALCE20RA 1OH·20 AMD~ LOGIC DIAGRAM DIP (PLCC JEDEC) Pinouts 1 (2) ....... ··· ·· a 1 23.5. 1 a 1101112131"'5 '.'71"11 201122232425262128213G31 32333435 3&313831 ~D- < r- 7 ~ ...." .. " ,."" . 21 23 ...... ,.• 7 211 311 31 5 (6) 34 35 3J ... ... ... . .... <7 •. 50 52 50 .. 56 8 (10) " 57 58 .. ...." ...... .... 511 9 (11) 17 " ... (21) .... (20) .... (19) ~D~ ~ 73 7< 75 71 77 (17) '" 11 a 1 2 3 4 5 '7 • • 1a 11 12131415 11" 18,. 20 21 2223 242526 27 28 2130 31 3233 '" 35 313738 3D ---I 4-@] (16) [12l-, (i4i (23) ~, 7. 11 (13) 1 (18) " 0 (12) 1 :~D 35 .... 7 (9) '" ~ . 33 6 (7) I'lAII P ~D- " 17 4 (5) ~ ~L~ ~ ~I ~D ~~ "'I ~D ~~ ~~ "'I ~D~~ "I ~D~~ "1 ~D- ~ ~ "'I ~D ~~ 1D 3 (4) _ .... 154340-7 PALCE20RA 1OH-20 2-241 ~AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied ............. -55°C to + 125°C Supply Voltage with ' Respect to Ground ............. -0.5 V to +7.0 V Ambient Temperature (TA) Operating in FreeAir .............. O°C to +75°C Supply Voltage (Vee) with Respect to Ground ........ +4.75 V to +5.25 V DC Input Voltage ........... -0.5 V to Vee + 0.5 V (Except Pin 5) Industrial (I) Devices DC Input Voltage (Pin 5) ........ -0.6 V to + 11.0 V Supply Voltage (Vee) with Respect to Ground .............. +4.5 V to +5.5 V DC Output or I/O Pin Voltage .................. -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Operating Case ................ -40°C to +85°C Operating ranges define those limits between which the functionality of the device is guaranteed. Latchup Current (Te = -40°C to +85°C) . . . . . . . . . . . . . . . . .. 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges unless otherwise specified Parameter Symbol VOH Parameter Description Test Conditions Min Output HIGH Voltage IOH =-3.2 mA VIN = VIH or Vil Vee = Min 2.4 VOL Output LOW Voltage IOl = 8 mAVIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Voltage for ail Inputs (Note 1) Guaranteed Input Logical HIGH Vil Input LOW Voltage Voltage for ail Inputs (Note 1) Guaranteed Input Logical LOW IIH Input HIGH Leakage Current VIN III Input LOW Leakage Current VIN = 0 V, Vee IOZH Off-State Output Leakage Current HIGH IOZl Off-State Output Leakage Current LOW Ise Output Short-Circuit Current Icc Supply Current Vec = Max 2.0 -30 I COM'L liND Unit V 0.4 = 5.5 V, Vee = Max (Note 2) = Max (Note 2) VOUT = 5.5 V, Vce = Max VIN = Vil or VIH (Note 2) VOUT = 0 V, Vee = Max VIN = Vil or VIH (Note 2) VOUT = 0.5 V, Vee = Max (Note 3) VIN = 0 V, Outputs Open· (lOUT = 0 mA) Max V V 0.8 V 10 ~ -100 ~ 10 ~ -100 ~ -150 mA 90 130 mA mA Notes: 1. These are absolute values with respect to device ground and all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of IJL and IOZL (or IJH and IOZH). 3. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. ' 2-242 PALCE20RA10H-20 (Com'I, IND) AMO ;r1 CAPACITANCE (Note 1) Parameter Symbol CIN Parameter Description Test Conditions Input Capacitance VIN = 2.0 V \ Inputs IOE COUT Output Capacitance VOUT = 2.0 V Typ Vee = 5.0 V Unit 5 TA = +25°C 9 f = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges (Note 2) -20 Parameter Symbol Min (Note 3) Parameter Description tPD Input or Feedback to Combinatorial Output ts Setup Time from Input, Feedback or SP to Clock 7 Hold Time 4 tH Max Unit 20 ns ns ns teo Clock to Output or Feedback 20 ns tAp Asynchronous Preset to Registered Output 20 ns 12 ns tApw AsynchronoLis Preset Width tAPR Asynchronous Preset Recovery Time (Note 4) tAR Asynchronous Reset to Registered Output 12 tARW Asynchronous Reset Width tARR Asynchronous Reset Recovery Time (Note,4) twL tWH fMAX Clock Width Maximum Frequency (Note 5) ns 20 12 ns ns 12 ns LOW 12 HIGH 12 ns ns External Feedback 11/(ts + teo) 37 MHz No Feedback \1/(tWH + tWL) 41.6 MHz tpzx OE to Output Enable 15 ns tpxz OE to Output Disable 15 ns tEA Input to Output Enable Using Product Term Control 20 ns tER Input to Output Disable Using Product Term Control 20 ns Notes: 2. See Switching Test Circuit for test conditions. 3. Output delay minimums are measured under best-case conditions. 4. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. 5. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where the frequency may be affected. PALCE20RA10H-20 (Com'l, INO) 2-243 ~AMD SWITCHING WAVEFORMS Input or Feedback Combinatorial Output --~-vr154340-8 Input or Feedback _ _ _--' Vr Clock Registered Output _ _ _ _ _ _ _ __ 154340-9 Registered Output Combinatorial Output Input Asserting --~I_----Asynchronous Preset - - - - ' '-----.---'1 ' - - - - . . I,..--- Registered -4_--- Output ----~~~~.... Input Asserting Asynchronous Reset Registered Output _ _ _"--11;....£."""",,-...1 Clock tARR 154340-10 Vr Clock 154340-11 Asynchronous Preset Asynchronous Reset Clock 154340-12 Clock Width OE Input or Feedback tER Vr Output tpxz tEA Vr Output 154340-14 154340-13 Input to Output Disable/Enable Notes: 1. VT=1.5V 2. Input pulse amplitude 0 V to 3.0 V 3. Input rise and fall times 2 ns - 5 ns typical. 2·244 tpzx PALCE20RA1OH·20 OE to Output Disable/Enable AMD~ KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L ///// May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State WAVEFORM xxxxxx KSOOOO10-PAL SWITCHING TEST CIRCUIT 5V Output o--+----....-~) Test Point 154340-15 Commercial Specification Sl CL tPD, teo Closed tpzx, tEA Z~ Z~ tpxz, tER H: Open L: Closed H ~Z: Open R1 R2 Measured Output Value 1.5 V 50 pF 1.5V 560n 1.1 kn H ~Z: VOH -0.5 V 5 pF L ~ Z: VOL + 0.5 V L ~Z: Closed PALCE20RA 1OH·20 2·245 ~ AMD ENDURANCE CHARACTERISTICS The PALCE20RA10 is manufactured using AMD's advanced Electrically Erasable process. This technology uses an EE cell to replace the fuse link used in bipolar Symbol tOR N parts. As a result, the device can be erased and reprogrammed-a feature which allows 100% testing at the factory. Parameter Test Conditions Min Pattern Data Retention Time Min Reprogramming Cycles Min Unit Max Storage Temperature 10 Years Max Operating Temperature 20 Years Normal Programming Conditions 100 Cycles Robustness The PALCE20RA1OH-20 has been designed with some unique features that make it extremely robust, even when operating in high-speed design environments. Pull-up resistors on the inputs and I/0s cause unconnected pins to default to the HIGH state. Input-clamping circuitry limits negative overshoot, eliminating the possibility of false clocking caused by subsequent ringing. A special noise filter makes the programming circuitry completely insensitive to any positive overshoot that has a pulse width of less than about 100 ns. INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee Vee >50kn I I I I ESD Protection and Clamping I II Programmlng . -= L !2n,! £n.!t ____ _ Typical Input Vee Vee >50kn Provides ESD Protection and Clamping Preload Circuitry Typical Output 2·246 PALCE20RA1 OH·20 Feedback Input 15434D-17 AMD~ POWER-UP RESET The PALCE20RA1 0 has been designed with the capability to reset during system power-up. Following powerup, all flip-flops will be reset to LOW. The output state will be HIGH independent of the logic polarity. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset Parameter Symbol tPR and the wide range of ways Vec can rise to its steady state, two conditions are required to ensure a valid power-up reset. These conditions are: • The Vee rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Descriptions Max Unit Power-Up Reset Time 1000 ns Input or Feedback Setup Time ts See Switching Characteristics Clock Width LOW tWL Vee Power Registered Output Clock 154340-18 Power-Up Reset Waveform PALCE20RA1OH-20 2-247 ~ AMD 2-248 - ~ COM'L Advanced Micro Devices AmPAL22P10B/AL/A 24-Pin Combinatorial TTL Programmable Array Logic DISTINCTIVE CHARACTERISTICS • As fast as 15 ns maximum propagation delay • Universal combinatorial architecture • Programmable output polarity • Programmable replacement for high-speed TIL logic • Extensive third-party software and programmer support through FusionPLD partners • 24-pln SKINNYDIP and 28-pln PLCC packages save space GENERAL DESCRIPTION The AmPAL22P10 utilizes Advanced Micro Devices' advanced oxide-isolated bipolar process and fuse-link technology. The devices provide user-programmable logic for replacing conventional SSIIMSI gates and f1ipflops at a reduced chip count. The AmPAL22P1 0 allows the systems engineer to implement the design on-chip, by opening fuse links to configure AND and OR gates within the device, according to the desired logic function. Complex intercorinec- , tions between gates, which previously required time-consuming layout, are lifted from the PC board and placed on silicon, where they can be easily modified during prototyping or production. The PAL device implements the familiar Boolean logic transfer function, the sum of products. The PAL device is a programmable AND array driving a fixed OR array. The AND array is programmed to create custom product terms, while the OR array sums selected terms at the BLOCK DIAGRAM I/O 110 outputs. In addition, the PAL device provides the following options: - Variable inpuVoutput pin ratio - Programmable three-state outputs Product terms with all fuses opened assume the logical HIGH state; product terms connected to both true and complement of any single input assume the logical LOW state. Unused input pins should be tied to Vcc or GND. The entire PAL device family is supported by the FusionPLD partners. The PAL family is programmed on conventional PAL device programmers with appropriate personality and socket adapter modules. See the Programmer Reference Guide for approved programmers. Once the PAL device is programmed and verified an additional fuse may be opened to prevent pattern readout. This feature secures proprietary circuits. AmPAL22P10 Inputs 110 110 110 110 110 110 110 110 12984E-1 Publication# 12984 Rev. E Issue Date: January 1992 AmendmentlO 2-249 ~ AMD PRODUCT SELECTOR GUIDE tPD ns (Max) Family Icc mA (Max) IOL mA(Mln) Very High Speed (liB") Versions 15 210 24 High Speed ("A") Versions 25 210 24 High Speed, Half Power ("AL") Versions 25 105 24 CONNECTION DIAGRAMS Top View PLCC SKINNYDIP () 0 z ~ ~ ~ Vee 110 110 110 110 25 110 24 110 23 110 22 NC 110 21 110 110 20 110 110 NC 110 110 110 110 0 12984E-3 12984E-2 Note: Pin 1 is marked for orientation PIN DESIGNATIONS 2·250 ~ ~ (!) GND GND I I/O NC Vee () z z Ground Input InpuVOutput No Connect Supply Voltage AmPAL22P10B/AUA AMD~ ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: AmPAL 22 P 10 A L P C ____T---' -- -- -- -_ - r FAMILY TYPE Am PAL = Programmable Array Logic L OPTIONAL PROCESSING Blank = Standard Processing NUMBER OF ARRAY INPUTS OUTPUT TYPE - - - - - - - -...... P = Programmable Polarity OPERATING CONDITIONS C = Commercial (OOC to +75°C) NUMBER OF OUTPUTS - - - - - - - - - ' SPEED B = 15 ns tPD A = 25 ns tPD POWER _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ PACKAGE TYPE P = 24-Pin 300 mil Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) D = 24-Pin 300 mil Ceramic SKINNYDIP (CD3024) ~ L = Low Power (105 rnA Icc) Blank = Full Power (210 rnA Icc) Valid Combinations AmPAL22P10 Valid Combinations lists configurations planned to be supported in volume forthis device. Consult the local AMD sales office to confirm availability of specific valid combinations, and to check on newly released combinations. Note: Marked with AMD logo. AmPAL22P10B/ALlA (Com'l) 2-251 ~ AMD FUNCTIONAL DESCRIPTION All parts are produced with a fuse link at each input to the AND gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Utilizing an easily-implemented programming algorithm, these products can be rapidly programmed to any customized pattern. Information on approved programmers can be found in the Programmer Reference Guide. Extra test words are pre-programmed during manufacturing to ensure extremely high field programming yields, and provide extra test paths to achieve excellent parametric correlation. Variable Input/Output Pin Ratio The AmPAL22P10 has twelve dedicated input lines, and all ten combinatorial outputs are 1/0 pins. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vee or GND. Programmable Three-State Outputs Each output has a three-state output buffer with threestate control. A product term controls the buffer, allowing enable and disable to be a function of any product of device inputs or output feedback. The combinatorial output provides a bidirectional 1/0 pin, and may be cO.nfigured as a dedicated input if the buffer is always dISabled. Programmable Polarity The polarity of each output can be active-high or activelow, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean 2·252 expressions to be written in their most compact form (true or inverted). and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is through a programmable fuse which controls an exclusive-OR gate at the output of the AND/OR logiC. The output is active high if the fuse is 1 (programmed) and active low if the fuse is 0 (intact). Security Fuse After programming and verification, an AmPAL22P10 design can be secured by programming the security fuse. Once programmed, this fuse defeats readback of the internal programmed pattern by a device programmer, securing proprietary designs from competitor~. When the security fuse is programmed, the array Will read as if every fuse is programmed. Quality and Testability The AmPAL22P10 offers a very high level of built-in quality. Extra programmable fuses provide a means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Technology The AmPAL22P10 is fabricated with AMD's advanced oxide-isolated bipolar process. This process reduces parasitic capacitances and minimum geometries to provide higher performance. The array connections are formed with proven PtSi fuses for reliable operation. AmPAL22P10B/AUA AMD~ LOGIC DIAGRAM AmPAL22P10 Inputs (0-43) • 1 II .. i. J • ".,1 12131'415 "'71'" fl""" Mann "n3D!1 3Ill!Il436 ··, ···· H!7:tI31 40"410 7 0--tf= ,. 17 ,. -, ~J)l ., ... .... .... ~ II ~. 17 II 7 ....".. :! .. .... .. .. .. ..., " i '=:J ..........l 47 1 :Bn ~-/ ... r . 21 20 19 j ~D~[ 51 7 22 ~ 17 6 .. I 1 ;~D~[ " ...."" 5 I ~D~l ,. ,. 4 2 1 ,. f3l. ......... rill -~D1... 18 ~ 54 II i7 "\ :BTJ-l .,,-/ ... r II II ~ 17 8 " 1 Ii II 17 II II 10 71 9 ......, ).:»))"[ 16 ~ n ,." 74 :B1)1 ~ .. ..., 7t 10 U 15 r ~ :: -::2 Ii II ., :: 11 r-/'" ~ > 1 ~D"[ 14 13 12984E-4 AmPAL22P10B/AUA 2-253 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Commercial (C) Devices Ambient Temperature With Power Applied ............. -55°C to + 125°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Ambient Temperature (TA) Operating in Free Air ........... , O°C to +75°C Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . . . . . .. -0.5 V to +5.5 V DC Input Current ............ " -30 rnA to +5 rnA Operating ranges define those limits between which the func- ' tionality of the device is guaranteed. DC I/O Pin Voltage ............ -0.5 V to Vee Max Static Discharge Voltage ................ 2001 V Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH = -3.2 mA VIN = VIH or VIL Vcc = Min Min VOL Output LOW Voltage IOL = 24 mA VIN = VIH or VIL Vcc = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) Max 2.0 Unit V 2.4 0.5 V 5.5 V 0.8 V VI Input Clamp Voltage IIN= -18 mA, Vee = Min -1.2 V IIH Input HIGH Current VIN = 2.7 V, Vee = Max (Note 2) 25 ~ IlL Input LOW Current VIN = 0.4 V, Vee = Max (Note 2) -100 ~ Ii Maximum Input Current VIN = 5.5 V, Vee = Max 1 mA loZH Off-State Output Leakage Current HIGH VOUT= 2.7 V, Vee = Max VIN = VIH or VIL (Note 2) 100 ~ loZL Off-State Output Leakage Current LOW VOUT = 0.4 V, Vcc = Max VIN = VIH or VIL (Note 2) -100 ~ Isc Output Short-Circuit Current VOUT = 0.5 V, Vcc = Max (Note 3) -90 mA Icc Supply Current VIN = 0 V, Outputs Open (loUT = 0 rnA) Vee = Max B,A 210 rnA AL 105 rnA -30 Notes: 1. These are absolute values with respect to device ground and all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of IiL and IOZL (or IiH and IOZH). 3. Not more than one output should be shorted at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-254 AmPAL22P10B/ALlA (Com'l) AMD 11 CAPACITANCE (Note 1) Parameter Symbol CIN Parameter Description Input Capacitance I I COUT Test Conditions Pins 1 13 Others Output Capacitance VIN = 2.0 V Typ Vee = 5.0 V TA VOUT = 2.0 V Unit 11 6 = +25°C pF 9 f = 1 MHz Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol A,AL B Parameter Description Min Max Min Max Unit tPD Input or Feedback to Combinatorial Output 15 25 ns tEA Input to Output Enable Using Product Term Control 18 25 ns tER Input to Output Disable Using Product Term Control 15 25 ns Note: 2. See Switching Test Circuit for test conditions. SWITCHING WAVEFORMS Input or Feedback Combinatorial Output ;g Input or Feedback VT -xxmtP~ _ _ _---I tER -VT tEA Combinatorial >-+-+++-----+~f-t- VT Output ----.....&....f....I.-1 12984E-5 Combinatorial Output 12984E-6 Input to Output Disable/Enable Notes: 1. VT= 1.5 V 2. Input pulse amplitude 0 V to 3.0 V 3. Input rise and fall times.2 ns-5 ns typical. AmPAL22P10B/ALlA (Com'l) 2-255 ~ AMD KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be \\\\\ May Change from H to L Will be 1//71 May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing. State Unknown Does Not Apply Center Line is HighImpedance "Off" State WAVEFORM xxxxxx Steady Changing from H to L KSOOOO10-PAL SWITCHING TEST CIRCUIT 5V Output O--+----......-~l Test Point 12984E-7 Specification S1 tPD Closed tEA Z~ H:Open CL R1 2-256 H ~Z: Open L ~Z: Closed Measured Output Value 1.5 V 50 pF Z ~ L: Closed tER R2 1.5 V 200n 5 pF AmPAL22P1 OBI AU A 390n H ~ Z: VOH - 0.5 V L ~ Z: VOL + 0.5 V AMD~ INPUT/OUTPUT EQUIVALENT SCHEMATICS -----------------.---oVCC Input ProgramNerify Circuitry 12984E-8 Typical Input ---------------~I--------o VCC 40QNOM :>1-......---+---------+__0 Input. I/O Pins Output ProgramNerify/ Test Circuitry 12984E-9 Typical Output AmPAL22P10B/AUA 2·257 _ COM'L:-7/10/15 ~ MIL: -12/20 Advanced Micro Devices PAL22V10 Family, AmPAL22V10/A 24-Pin TTL Versatile PAL Device DISTINCTIVE CHARACTERISTICS • • • As fast as 7.5 ns propagation delay and 91 MHz fMAx (external) 10 Macrocells programmable as registered or combinatorial, and active high or active low to match application needs Varied product term distribution allows up to 16 product terms per output for complex functions • • • • Global asynchronous reset and synchronous preset for Initialization Power-up reset for initialization and register preload for testability Extensive third-party software and programmer support through FusionPLD partners 24-Pin SKINNVDIP, 24-pin Flatpack and 28-pln PLCC and LCC packages save space GENERAL DESCRIPTION The PAL22V10 provides user-programmable logic for replacing conventional SSI/MSI gates and flip-flops at a reduced chip count. . The PAL22V1 0 device implements the familiar Boolean logic transferfunction, the sum of products. The PAL device is a programmable AND array driving a fixed OR array. The AND array is programmed to create custom product terms, while the OR array sums selected terms at the outputs. The product terms are connected to the fixed OR array with a varied distribution from 8 to 16 across the outputs (see Block Diagram). The OR sum of the products feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial, and active high or active low. The output configuration is determined by two fuses controlling two multiplexers in each macrocell. AMD's FusionPLD program allows PAL22V10 designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a designer can use a tool that is already installed and familiar. Please referto the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. BLOCK DIAGRAM 11 - 11, elK/lo Programmable AND Array (44 x 132) 2-258 Publication# 16559 Rev. B Issue Date: June 1993 AmendmentlO AMD~ CONNECTION DIAGRAMS Top View PLCC/LCC SKINNYDIP/FLATPACK CLKllo 11 IIQg 12 1I0a 13 1107 1I0s 1105 1104 14 15 "i Vee ..!'I : () Z o CJ) ° CIO >00 .::.. ::::. 1/07 13 IIOs 14 15 23 1/°5 NC 22 NC 1/03 Is 19 1/02 1/01 110 17 la 1/00 GND ....J () 1/°4 1/03 19 1/°2 12 13 14 15 16 17 18 111 ..2' 165598-2 .:t- oz CD () Z .£0 .::.. g165598-3 Note: Pin 1 is marked for orientation. PIN DESIGNATIONS ClK GND Clock Ground I/O Input Input/Output NC No Connect Vee Supply Voltage PAL22V10 Family 2-259 ~ AMD ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL T FAMILY TYPE 22 V 10 -7 P C PAL or AmPAL = Programmable Array Logic NUMBER OF ARRAY INPUTS OUTPUTTYPE--------------------~ V = Versatile NUMBER OF OUTPUTS ----------~ SPEED ------------------------------~ ~. OPTIONAL PROCESSING Blank = Standard Processing OPERATING CONDITIONS C = Commercial (O°C to +75°C) PACKAGE TYPE P = 24-Pin 300 mil Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) -7 = 7.5 ns tPD -10= 10 ns tPD -15= 15 ns tPD A= 25 ns tPD Valid Combinations PAL22V10-7 PAL22V10-10 PAL22V10-15 PC,JC Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. AmPAL22V10A 2-260 PAL22V10-7/10/15, AmPAL22V10A (Com'l) ~MD~ ORDERING INFORMATION APL Products AMD programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STD-883 requirements. The order number (Valid Combination) is formed by a combination of: PAL 22 V 10 -12 IB L A J FAMILY TYPE PAL or Am PAL = Programmable Array Logic NUMBER OF ARRAY INPUTS ~ OUTPUT TYPE - - - - - - - - - - - - - ' V = Versatile LEAD FINISH A = Hot Solder Dip PACKAGE TYPE L = 24-Pin 300 mil Ceramic SKINNYDIP (CD3024) K = 24-Pin Ceramic Flatpack (CFL024) 3 = 28-Pin Ceramic Leadless Chip Carrier (CL 028) DEVICE CLASS IB = MIL-STD-883C Class B NUMBER OF OUTPUTS SPEED --------------------~ -12 = 12 ns tPD -20 = 20 ns tPD A = 30 ns tPD Blank = 40 ns tPD Valid Combinations PAL22V10-12 PAL22V10-20 AmPAL22V10A IBLA, IBKA, IB3A Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. AmPAL22V10 Group A Tests Group A tests consist of Subgroups 1, 2, 3, 7, 8, 9, 10, 11. Military Burn-In Military burn-in is in accordance with the current revision of MIL-STD-883, Test Method 1015, Conditions A through E. Test conditions are selected at AMD's option. PAL22V10-12/20, AmPAL22V10/A (Mil) 2-261 ~ AMD FUNCTIONAL DESCRIPTION The PAL22V10 allows the systems engineer to implement the design on-chip, by opening fuse links to configure AND and OR gates within the device, according to the desired logic function. Complex interconnections between gates, which previously required timeconsuming layout, are lifted from the PC board and placed on silicon, where they can be easily modified during prototyping or production. Product terms with all fuses opened assume the logical HIGH state; product terms connected to both true and complement of any single input assume the logical LOW state. The PAL22V10 has 12 inputs and 10 I/O macrocells. The macrocell (Figure 1) allows one of four potential output configurations; registered output or combinatorial I/O, active high or active low (see Figure 2). The configuration choice is made according to the user's design specification and corresponding programming of the configuration bits So - S1. Multiplexer controls initially are connected to ground (0) through a programmable fuse, selecting the "0" path through the multiplexer. Programming the fuse disconnects the control line from GND and it is driven to a high level, selecting the "1" path. The device is produced with a fuse link at each input to the AN D gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Variable Input/Output Pin Ratio The PAL22V10 has twelve dedicated input lines, and each macrocell output can be an I/O pin. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vee or GND. I/On -=- 51 50 0 0 1 1 0 1 0 1 Output Configuration Registered/Active Low Re istered/Active Hi h Combinatorial/Active Low Combinatorial/Active Hi h o =Unprogrammed fuse 1 = Programmed fuse 166598-4 Figure 1. Output Logic Macrocell Diagram 2·262 PAL22V10 Family AMD~ Registered Output Configuration Combinatorial I/O Configuration a Each macrocell of the PAL22V1 includes a D-type flipflop for data storage and synchronization. The flip-flop is loaded on the LOW-to-HIGH transition of the clock input. In the registered configuration (S1 = a), the array feedback is from of the flip-flop. a Any macrocell can be configured as combinatorial by selecting the multiplexer path that bypasses the flip-flop (S1 =1). In the combinatorial configuration the feedback is from the pin. Combinatorial/Active Low Registered/Active Low Combinatorial/Active High Registered/Active High 165598-5 Figure 2. Macrocell Configuration Options Programmable Three-State Outputs Preset/Reset Each output has a three-state output buffer with threestate control. A product term controls the buffer, allowing enable and disable to be a function of any product of device inputs or output feedback. The combinatorial output provides a bidirectional 110 pin, and may be configured as a dedicated input if the buffer is always disabled. For initialization, the PAL22V10 has Preset and Reset product terms. These terms are connected to all registered outputs. When the Synchronous Preset (SP) product term is asserted high, the output registers will be loaded with a HIGH on the next LOW-to-HIGH clock transition. When the Asynchronous Reset (AR) product term is asserted high, the output registers will be immediately loaded with a LOW independent of the clock. Programmable Output Polarity Note that preset and reset control the flip-flop, not the output pin. The output level is determined by the output polarity selected. The polarity of each macrocell output can be active high or active low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is controlled by programmable bit So in the output macrocell, and affects both registered and combinatorial outputs. Selection is automatic, based on the design speCification and pin definitions. Power-Up Reset All flip-flops power-up to a logic LOW for predictable system initialization. Outputs of the PAL22V10 will depend on the programmed output polarity. The Vee rise must be monotonic and the reset delay time is 1000 ns maximum. .. PAL22V10 Family 2-263 ~ AMO' Register Preload Quality and Testability The register on the PAL22V10 can be preloaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. The PAL22V1 ooffers a very high level of built-in quality. Extra programmable fuses, test words and test columns provide a means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Security Fuse After programming and verification, a PAL22V1 0 design can be secured by programming the security fuse. Once programmed, this fuse defeats readback of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. When the security fuse is programmed, the array will read as if every fuse is programmed, and preload will be disabled. Programming Technology The PAL22V10A is fabricated with AMD's Junctionisolated process. The array connections are formed with highly reliable PtSifuse. The PAL22V10-15, -10 and -7 are fabricated with AMD's advanced oxide-isolated bipolar process. This process reduces parasitic capacitances and minimum geometries to provide higher performance. The array connections are formed with PtSi fuses on the -15, and TiW fuses on the -7 and -10 for reliable operation. The PAL22V10 can be programmed on standard logiC programmers. Approved programmers are listed in the Programmer Reference Guide. 2·264 PAL22V10 Family AMD~ LOGIC DIAGRAM SKINNYDIP (PLCC/LCC) Pinouts IE (28) Vee 1111111 33 34 48 49 1111111111111111111111111 I I LLIIII II II I II I II II II II II II I I II I III II IILlI I 1I III II II IIII I II I II III IIII III l_li I I I I I I I I I I II I I I I I I I I I I I I I I I I I I I I II I I I I I I I I r- ~~l I !V05 r- ~~ 1 I ~ro. t1=I=F:f::t::::t+:f::t::::t+:t:t=t:t:f::t::::t++t=+++t==f;~rtt--i"rh \~ l!J 1 85 I IIIII IIII IIII ITIT IIII IIII IIII IIII IIII IIII 88- ~ '5 ~m(~-~83Ii=t:I+:j 11~1=t1:t:J11~1~1~1:t:tII~I~I~II;::~II+III=~ 1,t1 1';::~I'+IIt=~'I~II+=+II*=*'+I:::t:1I~,~,=I,:::t:II*=*,::::.ljI---ttt------t~ 18 0w-~97R11+"J::t4I+1II=t+II~+It+I=t":pII:t:I::t:I:t:tI:t:I::ul t I rIII 1111 III I II 7 I 981.1II1II.1IIf=.r-MAA~~'~I I o SP 110 ' L_i:l r~o ~ \~ 11::t:t:t=t:***1~1l1I IIII III -tR:t=t:F+Ff:t=t:t::ij::::j:::t:1t:t=lm 111:t=t (180}-) ~ : : 0 J.. ,. :::1,1"1"1,,1,1,,,1,,,1,,1,"1",I'1111ti=t:lI1~J,Ittt:J:,,,~:r-!~~,~:~Llt[p~_~1 I 122 II II IIII II :E VO, (18) 1.!J I I I I I I I I I I I I I I I I I I I I I TIT I I I I I I I I r--~~l 130 I :E!VO o (17) 1 131-+H+-tt-t~~-H++-t+t+-+t-t+-++-t-t--ir+H-H++-++++-++++_ _~SIP .A r---' o 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 32 35 36 39 40 43 16559B-6 PAL22V10 Family 2-265 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied . . . . . . . . . . . . . . . . . -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air ............ DoC to +75°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . .. -1.2 V to Vee + 0.5 V DC Output or I/O Pin Voltage . -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliabifity. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 mA VIN = VIH or Vil Vee = Min VOL Output lOW Voltage IOl = 16 rnA VIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) VI Input Clamp Voltage liN = -18 mA, Vee = Min IIH Input HIGH Current VIN = 2.7 V, Vee = Max (Note 2) III Input LOW Current VIN = 0.4 V, Vee = Max Min (Note 2) Ii Max 2.4 V 0.5 2.0 V V 0.8 -1.2 25 I Input I ClK Unit -100 -150 V V ~ J.lA Maximum Input Current VIN = 5.5 V, Vee = Max 1 mA IOZH Off-State Output Leakage Current HIGH VOUT = 2.7 V, Vee = Max VIN = VIH or VIL (Note 2) 100 ~ IOZl Off-State Output Leakage Current LOW VOUT = 0.4 V, Vee = Max VIN = VIH or Vil (Note 2) -100 ~ Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) -130 mA Icc Supply Current VIN = 0 V, Outputs Open (lOUT = 0 mA) Vee = Max 220 mA -30 Notes: 1. These are absolute values with respect to device ground and all overshoots due to system andlor tester noise are included. 2. 110 pin leakage is the worst case of liL and lozL (or I/H and IOZH ). 3. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT has been chosen to avoid test problems caused by tester ground degradation. 2-266 PAL22V10-7 (Com'l) =0.5 V AMD~ CAPACITANCE (Note 1) Parameter Symbol Parameter Description Test Conditions CIN Input Capacitance VIN =2.0 V CoUT Output Capacitance VOUT= 2.0 V Typ I I Vee =5.0 V TA = 25°C f = 1 MHz Unit 6 5 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol Parameter Description Min (Note 3) Max 7.5 tpo Input or Feedback to Combinatorial Output 1 ts Setup Time from Input, Feedback or SP to Clock 5 Hold Time 0 Clock to Output 1 tH teo tSKEWR tAR Skew Between Registered Outputs (Note 5) Asynchronous Reset to Registered Output Unit ns ns ns 6 ns 1 ns 12 8 ns tARW Asynchronous Reset Width tARR Asynchronous Reset Recovery Time 8 ns tSPR Synchronous Preset Recovery Time 5 ns LOW 4 ns HIGH 4 ns 91 MHz twL Clock Width tWH fMAX Maximum Frequency (Note 4) External Feedback I 1/(ts + teo) Internal Feedback (feNT) No Feedback I 1/(tWH + tWL) ns 100 MHz 125 MHz tEA Input to Output Enable Using Product Term Control 8 ns tER Input to Output Disable Using Product Term Control 7.5 ns Notes: 2. See Switching Test Circuit for test conditions. 3. Output delay minimums are measured under best-case conditions. 4. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. 5. Skew is measured with all outputs switching in the same direction. PAL22V10-7 (Com'l) 2-267 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied ................. -55°C to +125°C Ambient Temperature (TA) Operating in Free Air ............ DoC to +75°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage ......... " -1.2 V to Vee + 0.5 V DC Output or I/O Pin Voltage ............... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 rnA VIN = VIH or Vil Vee = Min VOL Output lOW Voltage IOl=16mA VIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Guaranteed Input logical HIGH Voltage for all Inputs (Note 1) Vil Input lOW Voltage Guaranteed Input logical lOW Voltage for all Inputs (Note 1) Min VI Input Clamp Voltage liN = -18 rnA, Vee = Min IIH Input HIGH Current VIN = 2.7 V, Vee = Max (Note 2) III Input lOW Current II Max 2.4 V 0.5 2.0 V 25 ~ VIN = 0.4 V, Vee = Max I Input -100 IClK -150 VIN = 5.5 V, Vee = Max IOZH IOZl Off-State Output leakage Current LOW Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) Icc Supply Current VIN = 0 V, Outputs Open (lOUT = 0 rnA) Vee = Max V -1.2 (Note 2) Maximum Input Current V V 0.8 Off-State Output leakage Current HIGH Unit JlA 1 rnA VOUT = 2.7 V, Vee = Max VIN = VIH or Vil (Note 2) 100 ~ VOUT = 0.4 V, Vee = Max VIN = VIH or Vil (Note 2) -100 ~ -30 -130 rnA 180 rnA Notes: 1. These are absolute values with respect to device ground and all overshoots due to system andlor tester noise are included. 2. 110 pin leakage is the worst case of IJL and lozL (or hH and IOZH ). 3. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-268 PAL22V10-10 (Com'I)' AMD~ CAPACITANCE (Note 1) Parameter Symbol Parameter Description Test Conditions CIN Input Capacitance VIN =2.0 V Cour Output Capacitance VOUT=2.0 V Note: Typ I I Vcc =5.0 V TA = 25°C f = 1 MHz Unit 6 5 pF 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol Parameter Description Min (Note 3) Max 10 Unit tPD Input or Feedback to Combinatorial Output 1 ts Setup Time from Input, Feedback or SP to Clock 7 ns tH Hold Time 0 ns tco Clock to Output 1 tAR Asynchronous Reset to Registered Output 7 15 ns ns ns tARw Asynchronous Reset Width 10 ns tARR Asynchronous Reset Recovery Time 8 ns tSPR Synchronous Preset Recovery Time tWL 8 ns LOW 5 ns HIGH 5 ns 71 MHz 80 MHz Clock Width tWH fMAX Maximum Frequency (Note 4) External Feedback I 1/(ts + tco) Internal Feedback (fCNT) No Feedback I 1/(tWH + tWL) 100 MHz tEA Input to Output Enable Using Product Term Control 11 ns tER Input to Output Disable Using Product Term Control 9 ns Notes: 2. See Switching Test Circuit for test conditions. 3. Output delay minimums are measured under best-case conditions. 4. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PAL22V10-10 (Com'I) 2-269 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied ................. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air .......... " Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 0.5 V DC Input Current ............. -30 rnA to +5 rnA O°C to +75°C Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or 1/0 Pin Voltage ............... -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 mA VIN = VIH or Vil Vee = Min VOL Output LOW Voltage IOl = 16 mA VIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) VI Input Clamp Voltage liN = -18 mA, Vee = Min hH Input HIGH Current III Input LOW Current II Maximum Input Current VIN = 5.5 V, Vee = Max IOZH Off-State Output Leakage Current HIGH IOZl Min Max Unit V 2.4 0.5 V V 2.0 0.8 V -1.2 V VIN = 2.7 V, Vee = Max (Note 2) 25 ~ VIN = 0.4 V, Vee = Max (Note 2) -100 ~ 1 mA VOUT = 2.7 V, Vee = Max VIN = VIH or Vil (Note 2) 100 ~ Off-State Output Leakage Current LOW VOUT = 0.4 V, Vee = Max VIN = VIH or Vil (Note 2) -100 ~ Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) -130 mA Icc Supply Current VIN = 0 V, Outputs Open (lOUT = 0 mA) Vee = Max 180 mA -30 Notes: 1. These are absolute values with respect to device ground and all overshoots due to system andlor tester noise are included. 2. /10 pin leakage is the worst case of hL and IOZL (or hH and IOZH ). 3. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-270 PAL22V10-15 (Com'I) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Description Test Conditions Input Capacitance VIN =2.0 V Output Capacitance VOUT=2.0 V Typ Unit 9 6 5 pF Vee =5.0 V TA = 25°C f = 1 MHz Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol Min (Note 3) Parameter Description Max Unit 15 ns tPD Input or Feedback to Combinatorial Output ts Setup Time from Input, Feedback or SP to Clock 10 ns tH Hold Time 0 ns teo Clock to Output 10 tAR Asynchronous Reset to Registered Output 20 ns ns tARW Asynchronous Reset Width 15 tARR Asynchronous Reset Recovery Time 10 ns tSPR Synchronous Preset Recovery Time 10 ns tWL Clock Width tWH fMAX Maximum Frequency (Note 4) ns LOW 6 ns HIGH 6 ns 50 MHz External Feedback I 1/(ts + teo) Internal Feedback (teNT) No Feedback I 1/(tWH + tWL) 80 MHz 83 MHz tEA Input to Output Enable Using Product Term Control 15 ns tER Input to Output Disable Using Product Term Control 15 ns Notes: 2. See Switching Test Circuit for test conditions. 3. Output delay minimums are measured under best-case conditions. 4. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PAL22V10·15 (Com'l) 2·271 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied ................. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . .. O°C to +75°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage ............... -0.5 V to +5.5 V DC Input Current .............. -30 rnA to +5 rnA Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or 1/0 Pin Voltage ... -0.5 V to Vee Max Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol VOH Parameter Description Test Conditions Output HIGH Voltage IOH m-3.2 mA Min VIN .. VIH or VIL Vee = 16 mA VIN = VIH or VIL Vee = Min Output LOW Voltage IOL VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) VI Input Clamp Voltage liN = -18 mA, Vee hH Input HIGH Current = 2.7 V, Vee = Max (Note 2) VIN = 0.4 V, Vee .. Max (Note 2) VIN = 5.5 V, Vee = Max VOUT = 2.7 V, Vee = Max Input LOW Current II Maximum Input Current IOZH IOZl Off-State Output Leakage Current HIGH Off-State Output Leakage Current LOW Unit V = Min VOL ilL Max 2.4 0.5 2.0 V 0.8 = Min VIN = VIH or VIL (Note 2) VOUT = 0.4 V, Vee = Max V V -1.2 V 25 ~ -100 ~ 1 rnA 100 ~ -100 ~ -90 rnA 180 rnA VIN VIN = VIH or Vil (Note 2) =0.5 V, Vee = Max (Note 3) Isc Output Short-Circuit Current VOUT Icc Supply Current VIN = 0 V, Outputs Open (lOUT = 0 mA) Vee = Max -30 Notes: 1. These are absolute values with respect to device ground and all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of IiL and IOZL (or IIH and IOZH ). 3. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT has been chosen to avoid test problems caused by tester ground degradation. 2-272 AmPAL22V10A (Com'I) =0.5 V AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Description Test Conditions Input Capacitance VIN =2.0 V Output Capacitance VOUT=2.0 V Typ Unit 11 6 pF Vee =5.0 V TA = 25°C f = 1 MHz 9 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol Parameter Description Min tPD Input or Feedback to Combinatorial Output ts Setup Time from Input, Feedback or SP to Clock tH Hold Time Max 25 Unit ns 20 ns 0 ns teo Clock to Output 15 ns tAR Asynchronous Reset to Registered Output 30 ns Asynchronous Reset Width 25 ns tARR Asynchronous Reset Recovery Time 35 ns tSPR Synchronous Preset Recovery Time 20 ns LOW 15 ns HIGH 15 ns 28.5 MHz tARW twL Clock Width tWH fMAX Maximum Frequency (Note 3) External Feedback 1/(ts + teo) tEA Input to Output Enable Using Product Term Control 25 ns tER Input to Output Disable Using Product Term Control 25 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. AmPAL22V10A (Com'I) 2-273 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Military (M) Devices (Note 1) . Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V DC Input Voltage ............... -1.2 V to +7.0 V DC Output or 1/0 Pin Voltage ..... -0.5 V to +7.0 V Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Absolute Maximum Ratings are for system design reference; parameters given are not tested. Ambient Temperature (TA) Operating in Free Air ............ " -55°C Min Operating Case (Te) Temperature .................. " 125°C Max Supply Voltage (Vee) with Respect to Ground ..... +4.50 V to +5.50 V Note: 1. Military products are tested at Te and -5SOC, per MIL-STD-883. =+2SOC, +12SOC, Operating ranges define those limits between which the functionality of the device is guaranteed. DC CHARACTERISTICS over MILITARY operating ranges unless otherwise specified (Note 2) Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-2 mA VIN = VIH or Vil Vee = Min Min VOL Output LOW Voltage IOl = 12 mA VIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 3) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 3) VI Input Clamp Voltage liN = -18 mA, Vee = Min IIH Input HIGH Current III Input LOW Current = 2.7 V, Vee = Max (Note 4) VIN = 0.4 V, Vee = Max (Note 4) Max 0.5 0.8 1.lnput lCLK = 5.5 V, Vee = Max V V 2.0 VIN Unit V 2.4 V -1.2 V 25 ~ -100 -150 ~A Maximum Input Current VIN 1 mA IOZH Off-State Output Leakage Current HIGH VOUT = 2.7 V, Vee = Max VIN = VIH or Vil (Note 4) 100 ~ IOZl Off-State Output Leakage Current LOW VOUT = 0.4 V, Vee = Max VIN = VIH or Vil (Note 4) -100 ~ Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 5) -130 mA lee Supply Current VIN = 0 V, Outputs Open (lOUT = 0 mA) Vee = Max 200 mA II -30 Notes: 2. For APL Products, Group A, Subgroups 1,2, and 3 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. ViL and V/H are input conditions of output tests and are not themselves directly tested. V/L and V/H are absolute voltages with respect to device ground and include all overshoots due to system and/or tester noise. Do not attempt to test these values without suitable equipment. 4. 110 pin leakage is the worst case of IlL and IOZL (or I/H and IOZH ). 5. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-274 PAL22V10-12 (Mil) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN '" 2.0 V Output Capacitance VOUT ... 2.0 V Typ I I Vee = 5.0 V TA" 25°C f = 1 MHz Unit 6 pF 8 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over MILITARY operating ranges (Note 2) Parameter Symbol tPD ts tH Min Parameter Description Unit 12 Input or Feedback to Combinatorial Output Setup Time from Input or Feedback to Clock 10 Hold Time 0 ns ns ns 10 teo Clock to Output tAR Asynchronous Reset to Registered Output tARW Max ns 20 Asynchronous Reset Width (Note 3) ns 15 ns tARR Asynchronous Reset Recovery Time (Note 3) 10 ns tSPR Synchronous Preset Recovery Time (Note 4) 10 ns LOW 6 ns HIGH 6 ns 50 MHz 58.8 MHz tWL tWH fMAX Clock Width Maximum Frequency (Note 4) External Feedback I 1/(ts + teo) Internal Feedback (feNT) No Feedback I 1/(twH + tWL) 83.3 MHz tEA Input to Output Enable Using Product Term Control (Note 4) 15 ns tER Input to Output Disable Using Product Term Control (Note 4) 12.5 ns Notes: 2. See Switching Test Circuit for test conditions. For APL products Group A, Subgroups 9, 10, and 11 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. tARW and tARR are not directly tested, but are guaranteed by the testing of ts and tAR. 4. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. PAL22V10-12 (Mil) 2-275 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Military (M) Devices (Note 1) Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . . . .. -55°C Min DC Input Voltage ............... -0.5 V to +5.5 V Operating Case (Te) Temperature .............. ; . . . .. 125°C Max DC Output or liO Pin Voltage ..... -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ..... +4.50 V to +5.50 V DC Input Current . . . . . . . . . . . . .. -30 rnA to +5 mA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Absolute Maximum Ratings are for system design reference; parameters given are not tested. Note: 1. Military products are tested at Tc and -55"C, per MIL-STD-883. = +25" C, + 125" C, Operating ranges define those limits between which the functionality of the device is guaranteed. DC CHARACTERISTICS over MILITARY operating ranges unless otherwise specified (Note 2) Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-2 mA VIN = VIH or VIL Vee = Min Min Max VOL Output LOW Voltage IOL=12mA VIN = VIH or VIL Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 3) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 3) VI Input Clamp Voltage liN = -18 mA, Vee = Min -1.2 V hH Input HIGH Current VIN = 2.7 V, Vee = Max (Note 4) 25 ~ ilL Input LOW Current VIN = 0.4 V, Vee = Max (Note 4) -100 ~ II Maximum Input Current VIN 0= 5.5 V, Vee = Max 1 mA IOZH Off-State Output Leakage Current HIGH VOUT = 2.7 V, Vee = Max VIN = VIH or Vil (Note 4) 100 ~ IOZl Off-State Output Leakage Current LOW VOUT = 0.4 V, Vee = Max VIN = VIH or VIL (Note 4) -100 ~ Isc Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 5) -90 mA Icc Supply Current VIN = 0 V, Outputs Open (lOUT = 0 mA) Vee = Max 200 mA 2.4 V 0.5 2.0 V V 0.8 -30 Unit V Notes: 2. For APL Products, Group A, Subgroups 1,2, and 3 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. \ilL and V/H are input conditions of output tests and are not themselves directly tested. V/L and V/H are absolute voltages with respect to device ground and include all overshoots due to system and/or tester noise. Do not attempt to test these values without suitable equipment. 4. 110 pin leakage is the worst case of IlL and IOZL (or ItH and lozH ). 5. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-276 PAL22V10-20 (Mil) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN =2.0 V Output Capacitance VOUT = 2.0 V Typ Unit 9 Vee = 5.0 V TA = 25°C f = 1 MHz pF 6 9 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over MILITARV operating ranges (Note 2) Parameter Symbol tPD Min Parameter Description Input or Feedback to Combinatorial Output Max Unit 20 ns ts Setup Time from Input or Feedback to Clock 17 ns tH Hold Time 0 ns 15 teo Clock to Output tAR Asynchronous Reset to Registered Output tARW ns 25 ns 20 Asynchronous Reset Width (Note 3) ns tARR Asynchrono~s Reset Recovery Time (Note 3) 20 ns tSPR Synchronous Preset Recovery Time (Note 4) 20 ns tWL tWH fMAX Clock Width Maximum Frequency (Note 4) LOW 15 ns HIGH 15 ns 31.2 MHz External Feedback I 1/(ts + teo) Internal Feedback (feNT) 33.3 MHz tEA Input to Output Enable Using Product Term Control (Note 4) 20 ns tER Input to Output Disable Using Product Term Control (Note ~) 20 ns Notes: 2. See Switching Test Circuit for test conditions. For APL products Group A, Subgroups 9, 10, and 11 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. tARW and tARR are not directly tested, but are guaranteed by the testing of ts and tAR. 4. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. PAL22V10-20 (Mil) 2-277 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Military (M) Devices (Note 1) Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . . . .. -55°C Min DC Input Voltage ............... -0.5 V to +5.5 V Operating Case (Te) Temperature . . . . . . . . . . . . . . . . . . .. 125°C Max DC Output or I/O Pin Voltage ... -0.5 V to Vee Max Supply Voltage (Vee) with Respect to Ground ..... "+4.50 V to +5.50 V DC Input Current . . . . . . . . . . . . .. -30 rnA to +5 rnA Output Sink Current ............ 100 rnA (Note 6) Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Absolute Maximum Ratings are for system design reference; parameters given are not tested. Note: 1. Military products are tested at Tc and -5S'C, per MIL-STD-883. =+2S'C, + 12S'C, Operating ranges define those limits between which the functionality of the device is guaranteed. DC CHARACTERISTICS over MILITARY operating ranges unless otherwise specified (Note 2) Parameter Symbol Min Max Parameter Description Test Conditions VOl-! Output HIGH Voltage IOH =-2 mA VIN = VIH or Vil Vee = Min VOL Output LOW Voltage IOl=12mA VIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 3) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 3) VI Input Clamp Voltage liN = -18 mA, Vee = Min hH Input HIGH Current VIN = 2.7 V, Vee = Max (Note 4) 25 III Input LOW Current ' VIN = 0.4 V, Vee = Max (Note 4) -100 ~ II Maximum Input Current VIN = 5.5 V. Vee = Max 1 mA IOZH Off-State Output Leakage Current HIGH VOUT = 2.7 V. Vee = Max VIN = VIH or Vil (Note 4) 100 ~ IOZl Off-State Output Leakage Current LOW VOUT = 0.4 V. Vee = Max VIN = VIH or Vil (Note 4) -100 ~ Isc Output Short-Circuit Current VOUT = 0.5 V. Vee = Max (Note 5) Icc Supply Current VIN = 0 V. Outputs Open (lOUT = 0 mA) Vee = Max 2.4 V 0.5 2.0 V V 0.8 ' -1.2 -30 Unit V V ~ -90 mA 180 mA Notes: 2. For APL Products, Group A, Subgroups 1,2, and 3 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. VtL and VtH are input conditions of output tests and are not themselves directly tested. VtL and litH are absolute voltages with respect to device ground and include all overshoots due to system and/or tester noise. Do not attempt to test these values without suitable equipment. 4. I/O pin leakage is the worst ca 78 of liL and lozL (or ItH and IozH ). 5. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT has been chosen to avoid test problems caused by tester ground degradation. 2·278 AmPAL22V10/A (Mil) =0.5 V AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN Parameter Description Test Conditions I Pins 1, 13 Input Capacitance VIN Typ = 2.0 V Vee = 5.0 V TA = 25°C f = 1 MHz 1Others COUT Output Capacitance VOUT = 2.0 V Unit 11 pF 6 9 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over MILITARY operating ranges (Note 2) Std A Parameter Symbol Parameter Description Min tPD Input or Feedback to Combinatorial Output ts Setup Time from Input or Feedback to Clock tH Hold Time Max Min 30 25 0 Max Unit 40 ns 35 ns ns 0 teo Clock to Output 20 25 ns tAR Asynchronous Reset to Registered Output 35 45 ns tARW Asynchronous Reset Width (Note 3) 30 40 ns tARR Asynchronous Reset Recovery Time (Note 3) 30 40 ns tWL Clock Width tWH fMAX Maximum Frequency (Note 4) LOW 20 30 ns HIGH 20 30 ns 22 16.5 MHz External Feedback 1/(ts + teo) tEA Input to Output Enable Using Product Term Control (Note 5) 30 40 ns tER Input to Output Disable Using Product Term Control (Note 5) 30 40 ns Notes: 2. See Switching Test Circuit for test conditions. For APL products Group A, Subgroups 9, 10, and 11 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. tARW and tARR are not directly tested, but are guaranteed by the testing of ts and tAR. 4. These parameters are not 100% tested, but are calculated at initial charzation ond at any time the design is modified where frequency may be affected. 5. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. ' AmPAL22V10/A (Mil) 2-279 ~ AMD SWITCHING WAVEFORMS Input or Input or Feedback Combinatorial Output -~-VT F•• r t :~::---"""I:=~:t:~' Vr ---~teo VT Registered Output _ _ _ _ _ _ _ _...... 165598-8 165598-7 Registered Output Combinatorial Output Input or Feedback _ _ _- - I tER Clock Output ----.....L..L..I.-1 165598-10 165598-9 Input to Output Disable/Enable Clock Width Input Input Asserting Asynchronous Reset Asserting Synchronous Preset =l= _ _ _ __ ts Registered Output _ _ _"'-"'--K............., Clock tARR Clock VT Registered Output _ - ' - -_ _ _ _..J -----------xXffi· teo VT ~---;... \.-.._ _ __ 165599-12 165598-11 Asynchronous Reset Synchronous Preset Notes: 1. Vr = 1.5 V. 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns - 4 ns typical. 2-280 PAL22V10 Family AMD~ KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L ///// May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State WAVEFORM XXXXXX J1) EK KSOOOO10-PAL SWITCHING TEST CIRCUIT 5V Output D--+----.....--te) Test Point 165598-13 Military Commercial Specification tpo. teo tEA - S1 R1 R2 R1 R2 Measured Output Value 50 pF 300n All except -7: 390n 390n 750n 1.5 V Closed Z~ Z~ tER CL H:Open L: Closed H ~Z: Open L~Z: Closed 5 pF -7: 300n PAL22V10 Family 1.5 V H ~Z: VOH-0.5 V L ~Z: VOL + 0.5 V 2·281 ~ AMD MEASURED SWITCHING CHARACTERISTICS for the PAL22V10-10 Vee = 4.75 V, TA = 75°C (Note 1) 10 9 tpo, ns 8 7+---r-~~-+--~--+---r--+--~~ 234 5 6 7 8 9 10 Number of Outputs Switching bo VS. Number of Outputs Switching 165598-14 13 12 11 10 tpo, ns 9 8 7 0 40 120 80 160 200 el, pF tpo VS. Load Capacitance 165598-15 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where tPD may be affected. 2-282 PAL22V10-10 AMD~ INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee Input ProgramNerify Circuitry 165598-16 Typical Input ------------------.-------~ Vee 400 NOM ......I - - -.....--.-_o ~-+- Input, liD Pins Output ProgramNerifyl Test Circuitry Preload Circuitry 165598-17 Typical Output PAL22V10 Family 2·283 ~ AMD POWER-UP RESET The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. The output state will depend on the programmed pattern. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vcc Parameter Symbol can rise to its steady state, two conditions are required to ensure a valid power-up reset. These conditions are: • The Vcc rise must be monotonic. • Following reset, the clock input must not be driven from LOWto HIGH until all applicable input and feedback setup times are met. Parameter Description Max Unit tPR Power-up Reset Time 1000 ns ts Input or Feedback Setup Time tWL Clock Width LOW See Switching Characteristics Vcc Power Registered Active-Low Output Clock 165598-18 Power-Up Reset Waveform 2-284 PAL22V10 Family AMD~ 2-285 COM'L: H-5/7/10/15/25, Q-10/15/25 _ MIL: H-10/15/20/25/30 Advanced Micro Devices PALCE22V10 Family 24-Pin EE CMOS Versatile PAL Device DISTINCTIVE CHARACTERISTICS • • • • As fast as 5 ns propagation delay and 142.8 MHz fMAx (external) Low-power EE CMOS 10 macrocells programmable as registered or combinatorial, and active high or active low to match application needs Varied product term distribution allows up to 16 product terms per output for complex functions ~ • • • • • Global asynchronous reset and synchronous preset for Initialization Power-up reset for Initialization and register preload for testability Extensive third-party software and programmer support through FusionPLD partners 24-pln SKINNYDIP, 24-pln SOIC, 24-pln Flatpack and 28-pln PLCC and LCC packages save space 5 ns and 7.5 ns versions utilize split lead frames for Improved performance GENERAL DESCRIPTION The PALCE22V10 provides user-programmable logic for replacing conventional SSI/MSI gates and flip-flops at a reduced chip count. The PAL device implements the familiar Boolean logic transfer function, the sum of products. The PAL device is a programmable AND array driving a fixed OR array. The AN D array is programmed to create custom product terms, while the OR array sums selected terms at the outputs. The product terms are connected to the fixed OR array with a varied distribution from 8 t016 across the outputs (see Block Diagram). The OR sum ofthe products feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial, and active high or active low. The output configuration is determined by two bits controlling two multiplexers in each macrocell. AMD's FusionPLD program allows PALCE22V10 designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accu rate, quality support. By ensu ring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. Th,e FusionPLD program also greatly reduces design time since a designer can use a tool that is already installed and familiar. Please refer to the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. BLOCK DIAGRAM VOo 2-286 1/02 VOs VOs V07 VOe V()g Publication# 16564 Rev, B Issue Date: June 1993 Amendment/O AMD CONNECTION DIAGRAMS Top View PLCC/LCC SKINNYDIP/SOIC/FLATPACK Vee 110g II0a 1/07 1105 13 14 15 1/04 NC 1103 16 17 18 1106 1/02 1/01 1100 Note: Pin 1 is marked for orientation. PIN DESIGNATIONS = Clock GND = Ground Input = Input/Output NC = No Connect Vee = Supply Voltage 1/06 1105 NC 1/04 1/03 1/02 -=zz=OO (!J ::::. ::::. 165648-2 1/0 co 1/07 0)000..-0..- 111 ClK C\J 25 24 23 22 21 20 19 PALCE22V10 Family 165648-3 ;t1 ~ AMD ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL _____T-l p~ £.a _V,... 10 H -5 FAMILY TYPE PAL = Programmable Array Logic TECHNOLOGY--------~ CE =CMOS Electrically Erasable NUMBER OF ARRAY INPUTS OUTPUT TYPE --------~ V = Versatile NUMBER OF OUTPUTS P C 15 r OPTIONAL PROCESSING Blank = Standard processing PROGRAMMING DESIGNATOR Blank = Initial Algorithm 14 First Revision 15 Second Revision (Same Algorithm as 14) OPERATING CONDITIONS C = Commercial (O°C to +75°C) PACKAGE TYPE POWER-------------~ Q = Quarter Power (55 mA Icc) H = Half Power (90-140 mA Icc) P = 24-Pin 300 mil Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) S = 24-Pin Plastic Gull-Wing Small Outline Package (SO 024) SPEED -5 -7 -10 -15 -25 Valid Combinations PALCE22V10-5 JC PALCE22V10H-7 15 PALCE22V10H-10 PALCE22V1OQ-10 PC,JC, SC PALCE22V10H-15 2-288 PC,JC PALCE22V10H-25 PC,JC, SC PALCE22V1oo-25 PC,JC = 10 ns tPD = 15 ns tPD = 25 ns tPD Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. Blank,/4 PALCE22V10Q-15 = 5 ns tPD = 7.5 ns tpD 15 Blank,/4 PALCE22V10H-5/7/10/15/25, Q-10/15/25 (Com'l) AMD~ ORDERING INFORMATION APL Products AMD programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STD-883 requirements. The order number (Valid Combination) is formed by a combination of: _____T..IAL ~~ ~~ -Y- 10 H -15 E4 IB FAMILY TYPE PAL = Programmable Array Logic TECHNOLOGY------------~ L A ~ CE = CMOS Electrically Erasable NUMBER OF ARRAY INPUTS OUTPUT TYPE ----------~ V = Versatile LEAD FINISH A = Hot Solder Dip PACKAGE TYPE L = 24-Pin 300-mil Ceramic SKINNYDIP (CD3024) K = 24-Pin Ceramic Flatpack (CFL024) 3 = 28-Pin Ceramic Leadless Chip Carrier (CL 028) DEVICE CLASS !8 = MIL-STD-883C Class B NUMBER OF OUTPUTS PROGRAMMING DESIGNATOR Blank = Initial Algorithm E4 First Revision POWER--------------------------~ H = Half Power (100 - 150 mA Icc) SPEED -15 == 15 ns tPD -20 = 20 ns tPD -25 = 25 ns tPD -30 = 30 ns tPD Valid Combinations PALCE22V10H-15 PALCE22V10H-20 PALCE22V10H-25 E4 Blank, E4 IBLA,/BKA, IB3A Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE22V10H-30 Group A Tests Group A tests consist of Subgroups 1, 2, 3, 7, 8, 9, 10, 11. Military Burn-In Military burn-in is in accordance with the current revision of MIL-STD-883, Test Methods 1015, Conditions A through E. Test conditions are selected at AMD's option. PALCE22V10H-1S/20/2S/30 (Mil) 2-289 ~ AMD FUNCTIONAL DESCRIPTION The PALCE22V10 allows the systems engineer to implement the design on-chip, by programming EE cells to configure AND and OR gates within the device, according to the desired logic function. Complex interconnections between gates, which previously required timeconsuming layout, are lifted from the PC board and placed on silicon, where they can be easily modified during prototyping or production. Product terms with all connections opened assume the logical HIGH state; product terms connected to both true and complement of any single input assume the logical LOW state. The PALCE22V1 0 has 12 inputs and 10 110 macrocells. The macrocell Figure 1allows one of four potential output configurations; registered output or combinatorial 110, active high or active low (see Figure 1). The configuration choice is made according to the user's design . specification and co~responding programming of the configuration bits So - S1. Multiplexer controls are connected to ground (0) through a programmable bit, selecting the "0" path through the multiplexer. Erasing the bit disconnects the control line from GND and it is driven to a high level, selecting the "1" path. The device is produced with a EE cell link at each input to the AND gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Utilizing an easily-implemented programming al- ' gorithm, these products can be rapidly programmed to any customized pattern. Variable Input/Output Pin Ratio The PALCE22V1 0 has twelve dedicated input lines, and each macrocell output can be an 1/0 pin. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vee or GND . I/On Output Configuration 51 50 0 0 Registered/Active Low 0 1 Registered/Active High 1 0 Combinatorial/Active Low 1 Combinatorial/Active High 0= Programmed EE bit 1 = Erased (charged) EE bit 165648-4 Figure 1. Output Logic Macrocell Diagram 2·290 PALCE22V10 Family AMD l1 Registered Output Configuration Combinatorial I/O Configuration Each macrocell of the PALCE22V10 includes aD-type flip-flop for data storage and synchronization. The flipflop is loaded on the LOW-to-HIGH transition of the clock input. In the registered configuration (51 = 0), the array feedback is from of the flip-flop. Any macrocell can be configured as combinatorial by selecting the multiplexer path that bypasses the flip-flop (51 =1). In the combinatorial configuration the feedback is from the pin. a Combinatorial/Active Low Reg Istered/Active Low Combinatorial/Active High Registered/Active High 165648-5 Figure 2. Macrocell Configuration Options PALCE22V10 Family 2·291 ~ AMD Programmable Three-State Outputs Register Preload Each output has a three-state output buffer with threestate control. A product term controls the buffer, allowing enable and disable to be a function of any product of . device inputs or output feedback. The combinatorial output provides a bidirectional 110 pin, and may be configured as a dedicated input if the buffer is always disabled. Programmable Output Polarity The polarity of each macrocell output can be active high or active low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is controlled by programmable bit So in the output macrocell, and affects both registered and combinatorial outputs. Selection is automatic, based on the design specification and pin definitions. If the pin definition and output equation have the same polarity, the output is programmed to be active high (So = 1). Preset/Reset For initialization, the PALCE22V10 has Preset and Reset product terms. These terms are connected to all registered outputs. When the Synchronous Preset (SP) product term is asserted high, the output registers will be loaded with a HIGH on the next LOW-to-HIGH clock transition. When the Asynchronous Reset (AR) product term is asserted high, the output registers will be immediately loaded with a LOW independent of the clock. Note that preset and reset control the flip-flop, not the output pin. The output level is determined by the output polarity selected. Security Bit After programming and verification, a PALCE22V1 0 design can be secured by programming the security EE bit. Once programmed, this bit defeats readback of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. When the security bit is programmed, the array will read as if every bit is erased, and preload will be disabled. The bit can only be erased in conjunction with erasure of the entire pattern. Programming and Erasing The PALCE22V10 can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erase operation is required. Quality and Testability The PALCE22V100ffers a very high level of built-in quality. The erasability of the device provides a direct means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Technology Power-Up Reset All flip-flops power-up to a logic LOW for predictable system initialization. Outputs of the PALCE22V10 will depend on the programmed output polarity. The Vee rise must be monotonic and the reset delay time is 1000 ns maximum. 2·292 The register on the PALCE22V10 can be preloaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. The high-speed PALCE22V10 is fabricated with AMD's advanced electrically erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are deSigned to be compatible with TIL devices. This technology provides strong input clamp diodes, output slew-rate control, and a grounded substrate for clear switching. PALCE22V10 Family AMD~ LOGIC DIAGRAM SKINNVDIP/SOIC/FLATPACK (PLCC/LCC) Pinouts ... ill (2) ], ,~ t I I T f"V fu? If,,19 2O u 2t 2t 27 28 ,3,1 12, 35 0 I 1111 1111 1111 1111 III III 36 II I (28) 3l fu 4t AR 1111 r- 9 1 1. -~' IS 01 2 IlO g (27) SP o ~ ':' I 10 ~J o 20 0 0 01 IS .. ~ (26) SP m 'i3i o - - 21 ~":'lJ ~ IS o 33 3 (4) lJ I ... I I I I I 34 01 o l.!..J I " 48 49 .. -, 0 IS ta 01 II o " l.!..J (25) -@ 1/06 (24) .". ~ L 0 OJ .. ~"tj o -ffi] .,.. ~:: 1 J SP ~ (5) .". I 0 19 1105 (23) IS rn(6) ~ 65 o - I T I 66 l!..J - .r82 6 83 II I " I li! ' ~::J ' (7) - 0 Q SP 0 o -ill] (21) - L...!J "" I" I 1 )- ~ ,~ ~"! ,AAo IS :>"' 01 17 (20) SP 97 7 (9) I I I I I I J 11 IIII I I I o ~::J.. ~' IS 0 , 110 8 (10) III III II lL11 i l 122 l.!J IS 11 II II I o I lbJ. I r- 130 10 (12) - 131 .A [j] r---<" (13) GND !W---:t (14) 3 4 7 8 II 12 15 16 19 20 23 24 27 28 31 32 35 36 39 40 43 liL' IS .00 lii' 01 o :IE (18) - ~::l SP '9 .". 01 , SP 0 II II Jill 16 (19) oJ.". ~::J I I m- 01 SPJ~ II III 121 (;1) .". l..!J I 98 -{ill 110 0 (17) ':'- I ...sf' 1 (16) 165648-6 .". PALCE22V10 Family 2-293 ~ AMD PRELIMINARY ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied . . . . . . . . . . . . . . .. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . .. O°C to +75°C Supply Voltage with Respect to Ground .................... -0.5 V to +7.0 V Supply Voltage (Vce) with Respect to Ground ......... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 1.0 V DC Output or 1/0 Pin Voltage .................. -0.5 V to Vee + 1.0 V Operating Ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Latchup Current (TA = O°C to +75°C) ...... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to absolute maximum ratings for extended periods may affect device reliability. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified PRELIM NARY Parameter Symbol Parameter Description Test Conditions Voo Output HIGH Voltage IOH =-3.2 mA VIN = VIH or Vil Vee = Min Min VOL Output LOW Voltage IOL = 16 mA VIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) IIH Input HIGH Leakage Current III Input LOW Leakage Current IOZH Off-State Output Leakage Current HIGH IOZl Max Unit V 2.4 V 0.4 V 2.0 0.8 V VIN = Vce. Vec = Max (Note 2) 10 ~ VIN = 0 V. Vec = Max (Note 2) -100 ~ Your = Vee. Vee - Max. VIN = Vil or VIH (Note 2) 10 ~ Off-State Output Leakage Current LOW VOUT = 0 V. Vee = Max. VIN = Vil or VIH (Note 2) -100 ~ Isc Output Short-Circuit Current Vour = 0.5 V. Vee .. Max (Note 3) -130 mA Icc Supply Current (Static) Outputs Open. (lOUT = 0 mA). Vee = Max 115 mA Supply Current (Dynamic) Outputs Open. (lOUT - 0 mA). Vec = Max. f = 25 MHz 140 mA -30 Notes: 1. These are absolute values with respect to the device ground and all overshoots due to system and tester noise are included. 2. 110 pin leakage is the worst case of hL and IOZL (or hH and IOZH). 3. Not more than one output should be tested at a time. Duration of the short-circuit test should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-294 PALCE22V10H-5 (Com'l) AMD~ PRELIMINARY CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance VOUT = 2.0 V Typ Unit 5 Vee =5.0 V TA = 25°C f = 1 MHz pF: 8 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol PRELIMINARY -5 Parameter Description Min Max Unit 1 5.0 ns tPD Input or Feedback to Combinatorial Output t51 Setup Time from Input or Feedback 3.0 ns t52 Setup Time from SP to Clock 4.0 ns tH Hold Time 0 Clock to Output 1 teo tS"KEWA tAA Skew Between Registered Outputs (Note 3) Asynchronous Reset to Registered Output ns 4.0 ns 1 ns 7.5 ns tARW Asynchronous Reset Width 4.5 tARR Asynchronous Reset Recovery Time 4.5 ns tSPR Synchronous Preset Recovery Time 4.5 ns tWL tWH Clock Width LOW 2.5 ns HIGH 2.5 ns 142.8 MHz External Feedback fMAX Maximum Frequency (Note 4) ns I 1/(ts + teo) Internal Feedback (feNT) I No Feedback 1/(tWH + tWL) 150 MHz 200 MHz tEA Input to Output Enable Using Product Term Control 5.0 ns tEA Input to Output Disable Using Product Term Control 5.0 ns Notes: 2. See Switching Test Circuit for test conditions. 3. Skew is measured with all outputs switching in the same direction. 4. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PALCE22V10H-5 (Com'l) 2-295 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied . . . . . . . . . . . . . . .. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . .. O°C to +75°C Supply Voltage with Respect to Ground .................... -0:5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ......... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 1.0 V DC Output or I/O Pin Voltage .................. -0.5 V to Vee + 1.0 V Operating Ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Latchup Current (TA = O°C to +75°C) ...... 100 rnA Stresses abOve those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to absolute maximum ratings for extended periods may affect device reliability. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol VOH Parameter Description Test Conditions Output HIGH Voltage IOH =-3.2 mA Min VIN = VIH or VIL Max 2.4 Unit V Vee = Min VIN = VIH or VIL Vee = Min 0.4 V VOL Output LOW Voltage IOL = 16 mA VIH Input HIGH Voltage Guaranteed Input Logical HIGH Vohage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) 0.8 V IIH Input HIGH Leakage Current VIN = Vee, Vee = Max (Note 2) 10 ~ ilL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) -100 ~ Off-State Output Leakage 10 ~ Current HIGH VOUT = Vee, Vee = Max, VIN = VIL or VIH (Note 2) Off-State Output Leakage VOUT = 0 V, Vee = Max, -100 ~ Current LOW VIN = VIL or VIH (Note 2) Output Short-Circuit Current -130 mA TA = 25°C (Note 3) 115 mA 140 mA IOZH IOZL Ise lee VOUT = 0.5 V, Vee = Max Supply Current (Static) Outputs Open, (lOUT = 0 mA), Supply Current (Dynamic) Outputs Open, (lOUT 2.0 -30 V Vee = Max Vee = Max, f = 0 mA), = 25 MHz Notes: 1. These are absolute values with respect to the device ground and all overshoots due to system and tester noise are included. 2. 110 pin leakage is the worst case of IJL and IOZL (or IJH and IOZH). 3. Not more than one output should be tested at a time. Duration of the short-circuit test should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-296 PALCE22V10H-7 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN GoUT Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance VOUT= 2.0 V Typ Unit 5 Vcc = 5.0 V TA = 25°C f = 1 MHz pF 8 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) -7 Parameter Symbol PDIP Max Min Parameter Description Input or Feedback to Combinatorial Output 1 tS1 Setup Time from Input or Feedback 5 4.5 ns tS2 Setup Time from SP to Clock 6 6 ns tH Hold Time 0 tco Clock to Output 1 tAR Skew Between Registered Outputs (Note 3) Asynchronous Reset Width tARR tSPR fMAX 1 10 ns ns 4.5 ns 1 ns 10 ns 7 7 Asynchronous Reset Recovery Time ·7 7 ns Synchronous Preset Recovery Time 7 7 ns ns tWL tWH 7.5 0 5 1 Asynchronous Reset to Registered Output tARW 1 Unit tPD tSKEWA 7.5 PLCC Min Max Clock Width Maximum Frequency (Note 4) ns LOW 3.5 3.0 HIGH 3.5 3.0 ns External Feedback 11/(ts + tco) 100 111 MHz 125 133 MHz 142.8 166 Internal Feedback (tCNT) I 1/(twH + tWL) No Feedback MHz tEA Input to Output Enable Using Product Term Control 7.5 7.5 ns tEA Input to Output Disable Using Product Term Control 7.5 7.5 ns Notes: 2. See Switching Test Circuit for test conditions. 3. Skew is measured with all outputs switching in the same direction. 4. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PALCE22V10H-7 (Com'l) 2-297 ~AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied . . . . . . . . . . . . . . .. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air ............ O°C to +75°C Supply Voltage with Respect to Ground .................... -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ......... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 1.0 V DC Output or 1/0 Pin Voltage .................. -0.5 V to Vee + 1.0 V Operating Ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Latehup Current (TA = O°C to +75°C) ...... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to absolute maximum ratings for extended periods may affect device reliability. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 mA VIN = VIHor Vil Vee = Min VOL Output LOW Voltage IOl = 16 mA VIN = VIH or Vil Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) IIH Input HIGH Leakage Current III Input LOW Leakage CUrrent IOZH Off-State Output Leakage Current HIGH IOZl Min Max 2.4 Unit V 0.4 2.0 V V 0.8 V VIN = Vee, Vee = Max (Note 2) 10 ~ VIN = 0 V, Vee = Max (Note 2) -100 ~ VOUT = Vee, Vee =Max, VIN = Vil or VIH (Note 2) 10 ~ Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = Vil or VIH (Note 2) -100 ~ Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max TA = 25° C (Note 3) -130 mA Icc Supply Current (Dynamic) Outputs Open, (lOUT = 0 mAl, Vee = Max, f = 25 MHz 120 mA -30 Notes: 1. These are absolute values with respect to the device ground and all overshoots due to system and tester noise are included. 2. 110 pin leakage is the worst case of IlL and IOZL (or hH and IOZH). 3. Not more than one output should be tested at a· time. Duration of the short-circuit test should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-298 PALCE22V10H-10 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN GoUT Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance VOUT =2.0 V Vee =5.0 V TA = 25°C f = 1 MHz Typ Unit 5 pF 8 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol -10 Min Parameter Description Max Unit 10 ns tPD Input or Feedback to Combinatorial Output t51 Setup Time from Input or Feedback 6 t52 Setup Time from SP to Clock 7 ns tH Hold Time 0 ns ns teo Clock to Output 6 ns tAR Asynchronous Reset to Registered Output 13 ns tARW Asynchronous Reset Width 8 tARR Asynchronous Reset Recovery Time 8 ns tSPR Synchronous Preset Recovery Time 8 ns tWL tWH fMAX Clock Width Maximum Frequency (Note 3) ns LOW 4 ns HIGH 4 ns External Feedback 11/(tS + teo) 83.3 MHz Internal Feedback (teNT) 110 MHz 125 MHz I 1/(tWH + tWL) No Feedback tEA Input to Output Enable Using Product Term Control 10 ns tER Input to Output Disable Using Product Term Control 9 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PALCE22V10H-10 (Com'l) 2-299 ~ AMD PRE LIM I N A R Y ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to +150°C Commercial (C) Devices Ambient Temperature with Power Applied . . . . . . . . . . . . . . .. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air .............. O°C to +75°C Supply Voltage with Respect to Ground .................... -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ............ +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vee + 1.0 V Operating Ranges define those limits between which the functionality of the device is guaranteed. DC Output or 1/0 Pin Voltage .................. -0.5 V to Vee + 1.0 V Static Discharge Voltage ................. 2001 V Latchup Current (TA= O°C to +75°C) ...... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to absolute maximum ratings for extended periods may affect device reliability. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified PRELIMINARY Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-3.2 rnA VIN = VIH or VIL Vee = Min VOL Output LOW Voltage IOL = 16 rnA VIN = VIH or VIL Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) hH Input HIGH Leakage Current ilL Input LOW Leakage Current IOZH Off-State Output Leakage Current HIGH VOUT = Vee, Vee = Max VIN = VIL or VIH (Note 2) IOZL Off-State Output Leakage Current LOW Vour = 0 V, Vee = Max VIN = VIL or VIH (Note 2) Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = 5 V TA = 25°C (Note 3) lec Supply Current (Static) VIN = 0 V, Outputs Open (lOUT = 0 rnA), Vee = Max Min Max Unit 2.4 V 0.4 V 2.0 V O.B V VIN = Vee, Vec = Max (Note 2) 10 VIN = 0 V, Vee = Max (Note 2) -100 10 JlA JlA JlA -100 JlA -130 rnA 55 rnA -30 Notes: 1. These are absolute values with respect to the device ground and all overshoots due to system and tester noise are included. 2. 110 pin leakage is the worst case of ItL and lozL (or I/H and lozH). 3. Not more than one output should be tested at a time. Duration of the short-circuit test should not exceed one second. Your = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-300 PALCE22V10Q-10 (Com'l) AMD~ PRELIMINARY CAPACITANCE (Note 1) Parameter Symbol Parameter Description Test Conditions CIN Input Capacitance VIN =2.0 V COUT Output Capacitance VOUT = 2.0 V Typ I Vee =5.0 V I TA=25°C f 1 MHz Unit 5 pF 8 = Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) PRELIMINARY Parameter Symbol -10 Parameter Description Min tpo Input or Feedback to Combinatorial Output ts Setup Time from Input, Feedback or SP to Clock 6 tH Hold Time 0 Max Unit 10 ns ns ns 6 teo Clock to Output tAR Asynchronous Reset to Registered Output 13 ns ns tARW Asynchronous Reset Width 8 ns tARR Asynchronous Reset Recovery Time 8 ns tSPR Synchronous Preset Recovery Time 8 ns LOW 4 ns HIGH 4 ns 83 MHz 110 MHz tWl tWH fMAX Clock Width Maximum Frequency (Note 3) External Feedback I l/(ts + teo) Internal Feedback (feNT) I No Feedback l/(tWH + tWl) 125 MHz tEA Input to Output Enable Using Product Term Control 10 ns tER Input to Output Disable Using Product Term Control 9 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PALCE22V10Q-10 (Com'l) 2-301 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to +150°C Ambient Temperature with Power Applied . . . . . . . . . . . . . . .. -55°C to +125°C Supply Voltage with Respect to Ground ................... -0.5 V to +7.0 V Commercial (C) Devices DC Input Voltage .......... -0.5 V to Vee + 0.5 V DC Output or I/O Pin Voltage .................. -0.5 V to Vee + 0.5 V Statie Discharge Voltage ................. 2001 V Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . .. O°C to +75°C Supply Voltage (Vce) with IRespect to Ground (H/Q-15) .. +4.75 V to +5.25 V Supply Voltage (Vee) with Respect to Ground (H/Q-25) ... , +4.5 V to +5.5 V Operating Ranges define those limits between which the functionality of the device is guaranteed. Latchup Current (TA = O°C to +75°C) ...... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to absolute maximum ratings for extended periods may affect device reliability. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions Min Max 2.4 Unit V VOH Output HIGH Voltage IOH =-3.2 rnA VIN = VIHor VIL Vee = Min VOL Output LOW Voltage IOL = 16 rnA VIN = VIH or VIL Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) 0.8 V hH Input HIGH Leakage Current VIN = Vee, Vee = Max (Note 2) 10 ~ ilL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) -100 ~ IOZH Off-State Output Leakage Current HIGH VOUT = Vee, Vee = Max, VIN = VIL or VIH (Note 2) 10 ~ IOZl Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max, VIN = VIL or VIH (Note 2) -100 ~ Ise Output Short-Circuit Current -130 rnA TA = 25°C (Note 3) 90 55 rnA lec Supply Current 0.4 2.0 -30 VOUT = 0.5 V, Vee = 5 V VIN = 0 V, Outputs Open (lOUT = 0 rnA), Vee = Max Notes: k- V V 1. These are absolute values with respect to the device ground and all overshoots due to system and tester noise are included. 2. 110 pin leakage is the worst case of IJL and IOZL (or IJH and IOZH). 3. Not more than one output should be tested at a time. Duration of the short-circuit test should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-302 PALCE22V10H-15/25, Q-15/25 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Typ Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Vee = 5.0 V 5 Output Capacitance VOUT=2.0 V TA = 25°C f = 1 MHz 8 Unit pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS O\ler COMMERCIAL operating ranges (Note 2) -25 -15 Parameter Symbol Parameter Description Min Max Min 15 tPD Input or Feedback to Combinatorial Output ts Setup Time from Input. Feedback or SP to Clock 10 15 tH Hold Time 0 0 teo Clock to Output 10 tAR Asynchronous Reset to Registered Output 20 Max Unit 25 ns ns ns 15 ns 25 ns tARW Asynchronous Reset Width 15 25 ns tARR Asynchronous Reset Recovery Time 10 25 ns tSPR Synchronous Preset Recovery Time 10 25 ns 8 13 ns LOW tWL tWH fMAX Clock Width Maximum Frequency (Note 3) HIGH External Feedback I 1/(ts + teo) Internal Feedback (teNT) 8 13 ns 50 33.3 MHz 35.7 58.8 MHz tEA Input to Output Enable Using Product Term Control 15 25 ns tER Input to Output Disable Using Product Term Control 15 25 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALCE22V10H-15/25, 0-15/25 (Com'I) 2-303 ~ AMD ABSOLUTE MAXIMUM RATINGS Storage Temperature .......... -65°C to +150°C OPERATING RANGES Military (M) Devices (Note 1) Ambient Temperature with Power Applied ............ -55°C to + 125°C Operating Case Temperature (Te) ........... -55°C to +125°C Supply Voltage with Respect to Ground ............. -0.5 V to+7.0 V Supply Voltage (Vee) with Respect to Ground ....... +4.5 V to +5.5 V DC Input Voltage ......... " -0.5 V to Vee + 1.0 V DC Output or 1/0 Pin Voltage ............... -0.5 V to Vee + 0.5 V Note: 1. 'Military products are tested at Tc = +25' C, + 12SO C and -SSOC, per MIL-STD-883. Static Discharge Voltage ................. 2001 V Latchup Current (TA =-55°C to +125°C) ... 100 rnA Operating Ranges define those limits between which the functionality of the device is guaranteed. Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Absolute Maximum Ratings are for system design reference; parameters given are not tested. DC CHARACTERISTICS over MILITARY operating ranges unless otherwise specified (Note 2) Parameter Symbol Min Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-2.0 rnA Vee = Min VOL Output LOW Voltage IOl = 12 rnA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 3) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 3) IIH Input HIGH Leakage Current III Input LOW Leakage Current = 5.5 V, Vee = Max (Note 4) = 0 V. Vec = Max (Note 4) VOUT = 5.5 V. Vee = Max. VIN = Vil or VIH (Note 4) VOUT = 0 V. Vee = Max. VIN = VIH or Vil (Note 4) VOUT = 0.5 V. Vee = 5 V TA = 25°C (Note 5) VIN = 0 V. Outputs Open I (lOUT = 0 rnA). Vee = Max I IOZH Off-State Output Leakage Current HIGH IOZl Off-State Output Leakage Current LOW Isc Output Short-Circuit Current Icc Supply Current VIN VIN = VIH or Vil Max 2.4 = VIH or Vil Unit V 0.4 2.0 V V 0.8 V VIN 10 ~ VIN -10 ~ 10 ~ -10 ~ -135 rnA 120 rnA -50 -15/-20 -25/-30 100 Notes: 2. For APLproducts, Group A, Subgroups 1,2 and 3 are tested per MIL-STD-883. Method 5005, unless otherwise noted. 3. V/L and V/H are input conditions of output tests and are not themselves directly tested. V/L and V,H are absolute voltages with respect to device ground and include all overshoots due to system and/or tester noise. Do not attempt to test these values without suitable equipment. 4. //0 pin leakage is the worst case of hL and lozL (or I/H and IOZH). 5. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. This parameter is not 100% tested, but is evaluated at initial characterization and at any time the design is modified where Isc may be affected. 2·304 PALCE22V10H-15/20/25/30 (Mil) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN GoUT Typ Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Vee =5.0 V TA =25°C 8 Output Capacitance VOUT .. 2.0 V f ... 1 MHz 9 Unit pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over MILITARY operating ranges (Note 2) Parameter Symbol -15 Min Max Parameter Description tPD Input or Feedback to Combinatorial Output ts Setup Time from Input, Feedback or SP to Clock 12 Hold Time (Note 3) 0 tH teo Clock to Output tAR Asynchronous Reset to Registered Output Min -20 Max -25 Min Max 20 15 25 0 -30 Max 30 20 18 15 Min ns ns 0 0 Unit ns 8 15 20 20 ns 20 25 25 30 ns 20 30 tARW Asynchronous Reset Width (Note 3) tARR Asynchronous Reset Recovery Time (Note 3) 15 20 25 30 ns tSPR Synchronous Preset Recovery Time 15 20 25 30 ns tWL tWH LOW 8 10 15 15 Clock Width ns HIGH 8 10 15 15 ns Maximum Frequency (Note 3) External Feedback 1/(ts + teo) 50 33.3 26.3 25 MHz fMAX Internal Feedback (feNT) 53 15 25 40 32.2 ns 25 MHz tEA Input to Output Enable Using Product Term Control (Note 3) 15 20 25 25 ns tER Input to Output Disable Using Product Term Control (Note 3) 15 20 25 25 ns Notes: 2. See Switching Test Circuit for test conditions. For APL products Group A, Subgroups 7, 8, 9, 10, and 11 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where these parameters may be affected. PALCE22V10H-15/20/25/30 (Mil) 2-305 ~AMD SWITCHING WAVEFORMS Input or Feedback e _Xffi1tPD. t 'npulo, ~ Clock VT -VT Combinatorial Output Registered Output _ _ _ _ _ _ _ __ 165648-8 165648-7 Registered Output Combinatorial Output Input _ _ _---J tER Clock Output tEA )-+o-ttJ-~---4i-+-f-f_ _ _ _-L..L.J.'-4 VT 165648-10 165648-9 Input to Output Disable/Enable Clock Width Input Input Asserting Asynchronous Reset Asserting Synchronous Preset =f' _ _ _ __ t5 Registered Output _ _ _""--IOt.....K........_ Clock _ _ _ _ _ _..J tARR Clock VT Registered Output ~~V-T---- 165648-11 Asynchronous Reset Notes: 1. Vr = 1.5 V. 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns - 5 ns typical. 2-306 165648-12 Synchronous Preset PALCE22V10 Family AMD~ KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L /77// May Change from L to H Will be Changing from L to H xxxxxx Don't Care, Any Change Permitted Changing, State Unknown Jl)(K Does Not Apply Center Line is HighImpedance "Off"State WAVEFORM KSOOOO10-PAL SWITCHING TEST CIRCUIT Output O--..----4--~) Test Point 165648-13 Commercial Specification tPD, teo tEA S1 CL R1 R2 300n All except H-517: 390n Closed Z~ H: Open 50 pF Military R1 R2 Measured Output Value 1.5 V 390n 750n 1.5 V Z ~ L: Closed tEA H ~Z: Open L ~Z: Closed 5 pF H-517: 300n PALCE22V10 Family H ~Z: VOH-0.5 V L ~ Z: VOL + 0.5 V 2·307 ~ AMD TYPICAL lee CHARACTERISTICS Vee =5.0 V, TA =25°C 150 125 22V10H-7 22V10H-10 100 lee (mA) 22V10H-15 22V10H-25 75 22V10Q-25 50 25 O-r----+-----~--_+----~----+_--~----_+----~----~--~ o 10 20 30 Frequency (MHz) 40 50 165648-14 lee vs. Frequency The selected "typical" pattern utilized 50% of the device resources. Half of the macrocells were programmed as registered, and the other half were programmed as combinatorial. Half of the available product terms were used for each macrocell. On any vector, half of the outputs were switching. By utilizing 50% of the device, a midpoint is defined for Icc. From this midpoint, a designer may scale the Icc graphs up or down to estimate the Icc requirements for a particular design. 2-308 PALCE22V10 Family AMD~ ENDURANCE CHARACTERISTICS parts. As a result, the device can be erased and reprogrammed-a feature which allows 100% testing at the factory. The PALCE22V10 is manufactured using AMD's advanced Electrically Erasable process. This technology uses an EE cell to replace the fuse link used in bipolar Endurance Characteristics Symbol tOR N Test Conditions Parameter Min Pattern Data Retention Time Min Reprogramming Cycles Min Unit Max Storage Temperature 10 Years Max Operating Temperature (Military) 20 Years Normal Programming Conditions 100 Cycles INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee ESD ProgramNerify Protection Circuitry Typical Input Vee = Preload Circuitry Feedback Input 16564B-15 Typical Output PALCE22V10 Family 2-309 ~AMD ROBUSTNESS FEATURES A special noise filter makes the programming circuitry completely insensitive to any positive overshoot that has a pulse width of less than about 100 ns for the /5 version. Selected /4 devices are also being retrofitted with these robustness features. See the chart below for device listing. The PALCE22V10X-Xl5 devices have some unique features that make them extremely robust, especially when operating in high-speed design environments. Pull-up resistors on inputs and I/O pins cause unconnected pins to default to a known state. Input clamping circuitry limits negative overshoot, eliminating the possibility of false clocking caused by subsequent ringing. INPUT/OUTPUT EQUIVALENT SCHEMATICS FOR /5 VERSION AND SELECTED 14 DEVICES· Vee Vee >50kn I I I I I . II Programmlng ESD Protection and Clamping L'2n.=.~2:. =- ____ _ Typical Input Vee Vee >50kn Provides ESD Protection and Clamping Preload Circuitry Feedback Input 165648-16 Typical Output • Device Rev Letter PALCE22V10H-15 D PALCE22V10H-25 D PALCE22V1oo-25 8 Tops/de Marking: AMO CMOS PLO's are marked on top of the package in the following manner: PALCEXXXX Oatecode (3 numbers) Lot /D (4 characters)- -(Rev Letter) The Lot 10 and Rev Letter are separated by two spaces. 2-310 PALCE22V10 Family AMD~ POWER-UP RESET The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. The output state will depend on the programmed pattern. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vce can rise to its steady state, two conditions are Parameter Symbol tPR Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Max Unit 1000 ns Input or Feedback Setup Time Clock Width LOW 4? I Clock The Vee rise must be monotonic. • Power-up Reset Time tWL Registered Active-Low Output • Parameter Description ts Power required to ensure a valid power-up reset. These conditions are: See Switching Characteristics Vee ~14L.1---- tPR --' // 9= 165648-17 tWL Power-Up Reset Waveform PALCE22V10 Family 2-311 ~AMD TYPICAL THERMAL CHARACTERISTICS PALCE22V10/4 (PALCE22V10H-15) Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Typ PLCC Unit Sjc Thermal impedance, junction to case 15 16 Sja Thermal impedance, junction to ambient 72 54 200 Ifpm air 67 49 400 Ifpm air 60 43 600 Ifpm air 53 37 800 Ifpm air 46 31 °elW °elW °elW °elW °elW °elW Sjma Parameter Description Thermal impedance, junction to ambient with air flow SKINNYDIP PALCE22V10/5 {PALCE22V1OH-10) Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Typ PLCC Unit Sjc Thermal impedance, junction to case 20 18 Sja Thermal impedance, junction to ambient 73 55 °elW °elW °elW °elW °elW °elW Sjma Parameter Description Thermal impedance, junction to ambient with air flow SKINNYDIP 200 Ifpm air 66 48 400 Ifpm air 61 43 600 Ifpm air 55 40 800 Ifpm air 52 37 Plastic Sic Considerations The data listed for plastic Sjc are for reference only and are not recommended for use in calculating junction temperatures. The heat-flow paths in plastic-encapsulated devices are complex, making the Ojc measurement relativo to a specific location on the package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom center of the package. Furthermore, Sjc tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a siml7ar environment. 2·312 PALCE22V10 Family _ COM'L:-25 ~ INO: -15/25 PALCE22V10Z Family Zero-Power 24-Pin EE CMOS Versatile PAL Device Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • Zero-power CMOS technology - 15 ~ standby current - As fast as 15 ns first-access propagation delay and 50 MHz fMAX (external) • 10 macrocells programmable as registered or combinatorial, and active high or active low to match application needs • • Unused product term disable for reduced power consumption Varied product term distribution allows up to 16 product terms per output for complex functions • • Available In Industrial operating range - Tc = -40°C to +85°C - Vee = +4.5 V to +5.5 V Global asynchronous reset and synchronous preset for Initialization • Power-up reset for Initialization and register preload for testability • HC- and HCr-compatible Inputs and outputs • • Electrically-erasable technology provides reconfigurable logic and full testability Extensive third-party software and programmer support through FusionPLO partners • 24-pln SKINNYOIP and 28-pln PLCC packages save space GENERAL DESCRIPTION The PALCE22V10Z is an advanced PAL device built with zero-power, high-speed, electrically-erasable CMOS technology .It provides user-programmable logiC for replacing conventional zero-power CMOS SSIIMSI gates and flip-flops at a reduced chip count. the output macrocell. Each macrocell can be programmed as registered or combinatorial, and active high or active low. The output configuration is determined by two bits controlling two multiplexers in each macrocell. The PALCE22V10Z provides zero standby power and high speed. At 15 ~ maximum standby current, the PALCE22V1 OZ allows battery powered operation for an extended period. AMD's FusionPLD program allows PALCE22V10Z designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensu ring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces deSign time since a designer can use a tool that is already installed and familiar. Please refer to the Software Reference Guide to PLD Compliers for certified development systems, and the Programmer Reference Guide for approved programmers. The ZPAL™ device implements the familiar Boolean logic transfer function, the sum of products. The PAL device is a programmable AND array driving a fixed OR array. The AND array is programmed to create custom product terms, while the OR array sums selected terms at the outputs. The product terms are connected to the fixed OR array with a varied distribution from 8 t016 across the outputs (see Block Diagram). The OR sum of the products feeds Publication# 15700 Rev. Issue Date: June 1993 e Amendment/O 2-313 ~AMD BLOCK DIAGRAM 11·111 CLKllo 15700C-1 CONNECTION DIAGRAMS Top View SKINNYDIP/SOIC 0 ClKllo Vee 11 II0g 12 I/0a 13 14 1/07 1/06 15 16 1/04 17 1/03 la 1/02 Ig 110 1/00 GND PLCC 1105 ~ ~-=-d 1/06 1105 NC 1/04 1/03 1/02 1/01 - 0'10 Cl ..- = Ground = Input = Input/Output = No Connect = Supply Voltage PALCE22V10Z Family 0 ..- 0 ..- z z -=-~~ " 111 PIN DESCRIPTION ClK = Clock 2-314 1/07 24 23 22 21 20 16 17 18 Pin 1 is marked for orientation. Vee CX) ;:::, NC Note: NC 0'1 Z>;:::, 13 14 15 15700C-2 GND I I/O o 000 0 15700C-3 AMOl1 ORDERING INFORMATION Commercial and Industrial Products AMD programmable logic products for industrial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of these elements: _______T...JAL ~.§ ~~ _V,.- 10 Z -15 FAMILY TYPE PAL = Programmable Array Logic P I l OPERATING CONDITIONS I = Industrial (-40 oC to +85°C) C = Commercial (OOC to +75°C) TECHNOLOGY-------~ PACKAGE TYPE P = 24-Pin 300 mil Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) S = 24-Pin Plastic Gull-Wing Small Outline Package (SO 24) CE = CMOS Electrically Erasable NUMBER OF ARRAY INPUTS OUTPUT TYPE _ _ _ _ _ _ _ _ _ _....J V = Versatile NUMBER OF OUTPUTS POWER _ _ _ _ _ _ _ _ _ _ _ _ _ _ _....J SPEED -15 = 15 ns tPD -25 = 25 ns tPD Z = Zero Power (15 J,LA Icc standby) Valid Combinations PALCE22V1 OZ-15 PI,JI, SI, PALCE22V10Z-25 PC,JC, SC, PI,JI, SI Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations, and to check on newly released combinations. PALCE22V10Z-15/25 (Com'I, INO) 2-315 ~AMD FUNCTIONAL DESCRIPTION The PALCE22V10Z is the zero-power version of the PALCE22V10. It has all the architectural features of the PALCE22V10. In addition, the PALCE22V1 OZ has zero standby power and unused product term disable. The PALCE22V10Z allows the systems engineer to implement the design on-chip, by programming EE cells to configure AND and OR gates within the device, according to the desired logic function. Complex interconnections between gates, which previously required time-consuming layout, are lifted from the PC board and placed on Silicon, where they can be easily modified during prototyping or production. Product terms with all connections opened assume the logical HIGH state; product terms connected to both true and complement of any single input assume the logical LOW state. The PALCE22V10Z has 12 inputs and 10 I/O macrocells. The macrocell (Figure 1) allows one of four potential output configurations; registered output or combinatorial I/O, active high or active low (see Figure 2). The configuration choice is made according to the user's design specification and corresponding programming of the configuration bits 80-81. Multiplexer controls are connected to ground (0) through a programmable bit, selecting the "0" path through the multiplexer. Erasing the bit disconnects the control line from GND and it floats to Vcc (1), selecting the "1" path. The device is produced with a EE cell link at each input to the AND gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Utilizing an easily-implemented programming algorithm, these products can be rapidly programmed to any customized pattern. Variable Input/Output Pin Ratio The PALCE22V10Z has twelve dedicated input lines, and each macrocell output can be an I/O pin. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vcc or GND. Registered Output Configuration Each macrocell of the PALCE22V1 OZ includes aD-type flip-flop for data storage and synchronization. The flipflop is loaded on the LOW-to-HIGH transition of the clock input. In the registered configuration (81 = 0), the array feedback is from Q of the flip-flop. Combinatorial I/O Configuration Any macrocell can be configured as combinatorial by selecting the multiplexer path that bypasses the flip-flop (81 = 1). In the combinatorial configuration the feedback is from the pin. 1---~"""x)4-~~ I/On Output Configuration So So o o o Registered/Active Low o Combinatorial/Active Low 1 Combinatorial/Active High Registered/Active High o = Programmed EE bit 1 = Erased (charged) EE bit 1S700C-4 Figure 1. Output Lagle Macrocell 2-316 PALCE22V10Z Family AMD~ Combinatorial/Active Low Registered/Active Low Combinatorial/Active High Registered/Active High 15700C-5 Figure 2. Macrocell Configuration Options Programmable Three-State Outputs Each output has a three-state output buffer with threestate control. A product term controls the buffer, allowing enable and disable to be a function of any product of device inputs or output feedback. The combinatorial _output provides a bidirectional 1/0 pin, and may be configured as a dedicated input if the buffer is always disabled. Programmable Output Polarity The polarity of each macrocell output can be active high or active low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is controlled by programmable bit So in the output macrocell, and affects both registered and combinatorial outputs. Selection is automatic, based on the design specification and pin definitions. If the pin definition and output equation have the same polarity, the output is programmed to be active high (So = 1). Preset/Reset For initialization, the PALCE22V10Z has additional Preset and Reset product terms. These terms are connected to all registered outputs. When the Synchronous Preset (SP) product term is asserted high, the output registers will be loaded with a HIGH on the next LOW-toHIGH clock transition. When the Asynchronous Reset (AR) product term is asserted high, the output registers will be immediately loaded with a LOW independent of the clock. Note that preset and reset control the flip-flop, not the output pin. The output level is determined by the output polarity selected. Zero-Standby Power Mode The PALCE22V10Z features a zero-standby power mode. When none of the inputs switch for an extended period (typically 50 ns). the PALCE22V10Z will go into standby mode, shutting down most of its internal circuitry. The current will go to almost zero (Icc < 15 J,JA). The outputs will maintain the states held before the device went into the standby mode. PALCE22V10Z Family 2-317 ~AMD When any input switches, the internal circuitry is fully enabled and power consumption returns to normal. This f~ature results in ~nsiderable power savings for operation at low to medium frequencies. This savings is illustrated in the Icc vs. frequency graph. Product-Term Disable On a·programmed PALCE22V10Z, any product terms that are not used are disabled. Power is cut off from these product terms so that they do not draw current. As shown in the lee vs. frequency graph, product-term disabling results in considerable power savings. This savings is greater at the higher frequencies. Further hints on minimizing power consumption can be found in the Application Note "Minimizing Power Consumption with Zero-Power PLDs." Power-Up Reset All flip-flops power-up to a logic LOW for predictable system initialization. Outputs of the PALCE22V10Z will depend on the programmed output polarity. The Vee rise must be monotonic and the reset delay time is 1000 ns maximum. Register Preload The registers on the PALCE22V10Z can be pre loaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. 2-318 Security Bit After programming and verification, a PALCE22V10Z design can be secured by programming the security EE bit. Once programmed, this bit defeats readback of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. When the security bit is programmed, the array will read as if every bit is erased, and preload will be disabled. The bit can only be erased in conjunction with erasure of the entire pattern. Programming and EraSing The PALCE22V10Z can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erase operation is required. Quality and Testability The PALCE22V10Z offers a very high level of built-in quality. The erasability of the CMOS PALCE22V10Z allows direct testing of the device array to guarantee 100% programming and functional yields. Technology The high-speed PALCE22V10Z is fabricated with AMD's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with HC and HCT devices. This technology provides strong input-clamp diodes, output slew-rate control, and a grounded substrate for clean switching. PALCE22V10Z Family AMol1 LOGIC DIAGRAM SKINNYDIP (PLCC) Pinouts CLKIIO ... I (2) 0 I ~I d IIII 4 ~ TI II II I II 'F I I'r 16 III 'r I 2? I 12? IIII 2t 2r 2r I 131' I I IIII 3F I ,35 1, ,3r J I II II 4f, E!I (28) Vcc ,4r Afl II II ~. o ~ "; ~l o 20 0 15 0 0 0 I o "; I -r,Ql 21 ~ ~ ~J OAflg o W. '" Fr~ll 0 15 I 1J I II II I I II IIII III II L JJ I I IIII IIII 49 1 '" L...!.J 10 ~ ~ 0 OJ 0 I oARo 15 o I I I 11 I I ~ ~J o I I >98 I I I I II I III I II I II I II IIII I 1 JJ I 11111111 IIII IIII IIII 111 I II I I I I II I II I I 11 r Ir II I 15 , I I 11 16 (19) 0 I ~ ';" .. 1 ~I 1 ~::j' 15 (18) SP~rh ~ I II 10 L!.J III )- 122 17 I (20) Fe~ .. o I l U .. L!.J 121 m ~" SP~rh ~ 0 SP I I II TO oARo ....... I r110 18 I (21) - L..!J I 1 ~ SP 82 97 (23) .". 10 83 19 110 5 .. L...!.J II OARg f6l. 1 ~ SP 66 (24) '\ o 65 20 I 0 OJ 0 I SP 48 4 (5) (II) (251 L...!.J 34 o 8 (10) 21 p i!. SP 33 7 (9) 22 I (26) .. SP 2 (3) m 110 9 I 10 5 (6) 2 (27) ~ SP (4) 1 10 9 ~ III - >- 130 ~ TIl l ~. 14 110 0 (17) ,~ SP .:l;- '9 10 (12) o I 131 .... 11 (13) GND ffi---:L (14) -- '" SP 3 4 7 8 II 12 15 16 19 20 23 24 27 28 31 32 35 36 39 40 43 .../131 (;6) 'II 1S700C-6 ":;' PALCE22V10Z Family 2-319 ~ AMD PRE LIM I N A R Y ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to +150°C Industrial (I) Devices Ambient Temperature with Power Applied . . . . . . . . . . . . . . .. -55°C to + 125°C Operating Case Temperature (Te) ............ -40°C to +85°C Supply Voltage with Respect to Ground ................... -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ........... +4.5 V to +5.5 V DC Input Voltage .......... -0.5 V to Vee + 0.5 V Operating Ranges define those limits between which the functionality of the device is guaranteed. DC Output or 1/0 Pin Voltage ............... -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Latchup Current (Te = 40°C to +85°C) ..... 100 rnA' Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over INDUSTRIAL operating ranges unless otherwise specified PRELIMINARY Parameter Symbol VOH Parameter Description Test Conditions Output HIGH Voltage VIN = VIH or Vil = Min Vee Min IOH IOH =6 rnA =20 JlA Max Unit 3.84 V Vcc- V 0.1 VOL Output LOW Voltage VIN = VIH or Vil IOl Vee = Min = 16 rnA 0.5 V IOl = 6 rnA 0.33 V IOl = 20 JlA 0.1 V VIH Input HIGH Voltage Guaranteed Input Logical HIGH Vo~age for all Inputs (Notes 1, 2) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Notes 1, 2) IIH Input HIGH Leakage Current III Input LOW Leakage Current = Vee, Vee = Max (Note 3) VIN =0 V, Vee =Max (Note 3) VOUT = Vee, Vee = Max VIN = VIH or Vil (Note 3) VOUT =0 V, Vee = Max VIN = VIH or Vil (Note 3) VOUT =0.5 V, Vee = Max (Note 4) f =0 MHz Outputs Open (lOUT =0 rnA) f = 25 MHz Vee = Max IOZH Off-State Output Leakage Current HIGH IOZl Off-State Output Leakage Current LOW Isc Output Short-Circuit Current Icc Supply Current V 2.0 VIN 0.9 V 10 10 JlA JlA JlA -10 JlA -150 rnA -10 -30 15 JlA 90 rnA Notes: 1. 2. 3. 4. These are absolute values with respect to device ground and all overshoots due to system or tester noise are included. Represents the worst case of HC and HCT standards, allowing compatibility with either. 110 pin leakage is the worst case of IJL and IOZL (or IJH and IOZH). Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-320 PALCE22V10Z-15 (INO) AMO~ PRELIMINARY CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Condition Input Capacitance VIN Output Capacitance VOUT = 2.0 V = 2.0 V Typ = 5.0 V TA = 25°C f = 1 MHz Unit 5 Vee pF 8 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over INDUSTRIAL operating ranges (Note 2) PRELIMINARY Parameter Symbol Parameter Description Min tPD Input or Feedback to Combinatorial Output ts Setup Time from Input, Feedback or SP to Clock 10 tH Hold Time 0 teo Clock to Output tAR Asynchronous Reset to Registered Output tARW Asynchronous Reset Width Max Unit 15 ns ns ns 10 ns 20 ns 15 ns tARR Asynchronous Reset Recovery Time 10 ns tSPR Synchronous Preset Recovery Time 10 ns tWL Clock Width LOW 8 ns HIGH 8 ns 50 MHz tWH fMAX Maximum Frequency (Note 3) External Feedback I 1/(ts + teo) Internal Feedback (br) No Feedback I 1/(twH + tWL) 58.8 MHz 62.5 MHz tEA Input to Output Enable Using Product Term Control 15 ns tER Input to Output Disable Using Product Term Control 15 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALCE22V10Z-15 (INO) 2-321 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to + 150°C Commercial (C) DevIces Ambient Temperature with Power Applied . . . . . . . . . . . . . . .. -55°C to + 125°C Supply Voltage with Respect to Ground .................... -0.5 V to +7.0 V Ambient Temperature (TA) ....... O°C to +75°C Supply Voltage (Vcc) with Respect to Ground ......... +4.75 V to +5.25 V Industrial (I) DevIces DC Input Voltage .......... -0.5 V to Vcc + 0.5 V Operating Case Temperature (Tc) ............ -40°C to +85°C DC Output or I/O Pin Voltage .................. -0.5 V to Vcc + 0.5 V Static Discharge Voltage ................. 2001. V Latchup Current (Tc =-40°C to +85°C) .... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Supply Voltage (Vee) with Respect to Ground ........... +4.5 V to +5.5 V Operating Ranges define those limits between which the functionality of the device is guaranteed. DC CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges unless otherwise specified Parameter Symbol VOH VOL Parameter Description Test Conditions Output HIGH Voltage VIN = VIH or Vil IOH = 6 rnA 3.84 V Vee = Min IOH = 20 JlA Vee0.1 V VIN = VIH or Vil IOL = 16 rnA 0.5 V Vee = Min IOl = 6 rnA 0.33 V IOl = 20 JlA 0.1 V Output LOW Voltage Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Notes 1, 2) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Notes 1, 2) IIH Input HIGH Leakage Current ilL IOZH lozl Max 2.0 Unit V 0.9 V VIN = Vec, Vec = Max (Note 3) 10 Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 3) -10 Off-State Output Leakage VOUT = Vee, Vee = Max 10 JlA JlA JlA Current HIGH VIN = VIH or Vil (Note 3) Off-State Output Leakage VOUT = 0 V, Vec = Max -10 JlA Current LOW VIN = VIH or Vil (Note 3) -150 rnA Isc Output Short-Circuit Current VOUT = 0.5 V, Vec = Max (Note 4) lec Supply Current Outputs Open (lOUT = 0 rnA) f = 0 MHz -30 15 JlA Vee = Max f=25MHz 120 rnA Notes: 1. 2. 3. 4. These are absolute values with respect to device ground and all overshoots due to system or tester noise are included. Represents the worst case of HC and HCT standards, allowing compatibility with either. 110 pin leakage is the worst case of flL and IOZL (or flH and IOZH). Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-322 PALCE22V10Z-25 (Com'l, IND) AMO~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Typ Parameter Description Test Condition Input Capacitance VIN = 2.0 V Vee = 5.0 V 5 VOUT = 2.0 V TA = 25°C f = 1 MHz 8 Output Capacitance Unit pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL and INDUSTRIAL operating ranges (Note 2) Parameter Symbol Min Parameter Description tpo Input or Feedback to Combinatorial Output (Note 3) ts Setup Time from Input. Feedback or SP to Clock 15 0 tH Hold Time teo Clock to Output tAR Asynchronous Reset to Registered Output Max Unit 25 ns ns ns 15 ns 25 ns tARW Asynchronous Reset Width 25 ns tARR Asynchronous Reset Recovery Time 25 ns tSPR Synchronous Preset Recovery Time 25 ns tWL Clock Width LOW 10 ns HIGH 10 ns 33.3 MHz 35.7 MHz tWH fMAX Maximum Frequency (Note 4) External Feedback I 1/(ts + teo) Internal Feedback (fCNT) No Feedback I 1/(twH + tWL) 50 MHz tEA Input to Output Enable Using Product Term Control 25 ns tER Input to Output Disable Using Product Term Control 25 ns Notes: 2. See Switching Test Circuit for test conditions. 3. This parameter is tested in Standby Mode. When the device is not in Standby Mode, the tpo will typically be 5 ns faster. 4. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALCE22V10Z-25 (Com'l, INO) 2-323 ~AMD SWITCHING WAVEFORMS Input or Feedback Combinatorial Output -~-vr 15700C-7 15700C-8 Combinatorial Output Registered Output Input _ _ _-..I tER Clock Output tEA _ _ _ _--L...6-I.-1 15700C-10 15700C-9 Input to Output Disable/Enable Clock Width Input Input Asserting Asynchronous Reset Assertirig~ Synchronous Preset _ _ _ __ ts Registered Output _ _ _"""-liI~"""'JI Clock _ _ _ _ _ _..J tARR Clock Vr Registered Output _____ ~~~t.co-v-r---1S700C-12 15700C-11 . Asynchronous Reset Synchronous Preset Notes: 1. Vr = 1.5 V for Input Signals and 2.5 V for Output Signals. 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns-5 ns typical. 2-324 PALCE22V10Z Family AMD~ KEY TO SWITCHING WAVEFORMS WAVEFORM INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be IIIZI May Change from L to H Will be Don't Care, Any Change Permitted Changing, State Unknown xxxxxx Does Not Apply Changing from H to L Changing from L to H Center Line is High. Impedance "Off" State KSOOOO 1O-PAL SWITCHING TEST CIRCUIT Output Test Point 1-. Specification tPD, teo tEA tER 51 S2 52 Closed Closed Z~H:Open Z~ Z~ Z ~L:Open L: Closed H ~Z: Open L ~Z: Closed H: Closed H ~Z: Closed L ~Z:Open 15700C-13 CL R1 R2 30 pF 820n 820n Measured Output Value 2.5V 5 pF PALCE22V10Z Family 2.5 V H ~ Z: VOH - 0.5 V L ~ Z: VOL + 0.5 V 2·325 ~AMD TYPICAL Icc CHARACTERISTICS FOR THE PALCE22V10Z-25 Vee = 5.0 V, TA = 25°C 110 90 70 50 30 Icc (mA) 10 0.1 0.Q1 0.1k 1k 10k 100k 1M 10M 30M 50M Frequency (Hz) 15700C-15 ·Percent of product terms used. Icc vs. Frequency Graph for the PALCE22V10Z·25 2·326 PALCE22V10Z·25 AMD~ ENDURANCE CHARACTERISTICS The PALCE22V10Z is manufactured using AMO's advanced Electrically Erasable process. This technology uses an EE cell to replace the fuse link used in bipolar parts. As a result, the device can be erased and reprogrammed-a feature which allows 100% testing at the factory. Endurance Characteristics Symbol tOR N Parameter Test Conditions Min Pattern Data Retention Time Max Storage Temperature Min Unit 10 Years Max Operating Temperature 20 Years Min Reprogramming Cycles Normal Programming Conditions 100 Cycles ROBUSTNESS FEATURES The PALCE22V10Z has some unique features that make it extremely robust, especially when operating in high speed design environments. Input clamping circuitry limits negative overshoot, eliminating the possi- bility of false clocking caused by subsequent ringing. A special noise filter makes the programming circuitry completely insensitive to any positive overshoot that has a pulse width of less than about 100 ns. INPUT/OUTPUT EQUIVALENT SCHEMATICS Vcc I I I I I ESD Input Protection Transition and Detection Clamping I Programming =~~s~~___ _ Typical Input 1 Preload Circuitry Feedback Input Input Transition Detection Typical Output 15700C-16 PALCE22V10Z Family 2-327 ~AMD POWER-UP RESET FOR THE PALCE22V10Z-15 The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. The output state will depend on the programmed pattern. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vee can rise to its steady state, two conditions are Parameter Symbol required to ensure a valid power-up reset. These conditions are: • The Vee rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Description Max Unit tPR Power-Up Reset Time 1000 ns ts Input or Feedback Setup Time twL See Switching Characteristics Clock Width LOW Power Registered Active-Low Output 4Vf --'114--'··- Vcc tPR -7--"~ Clock 15700C-18 Power-Up Reset Waveform 2-328 PALCE22V10Z-15 AMD~ POWER-UP RESET FOR THE PALCE22V10Z-25 The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. The output state will depend on the programmed pattern. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vee can rise to its steady state, four conditions are required to ensure a valid power-up reset. These conditions are: Parameter Symbol tPR • The supply voltage prior to the Vee rise must not exceed Vee off. • The Vcc rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. • If inputs are not switching at the time of power-up, an input transition must take place to assure proper data is set-up in registers or to outputs. Parameter Description Max Unit Power-Up Reset Time 1000 ns Input or Feedback Setup Time ts twL See Switching Characteristics Clock Width LOW Vee Off Supply Voltage Prior to Power-Up mV 100 Vec Power ______4_V{ VccOff Registered Active-Low Output Clock 15700C-19 Power-Up Reset Waveform PALCE22V10Z-25 2-329 ~ AMD ?I 2·330 GL'. Devices ~ PALL V22V1 OZ-25 Low-Voltage, Zero-Power 24-Pin EE CMOS Versatile PAL Device Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • • • • • Low-voltage operation, 3.3 V JEDEC compatible Zero-power CMOS technology - 15 J..LA standby current - 25 ns first-access propagation delay Unused product term disable for reduced power consumption Industrial operating temperature range - Tc =-40°C to +85°C 3.3 V (CMOS) and 5 V (CMOS and TTL)compatible inputs and I/O Electrically-erasable technology provides reconfigurable logic and full testability • • • • • • 10 macrocells programmable as registered or combinatorial, and active high or active low to match application needs Varied product term distribution allows up to 16 product terms per output for complex functions Global asynchronous reset and synchronous preset for Initialization Power-up reset for Initialization and register preload for testability ExtenSive third-party software and programmer support through FusionPLD partners 24-pin SKINNYDIP, 24-pin SOIC and 28-pin PLCC packages save space GENERAL DESCRIPTION The PALLV22V10Z is an advanced PAL device built with low-voltage, zero-power, high-speed, electricallyerasable CMOS technology. It provides user-programmable logic for replacing conventional zero-power CMOS SSI/MSI gates and flip-flops at a reduced chlp count. (see Block Diagram). The OR sum of the products feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial, and active high or active low. The output configuration is determined by two bits controlling two multiplexers in each macrocell. The PALLV22V1 OZ provides high speed at low voltage and zero standby power. At 15 J..LA maximum standby current, the PALLV22V10Z allows battery powered operation for an extended period. AMD's FusionPLD program allows PALLV22V10Z designs to be implemented using a wide variety of popular industry-standard deSign tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuring thatthirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a designer can use a tool that is already installed and familiar. Please refer to the Software Reference Guide to PLD Compliers for certified development systems, and the Programmer Reference Guide for approved programmers. The ZPAL device implements the familiar Boolean logic transfer function, the sum of products. The PAL device is a programmable AND array driving a fixed OR array. The AN D array is programmed to create custom product terms, while the OR array sums selected terms at the outputs. The product terms are connected to the fixed OR array with a varied distribution from 8 t016 across the outputs Publication# 17661 Rev. A Amendment/O L..;1.;..;ss.;..;ue;...;D;;.;;a;,;.;;te..;..:..;..Fe;.;;b.;.;ru;.;;;ary~19;.;;9.;;..3_ _ _ _....J This document contains information on a product under development at Advanced Micro Devices. Inc. The ~~~r:ifh~si~~=:~ ~~~~~ ~~h~~:~~~c:~ this product. AMD reserves the right to change or discontinue 2-331 ~ PRELIMINARY AMD BLOCK DIAGRAM II -111 CLKllo Programmable AND Array (44 x 132) 17661A-1 CONNECTION DIAGRAMS Top View SKINNYDIP/SOIC CLKllo Vee 11 II0g 12 II0s 13 1/07 14 1/05 16 1/04 17 1/03 Is 1/02 Ig 1/01 110 1/00 Note: Pin 1 is marked for orientation_ o ~ NC Clock Ground Input Input/Output No Connect Vee Supply Voltage GND I I/O 2-332 OJ ex> 1/07 24 1/06 23 1/05 NC 1/04 1/03 1/02 22 21 111 17661A-3 17661A-2 PIN DESCRIPTION ClK () ~=d~~~~ 1/06 15 GND PLCC PALLV22V10Z-25 AMD~ PRELIMINARY ORDERING INFORMATION Industrial Products AMD programmable logic products for industrial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of these elements: - PAL LV 22 V 10 Z -25 P I L _______T-' -- -- -FAMILY TYPE PAL = Programmable Array Logic TECHNOLOGY------------~ OPERATING CONDITIONS I = Industrial (-40°C to +85°C) PACKAGE TYPE LV = Low-Voltage V = Versatile P = 24-Pin 300 mil Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) S = 24-Pin Plastic Gull-Wing Small Outline Package (SO 024) NUMBER OF OUTPUTS SPEED NUMBER OF ARRAY INPUTS OUTPUT TYPE -----------~ -25 = 25 ns tPD POWER --------------------------------' Z = Zero Power (15 JlA Icc standby) Valid Combinations PALLV22V10Z-25 PI,JI, SI Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations, and to check on newly released combinations. PALLV22V10Z-25 (IND) 2-333 ~ AMD PRELIMINARY FUNCTIONAL DESCRIPTION The PALLV22V10Z is the low-voltage, zero-power version of the PALCE22V1 O. It has all the architectural features of the PALCE22V10. In addition, the PALLV22V10Z has zero standby power and unused product term disable. The PALLV22V10Z allows the systems engineer to implement the design on-chip, by programming EE cells to configure AND and OR gates within the device, according to the desired logic function. Complex interconnec~ tions between gates, which previously required time-consuming layout, are lifted from the PC board and placed on silicon, where they can be easily modified during prototyping or production. Product terms with all connections opened assume the logical HIGH state; product terms connected to both true and complement of any single input assume the logical LOW state. The PALL V22V1 OZ has 12 inputs and 10 1/0 macrocells. The macrocell (Figure 1) allows one of fou r potential output configurations; registered output or combinatorial 1/0, active high or active low (see Figure 2). The configuration choice is made according to the user's design specification and corresponding programming of the configuration bits 80 - 81. Multiplexer controls are connected to ground (0) through. a programmable bit, selecting the "0" path through the multiplexer. Erasing the bit disconnects the control line from GND and it floats to Vee (1), selecting the "1" path. The device is produced with a EE cell link at each input to the AND gate array, and connections may be selectively removed by applying appropriate voltages to the circuit. Utilizing an easily-implemented programming algorithm, these products can be rapidly programmed to any customized pattern. Variable Input/Output Pin Ratio The PALLV22V10Z has twelve dedicated input lines, and eaeh macrocell output can be an 1/0 pin. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Unused input pins should be tied to Vee or GND. Reg.istered Output Configuration Each macrocell of the PALLV22V1 OZ includes aD-type flip-flop for data storage and synchronization. The flipflop is loaded on the LOW-to-HIGH transition of the clock input. In the registered configuration (81 = 0), the array feedback is from Q of the flip-flop. Combinatorial I/O Configuration Any macrocell can be configured as combinatorial by selecting the multiplexer path that bypasses the flip-flop (81 = 1). In the combinatorial configuration the feedback is from the pin. I/On Sl So 0 0 Output Configuration Registered/Active Low 0 1 Registered/Active High 1 0 Combinatorial/Active Low 1 Combinatorial/Active High o = Programmed EE bit 1 = Erased (charged) EE bit 17661A-4 Figure 1. Output Logic Macrocell 2-334 PALLV22V10Z-25 AMD~ PRELIMINARY Combinatorial/Active Low Registered/Active Low Combinatorial/Active High Registered/Active High 17661A-5 Figure 2. Macrocell Configuration Options Programmable Three-State Outputs Each output has a three-state output buffer with threestate control. A product term controls the buffer, allowing enable and disable to be a function of any product of device inputs or output feedback. The combinatorial output provides a bidirectional I/O pin, and may be configured as a dedicated input if the buffer is always disabled. Programmable Output Polarity The polarity of each macrocell output can be active high or active low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is controlled by programmable bit So in the output macrocell, and affects both registered and combinatorial outputs. Selection is automatic, based on the design specification and pin definitions. If the pin definition and output equation have the same polarity, the output is programmed to be active high (So = 1). Preset/Reset For initialization, the PALLV22V10Z has additional Preset and Reset product terms. These terms are connected to all registered outputs. When the Synchronous Preset (SP) product term is asserted high, the output registers will be loaded with a HIGH on the next LOW-toHIGH clock transition. When the Asynchronous Reset (AR) product term is asserted high, the output registers will be immediately loaded with a LOW independent of the clock. Note that preset and reset control the flip-flop, not the output pin. The output level is determined by the output polarity selected. Benefits of Lower Operating Voltage The PALL V22V1 OZ has an operating voltage range of 3.0 V to 3.6 V. Low voltage allows for lower operating power consumption, longer battery life, and/or smaller batteries for notebook applications. Because power is proportional to the square of the voltage, reduction of the supply voltage from 5.0 V to 3.3 V significantly reduces power consumption. This directly translates to longer battery life for portable applications. Lower power consumption can also be used to reduce the size and weight of the battery. Thus, 3.3 V designs facilitate a reduction in the form factor. A lower operating voltage results in a reduction of I/O voltage swings. This reduces noise generation and provides a less hostile environment for board design. Lower operating voltage also reduces electromagnetic radiation noise and makes obtaining FCC approval easier. PALLV22V10Z·25 2·335 ~ AMD PRELIMINARY zero-Standby Power Mode Register Preload The PALLV22V10Z features a zero-standby power mode. When none of the inputs switch for an extended period (typically 30 ns), the PALLV22V10Z will go into standby mode, shutting down most of its internal circuitry. The current will go to almost zero (Icc < 15 JJ.A). The outputs will maintain the states held before the device went into the standby mode. The registers on the PALLV22V10Z can be preloaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. If a macrocell is used in registered mode, switching pin ClK/lo will not affect standby mode status for that macrocell. If a macrocell is used in combinatorial mode, switching pin CLK/lo will affect standby mode status for that macrocell. This feature reduces dynamic Icc proportional to the number of registered macrocells used. If all macrocells are used as registers, and only ClK/lo is switching, the device will not be in standby mode but dynamic Icc will typically be <2 mA. This is because only the ClK/lo buffer will draw current. When any input switches, the internal circuitry is fully enabled and power consumption returns to normal. This feature results in considerable power savings for operation at low to medium frequencies. Product-Term Disable On a programmed PALLV22V10Z, any product terms that are not used are disabled. Power is cut off from these product terms so that they do not draw current. Product-term disabling" results in considerable power savings. This savings is greater at the higher frequencies. Further hints on minimizing power consumption can be found in the Application Note "Minimizing Power Consumption with Zero-Power PLOs." 3.3 V (CMOS) and 5 V (CMOS and TTL)Compatible Inputs and 1/0 Input voltages can be at TTL levels without the device drawing more current than true 3.3 V CMOS levels. Additionally, the PALLV22V1 OZ can be driven with true 5 V CMOS levels due to special input and I/O buffer circuitry. Security Bit After programming and verification, a PALLV22V10Z design can be secured by programming the security EE bit. Once programmed, this bit defeats readback of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. When the security bit is programmed, the array will read as if every bit is erased, and preload will be disabled. The bit can only be erased in conjunction with erasure of the entire pattern. Programming and Erasing The PAlLV22V10Z can be programmed on standard logiC programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No'special erase operation is required. Quality and Testability The PAlLV22V10Z offers a very high level of built-in quality. The erasability of the CMOS PALLV22V10Z allows direct testing of the device array to guarantee 100% programming and functional yields. Technology The high-speed PALLV22V10Z is fabricated with AMO's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be 3.3 and 5 V device compatible. This technology provides strong input-clamp diodes, output slew-rate control, and a grounded substrate for clean switching. Power-Up Reset All flip-flops power-up to a logic LOW for predictable system initialization. Outputs of the PALLV22V10Z will depend on the programmed output polarity. The Vcc rise must be monotonic and the reset delay time is 1000 ns maximum. 2-336 PALLV22V10Z-25 AMD~ PRELIMINARY LOGIC DIAGRAM SKINNYDIP (PLCC) Pinouts CLKIIO W 1°1l_ _ _ _ _ 5=t1§~~ARQWfll1~ J' 21108 ~~ 20 mI1:2}--Ci!l:::::t:t:t~U +t=f::!±i::i:f±I:::t:f±I::t:f±I=t::t:t:I=t::t:t:I=t:~=t:~=t:~=::!I2--ttlr--i ~ (3) 12 (26) -= m(4) 49 16 ill (9) 110 18 m(11) -@IIO O 130 19 (17) 1m- (12) - P----"" 3 4 7 8 11 12 15 16 19 20 23 24 27 28 31 32 35 36 39 40 ..Ii3II 11 (16) 43 17661A-6 PALLV22V10Z·25 2·337 ~ AMD PRE LIM I N A R Y ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature .......... -65°C to +150°C Industrial (I) Devices Ambient Temperature with Power Applied ................ -55°C to +125°C Operating Case Temperature (Te) ............ -4O°C to +85°C Supply Voltage with Respect to Ground .................... -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ........... +3.0 V to +3.6 V DC Input Voltage .............. -0.5'V to +5.5 V DC Output or 1/0 Pin Voltage ..... -0.5 V to +5.5 V Operating Ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage .................. 2001 V Latchup Current (Te =-4O°C to +85°C) .... 100 mA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over INDUSTRIAL operating ranges unless otherwise specified PRELIMINARY Parameter Symbol VOH VOL Parameter Description Test Conditions Output HIGH Voltage VIN = VIH or Vil IOH =-2 rnA 2.4 V Vee = Min IOH = 100 IlA Vee0.1 V Output LOW Voltage VIN = VIH or Vil Min Max Unit IOl = 2 rnA 0.4 V IOl = 100 IlA 0.1 V VIH Input HIGH Voltage Guaranteed Input Logical HIGH 2.0 Voltage for all Inputs (Note 1) 5.5 V VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) 0.8 V IIH Input HIGH Leakage Current VIN = Vee, Vee = Max 10 jlA III Input LOW Leakage Current VIN = 0 V, Vee = Max -10 jlA Off-State Output Leakage VOUT = Vee, Vee = Max 10 jlA Current HIGH VIN = VIH or Vil (Note 2) Off-State Output Leakage VOUT = 0 V, Vee = Max -10 jlA Current LOW VIN = VIH or Vil (Note 2) -75 rnA IOZH IOZl Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) Icc Supply Current Outputs Open (lOUT = 0 rnA) f = 0 MHz -15 15 jlA Vec= Max f = 25 MHz 55 rnA Notes: 1. These are absolute values with respect to device· ground and all overshoots due to system or tester noise are inCluded. 2. 110 pin leakage is the worst case of hL and IOZL (or hH and IOZH). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-338 PALLV22V10Z-25 (IND) AMO~ PRELIMINARY CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Typ Parameter Description Test Condition Input Capacitance VIN = 2.0 V Vee = 3.3 V 5 Output Capacitance VOUT .. 2.0 V TA = 25°C f = 1 MHz 8 Unit pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over INDUSTRIAL operating ranges (Note 2) PRELIMINARY Parameter Symbol Min Parameter Description tPD Input or Feedback to Combinatorial Output (Note 3) ts Setup Time from Input, Feedback or SP to Clock 15 tH Hold Time 0 Max Unit 25 ns ns ns teo Clock to Output 15 ns tAR Asynchronous Reset to Registered Output 25 ns tARw Asynchronous Reset Width 25 ns tARR Asynchronous Reset Recovery Time 25 ns tSPR Synchronous Preset Recovery Time 25 ns 10 ns tWL Clock Width tWH fMAX LOW HIGH Maximum Frequency (Note 4) External Feedback I 1/(ts + teo) Internal Feedback (feNT) No Feedback I 1/(twH + tWl) 10 ns 33.3 MHz 35.7 MHz 50 MHz tEA Input to Output Enable Using Product Term Control 25 ns tER Input to Output Disable Using Product Term Control 25 ns Notes: 2. See Switching Test Circuit for test conditions. 3. This parameter is tested in Standby Mode. When the device is not in Standby Mode, the tpo may be slightly faster. 4. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALLV22V10Z·25 (INO) 2·339 ~ PRELIMINARY AMD SWITCHING WAVEFORMS Input or Feedback Combinatorial Output -~-VT 17661A-7 17661A-8 Registered Output Combinatorial Output Input tER Clock tEA Output ------'-J.-I.-1 17661A-10 17661A-9 Input to Output Disable/Enable Clock Width Input Input Asserting Asynchronous Reset Asserting=f Synchronous Preset _ _ _ __ t5 Registered Output _ _ _"--11~...¥...x Clock tARR Clock VT Registered Output ------...1 ------~xXxx1tco VT ,-----"--_____ 17661A-11 Asynchronous Reset Synchronous Preset Notes: 1. Vr = 1.5 V for Input Signals and 1.65 V for Output Signals. 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns - 5 ns typical. 2-340 17661A-12 PALLV22V10Z-25 AMD~ PRELIMINARY KEY TO SWITCHING WAVEFORMS WAVEFORM INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L /77// May Change from L to H Will be Changing from L to H XXXXXX Don't Care, Any Change Permitted Changing, State Unknown J])(K Does Not Apply Center Line is HighImpedance "Off" State KSOOOO10-PAL SWITCHING TEST CIRCUIT 3.3 V Output O - -.....- - -....-~J Test Point 17661A-13 Measured Specification tPD, teo tEA CL R1 R2 Output Value H: Closed L: Open 30 pF 1.6Kn 1.6Kn 1.65 V H ~Z: Closed 5 pF Closed Closed Z~ Z~ Z~ tER 52 51 H: Open L: Closed H ~Z: Open L ~Z: Closed Z ~ 1.65 V L~Z:Open PALLV22V10Z-25 H ~ Z: VOH - 0.5 V L ~Z: VOL + 0.5 V 2-341 ~ PRELIMINARY AMD ENDURANCE CHARACTERISTICS The PALLV22V10Z is manufactured using AMO's advanced Electrically Erasable process. This technology uses an EE cell to replace the fuse link used in bipolar Symbol parts. As a result, the device can be erased and reprogrammed-a feature which allows 100% testing at the factory. Parameter Test Conditions tOR Min Pattern Data Retention Time Max Storage Temperature Min Unit 10 Years Max Operating Temperature 20 Years N Min Reprogramming Cycles Normal Programming Conditions 100 Cycles ROBUSTNESS FEATURES The PALLV22V10Z-25 has some unique features that make it extremely robust, especially when operating in high speed design environments. Input clamping circuitry limits negative overshoot, eliminating the possi- bility of false clocking caused by subsequent ringing. A special noise filter makes the programming circuitry completely insensitive to any· positive overshoot that has a pulse width of less than about 100 ns. INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee I I I I I ESD Input Protection Transition and Detection Clamping IProgramming '=' ~~s~~___ _ Typical Input 1 Preload Circuitry Feedback Input Input Transition Detection Typical Output 17661A-16 2-342 PALLV22V10Z-25 AMD~ PRELIMINARY POWER-UP RESET The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. The output state will depend on the programmed pattern. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vcc can rise to its steady state, two conditions are reParameter Symbol quired to ensure a valid power-up reset. These conditions are: • The Vcc rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Description Max Unit tPR Power-Up Reset Time 1000 ns ts Input or Feedback Setup Time See Switching Characteristics Clock Width LOW tWL Vcc Power Registered Active-Low Output Clock 2.7VT '~...L..I-_- tPR ~ ZI st tWL 17661A-17 Power-Up Reset Waveform PALLV22V10Z-25 2-343 ~ AMD ~AL'· Devices . ~~ 2-344 _ ~ COM'L: H-15/25 Advanced Micro Devices PALCE24V10H-15/25 EE CMOS 28-Pin Universal Programmable Array Logic DISTINCTIVE CHARACTERISTICS • Electrically erasable CMOS technology provides reconfigurable logic and full testability &I High speed CMOS technology - 15 ns propagation delay for "-15" version - 25 ns propagation delay for "-25" version • Outputs Individually programmable as registered or combinatorial • Programmable output polarity • • Iii • • • Programmable enable/disable control Preloadable output registers for testability Automatic register reset on power-up Cost-effective 28-pin plastic SKINNYDIP and PLCC packages Extensive third-party support through FusionPLD partners Fully tested for 100% programming and functional yields and high reliability GENERAL DESCRIPTION The PALCE24V1 0 is an advanced PAL device built with low-power, high-speed, electrically-erasable CMOS technology. Its macrocells provide a universal device architecture. grammed as registered or combinatorial with an activehigh or active-low output. The output configuration is determined by two global bits and one local bit controlling four multiplexers in each macrocell. The PALCE24V10 utilizes the familiar sum-of-products (AND/OR) architecture that allows users to implement complex logic functions easily and efficiently. Multiple levels of combinatorial logic can always be reduced to sum-of-products form, taking advantage of the very wide input gates available in PAL devices. The equations are programmed into the device through floatinggate cells in the AND logic array that can be erased electrically. AMD's FusionPLD program allows PALCE24V10 designs to be implemented using a wide a variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuring that third-party tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces design time since a designer can use a tool that is already installed and familiar. Please refer to the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. The fixed OR array allows up to eight data product terms per output for logic functions. The sum of these products feeds the output macrocell. Each macrocell can be pro- CLK/lo BLOCK DIAGRAM OElI,3 1100 Publication# 12222 Rev. E Issue Date: Juno 1993 1/0, AmendmentlO 1/05 1/06 lIDs II0g 12222E-l 2-345 ~ AMD CONNECTION DIAGRAMS Top View PLCC SKINNYDIP CLKllo 115 11 114 12 \/09 \/Os \4 14 \/07 \5 15 \/Ce Vee Vee \6 \7 \s \9 \/05 16 GND 17 \/04 Is \/0:3 19 \/02 110 \/01 111 \/00 112 OE/113 Vee Supply Voltage 2-346 21 20 GND \/°4 T"'" C\I (') 0 .C\I ~~~~OOO -I~ ::. ::. ::. 12222E-3 PIN DESIGNATIONS OE 22 \/°3 o Note: Pin 1 is marked for orientation. Clock Ground Input InpuVOutput Output Enable \/Os \/°7 \/°6 \/05 23 8 9 -- 12222E-2 ClK GND I I/O 25 24 PALCE24V10H-15/25 AMD~ ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL FAMILYTYPE CE 24 V10H-15P C -_T----I-~ -- PAL= Programmable Array Logic 1L TECHNOLOGY - - CE = CMOS Electrically Erasable OPERATING CONDITIONS C = Commercial (DOC to +75°C) PACKAGE TYPE P = 28-Pin 300 mil Plastic SKINNYDIP (PD3028) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) NUMBER OF - - - - - - - - - - ' ARRAY INPUTS OUTPUT TYPE V = Versatile NUMBER OF FLIP-FLOPS - - - - - - - - ' POWER-------------~ H = Half Power (115 mA Icc) SPEED -15 15 ns tPD -25 = 25 ns tPD Valid Combinations PALCE24V10H-15 PALCE24V10H-25 I I PC,JC Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE24V10H-15/25 (Com'l) 2-347 ~ AMD FUNCTIONAL DESCRIPTION The PALCE24V10 is a universal PAL device. It has ten independently configurable macrocells (MCo .. MCg). Each macrocell can be configured as a registered output, combinatorial output, combinatorial 110, or dedicated input. The programming matrix implements a programmable AND logic array, which drives a fixed OR logic array. Buffers for device inputs have complementary outputs to provide user-programmable input signal polarity. Pins 1 and 15 serve either as array inputs or as clock (ClK) and output enable (OE) for all flip-flops. nected) assume the logical HIGH state and product terms with both true and complement of any input signal connected assume a logical LOW state. The programmable functions on the PAlCE24V10 are automatically configured from the user's design specification, which can be in a number of formats. The design specification is processed by development software to verify the design and create a programming file. This file, once downloaded to a programmer, configures the device according to the user's desired function. Unused input pins should be tied directly to Vee or GND. Product terms with all bits unprogrammed (discon- IIOx ~----~~D Q~----~ o·t-------+-...., Macrocells MC1 - MCa Vee IIOx ~---~~D Q~----... Ot--------+-....... Macrocells MCa and MCg From Adjacent Pin 12222E-4 PALCE24V10 Macrocell 2-348 PALCE24V10H-15/25 AMD Configuration Options Each macrocell can be configured as one of the following: registered output, combinatorial output, combinatorial 110 or dedicated input. In the registered ~tput configuration, the output buffer is enabled by the OE pin. In the combinatorial configuration, the buffer is either controlled by a product term or always enabled. In the dedicated input configuration, the buffer is always disabled. The macrocell configurations are controlled by the configuration control word. It contains 2 global bits (SGO and SG1) and 20 local bits (SlOo through SLOg and Sl1 0 through SL19). SGO determines whether registers will be allowed. SG1 determineswhetherthe output buffer is user-controlled or in a fixed state. Within each macrocell, SlOx, in conjunction with SG1 , selects the configuration of the macrocell and Sl1 xsets the output as either active low or active high. The configuration bits work by acting as control inputs forthe multiplexers in the macrocell. There are four multiplexers: a product term input, an enable select, an output select, and a feedback select multiplexer. SG1 and SlOx are the control signals for all four multiplexers. In MCo and MCg, SGO is added on the feedback multiplexer. l1 cause the macrocell is a dedicated output, the feedback is not used. Dedicated Input in a Non-Registered Device The control bit settings are SGO = 1, SG1 = 0 and SlOx = '1. The output buffer is disabled. The feedback signal is the I/O pin. Combinatorial I/O in a Non-Registered Device The control settings are SGO = 1 , SG 1 = 1 , and SlOx = 1. Only seven product terms are available to the OR gate. The eighth product term is used to enable the output buffer. The signal at the 110 pin is fed back to the AND array via the feedback multiplexer. This allows the pin to be used as an input. Combinatorial I/O in a Registered Device The control bit settings are SGO=0,SG1 =1 and SlOx =1. Only seven product terms are available to the OR gate. The eighth product term is used as the output enable. The feedback signal is the corresponding I/O signal. Table 1. Macrocell Configurations These configurations are summarized in table 1 and illustrated in figure 2. SGO If the PAlCE24 V1 0 is configured as a combinatorial device, the ClK and OE pins are available as inputs to the arraL!!. the device is configured with registers, the ClK and OE pins cannot be used as data inputs. 0 SG1 SLOx Cell Configuration Device has registers 0 1 0 1 1 Registered Output Combinatorial 1/0 Device has no registers Registered Output Configuration The control bit settings are SGO = 0, SG1 = 1 and SlOx = o. There is only one registered configuration. All eight product terms are available as inputs to the OR gate. Data polarity is determined by Sl1 x. Sl1 x is an input to the exclusive-OR gate which is the D input to the flipflop. Sl1x is programmed as 1 for inverted output or 0 for non-inverted output. The flip-flop is loaded on the lOW-to-HIGH transition of ClK. The feedback path is from Q on the register. The output buffer is enabled by OE. Combinatorial Configurations The PAlCE24V10 has three combinatorial output configurations: dedicated output in a non-registered device, I/O in a non-registered device and I/O in a registered device. Dedicated Output in a Non-Registered Device 1 0 0 1 0 1 Combinatorial Output Dedicated Input 1 1 1 Combinatorial 1/0 Programmable Output Polarity The polarity of each macrocell output can be active high or active low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is made through a programmable bit Sl1 x which controls an exclusive-OR gate at the output of the AND/OR logic. The output is active high if SL1x is a 0 and active low if Sl1 x is a 1. The control settings are SGO = 1, SG1 = 0, and SlOx = O. All eight product terms are available to the OR gate. Be- PAlCE24V10H-15/25 2-349 ~ AMD OE ____________________ OE--------------------~ a a Q Q Registered Active Low ~ Registered Active High Combinatorial 110 Active Low Combinatorial 1/0 Active High Combinatorial Output Active Low Combinatorial Output Active High Dedicated Input 12222E-5 Figure 2. Macrocell Configurations 2-350 PALCE24V10H-15/25 AMD~ Power-Up Reset Programming and EraSing All flip-flops power up to a logic LOW for predictable system initialization. Outputs of the PALCE24V10 depend on whether they are selected as registered or combinatorial.lf registered is selected, the output will be HIGH. If combinatorial is selected, the output will be a function of the logic. The PALCE24V10 can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erase operation is required. Quality and Testability Register Preload The register on the PALCE24V10 Series can be preloaded from the output pins to facilitate functional testing of complex state machine deSigns. This feature allows direct loading of arbitrary states, making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. Security Bit A security bit is provided on the PALCE24V10 as a deterrent to unauthorized copying of the array configuration patterns. Once programmed, this bit defeats readback of the programmed pattern by a device programmer, securing proprietary designs from competitors. However, programming and verification are also defeated by the security bit. The bit can only be erased in conjunction with the array during an erase cycle. The PAL24V1 0 offers a very high level of built-in quality. The erasability if the device provides a direct means of verifying performance of all the AC and DC parameters. In addition, is verifies complete programmability and functionality of this device to yield the highest programming yields and post-programming function yields in the industry. Technology The high-speed PALCE24V10 is fabricated with AMD's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with TTL devices. This technology provides strong inputclamp diodes, output slew-rate control, and a grounded substrate for clean switching. Electronic Signature Word An electronic signature word is provided in the PALCE24V10. It consists of 64 bits of programmable memory that can contain any user-defined data. The signature data is always available to the user independent of the security bit. PALCE24 V1 OH-15/25 2-351 ~ AMD LOGIC DIAGRAM ~M. '~-i~~~~+H~++HH~+H~++HH~+H~++HH++HH~+HH--------------------' I, • - ~'0l- ~ _ _ _-..., 1,9 vcclil-- 12222E-6 2-352 PALe E24 V1 OH-15/25 AMD . LOGIC DIAGRAM (continued) " " " . I 11 . :II " .. .. l - <1 l@-t II:!!}- 'IT r: ~~h .... ~~~ .~ ",: ~ >tP ~ 'IT >---t'0 ~ ",: >tP ~ ..... " VCe SGJ .• Sle~ ":; .. >fl? ~ ... - 'r' ~ en ~ .0 e, co ~ l ... >t1? - o 0 ~l CO 10 rt 17 a 0 @ -+- ~ " ,. .,. -S w, n 0' DC ... Ii= ~,e 0' l ~ -Gl 10, r;; .r:~ . ".... '1:-. ~~ .. SlO~ so,] ,e .. 20 " r D ".... .;.. l · Slo~ so,J '0 r: ~ ~ · · ;t1 ... ---;rnJ ~~h ~ l - ~ >t1? ~ D . ~l a 10 f-t .. 10, a . m " 0 .n .5 12222E-6 (concluded) PALCE24V10H-15/25 2-353 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Ambient Temperature with Power Applied ............. -55°C to + 125°C Commercial (C) Devices Temperature (TA) Operating in Free Air ....................... O°C to +75°C Supply Voltage with Respect to Ground ............ -0.5 V to + 7.0 V Supply Voltage (Vee) with Respect to Ground ........ +4.75 V to +5.25 V DC Input Voltage ........... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or I/O Pin Voltage ............ -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Latchup Current (TA = O°C to +75°C) .................... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise speciTIe d Parameter Symbol Test Conditions Parameter Description Min Max 2.4 Unit VOH Output HIGH Voltage IOH =-3.2 rnA Vee = Min VOL Output LOW Voltage IOl = 24 rnA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) hH Input HIGH Leakage Current VIN = 5.25 V, Vee = Max (Note 2) hL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) -10 ~ IOZH Off-State Output Leakage Current HIGH VOUT = 5.25 V, Vee = Max VIN = VIH or VIL (Note 2) 10 IlA IOZl Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or VIL (Note 2) -10 IlA Ise Output Short-Circuit Current Vee = Max VOUT = 0.5 V (Note 3) -150 rnA lee Supply Current Outputs Open (lOUT = 0 rnA) Vee = Max, f = 15 MHz 115 rnA VIN = VIH or Vil V 0.5 VIN = VIH or Vil 2.0 -30 V V 0.8 V 10 ~ Notes: 1. These are absolute values with respect to devic~ ground and all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of hL and lozL (or I,H and IOZH ). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-354 PALCE24V10H-15/25 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Descriptions Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance VOUT= 2.0 V If = 1 MHz I Vee = 5.0 V, TA = 25°C, Typ Unit 5 pF 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol tPD -15 Parameter Description Min Input or Feedback to Combinatorial Output 15 ts Setup Time from Input or Feedback to Clock 10 tH Hold Time teo twH Max 25 ns 6 HIGH External Feedback 1 1/(ts+tcO) Unit ns ns 0 10 LOW Clock Width Min 12 0 Clock to Output tWL -25 Max 12 8 ns ns 6 8 ns 50 41.6 MHz 66 50 MHz fMAX Maximum Frequency (Note 3) tpzx OE to Output Enable (Note 3) 15 20 ns tpxz OE to Output Disable (Note 3) 15 20 ns tEA Input to Output Enable Using Product Term Control (Note 3) 15 25 ns tER Input to Output Disable Using Product Term Control (Note 3) 15 25 ns Internal Feedback (feNT) No Feedback 1 1/(twH+twL) 83.3 62.5 MHz Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. PALCE24V10H-15/25 (Com'l) 2-355 ~ AMD SWITCHING WAVEFORMS Input or Feedback _ _ _-'" _ _ _ _ _--' Input or Feedback Combinatorial Output -~-VT Clock VT __________ Registered Output _ _ _ _ _ _ __ ~tco VT 12222E-7 12222E-8 Combinatorial Output Input Registered Output _ _ _ _--J tER tpzx tEA VT Output VT Output 12222E-9 12222E-10 OE to Output Disable/Enable Input to Output Disable/Enable Clock 12222E-11 Clock Width Notes: 1. Vr = 1.5 V 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns-5 ns typical. 2-356 VT PALCE24V10H-15/25 AMD~ KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L 17171 May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State WAVEFORM xxxxxx KSOOOO10-PAL SWITCHING TEST CIRCUIT Output ( } - - - - 4 I t - - - - - -. . 12222E-12 Specification CL S1 tPD, teo Closed tpzx, tEA Z~ Z~ tpXZ,tER H: Open L: Closed H ~Z: Open R1 R2 Measured Output Value 1.S V SO pF 200n S pF L~Z:Closed PALCE24V10H-15/25 390n 1.S V H ~Z: VOH-O.S V L~Z: Va.. + O.S V 2-357 ~ AMD ENDURANCE CHARACTERISTICS parts. As a result, the device can be erased and reprogrammed-a feature which allows 100% testing at the factory. The PALCE24V10 is manufactured using AMD's advanced electrically erasable process. This technology uses an EE cell to replace the fuse link used in bipolar Endurance Characteristics Symbol Parameter Test Conditions tOR Min Pattern Data Retention Time N Min Reprogramming Cycles Max Storage Temperature Min 10 Unit . Years Max Operating Temperature 20 100 Cycles Normal Programming Conditions Years INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee ESD ProgramNerify Protection Circuitry - . Typical Input Vee Preload Circuitry Feedback Input Typical Output 12222E-14 2-358 PALCE24V10H-15/25 AMD~ POWER-UP RESET The PALCE24V10 has been designed with the capability to reset during system power-up. Following powerup, all flip-flops will be reset to LOW. The output state will be HIGH independent of the logic polarity. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset Parameter Symbol and the wide range of ways Vcc can rise to its steady state, two conditions are required to insure a valid power-up reset. These conditions are: • • The Vcc rise must be monotonic. Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Min Parameter Description Power-Up Reset Time tPR ts Input or Feedback Setup Time twL Registered Output Clock- 4? Unit ns See Switching Characteristics Clock Width LOW Power Max 1000 Vee -------I~IIIIII._-- tPR J 71 ~ tWL 12222E-15 Power-Up Reset Waveform PALCE24V10H-15/25 2-359 ~ COM'L: H-15/20 _ Advanced Micro Devices PALCE26V12H-15/20 28-Pin EE CMOS Versatile PAL Device DISTINCTIVE CHARACTERISTICS • • • • • • 28-pln versatile PAL programmable logic device architecture Electrically erasable CMOS technology provides half power (only 105 rnA) at high speed (15 ns propagation delay) 14 dedicated Inputs and 121nput/output macrocelis for architectural flexibility Macrocells can be registered or combinatorial, and active high or active low Varied product term distribution allows up to 16 product terms per output Two clock Inputs for Independent functions • • • • • Global asynchronous reset and synchronous preset for Initialization Register preload for testability and built-In register reset on power-up Space-efficient 28-pln SKINNYDIP and PLCC packages Center Vee and GND pins to Improve signal characteristics Extensive third-party software and programmer support through FusionPLD partners GENERAL DESCRIPTION The PALCE26V12 is a 28-pin version of the popular PAL22V10 architecture. Built with low-power, highspeed, electrically-erasable CMOS technology, the PALCE26V12 offers many unique advantages. Device logic is automatically configured according to the user's design specification. Design is simplified by design software, allowing automatic creation of a programming file based on Boolean or state equations. The software can also be used to verify the design and can provide test vectors for the programmed device. The PALCE26V12 utilizes the familiar sum-of-products (AND/OR) architecture that allows users to implement complex logic functions easily and efficiently. Multiple levels of combinatorial logic can always be reduced to sum-of-products form, taking advantage of the very wide' input gates available in PAL devices. The functions are programmed into the device through electrically-erasable floating-gate cells in the AND logic array and the macrocells. In the unprogrammed state, all AND product terms float HIGH. If both true and complement of any input are connected, the term will be permanently LOW. 2-360 The product terms are connected to the fixed OR array with a varied distribution from 8 to 16 across the outputs (see Block Diagram). The OR sum of the products feeds the output macrocell. Each macrocell can be programmed as registered or combinatorial, active high or active low, with registered 1/0 possible. The flip-flop can be clocked by one of two clock inputs. The output configuration is determined by four bits controlling three multiplexers in each macrocell. AMD's FusionPLD program allows PALCE26V12 designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide accurate, quality support. By ensuring that thirdparty tools are available, costs are lowered because a designer does not have to buy a complete set of new tools for each device. The FusionPLD program also greatly reduces deSign time since a designer can use a tool that is already installed and familiar. Please refer to the PLD Software Reference Guide for certified development systems and the Programmer Reference Guide for approved programmers. Publication# 16072 Rev. C AmendmentlO Issue Date: June 1993 AMD~ BLOCK DIAGRAM 110 0 VOl 00 2 1104 1103 110 5 VO e 110 7 1106 V0 11 110 10 1109 16072C-1 CONNECTION DIAGRAMS Top View SKINNYDIP CLK1/10 11 12 CLK2I13 PLCC ~ ~ 1/011 ....J () 1/010 ~ .= '::i ....J () 0 .2 ~ ~ II0g 14 II0a 15 1/07 Vee 1/06 16 GND 17 1/05 18 1104 Ig 1/03 110 1/02 111 1101 112 1/00 Note: Pin 1 is marked for orientation. ~ (Y) 113 14 25 1109 15 VCC 24 23 1/07 Is 22 I/0s 17 21 GND Is 20 1/05 1/04 19 0 1S072C-2 I/0s .= (\j 0 (\j (Y) g § g g 1S072C-3 PIN DESCRIPTION ClK =Cloek GND = Ground = Input liD = InpuUOutput Vee = Supply Voltage PAlCE26V12H-15/20 2-361 ~ AMD ORDERING INFORMATION Commercial Products AMD commercial programmable logic products are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL FAMILY TYPE PAL = Programmable Array Logic T CE -- TECHNOLOGY CE = CMOS Electrically Erasable 26 V 12 H -15 -~ -:- -- -- P C 14 1L OPllONAL PROCESSING Blank = Standard Processing PROGRAMMING DESIGNATOR 14 = First Revision (May require programmer update) ' - - - OPERATING CONDITIONS C = Commercial (O°C to +75°C) NUMBER OF ARRAY INPUTS OUTPUT TYPE V= Versatile PACKAGE TYPE P = 28-Pin 300 mil Plastic SKINNYDIP (PD3028) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) NUMBER OF OUTPUTS POWER H= Half Power (105 mA Icc) SPEED -15=15nstPD -20 = 20 ns tPD Valid Combinations Valid Combinations PALCE26V12H-15 PALCE26V12H-20 2-362 I PC JC I I • 14 Valid Combinations list configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE26V12H-15/20 (Com'l) AMD~ FUNCTIONAL DESCRIPTION The PALCE26V12 has fourteen dedicated input lines, two of which can be used as clock inputs. Unused inputs should be tied directly to ground or Vcc. Buffers for device inputs and feedbacks have both true and complementary outputs to provide user-selectable signal polarity. The inputs drive a programmable AND logic array, which feeds a fixed OR logic array. The OR gates feed the twelve liD macrocells (see Figure 1). The macrocell allows one of eight potential output configurations; registered or combinatorial, active high or active low, with register or 110 pin feedback (see Figure 2). In addition, registered configurations can be clocked by either of the two clock inputs. The configuration choice is made according to the user's design specification and corresponding programming of the configuration bits 80-83 (see Table 1). Multiplexer controls initially float to Vcc (1) through a programmable cell, selecting the "1" path through the multiplexer. Programming the cell connects the control line to GND (0), selecting the "0" path. Table 1. Macrocell Configuration Table S3 1 Sl 0 So Output Configuration 0 Registered Output and Feedback, Active Low 1 0 1 Registered Output and Feedback, Active High 1 1 0 0 0 1 1 0 0 1 0 1 0 1 0 0 1 1 S2 Clock Input 1 CLK1/lo 0 CLKVb Combinatorial I/O, Active Low Combinatorial I/O, Active High OE-1~----------------------~ • When 53 = 1 (unprogrammed) the feedback is selected by 5,. When 53 ~ 0 (programmed). the feedback is the opposite of that sefected by 5 l' 16072C-4 Figure 1. PALCE26V12 Macrocell Registered or Combinatorial Each macrocell of the PALCE26V12 includes aD-type flip-flop for data storage and synchronization. The flipflop is loaded on the LOW-to-HIGH edge of the selected clock input. Any macrocell can be configured as combinatorial by selecting a multiplexer path that bypasses the flip-flop. Bypass is controlled by bit 81. Programmable Clock The clock input for any flip-flop can be selected to be from either pin 1 or pin 4. A 2:1 multiplexer controlled by bit 82 determines the clock input. Registered I/O, Active Low Registered I/O, Active High Programmable Feedback Combinatorial Output, Registered Feedback, Active Low A 2:1 multiplexer allows the user to determine whether the macrocell feedback comes from· the flip-flop or from the liD pin, independent of whether the output is registered or combinatorial. Thus, registered outputs may have internal register feedback for higher speed (fMAX internal), or 1/0 feedback for use of the pin as a direct input (fMAX external). Combinatorial outputs may have liD feedback, either for use of the signal in other equations or for use as another direct input, or register feedback. Combinatorial Output, Registered Feedback, Active High 1 = Unprogrammed EE bit 0= Programmed EE bit The feedback multiplexer is controlled by the same bit (81) that controls whether the output is registered or combinatorial, as on the 22Vl 0, with an additional control bit (83) that allows the alternative feedback path to be selected. When 83 = 1, 81 selects register feedback for registered outputs (81 = 0) and liD feedback for combinatorial outputs (81 = 1). When S3 = 0, the opposite is selected: liD feedback for registered outputs and register feedback for combinatorial outputs. PALCE26V12H-15/20 2-363 ~ AMD Programmable Enable and 1/0 Register Preload Each macrocell has a three-state output buffer controlled by an individual product term. Enable and disable can be a function of any combination of device inputs or feedback. The macrocell provides a bidirectional 1/0 pin if 1/0 feedback is selected, and may be configured as a dedicated input if the buffer is always disabled. This is accomplished by connecting all inputs to the enable term, forcing the AND of the complemented inputs to be always LOW. To permanently enable the outputs, all inputs are left disconnected from the term (the unprogrammed state). The register on the PALCE26V12 can be preloaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, thereby making it unnecessary to cycle through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. Programmable Output" Polarity " The polarity of each macrocell output can be active high or active low, either to match output signal needs or to reduce product terms. Programmable polarity allows Boolean expressions to be written in their most compact form (true or inverted), and the output can still be of the desired polarity. It can also save "DeMorganizing" efforts. Selection is controlled by programmable bit So in the output macrocell, and affects both registered and combinatorial outputs. Selection is automatic, based on the design specification and pin definitions. If the pin definition and output equation have the same polarity, the output is programmed to be active high. Security Bit After programming and verification, a PALCE26V12 design can be secured by programming the security bit. Once programmed, this bit defeats read back of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. Programming the security bit disables preload, and the array will read as if every bit is disconnected. The security bit can only be erased in conjunction with erasure of the entire pattern. Programming and Erasing The PALCE26V12 can be programmed on standard logiC programmers. It also may be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erase operation is required. Preset/Reset Quality and Testability For initialization, the PALCE26V12 has additional Preset and Reset product terms. These terms are connected to all registered outputs. When the Synchronous Preset (SP) product term is asserted high, the output registers will be loaded with a HIGH or the next LOW-toHIGH clock transition. When the Asynchronous Reset (AR) product term is asserted high, the output registers will be immediately loaded with a LOW independent of the clock. The PALCE26V12 offers a very high level of built-in quality. The erasability of the device provides a means of verifying performance of all AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to provide the highest programming yields and post-programming functional yields in the industry. Note that preset and reset control the flip-flop, not the output pin. The output level is determined by the output polarity selected. The high-speed PALCE26V12 is fabricated with AM D's advanced electrically erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with TIL devices. This technology provides strong input clamp diodes, output slew-rate control, and a grounded substrate for clean switching. Power-Up Reset All flip-flops power up to a logic LOW for predictable system inttialization. Outputs of the PALCE26V12 will be HIGH or LOW depending on whether the output is active low or active high, respectively. The Vee rise must be monotonic, and the reset delay time is 1000 ns maximum. 2-364 Technology PALCE26V12H-15/20 AMD~ Registered Active-High Output, Register Feedback Registered Active-Low Output, Register Feedback Registered Active-High 1/0 Registered Active-Low 1/0 Registered Outputs Combinatorial Active-High 1/0' Combinatorial Active-Low 1/0 Combinatorial Active-Low Output, Register Feedback Combinatorial Active-High Output, Register Feedback 16072C-5 Combinatorial Outputs Figure 2.PALCE26V12 Macrocell Configuration Options PALCE26V12H-15/20 2-365 ~ AMD LOGIC DIAGRAM PALCE26V12 ...t:::: 11::' I 4 12 II .. .. •1 .. .. 41 41 41 __ A5YNCH - ,RE5ET 28 113 CLK;/1 ~~U-r2 -::: 10 .., .. 2 ~~ I ~821~1 ~1 ... 11 ~SO' --- 1. 1. 12 - JTU- ~2 111 ~~ ~ 01 I! 082I~a~ 1. II 3 sal! R11- .... 11 ~SO' II 4 .. m:noW- ~I 4¥~" . 52 J> CLK2113 so ~o 42 - ... J> 5 14 \ 91 ~Jf~ .. m 92 R. SO 91 24 10. I 93' ~i 1,1 11 23 4Lt~D: ,,~I107 9~~Sl . 17 Is At 1 .., 6 91 o 1 ~93' . ... r;o /, 111 DA~OOH"""- 8 16 .. l'4 . 4t:¥f~." p 4 I 12 II 2D . '" m . .. .. 22 10. I 1 .1 * When S3 = 1 (unprogrammed) the feedback is selected by St. When S3 = 0 (programmed), the feedback is the opposite of AR 9 l~ ClK. 93' 21 .GND that selected by St. 16072C-6 2·355 PALCE26V12H-15/20 AMD~ LOGIC DIAGRAM (continued) .. . . . 12 11 . :M .. PALCE26V12 . . .. .. .. 5P elK • .. rTo'" L ~ .... - J11 52 '=t.. r -=- -51 RS - ts1 ~ .. 20 liDs 1 01 ~~~. o 5P so .- ~l ~ I~~~ ~ o 10 18 ..' ., r 1 -~ - ... Ki~~ , - 0 1 .. ., ~o .. - ~t~1 01 5P ""'- 52 0 so I _ 51 51 -.. Lc..irwW11 6 ~~II0 ... .... o SP 1 so ,-521~~1 \! 51 .. --- .. .. - , , .... R2 1 ~ dIt;J 14 ~2 I 51 '30 13 111 01 so R3 - 51 5P 52 - ~ .. 110 ~~ o .. 11 12 so R4 _ 51 51 "' 11 19 5P 52 '=t- -~~ o SP ...... - .A - '41 r1o~ ~1 5 ~Lc..i~ SYNCH '"~"r:" • When 53 = 1 (unprogrammed) the feedback is selected by 51When 53 = 0 (programmed), the feedback is the opposite of that selected by 51- so 00 II 521~7 0 Ro - 51 l,L b-51 16072C-6 (concluded) PALCE26V12H-1S/20 2-367 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Commercial (C) Devices Ambient Temperature with Power Applied ................. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air ... " ........ O°C to +75°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V DC Input Voltage. . . . . . . . . . . . . .. -0.6 V to +7.0 V DC Output or 1/0, Pin Voltage ............... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. Static Discharge Voltage ................. 2001 V Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage 10H = -3.2 rnA VIN = VIH or VIL Vee = Min Min VOL Output LOW Voltage 10L = 16 rnA VIN = VIH or VIL Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) IIH Input HIGH Leakage Current VIN = 5.5 V, Vee = Max (Note 2) IlL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) loZH Off-State Output Leakage Current HIGH VOUT = 5.5 V, Vee = Max VIN = VIHor VIL (Note 2) loZL Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIHor VIL (Note 2) Isc Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) Icc Supply Current VIN= 0 V, Outputs Open (loUT = 0 rnA) Vee = Max Max 0.4 2.0 -30 Unit V 2.4 V V 0.8 V 10 ~ -10 ~ 10 ~ -10 ~ -160 rnA 105 rnA Notes: 1. These are absolute values with respect to device ground and all overshoots due to system and/or tester noise are included. 2. 110 pin leakage is the worst case of IiL and IOZL (or IiH and loZH). 3. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-368 PALCE26V12H-15/20 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CaUT Parameter Description Test Conditions Input Capacitance VIN" OV Vcc .. 5.0 V 5 Output Capacitance VOUT" 0 V TA =+25°C f .. 1 MHz 8 Typ Unit pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) Parameter Symbol -15 Parameter Description Min -20 Max Min Max Unit tPD Input or Feedback to Combinatorial Output ts Setup Time from Input, Feedback, or SP to Clock 10 13 ns tH Hold Time 0 0 ns tco Clock to Output 10 12 ns tAR Asynchronous Reset to Registered Output 20 25 ns 15 20 ns tARW Asynchronous Reset Width 15 20 tARR Asynchronous Reset Recovery Time 15 20 ns tSPR Synchronous Preset Recovery Time 10 13 ns LOW 8 10 ns HIGH 8 10 ns 50 40 MHz 58.8 43 MHz tWL Clock Width tWH fMAX Maximum Frequency (Note 3) External Feedback j 1/(ts + teo) Internal Feedback (feNT) ns tEA Input to Output Enable Using Product Term Control 15 20 ns tER Input to Output Disable Using Product Term Control 15 20 ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are calculated at initial characterization and at any time the design is modified where frequency may be affected. PALCE26V12H-15/20 (Com'l) 2-369 ~ AMD SWITCHING WAVEFORMS Input or Feedback _ _ ___ Vr t s -....e-~ tH Input or Feedback Combinatorial Output -~-vr Clock _ _ _ _ _ _ _ _ _ ~V:co Vr Registered Output _ _ _ _ _ _ _ __ 16072C-8 16072C-7 Combinatorial Output Registered Output Input Clock tER tEA Output _ _ _ _ Vr --'-J-I,~ 16072C-10 16072C-9 Input to Output Disable/Enable Clock Width ---it----- Input Asserting Asynchronous Reset , - - - - - Input Asserting Synchronous Preset _ _--' ts Registered Outputs _ _ _~~-K"...K.....JI Clock Registered Outputs _ _ _ _ _......~......._ Clock 16072C-11 Asynchronous Reset Notes: 1. VT= 1.5V 2. Input pulse amplitude 0 V to 3.0 V. 3. Input rise and fall times 2 ns-5 ns typical. 2-370 16072C-12 Synchronous Preset PALCE26V12H-15/20 AMD~ KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be Steady May Change from H to L Will be Changing from H to L May Change from L to H Will be Changing from L to H XXX'tIX Don't Care, Any Change Permitted Changing, State Unknown JJ) Does Not Apply Center Line is HighImpedance ·OWState WAVEFORM \\\\\ /7111 CK KSOOOO1 O-PAL SWITCHING TEST CIRCUIT Output O--+----...~(e} Test Point 16072C-13 Specification tpo, teo tEA tER 51 CL Rl R2 1.5 V Closed Z~ H:Open Z~ L: Closed H ~Z: Open L~Z: Closed Measured Output Value 50 pF 1.5 V 3000 5 pF PALCE26V12H-15/20 3900 H ~Z: VOH-0.5 V L ~ Z: VOL + 0.5 V 2-371 ~ AMD ' ENDURANCE CHARACTERISTICS The PALCE26V12 is manufactured using AMD's advanced Electrically Erasable process. This technology uses an EE cell to replace the fuse link used in bipolar Symbol toR N parts. As a result, the device can be erased and reprogrammecl-a feature which allows 100% testing at the factory. Parameter Test Conditions Min Pattern Data Retention Time Min Reprogramming Cycles Min Unit Max Storage Temperature 10 Years Max Operating Temperature 20 Years Normal Programming Conditions 100 Cycles INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee ESD ProgramNerify Protection Circuitry Typical Input Vee Preload Circuitry Feedback Input Typical Output 16072C-15 2-372 PALCE26V12H-15/20 AMD~ ROBUSTNESS FEATURES The PALCE26V12H-15/20 Rev. B has some unique features that make it extremely robust, especially when operating in high speed design environments. Input clamping circuitry limits negative overshoot, eliminating the possibility of false clocking caused by subsequent ringing. A special noise filter makes the programming circuitry completely insensitive to any positive overshoot that has a pulse width of less than about 100 hs. INPUT/OUTPUT EQUIVALENT SCHEMATICS FOR REV. B VERSION Vee Vee >50kn ESD Protection and Clamping I I I I I II P rogrammlng . -=- L!'~~~Iy' ______ _ Typical Input Preload Circuitry Typical Output Device Rev. Letter PALCE26V12H-15 B PALCE26V12H-20 B Feedback Input 16072C-15 Tops/de Marking: AMO CMOS PLO's are marked on top of the package in the following manner: PALCExxxx Oatecode (3 numbers) LOT 10 (4 characters) - - (Rev. Letter) The Lot 10 and Rev. letter are separated by.two spaces. PALCE26V12H-1S/20 2-373 ~ AMD POWER-UP RESET The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. The output state will depend on the programmed configuration. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and the wide range of ways Vee can rise to its steady state, two conditions are Parameter Symbol required to ensure a valid power-up reset. These conditions are: • The Vee rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Description Max Unit tPR Power-Up Reset Time 1000 ns ts Input or Feedback Setup Time tWL Clock Width LOW Power Registered Active-Low Output Clock See Switching Characteristics 4? Vee '~.....---tPR ~ ZI ~ tWL Power-Up Reset Waveform 2-374 PALCE26V12H-1S/20 16072C-16 AMD~ TYPICAL THERMAL CHARACTERISTICS Measured at 25°C ambient. These parameters are not tested. PALCE26V12-15 Parameter Symbol Trp. SKINNYDIP PLCC Parameter Description Unit 9jc Thermal impedance, junction to case 19 18 °elW 9ja Thermal impedance, junction to ambient 65 55 °elW 200 Ifpm air 59 48 400 Ifpm air 54 44 600 Ifpm air 50 800 Ifpm air 50 39 37 °elW °elW °elW °elW 9jma Thermal impedance, junction to ambi.ent with air flow Plastic 9lc Considerations The The data listed for plastic 9jc are for reference only and are not recommended for use in calculating junction temperatures. heat-flow paths in plastic-encapsulated devices are complex, making the 9jc measurement relative to a specific location on the package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom center of the package. Furthermore, 9jc tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a similar environment. PALCE26V12-15 2-375 ~ AMD 2-376 _ ~ COM'L: H-25 Advanced Micro Devices PALCE29M16H-25 24-Pin EE CMOS Programmable Array Logic DISTINCTIVE CHARACTERISTICS semlcustom logic • High-performance replacement; Electrically Erasable (EE) • technology allows reprogrammabllity 16 bidirectional user-programmable I/O logic macrocells for Combinatorial/Registered/ Latched operation Output Enable controlled by a pin or product terms • Varied product term distribution for Increased • design flexibility Programmable clock selection with two clocks/ • latch enables ([Es) and LOW/HIGH clock/i:E Register/Latch Preload permits full logic • verification High speed (tPD = 25 ns, fMAX= 33 MHz and fMAX • Internal = 50 MHz) Full-function AC and DC testing at the factory • for high programming and functional yields • • and high reliability 24-Pln 300 mil SKINNVDIP and 2S-pln plastic leaded chip carrier packages Extensive third-party software and programmer support through FusionPLD partners polarity GENERAL DESCRIPTION The PALCE29M16 is a high-speed, EE CMOS Programmable Array Logic (PAL) device designed forgenerallogic replacement in TTL or CMOS digital systems. It offers high speed, low power consumption, high programming yield, fast programming and excellent reliability. PAL devices. combine the flexibility of custom logic with the off-the-shelf availability of standard products, providing major advantages over other semicustom solutions such as gate arrays and standard cells, including reduced development time and low up-front development cost. BLOCK DIAGRAM 1/°0 V02 VOl':! 08740G-1 Publication# 08740 Rev. G Amendment/O Issue Date: June 1993 2-3n ~AMD GENERAL DESCRIPTION (continued) The PALCE29M16 uses the familiar sum-of-products (AND-OR) structure, allowing users to customize logic functions by programming the device for specific applications. It provides up to 29 array inputs and 16 outputs. It incorporates AMD's unique inpuVoutput logic macrocell which provides flexible inpuVoutput structure and polarity, flexible feedback selection, multiple Output Enable choices, and a programmable clocking scheme. The macrocells can be individually programmed as combinatorial, registered, or latched with active-HIGH or active-LOW polarity. The flexibility of the logic macrocells permits the system designer to tailor the device to particular application requirements. Increased logic power has been built into the PALCE29M16 by providing a varied number of logic product terms per output. Eight outputs have 8 product terms each, four outputs have 12 product terms each, and the other four outputs have 16 product terms each. This varied product-term distribution allows complex functions to be implemented in a single PAL device. Each output can be dynamically controlled by a common Output Enable pin or Output Enable product terms per bank of four outputs. Each output can also be permanently enabled or disabled. System operation has been enhanced by the addition of common asynchronous-Preset and Reset prodilct terms and a power-up Reset feature. The PALCE29M16 also incorporates Preload and Observability functions which permit full logic verification of the design. The PALCE29M16 is offered in the space-saving 300-mil SKINNYDIP package as well as the plastic leaded chip carrier package. CONNECTION DIAGRAMS Top View SKINNYDIP CLKlLE PLCC Vcc 10 12 I/0Fo I/OF7 I/OF1 1I0F6 1/00 1/07 1101 110s 1/02 1/05 1103 1/04 I/OF2 I/0Fs I/OF3 1I0F4 IIOE 11 GND I/CLKlLE ~ 095 ~ () "u. ~~~ ~ I/OF6 1/07 1106 22 21 NC 1/05 1/04 I/OFS u.MIW 0 Cl z ~ ::: C) () Z IW-J-u. ...- ~ ~ ~ ~ 08740G-3 Pin 1 is marked for orientation. PIN DESIGNATIONS CLKlLE Clock/Latch Enable GND Ground I Input I/CLKlLE Input or Clock/Latch Enable I/O Input/Output I/OF Input/Output with Dual Feedback NC No Connection Vcc Supply Voltage 2-378 I~ I/OF1 1/0 0 1101 NC 1102 1/03 I/OF2 08740G-2 Note: 0 u. PALCE29M16H-25 AMD~ ORDERING INFORMATION Commercial Products _AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL CE 29 M 16 H ·25 PC 14 _____T"'" -----r- r' FAMILY TYPE PAL .. Programmable Array Logic TECHNOLOGY - - - - - - - - ' CE .. CMOS Electrically Erasable -- L OPTIONAL PROCESSING Blank .. Standard processing PROGRAMMING REVISION 14.. First Revision (Requires current programming Algorithm) NUMBER OF ARRAY INPUTS - - - - - ' TEMPERATURE RANGE C .. Commercial (O°C to +75°C) OUTPUT TYPE - - - - - - - - - - ' M .. Advanced Macrocell NUMBER OF FLIP·FLOPS - - - - - - - - ' PACKAGE TYPE P .. 24-Pin Plastic SKINNYDIP (PD3024) J .. 28-Pin Plastic Leaded Chip Carrier (PL 028) POWER - - - - - - - - - - - - - - - ' H = Half Power (100 mAl SPEED -25 = 25'ns Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. ' PALCE29M16H·25 (Com'l) 2·379 ~ AMD FUNCTIONAL DESCRIPTION Inputs The PALCE29M16 has 29 inputs to drive each product term (up to 58 inputs with both TRUE and complement versions available to the AND array) as shown in the block diagram in Figure 1. Of these 29 inputs, 3 are dedicated inputs, 16 are from 8 I/O logic macrocells with two feedbacks, 8 are from other I/O logic macrocells with single feedback, one is the l/OE input and one is the l/CLK/LE input. Initially the AND-array gates are disconnected from all the inputs. This condition represents a logical TRUE for the AND array. By selectively programming the EE cells, the AND array may be connected to either the TRUE input orthe complement input. When both the TRUE and complement inputs are connected, a logical FALSE results at the output of the AND gate. Product Terms The degree of programmability and complexity of a PAL device is determined by the number of connections that form the programmable-AND and OR gates. Each programmable-AND gate is called a product term. The PALCE29M16 has 188 product terms; 176 of these product terms provide logic capability and 12 are architectural or control product terms. Among the 12 control product terms, two are for common AsynchronousPreset and Reset, one is for Observability, and one is for Preload. The other eight are common Output Enable product terms. The Output Enable of each bank of four macrocells can be programmed to be controlled by a common Output Enable pin or two AND/XOR product terms. It may be also permanently enabled or permanently disabled. Each product term on the PALCE29M16 consists of a 58-input AND gate. The outputs of these AND gates are connected to a fixed-OR plane. Product terms are allocated to OR gates in a varied distribution across the device ranging from 8 to 16 wide, with an average of 11 logiC product terms per output. An increased number of product terms per output allows more complex functions to be implemented in a single PAL device. This flexibility aids in implementing functions such as counters, exclusive-OR functions, or complex state machines, where different states require different numbers of product terms. Common asynchronOUS-Preset and Reset product terms are connected to all Registered or Latched l/Os. When the asynchronous-Preset product term is asserted (HIGH) all the registers and latches will immediately be loaded with a HIGH, independent of the clock. When the asynchronous-Reset product term is asserted (HIGH) all the registers and latches will be immediately loaded with a LOW, independent of the clock. The actual output state will depend on the macrocell polarity selection. The latches must be in latched mode (not transparent mode) forthe Reset, Preset, Preload, and power-up Reset modes to be meaningful. Input/Output LogiC Macrocells The I/O logic macrocell allows the user the flexibility of defining the architecture of each input or output on an individual basis. It also provides the capability of using the associated pin either as an input or an output. The PALCE29M16 has 16 macrocells, one for each I/O pin. Each I/O macrocell can be programmed for combinatorial, registered or latched operation (see Figure 2). Combinatorial output is desired when the PAL device is used to replace combinatorial glue logic. Registers and Latches are used in synchronous logic applications. Common IIOE Pin OE PTs For Banks { of 4 Macrocells Common Asynchronous Preset P110rP15 CLKIT I/CLKii:E Common Asynchronous Reset To AND Array 08740G-4 Figure 2a. PALCE29M16 Macrocell (Single FeedbaCk) 2·380 PALCE29M16H·25 AMD~ The output polarity for each macrocell in each of the three modes of operation is user-selectable, allowing complete flexibility of the macrocell configuration. a dynamic I/O controlled by the Output Enable pin or by two AND-XOR product terms which are available for each bank of four I/O Logic Macrocells. Eight of the macrocells (I/OFo-l/OF7) have two independent feedback paths to the AND array (see Figure 2b). The first is a dedicated I/O pin feedback to the AND array for combinatorial input. The second path consists -of a direct register/latch feedback to the array. If the pin is used as a dedicated input using the first feedback path, the register/latch feedback path is still available to the AND array. This path provides the capability of using the register/latch as a buried state register/latch. The other eight macrocells have a single feedback path to the AND array. This feedback is user-selectable as either an I/O pin or a register/latch feedback (see Figure2a). 1/0 Logic Macrocell Configuration Each macrocell can provide true inpuVoutput capability. The user can select each macrocell register/latch to be driven by either the signal generated by the AND-OR array orthe 110 pin. When the I/O pin is selected as the input, the feedback path provides the register/latch input to the array. When used as an input, each macrocell is also user-programmable for registered, latched, orcombinatorial input. The PALCE29M16H has a dedicated CLKiL'Epin and an I/CLKlLE pin. All macrocells have a programmable switch to choose between these two pins as the clock or latch enable signal. These signals are clock signals for macrocells configured as registers and latch enable signals for macrocells configured as latches. The polarity of these CLKlLE signals is also individually programmable. Thus different registers or latches can be driven by different clocks and clock phases. The Output-Enable mode of each of the macrocells can be selected by the user. The I/O pin can be configured as an output pin (permanently enabled) or as an input pin (permanently disabled). It can also be configured as AMD's unique 110 macrocell offers major benefits through its versatile, programmable inpuVoutput cell structure, multiple clock choices, flexible Output Enable and feedback selection. Eight 110 macrocells with single feedback contain 9 EE cells, while the other eight macrocells contain 8 EE cells for programming the inpuV output functions (see Table 1). EE cell S1 controls whether the macrocell will be combinatorial or registered/latched. So controls the output polarity (active-HIGH or active-LOW). S2 determines whether the storage element is a register or a latch. S3 allows the use of the macrocell as an input registerllatch or as an output registerllatch. It selects the direction of the data path through the registerllatch. If connected to the usual AND-OR array output, the registerllatch is an output connected to the 1/0 pin. If connected to the I/O pin, the registerllatch becomes an input registerllatch to the AND array using the feedback data path. Programmable EE cells S4 and S5 allow the user to select one of the four CLKlLE signals for each macrocell. S6 and S7 are used to control Output Enable as pin controlled, two-product-term-controlled, permanently enabled or permanently disabled. Sa controls a feedback multiplexer for the macrocells with a single feedback path only. Using the programmable EE cells So-Sa various input and output configurations can be selected. Some of the possible configuration options are shown in Figure 3. In the unprogrammed state (charged, disconnected), an architectural cell is said to have a value of "1 "; in the programmed state (discharged, connected to GND), an architectural cell is said to have a value of "0." 1~~~-------------------------4 OE PTs For Banks { of 4 Macrocells Common Asynchr~~~~~ PO P7 ClKlIT IIClKlIT Common Asynchronous Reset To AND Array ~==~----------------I RFX To AND Array :---<0------------------------------' 08740G-5 Figure 2b. PALCE29M16 Macrocell (Dual Feedback) PALCE29M16H·25 2·381 ~AMD Table 1a. PALCE29M16110 Logic Macrocell Architecture Selections S3 110 Cell S2 Storage Element 1 Output Cell 1 Register 0 Input Cell 0 Latch S1 Output Type So 1 Combinatorial 1 Active LOW 0 RegisterlLatch 0 Active HIGH S8 Output Polarity Feedback* 1 RegisterlLatch 0 I/O ·Applies to macrocells with single feedback only. Table 1b. PALCE29M16 1/0 Logic Macrocell Clock Polarity and Output Enable Selections S4 S5 Clock Edge/Latch Enable Level 1 1 CLKlLE pin positive-going edge, active-LOW LE 1 0 CLKlLE pin negative-going edge, active-HIGH LE 0 1 I/CLKlLE pin positive-going edge, active-LOW LE 0 0 IICLKlLE pin negative-going edge, active-HIGH LE S6 S7 1 1 1 0 XOR PT-Controlled Three-State Enable 0 1 Permanently Enabled (Output only) 0 0 Permanently Disabled (Input only) Output Buffer Control Pin-Controlled Three-State Enable Notes: 1 = Erased State (charged or disconnected). 0= Programmed State (discharged or connected). ·Active-LOW LE means that data is stored when the LE pin is HIGH, and the latch is transparent when the LE pin is LOW. Active-HIGH LE means the opposite. 2-382 PALCE29M16H-25 AMD~ SOME POSSIBLE CONFIGURATIONS OF THE INPUT/OUTPUT LOGIC MACROCELL (For other useful configurations, please referto the macrocell diagrams in Figure 2. All macrocell architecture cells are independently programmable). OB740G-6 08740G-7 Output Registered/Active Low Output Combinatorial/Active Low OB740G-B OB740G-9 Output Registered/Active High Output Combinatorial/Active High Figure 3a. Dual Feedback Macrocells OB740G-11 OB740G-10 Output Registered/Active Low, I/O Feedback Output Combinatorial/Active Low, I/O Feedback 5 0 .0 5 0 =0 51 =0 53= 1 5 8 =0 5 2 =0 5 1 .1 53=1 5 8 =0 OB740G-13 OB740G-12 Output Latched/Active High, I/O Feedback Output Combinatorial/Active High, I/O Feedback Figure 3b. Single Feedback Macrocells PALCE29M16H-25 2-383 ~AMD POSSIBLE CONFIGURATIONS OF THE INPUT/OUTPUT LOGIC MACROCELL 5 0 =1 51- 0 53- 1 5a = 1 52- 1 08740G-14 08740G-15 Output Combinatorial/Active Low, Register Feedback Output Registered/Active Low, Register Feedback 50 = 1 51 =0 53= 1 5a= 1 52=0 08740G-16 08740G-17 Output Combinatorial/Active Low, Latched Feedback Output Latched/Active Low, Latched Feedback Figure 3b. Single Feedback Macrocells (continued) S3= 0 Sa = 1 (For, Single Feedback Only) S2 = 1 RegIster = 0 Latch Programmable-AND Array 08740G-18 Input Registered/Latched Figure 3c.' All Macrocells 2-384 PALCE29M16H-25 AMD~ Power-Up Reset All flip-flops power up to a logic LOW for predictable system initialization. The outputs of the PALCE29M16 depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be LOW if 'programmed as active LOW and HIGH if programmed as active HIGH. If combinatorial is selected, the output will be a function of the logic. Preload To simplify testing, the PALCE29M16 is designed with preload circuitry that provides an easy method for testing logical functionality. Both product-term-controlled and supervoltage-enabled preload modes are available. The TTL-level preload product term can be useful during debugging, where supervoltages may not be available. Preload allows any arbitrary state value to be loaded into the registers/latches of the device. A typical functionaHest sequence would be to verify all possible state transitions for the device being tested. This requires the ability to set the state registers into an arbitrary "present state" value and to set the device's inputs into an arbitrary ''present input" value. Once this is done, the state machine is clocked into a new state, or "next state," which can be checked to validate the transition from the "present state." In this way any transition can be checked. Since preload can provide tne capability to go directly to any desired arbitrary state, test sequences may be greatly shortened. Also, all possible states can be tested, thus greatly reducing test time and development costs and guaranteeing proper in-system operation. Observability The output register/latch observability product term, when asserted. suppresses the combinatorial output data from appearing on the I/O pin and allows the observation of the contents of the register/latch on the output pin for each of the logic macrocells. This unique feature allows for easy debugging and tracing of the buried state machines. In addition, a capability of supervoltage observability is also provided. Security Cell A security cell is provided on each device to prevent unauthorized copying of the user's proprietary logic design. Once programmed, the security cell disables the programming, verification, preload, and the observability modes. The only way to erase the protection cell is by erasing the entire array and architecture cells, in which case no proprietary design can be copied. (This cell should be programmed only after the rest of the device has been completely programmed and verified.) Programming and Erasing The PALCE29M16 can be programmed on standard logic programmers. It may also be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erasure operation is required. Quality and Testability The PALCE29M16 offers a very high level of built-in quality. The erasability of the device provides a direct means of verifying performance of all the AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to yield the highest programming yield and post-programming functional yield in the industry. Technology The high-speed PALCE29M16 is fabricated with AMD's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with TTL devices. This technology provides strong inputclamp diodes, output slew-rate control, and a grounded substrate for clean switching. PALCE29M16H·25 2·385 ~AMD LOGIC DIAGRAM DIP (PLCC) Pinouts CLKIlE (2) 1 0 (3)2 10 (4)3 VOFO 4 8 12 16 20 24 28 32 36 40 44 48 52 sa ~ r~Ll it .~,-< - OUTPUT MACRO c- - K r... rr~ ~ J" ""- OBSERVE PRODUCT TERM 23(27) ~--~ .>.>.AI H.:::II - (5) 4 VOF1 -1! - INPUTI OllTPllT MACRO NPUTI OllTPUT MACRO r- """I ~ PRESET PRODUCT TERM K ... r-t ~ t: - >- .... -~ t--I""" ~r 21 (25) VOFS """ INPUTI OUTPUT UACRO NPUTI OllTPUT MACRO .A ~~ 1100 VOF 7 1"""-- ~ (S) 5 J 12 ,...:22(2S) K ... - .. I-i I""" '-~ tJr 20(24) >- INPUT! OllTPUT MACRO :....l """ Tl .... VO] t-r-I""" - INPUTI OUTPUT UACRO K. ~. '-- ~ -r- '-~-~_1 9(23) > ~ ~ t-- NPUT! OllTPUT MACRO 0 VOS r--- 08740G-19 2-386 PALCE29M16H-25 AMD~ LOGIC DIAGRAM DIP (PLCC) Pinouts 12 -l INPUT! OUTPUT MACRO '--- 16 20 24 28 32 36 40 44 48 ~ ~s K f--t 1"'"1"'" ................. r- NPUT! OUTPUT MACRO 18(21) V05 LJ f- '" "'1"""1"'" ........ :-rt INPUT! OUTPUT MACRO I-< to.: '---- ... ---t 1"""1""" r - .... '>- ... ' ~ ........ ~ L: INPUT! OUTPUT MACRO t< ~ ..... +---t 1... ....... r:: 1..r'"1"'" ~ ...,~ ........ rt INPUT! OUTPUT MACRO NPUT! OUTPUT MACRO 1"'"1"'" ~r 16 (19) VOF 5 ..... t< ~ I-!t L.-- """-r- I~ \ .. .... ......... PRELOAD PRODUCT TERM J- >- (13l..!1 VOE V0 4 t--- ,...,... ....:1-- - 17(20) -~ INPUT! OUTPUT MACRO U (12) 10 VOF3 ~~ INPUT! OUTPUT MACRO [J" ~ 15(18) L'OF4 ~ .... . ~""r-t-- } := ..... 12 16 20 24 28 32 36 40 48 52 RESET PRODUCT TEAM 14 (17) 11 13(1~ vCLKlLE ~6 08740G-19 (concluded) PALCE29M16H-25 2-387 ~AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to + 150°C Commercial (C) Devices Ambient Temperature with Power Applied ............. -55°C to + 125°C Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . . .. O°C to 75°C Supply Voltage with Respect to Ground ............ ~ -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ....... 4.75 V to 5.25 V DC Input Voltage ........... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or I/O Pin Voltage ............... -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Latchup Current (TA = O°C to + 75°C) ..... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions VOH Output HIGH Voltage IOH =-2 mA VIN = VIH or VIL Vee = Min Min VOL Output LOW Voltage IOL = 8 mA VIN = VIH or VIL 0.5 IOL = 4 mA Vee = Min 0.33 Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) IIH Input HIGH Leakage Current VIN = 5.5 V, Vee = Max (Note 2) ilL Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 2) IOZH Off-State Output Leakage Current HIGH IOZL Unit V V 0.1 IOL = 20 IlA VIH Max 2.4 2.0 V 0.8 V 10 jlA -10 jlA VOUT = 5.5 V, Vee = Max VIN = VIH or VIL (Note 2) 10 jlA Off-State Output Leakage Current LOW VOUT = 5.5 V, Vee = Max VIN = VIH or VIL (Note 2) -10 jlA Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) -130 mA lee Supply Current VIN = 0 V, Outputs Open (lOUT = 0 mAl Vee = Max 100 mA -30 Notes: 1. These are absolute values with respect to device ground all overshoots due to system and/or tester noise are included. 2. I/O pin leakage is the worst case of IJL and IOZL (or IiH and IOZH). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-388 PALCE29M16H-25 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN CaUl Parameter Description Test Conditions Input Capacitance VIN= 0 V Output Capacitance VOUl= 0 V I I Typ Unit Vcc = 5.0 V, TA = 25°C, 5 pF f = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS Registered Operation Parameter Symbol Min Parameter Description Max Unit 25 ns Combinatorial Output tPD Input or I/O Pin to Combinatorial Output Output Register 15 ns tsoR Input or I/O Pin to Output Register Setup tCOR Output Register Clock to Output tHOR Data Hold Time for Output Register 0 ns tSIR I/O Pin to Input Register Setup 2 ns telR Register Feedback Clock to Combinatorial Output tHIR Data Hold Time for Input Register 15 ns Input Register 28 6 ns ns Clock and Frequency tCIS Register Feedback to Output Register/Latch Setup 20 ns fMAX Maximum Frequency 1/(tSOR + teoR) 33.3 MHz fMAXI Maximum Internal Frequency 1/tels 50 MHz tcwH Pin Clock Width HIGH 8 ns tCWl Pin Clock Width LOW 8 ns CLK~>---~---------------------, r----+,--, tels tSIR ••• AND-OR Array Output Register Inp.ut Register teOR I/O I/O tSOR , ----------,' '" I/O ' - - - - - - - - - - -.. teIR tPD - - - - - - - t ! - - - - - + - - - - - - - . - t P D 08740G-20 PALCE29M16H-25 (Com'l) 2-389 ~AMD SWITCHING WAVEFORMS Combinatorial Input Combinatorial Output --------~V-T-08740G-21 Combinatorial Output Combinatorial Input Clock ~ ~V _______'_I~~__~__OR_~C-R-~~,=T======= ~ Registered Output .... VT_ _ 08740G-22 Output Register Registered Input Clock Combinatorial Output 08740G-23 Input Register 08740G-24 Clock Width 2-390 PALCE29M16H-25 AMD~ SWITCHING CHARACTERISTICS L atc h e dO'perarIon Parameter Symbol Min Parameter Description Max Unit Combinatorial Output tPD Input or I/O Pin to Combinatorial Output 25 ns tpIO Input or I/O Pin to Output via One Transparent Latch 28 ns Output Latch 15 tsa.. Input or I/O Pin to Output Latch Setup tGOL Latch Enable to Output Through Transparent Output Latch ns tHOL Data Hold Time for Output Latch 0 ns tSTL Input or I/O Pin to Output Latch Setup via Transparent Input Latch 18 ns 2 15 ns Input Latch tSIL I/O Pin to Input Latch Setup tGIL Latch Feedback, Latch Enable Transparent Mode to Combinatorial Output tHIL Data Hold Time for Input Latch ns 28 ns 6 ns Latch Enable tGIS Latch Feedback to Output Register/Latch Setup 20 ns tGWH Pin Enable Width HIGH 8 ns tGWL Pin Enable Width LOW 8 ns AND-OR r--....I...--, !~~_____~~~~~ __ tSTL --+-r:-~-t----+---t----;~ I/O toOL ,,' tSOL - - - - - - - - - - , ' tPID - - - - - - ' tpm tPD - - - - - - - - I I - - - - t - - - - - - - . - tPD 08740G-25 Input/Output latch Specs PAlCE29M16H-25 (Com'l) 2-391 ~ AMD SWITCHING WAVEFORMS Latched Input Latched Transparent Combinatorial Input Combinatorial Output 08740G-27 LE Width Latched Output 08740G-26 Latch (Transparent Mode) LE ~Latched Vr ____T_r"""!an_s_p_a_re_n_t___ Input _ Latch tGiS Trans arent Latched Outp'ut Latch 08740G-29 Input and Output Latch Relationship Latched Input Combinatorial Input Latched Output 0874013-28 Output Latch Latched Input Combinatorial Output 08740G-30 Input Latch Note: 1. If the combinatorial input changes while LE is in the latched mode and LE goes into the transparent mode after tPTD ns has elapsed, the corresponding latched output will change tGOL ns after LE goes into the transparent mode. If the combinatorial input change while LE is in the latched mode and LE goes into the transparent mode before tPTD ns has elapsed, the corresponding latched output will change at the later of the following-tPTD ns after the combinatorial input changes or tGOL ns after LE goes into the latched mode. 2·392 PALCE29M16H·25 AMD~ SWITCHING CHARACTERISTICS Reset/Preset, Enable Parameter Symbol Min Parameter Description Max Unit 30 ns Combinatorial Output tAPO Input or I/O Pin to Output Register/Latch ResetlPreset tAW Asynchronous ReseUPreset Pulse Width 15 ns tARO Asynchronous ReseUPreset to Output RegisterlLatch Recovery 15 ns tARi Asynchronous ReseUPreset to Input RegisterlLatch Recovery 12 ns Output Enable Operation tpzX I/OE Pin to Output Enable 20 ns tpxz I/OE Pin to Output Disable (Note 1) 20 ns tEA Input or I/O to Output Enable via PT 25 ns Input or I/O to Output Disable via PT (Note 1) 25 ns tER Note: 1. Output disable times do not include test load RC time constants. SWITCHING WAVEFORMS Combinatorial _ _ _-,. ,._ _ _ _ _ _ _- - - - Asynchronous vr ReseUPreset _ _ _- I ' -_ _ _ _ _ _ '___ _ _ __ Reg iste red/Latched Output Clock Output Register/Latch Reset/Preset 08740G-31 Pin 11 Combinatorial/ Registered! Latched Output Combinatorial Asynchronous ResetlPreset Clock . --t?tAR'F LtAW~ --- Input Register/Latch Reset/Preset tpxz ....---I~-t PZX Pin 11 to Output Disable/Enable 08740G-32 08740G-33 Combinatorial Input Combinatorial! RegisteredlLatched Output Input to Output Disable/Enable PALCE29M16H-25 (Com'l) 08740G-34 2-393 ~AMD KEY TO SWITCHING WAVEFORMS INPUTS OUTPUTS Must be Steady Will be \\\\\ May Change from H to L Will be ///71 May Change from L to H Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State WAVEFORM xxxxxx Steady Changing from H to L Will be KSOOOO10 SWITCHING TEST CIRCUIT Output o--......~----.. 08740G-35 Specification Switch S1 tPD, teo, tGOl Closed tEA, tpzx tER, tpxz Cl R1 R2 Z--+H: open Z--+L: closed 35 pF 4700 390.0 H--+Z: open 5 pF 1.5 V 1.5 V H--+Z: VOH -0.5 V L--+Z: VOL +0.5 V L--+Z: closed 2-394 Measured Output Value PALCE29M16H-25 AMD~ PRELOAD The PALCE29M16 has the capability for product-term Preload. When the global-preload product term is true, the PALCE29M16 will enter the preload mode. This feature aids functional testing by allowing direct seUing of register states. The procedure for Preload is as follows: • Set the selected input pins to the user selected preload condition. • Apply the desired register value to the I/O pins. This sets a of the register. The value seen on the I/O pin, after Preload, will depend on whether the macrocell is active high or active low. Parameter Symbol Parameter Description • • • Pulse the clock pin (pin 1). Remove the inputs to the I/O pins. Remove the Preload condition. • Verify VoJVOH for all output pins as per programmed pattern. Because the Preload command is a product term, any input to the array can be used to set Preload (including I/O pins and registers). Preload itself will change the values of the I/O pins and registers. This will have unpredictable results. Therefore, only dedicated input pins should be used for the Preload command. Min Rec. Max Unit to Delay Time 0.5 1.0 5.0 Jls tw Pulse Width 250 500 700 ns tllO Valid Output 100 500 ns Inputs Preload Mode I/O Pins ~-- to ----.I elK - - - - - - - - VIH Pin 1 (2) VIL 08740G-37 Preload Waveform PALCE29M16H-25 2-395 ~·AMD OBSERVABILlTV The PALCE29M16 has the capability for product-term Observability. When the global-Observe product term is true, the PALCE29M16 will enter the Observe mode. This feature aids functional testing by allowing direct observation of register states. When the PALCE29M16 is in the Observe mode, the output buffer is enabled and the I/O pin value will be Q of the corresponding register. This overrides any OE inputs. The procedure for Observe is: • Remove the inputs to all the I/O pins. Parameter Symbol • Set the inputs to the, user selected, Observe configuration. • The register values will be sent to the corresponding I/O pins. • Remove the Observe configuration from the selected I/O pins. Because the Observe command is a product term, any input to the array can be used to set Observe (including I/O pins and registers). If I/O pins are used, the observe mode could cause a value change, which would cause the device to oscillate in and out of the Observe mode. Therefore, only dedicated input pins should be used for the Observe command. Parameter Description Min Ree. Max Unh to Delay Time 0.5 1.0 5.0 ~s tllO Valid Output 100 500 ns Input Pins k tD=fe~e MOOe I/O Pins tt~l VIH VIL VOH VOl -- - - VIH elK Pin 1 (2) VIL 08740G-38 Observability Waveform 2-396 PALCE29M16H-25 AMD l1 POWER-UP RESET The registered devices in the AMD PAL Family have been designed with the capability to reset during system power-up. Following power-up, all registers will be reset to LOW. The output state will depend on the polarity of the output buffer. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the asynchronous operation of the power-up reset, and the wide Parameter Symbol range of ways Vee can rise to its steady state, two conditions are required to ensure a valid power-up reset. These conditions are: • The Vec rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Description Min Max Unit 10 Il s Power-Up Reset Time tPR ts Input or Feedback Setup Time tw Clock Width tR Vee Rise Time tR Clock Ils 500 Power Registered Active LOW Output See Switching Characteristics . --- ~t~ tPR Vee --~.J ZI "= tw 08740G-39 Power-Up Reset Waveform PALCE29M16H-25 . 2-397 ~AMD TYPICAL THERMAL CHARACTERISTICS Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Typ. Parameter Description 9jc Thermal Impedance, Junction to Case 91a Thermal Impedance, Junction to Ambient 9jrna Thermal Impedance, Junction to Ambient with Air Flow SKINNYDIP PLCC Unh 17 11 °CIW °elW 63 51 200 Ifpm air 60 43 °elW 400 Ifpm air 52 38 °elW 600 Ifpm air 43 34 °elW 800 Ifpm air 39 30 °elW Plastic 9Jo Considerations The data listed for plastic 9,c are for reference only and are not recommended for use in calculating junction temperatures. The heat-floW paths in plastic-encapsulated devices are complex, making the 9,c measurement relative to a specific location on the package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom center of the package. Furthermore, 9,c tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a similar environment. 2·398 . PALCE29M16H·25 - ~ COM'L: H-25 Advanced Micro Devices PALCE29MA16H·25 24-Pin EE CMOS Programmable Array Logic DISTINCTIVE CHARACTERISTICS • • • • • High-performance semlcustom logic replacement; Electrically Erasable (EE) technology allows reprogrammability 16 bidirectional user-programmable 110 logic macrocells for Comblnatorlal/Reglstered/ Latched operation Output Enable controlled by a pin or product terms Varied product term distribution for Increased design flexibility Programmable clock selection with common pin clock!latch enable (IE) or Individual product term clocklLE with LOW/HIGH clock! LE polarity • Register/Latch Preload permits full logic verification 1:1 High speed (tPD • • • = = 25 ns, fMAX 33 MHz and fMAX Internal = 50 MHz) Full-function AC and DC testing at the factory for high programming and functional yields and high reliability 24-pln 300 mil SKINNVDIP and 28-pln plastic leaded chip carrier packages Extensive third-party software and programmer support through FusionPLD partners GENERAL DESCRIPTION The PALCE29MA16 is a high-speed, EE CMOS Programmable Array Logic (PAL) device designed for generallogic replacement in TTL or CMOS digital systems. It offers high speed, low power consumption, high pro- gramming yield, fast programming, and excellent reliability. PAL devices combine the flexibility of custom logic with the off-the-shelf availability of standard products, providing major advantages over other BLOCK DIAGRAM CLK/LE Programmable AND Array 58x178 08811G-l Publication# 08811 Rev. G Amendment/O Issue Date: June 1993 2-399 ~AMD GENERAL DESCRIPTION (continued) semicustom solutions such as gate arrays and standard cells, including reduced development time and low upfront development cost. . The PALCE29MA16 uses the familiar sum-of-products (AND-OR) structure, allowing users to customize logic functions by programming the device for specific applications. It provides up to 29 array inputs and 16 outputs. It incorporates AMD's unique inpuVoutput logic macrocell which provides flexible input/output structure and polarity, flexible feedback selection, multiple Output Enable choices, and a programmable clocking scheme. The macrocells can be individually programmed as combinatorial, registered, or latched with active-HIGH or active-LOW polarity. The flexibility of the logic macrocells permits the system designer to tailor the device to particular application requirements. Increased logic power has been built into the PALCE29MA16 by providing a varied number of logic product terms per output. Of the 16 outputs, 8 outputs have 4 product terms each, 4 outputs have 8 product terms each, and the other 4 outputs have 12 product terms each. This varied product-term distribution allows complex functions to be implemented in a single PAL device. Each output can be dynamically controlled by a common Output Enable pin or Output Enable product term. Each output can also be permanently enabled or disabled. System operation has been enhanced by the addition of common asynchronous-Preset and .Reset product terms and a power-up Reset feature. The PALCE29MA16 also incorporates Preload and Observability functions which permit full logic verification of the design. The PALCE29MA16 is offered in the space-saving 300-mil SKINNYDIP package as well as the plastic leaded chip carrier package. CONNECTION DIAGRAMS Top View SKINNYDIP CLKlLE PLCC Vcc 10 13 I/OF7 I/OFo I/OF1 I/OFs 1/00 1/07 1101 I/Os 1/05 1/02 1/03 I/OF2 1/00 1/0 1 NC 1/02 1/03 1/04 I/OFs I/OF4 I/OF3 IIOE Li'1~ 0 0 0;::. Z Z (!) 08811G-2 PIN DESIGNATIONS CLKlLE GND 110 1I0F Clock or Latch Enable Ground Input InpuVOutput Input/Output with Dual Feedback Vee Supply Voltage NC No Connection 2-400 PALCE29MA16H-25 1/07 23 22 21 I/OS NC 1/05 1/04 I/OFS ;::. Note: Pin 1 is marked for orientation. 24 20 I/OF2 12 11 GND I/OFS I/OF1 .= ~ ~ 0 ;::. 08811G-3 AMD~ ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of these elements: L PAL CE 29 MA 16 H -25 P C /4 OPTIONAL PROCESSING Blank = Standard Processing FAMILY TYPE PAL = Programmable Array Logic T PROGRAMMING REVISION 14 = First Revision (Requires current programming Algorithm) TECHNOLOGY CE = CMOS Electrically Erasable NUMBER OF ARRAY INPUTS _ _----I OUTPUT TYPE - - - - - - - - - ' " MA = Advanced Asynchronous Macrocell TEMPERATURE RANGE C = Com~ercial (OOC to +75°C) NUMBER OF FLIP-FLOPS - - - - - - - - - - ' POWER-----------------------------~ PACKAGE TYPE P = 24-Pin Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) H = Half Power (100 mA) SPEED -25 = 25 ns Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations and to check on newly released combinations. PALCE29MA16H-25 (Com'l) 2-401 ~AMD FUNCTIONAL DESCRIPTION Inputs The PALCE29MA16 has 29 inputs to drive each product term (up to 58 inputs with both TRUE and complement versions available to the AND array) as shown in the block diagram in Figure 1. Of these 29 inputs, 4 are dedicated inputs, 16 are from eight 110 logic macrocells with two feedbacks, 8 are from other 110 !2g!c macrocells with single feedback and one is the IIOE input. Initially the AND-array gates are disconnected from all the inputs. This condition represents a logical TRUE for the AND array. By selectively programming the EE cells, the AND array may be connected to either the TRUE input orthe complement input. When both the TRUE and complement inputs are connected, a logical FALSE results at the output of the AND gate. Product Terms The degree of programmability and complexity of a PAL device is determined by the number of connections that form the programmable-AND and OR gates. Each programmable-AND gate is called a product term. The PALCE29MA16 has 178 product terms; 112 of these product terms provide logic capability and others are architectural product terms. Among the control product terms, one is for Observability, and one is for Preload. The Output Enable of each macrocell can be programmed to be controlled by a common Output Enable pin or an individual product term. It may also be permanently disabled. In addition, independent product terms for each macrocell control Preset, Reset and CLK/LE. Each product term on the PALCE29MA16 consists of a 58-input AND gate. The outputs of these AND gates are connected to a fixed-OR plane. Product terms are allocated to OR gates in a varied distribution across the device ranging from 4 to 12 wide, with an average of 7 logic product terms per output. An increased number of product terms per output allows more complex functions to be implemented·in a single PAL device. This flexibility aids in implementing functions such as counters, exclusive-OR functions, or complex state machines, where different states require different numbers of product terms. Individual asynchronous-Preset and Reset product terms are connected to all. Registered or Latched lIDs. When the asynchronous-Preset product term is asserted (HIGH) the register or latch will immediately be loaded with a HIGH, independent of the clock. When the asynchronous-Reset product term is asserted (HIGH) the register or latch will be immediately loaded with a LOW, independent of the clock. The actual output state will depend on the macrocell polarity selection. The latches must be in latched mode (not transparent mode) for the Reset, Preset, Preload, and power-up Reset modes to be meaningful. Input/Output Logic Macrocells The 110 logic macrocell allows the user the flexibility of defining the architecture of each input or output on an individual basis. It also provides the capability of using the associated pin either as an input or an output. The PALCE29MA16 has 16macrocells, one for each 110 pin. Each 110 macrocell can be programmed for combinatorial, registered or latched operation (see Figure 2). Combinatorial output is desired when the PAL . device is used to replace combinatorial glue logic. Registers and Latches are used in synchronous logic applications. Registers and Latches with product term controlled clocks can also be used in asychronous application. Common IIOE (PIN) IndividualOE Individual Asynchronous Preset PO J-----t~...,--1ts..z1 IIOx P70R P11 Common ClKlLE (PIN) Individual ClK! LE Individual Asynchronous Reset Rx To AND Array Figure 2a. PALCE29MA16 Macroceii (Singie Feedbackj 2-402 PALCE29MA16H-25 08811G-4 AMD~ The output polarity for each macrocell in each of the three modes of operation is user-selectable, allowing complete flexibility of the macrocell configuration. Eight of the macrocells (I/OFo-l/OF7) have two independent feedback paths to the AND array (see Figure 2b). The first is a dedicated I/O pin feedback to the AND array for combinatorial input. The second path consists of a direct register/latch feedback to the array. If the pin is used as a dedicated input using the first feedback path, the register/latch feedback path is still available to the AND array. This path provides the capability of using the register/latch as a buried state register/latch. The other eight macrocells have a single feedback path to the AND array. This feedback is user-selectable as either an I/O pin or a register/latch feedback (see Figure 2a). Each macrocell can provide true input/output capability. The user can select each macrocell register/latch to be driven by either the signal generated by the AND-OR array or the corresponding I/O pin. When the I/O pin is selected as the input, the feedback path provides the register/latch input to the array. When used as an input, each macrocell is also user-programmable for registered, latched, or combinatorial input. The PALCE29MA 16 has a dedicated CLK/LE pin and one individual CLKlLE product term or macrocell. All macrocells have a programmable switch to choose between the CLKlLE pin and the CLK/LE product term as the clock or latch enable signal. These Signals are clock signals for macrocells configured as registers and latch enable signals for macrocells configured as latches. The polarity of these CLK/LE signals is also individually programmable. Thus different registers or latches can be driven by different clocks and clock phases. The Output-Enable mode of each of the macrocells can be selected by the user. The I/O pin can be configured as an output pin (permanently enabled) or as an input pin (permanently disabled). It can also be configu red as a dynamic I/O controlled by the Output Enable pin or by a product term. 1/0 Logic Macrocell Configuration AMD's unique I/O macrocell offers major benefits through its versatile, programmable inpuUoutput cell structure, multiple clock choices, flexible Output Enable and feedback selection. Eight I/O macrocells with single feedback contain 9 EE cells, while the other eight macroce lis contain 8 EE cells for programming the inpuU output functions (see Table 1). EE cell Sl controls whether the macrocell will be combinatorial or registered/latched. So controls the output polarity (active-HIGH or active-LOW). S2 determines whether the storage element is a register or a latch. S3 allows the use of the macrocell as an input register/latch or as an output register/latch. It selects the direction of the data path through the register/latch. If connected to the usual AND-OR array output, the register/latch is an output connected to the I/O pin. If connected to the I/O pin, the register/latch becomes an input register/latch to the AND array using the feedback data path. . Programmable EE cells S4 and Ss allow the user to select one of the four CLKlLE signals for each macrocell. S6 and S7 are used to control Output Enable as pin controlled, product-term controlled, permanently enabled or permanently disabled. Sa controls a feedback multiplexer for the macrocells with a single feedback path only. Using the programmable EE cells So-Sa various input and output configurations can be selected. Some of the possible configuration options are shown in Figure 3. In the erased state (charged, disconnected), an architectural cell is said to have a value of "1"; in the programmed state (discharged, connected to GND), an architectural cell is said to have a value of "0." Common IfOE PIN -------------------I~ Individual OE Individual Asynchronous Preset PO 1-----I>o-.---K>llfOFx P3 Common CLKlLE (PIN) --_.1 Individual CLKlLE Individual Asynchronous Reset To AND Array :===~ To AND Array _________ _________________ ---1 RFx ---.J :===~ 08811G-5, Figure 2b. PALCE29MA16 Macrocell (Dual Feedback) PALCE29MA16H-25 2-403 ~ AMD Table 1a. PALCE29MA16110 Logic Macrocell Architecture Selections S3 1/0 Cell S2 Storage Element 1 Output Cell 1 Register 0 Input Cell 0 Latch S1 Output Type So 1 Combinatorial 1 Active LOW 0 Register/Latch 0 Active HIGH S8 Output Po larlty Feedback· 1 RegisterILatch 0 I/O "Applies to macroce/ls with single feedback only. Table 1b. PALCE29MA16 1/0 Logic Macrocell Clock Polarity and Output Enable Selections Clock EdgelLatch Enable Level S4 S5 1 1 CLKlLE pin positive-going edge, active-LOW LE· 1 0 CLKlLE pin negative-going edge, active-HIGH LE· 0 1 CLKlLE PT positive-going edge, active-LOW LE· 0 0 CLKlLE PT negative-going edge, active-HIGH LE· 56 1 57 1 1 0 0 1 Permanently Enabled (Output only) 0 0 Permanently Disabled (Input only) Output Buffer Control Pin-Controlled Three-State Enable PT-Controlled Three-State Enable Notes: 1 = Erased State (Charged or disconnected). o= Programmed State (Discharged or connected). "Active-LOW LE means that data is stored when the LE pin is HIGH, and the latch is transparent when the LE pin is LOW. Active-HIGH LE means the opposite. 2-404 PALCE29MA 16H-25 AMDl1 SOME POSSIBLE CONFIGURATIONS OF THE INPUT/OUTPUT LOGIC MACROCELL (For other useful configurations, please refer to the macrocell diagrams in Figure 2. All macrocell architecture cells are independently programmable). 08811G-6 Output Reg Istered/Active Low 08811G-7 Output Combinatorial/Active Low 08811G-8 08811G-9 Output Registered/Active High Output Combinatorial/Active High Figure 3a. Dual Feedback Macrocells SO=1 S1" 0 S3=1 S8'"0 S2=1 08811G-10 Output Registered/Active L,ow, I/O Feedback 08811G-11 Output Combinatorial/Active Low, I/O Feedback SO=O S1 =0 S3= 1 S8=0 82=0 08811G-12 Output latched/Active High, I/O Feedback 08811G-13 Output Combinatorial/Active High, I/O Feedback Figure 3b. Single Feedback Macrocells PALCE29MA16H·25 2·405 ~AMD SOME POSSIBLE CONFIGURATIONS OF THE INPUT/OUTPUT LOGIC MACROCELL SO-1 S1- 0 S3- 1 Sa .. 1 S2- 1 08811-014A 08811-015A Output Registered/Active Low, Register Feedback Output Combinatorial/Active Low, Latched Feedback SO-1 S1"0 S3=1 sa" 1 S2=0 08811-016A Output latched/Active Low, Latched Feedback 08811-017A Output Combinatorial/Active Low, Latched Feedback Figure 3b. Single Feedback Macrocells (Continued) S3 =0 Sa .. 1 (FOR SINGLE FEEDBACK ONLV) S2" 1 REGISTER =0 LATCH 08811-018A Programmable-AND Array Figure 3c. All Macrocells 2-406 PALCE29MA 16H-25 AMD~ Power-Up Reset All flip-flops power up to a logic LOW for predictable systeminitialization. The outputs of the PALCE29MA16 depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be LOW if programmed as active LOW and HIGH if programmed as active HIGH. If combinatorial is selected, the output will be_a function of the logic. Preload To simplify testing, the PALCE29MA 16 is designed with preload circuitry that provides an easy method for testing logical functionality. Both product-term-controlled and supervoltage-enabled preload modes are available. The TIL-level preload product term can be useful during debugging, where supervoltages may not be available. Preload allows any arbitrary state value to be loaded into the registers/latches of the device. A typical functional-test sequence would be to verify all possible state transitions for the device being tested. This requires the ability to set the state registers into an arbitrary "present state" value and to set the device's inputs into an arbitrary "present input" value. Once this is done, the state machine is clocked into a new state, or "next state," which can be checked to validate the transition from the "present state." In this way any transition can be checked. Since preload can provide the capability to go directly to any desired arbitrary state, test sequences may be greatly shortened. Also, all possible states can be tested, thus greatly reducing test time and development costs and guaranteeing proper in-system operation. Observability The output register/latch observability product term, when asserted, suppresses the combinatorial output data from appearing on the I/O pin and allows the observation of the contents of the register/latch on the output pin for each of the logic macrocells. This unique feature allows for easy debugging and traCing of the buried state machines. In addition, a capability of supervoltage observability is also provided. Security Cell A security cell is provided on each device to prevent unauthorized copying of the user's proprietary logic design. Once programmed, the security cell disables the programming, verification, preload, and the observability modes. The only way to erase the protection cell is by erasing the entire array and architecture cells, in which case no proprietary design can be copied. (This cell should be programmed only after the rest of the device has been completely programmed and verified.) Programming and Erasing The PALCE29MA16 can be programmed on standard logic programmers. It may also be erased to reset a previously configured device back to its virgin state. Erasure is automatically performed by the programming hardware. No special erasure operation is required. Quality and Testability The PALCE29MA16 offers a very high level of built-in quality. The erasability of the device provides a direct means of verifying performance of all the AC and DC parameters. In addition, this verifies complete programmability and functionality of the device to yield the highest programming yield and post-programming functional yield in the industry. Technology The high-speed PALCE29MA16 is fabricated with AMD's advanced electrically-erasable (EE) CMOS process. The array connections are formed with proven EE cells. Inputs and outputs are designed to be compatible with TTL devices. This technology provides strong input-clamp diodes, output slew-rate control, and a grounded substrate for clean switching. PALCE29MA16H-25 2·407 ~ AMD LOGIC DIAGRAM SKINNY DIP (PLCC) Pinouts CLKlLE (2) 1 0 ........ (3) 2 10 r- ........ t IIOFo 8 12 16 20 24 28 32 36 40 44 48 ~ 56 K .A (4) 3 4 InpuV Output Macro "'-I .... L- ~ 1--1--1 t L....r-r- . .. .OBSERVE PRODUCT ~ TERM 23 (27) 13 ......... L.... -.-r ~ InpuV Output MauD tt .-:22(26) IIOF 7 .A ~--,-- --(5) 4 IIOF1 -Ii! .. .... 3;. ~ InpuV K Output MauD .... I-f ..... _L.... ~ -r-r- . .. ~ V- InpuV Output MaUD ~ ~ ';I L....,....r- ~r 21 (25) IIOF s .... '-'H Inpuv ~rt Output MaUD h L....--. ... f-I L....""'r- _I-L.... r ~~2Ovb 7 ) 24 InpuV Output MauD "1.... ' L...._ ..... '-rt~~~K InpuV Output MauD ' - ..... I-t '-""'r- D--~~[J~ , InpuV Output MaUD "1... I--- 088iiB-i9 2·408 PALCE29MA 16H-25 AMD~ LOGIC DIAGRAM SKINNY DIP (PLCC) Pinouts Continued from Previous Page 0 .A (9) 7 8 1 15 20 24 2. 32 38 40 44 156 K t 1/02 4 Input! Output Macro '--- .... ., t---t -r-- >-r-........ ~ >-1 8 (21) lJ Input! Output Macro .... .... V0 5 I-- ..... -r- r- .... 10)8 1/0 3 -~ K Input! Output Macro '----- .... -r-- f----.-t., r-"- S-~1 7(20) 1/0 4 >Input! Output Macro ~ .... r-"- [11)9 I/OF2 -l! Input! Output K Macro .... r - - '--- ..... r - - .. - _.... - r-1 ~ ... >- "- Input! Output ~ Macro ..... .t - :A 12) 10 -< Input! Output Macro '---- ---- .. r-t-t ~ . --~ [t -- r PRELOAD PRODUCT Input! Output TERM Macro (13)11 I/O E IIOF5 .... rt~-- K /OF3 --roo- 16 (19) ..... ..... . J-- ::1- 15 (18) I/OF4 - .... 14(17) 12 .A ::I 13(16) 11 12 16 20 24 28 32 36 40 44 48 52 56 088118-19 (concluded) PALCE29MA 16H-25 2-409 ~AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Commercial (C) Devices Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . .. DoC to +75°C Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V Ambient Temperature with Power Applied ............. -55°C to + 125°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V DC Input Voltage. . . . . . . . . .. -0.5 V to Vcc + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or 110 Pin Voltage ............... -0.5 V to Vee + 0.5 V Static Discharge Voltage ................. 2001 V Latchup Current (TA= DoC to +75°C) ...... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified Parameter Symbol Parameter Description Test Conditions Min Max VOH Output HIGH Voltage IOH =-2 rnA VIN = VIH or VIL Vee = Min VOL Output LOW Voltage IOL = 8 rnA VIN = VIH or VIL 0.5 IOL =4 rnA Vee = Min 0.33 Unit V 2.4 V 0.1 IOL = 20 JlA V VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) 2.0 VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) 0.8 V IIH Input HIGH Leakage Current VIN = 5.5 V, Vce = Max (Note 2) 10 ilL Input LOW Leakage Current VIN = 0 V, Vce = Max (Note 2) lozH Off-State Output Leakage Current HIGH VOUT = 5.5 V, Vee = Max VIN = VIH or VIL (Note 2) 10 !lA !lA !lA IOZL Off-State Output Leakage Current LOW VOUT = 5.5 V, Vee = Max VIN = VIH or VIL (Note 2) -10 !lA Isc Output Short-Circuit Current VOUT = 0.5 V, Vee = Max (Note 3) -130 rnA Icc Supply Current VIN = 0 V, Outputs Open (lOUT = 0 rnA) Vec= Max 100 rnA -10 -30 Notes: 1. These are absolute values with respect to device ground all overshoots due to system and/or tester noise are included. 2.. //0 pin leakage is the worst case of hL and IOZL (or IrH and IOZH ). 3. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-410 PALCE29MA16H-25 (Corn'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN COUT Parameter Description Test Conditions Typ Unit Input Capacitance VIN= 0 V I Vcc = 5.0 V, TA = 25°C, 5 pF Output Capacitance VOUT= 0 V If = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS Registered Operation Parameter Symbol Parameter Description Min Max Unit 25 ns Combinatorial Output tPD Input or 1/0 Pin to Combinatorial Output Output Register - Pin Clock tSOR Input or VO Pin to Output Register Setup tCOR Output Register Clock to Output tHOR Data Hold Time for Output Register 15 ns 15 0 ns ns Output Register - Product Term Clock tSORP 1/0 Pin or Input to Output Register Setup tcoRP Output Register Clock to Output tHORP Data Hold Time for Output Register 4 ns 29 10 ns ns Input Register - Pin Clock tSIR 1/0 Pin to Input Register Setup tCIR Register Feedback Clock to Combinatorial Output tHIR Data Hold time for Input Register 2 ns 28 ns 6 ns Clock and Frequency tCIS Register Feedback (Pin Driven Clock) to Output Registerllatch (Pin Driven) Setup 20 ns tCISPP Register Feedback (PT Driven Clock) to Output RegisterILatch (PT Driven) Setup 25 ns fMAX Maximum Frequency (Pin Driven) 1/(tSOR + tCOR) 33.3 MHz fMAXI Maximum Internal Frequency (Pin Driven) 1ltcis 50 MHz fMAXP Maximum Frequency (PT Driven) 1/(tSORP + tCORP) 30 MHz fMAXIPP Maximum Internal Frequency (PT Driven) 1ltclsPp 40 MHz 8 ns tCWH Pin Clock Width HIGH tcW\.. Pin Clock Width LOW 8 ns tCWHP PT Clock Width HIGH 12 ns tCWLP PT Clock Width LOW 12 ns PALCE29MA16H-25 (Com'l) 2-411 ~ AMD ClK r--......pr--... tCIS ------ IsIR ... AND-OR Array --------Output Register Input Register teOR , va " 1<"':>1--------1",; VO .. ~ - - - - - - - - -.. tCIR VO tSOR tpD 110 c: . . . t - - - - - - - t < -------I----t------- tpD 08811G-20 Input/Output Register Specs (Pin ClK Reference) ClK Inputr:>----------. Input Register -tCORP ,, VO , , VO " tSORP - - - - - - - - - - " VO 110 tpD - - - - - - - t - - - - t - - - - - - - t p D 08811G-21 Input/Output Register Specs (PT ClK Reference) 2-412 PAlCE29MA 16H-25 AMD~ SWITCHING WAVEFORMS )j(1iT- Combinatorial Input _ _ _ _ _ _ _ ~·'~~tPD ""V- - - Combinatorial Output T 08811G-22 Combinatorial Output Combinatorial *l'::tSOR--'evtHT-O-R~""':',_~T~~~~=== v 1{v Input - - - - , Clock . _,-VT___ Registered Output 08811G-23 Output Register (Pin Clock) Combinatorial Input ~~y. ~~VT J\ J\ T . t SORP.... Combinatorial Input as Clock ''I )~ V T tHORP teoR Registered Output 08811G-24 Output Register (PT Clock) Registered ~v Input - - - - , ~v I.:-Tt SiR ..... ~ tHIR-::j ,_T____ Clock Combinatorial Output VT _~VT___ 08811G-25 Input Register PALCE29MA 16H-25 2-413 ~AMD SWITCHING WAVEFORMS Clock 08811G-26 Pin Clock Width Combinatorial Input as Clock 08811G-27 PT Clock Width 2-414 PALCE29MA 16H-25 AMD~ SWITCHING CHARACTERISTICS Latched Operation Parameter Symbol Parameter Description Min Max Unit Combinatorial Output tPD Input or I/O Pin to Combinatorial Output 25 ns tPTD Input or I/O Pin to Output via Transparent Latch 28 ns Output Latch - Pin LE 15 tsOl Input or I/O Pin to Output Register Setup tGOL Latch Enable to Transparent Mode Output tHOL Data Hold Time for Output Latch 0 ns tSTL Input or I/O Pin to Output Latch Setup via Transparent Input Latch 18 ns ns 15 ns Output Latch - Product Term LE tSOLP Input or I/O Pin to Output Latch Setup tGOLP Latch Enable to Transparent Mode Output 4 ns 29 ns tHOLP Data Hold Time for Output Latch 10 ns tSTLP Input or I/O Pin to Output Latch Setup via Transparent Input Latch 10 ns Input Latch - Pin LE 2 tSIL I/O Pin to Input Latch Setup tGIL Latch Feedback, Latch Enable Transparent Mode to Combinatorial Output tHIL Data Hold Time for Input Latch ns 28 ns 6 ns Latch Enable tGIS Latch Feedback (Pin Driven) to Output Register/Latch (Pin Driven) Setup 20 ns tGISPP Latch Feedback (PT Driven) to Output Register/Latch (PT Driven) Setup 25 ns tGWH Pin Enable Width HIGH 8 ns tGWL Pin Enable Width LOW 8 ns tGWHP PT Enable Width HIGH 12 ns tGWLP PT Enable Width LOW 12 ns PALCE29MA16H-25 (Com") 2-415 ~AMD AND-OR r---'----, ~£ll§ Array __ ________ _ tSTL--~----~~--~------4_--~ VO tSOL tPTD tpD '" VO ,-----------~ tGIL ' - - - - - - - tpTD --------------4-------+-------------~tpD 08811G-28 Input/Output Latch Specs (Pin LE Reference) IT INPUT tSTLP --~--:---~~---t-------+----~ VO VO " t SOLP - - - - - - - - - - - ' , tpTD - - - - - - - - - - - " t PTD t PD --------------~------+-------------- tpD 08811G-29 Input/Output Latch Specs CPT LE Reference) 2-416 PALCE29MA 16H-25 AMD~ SWITCHING WAVEFORMS Latched Input -------[IPTD- IT ---~--EVT Input -t-G-IS-------- Latch Combinatorial Input Combinatorial Output Latched Output Transparent Latched VT tP_T_D_~~~ _____________________ IT ___ Tr_~_s~pa_re_n_t_ _ _ _J _ - - - - Output Latch Latched VT OBBllG-31 Input and Output Latch Relationship 08811G-30 Latch (Transparent Mode) Latched Input Latched Input = t v ._T______________________ ..-tSTLP Combinatorial Input Combinatorial Input ~--+--"'" Combinatorial Input as LE _ _+-__ 1'-______-'1 "'--_ _ __ Latched Output Latched Output ___-t"" 08811G-33 08811G-32 Output Latch (Pin IE) Output Latch (PT IE) Latched Input LATCHED TRANSPARENT 08811G-35 Combinatorial Output Pin LE Width 08811G-34 Input Latch (Pin 'IT) Latched Combinatorial Input as LE ~v I Transparent v PT LE Width Note: ~ --'1:..=tGWLP...;I~tGWHP-..:I~ 08811G-36 1. If the combinatorial input changes while LE is in the latched mode and LE goes into the transparent mode after tPTo ns has elasped, the corresponding latched output will change tGOL ns after LE goes into the transparent mode. If the combinatorial input changes while LE is in the latched mode and LE goes into the transparent mode before tPTD ns has elapsed, the corresponding latched output will change at the later of the following - tPTD ns after the combinatorial input changes or tGOL ns after LE goes into the latched mode. PALCE29MA 16H-25 2-417 ~AMD SWITCHING CHARACTERISTICS Reset/Preset , Enable Parameter Symbol Min Parameter Description Max Unit 30 ns tAPa Input or 1/0 Pin to Output Registerllatch ResetlPreset tAW Asynchronous Reset/Preset Pulse Width 15 ns tARO Asynchronous Reset/Preset to Output Registerllatch Recovery 15 ns tARI Asynchronous Reset/Preset to Input Registerllatch Recovery 12 ns tARPO Asynchronous Reset/Preset to Output Registerllatch Recovery PT ClocklLE 4 ns tARPI Asynchronous Reset/Preset to Input Registerllatch Recovery PT Clock!LE 6 ns Output Enable Operation 20 tpzx 1I0E Pin to Output Enable tpxz IIOE Pin to Output Disable (Note tEA Input or 1/0 to Output Enable via PT tER Input or 1/0 to Output Disable via PT (Note 1) 1) ns 20 ns 25 ns 25 ns Note: 1. Output disable times do not include test load RC time constants. SWITCHING WAVEFORMS ....-tAW~ Combinatorial Asynchronous Reset/Preset Registered! Latched Output r Clock _ _ _ _ _ _ _ _ _ _ _'_·__ Vr...... 08811G-37 Pin 11 Combinatorial! Registeredl Latched Output Output Register/Latch Reset/Preset Combinatorial Asynchronous ResetIPreset Clock k'AW-t- ----~ ~ ~-tA-R-I-~-·-I,---v-rlVr 08811G-38 Combinatorial Input Combinatoriall -----"'T""~i-\.I Registeredl HH-+...-:..:..........-+++-f vr Latched _----""""-"""-""""'1 Output 08811G-40 Input Register/Latch Reset/Preset 2·418 08811G-39 Pin 11 to Output Disable/Enable PALCE29MA16H-25 (Com'l) Input to Output Disable/Enable AMD~ KEY TO SWITCHING WAVEFORMS. WAVEFORM INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from H to L Will be Changing from H to L /7777 May Change from L to H Will be Changing from L to H Don't Care, Any Change Permitted Changing, State Unknown Does Not Apply Center Line is HighImpedance "Off" State xxxxxx KSOOOO10-PAL SWITCHING TEST CIRCUIT Sl 5V~o---....., Output 0 - - - - - 4 1 1 - - - - - _ 08811G-41 Specification Switch S1 tPD, teo, tGOL Closed tEA, tpzx Z~H: tER, tpxz open Z~L: closed H~Z: open L~Z: closed CL R1 R2 Measured Output Value 1.5 V 1.5 V 35 pF 470Q 5 pF 390Q H~Z: VOH -0.5 V L~Z: VOL PALCE29MA16H·25 +0.5 V 2·419 ~AMD PRELOAD The PALCE29MA 16 has the capability for product-term Preload. When the global-preload product term is true, the PALCE29MA16 will enter the preload mode. This feature aids functional testing by allowing direct setting of register states. The· procedure for Preload is as follows: Set the selected input pins to the user selected preload condition. • Apply the desired register value to the I/O pins. .. This sets a of the register. The value seen on the 110 pin, after Preload, will depend on whether the macrocell is active high or active low. • Parameter Symbol • • • Pulse the clock pin (pin 1). Remove the inputs to the 1/0 pins. Remove the Preload condition. • Verify VaNoH for all output pins as per programmed pattern. Because the Preload command is a product term, any input to the array can be used to set Preload (including I/O pins and registers). Preload itself will change the values of the I/O pins and registers. This will have unpredictable results. Therefore, only dedicated input pins should be used for the Preload command. Parameter Description Min Ree. Max Unit to Delay Time 0.5 1.0 5.0 Jis tw Pulse Width 250 500 700 ns tvo Valid Output 100 500 ns Inputs Preload Mode 110 Pins 14--- to ---~ - - - - - - - - VIH elK Pin 1 (2) ' - - - - - - - - - - - - VIL 08811G-43 Preload Waveform 2-420 PALCE29MA16H-25 AMD~ OBSERVABILITY The PALCE29MA 16 has the capability for product-term Observability. When the global-Observe product term is true, the PALCE29MA 16 will enter the Observe mode. This featu re aids functional testing by allowing direct observation of register states. When the PALCE29MA16 is in the Observe mode, the output buffer is enabled and the I/O pin value will be Q of the corresponding register. This overrides any OE inputs. The procedure for Observe is: • Remove the inputs to all the I/O pins. Parameter Symbol • Set the inputs to the, user selected, Observe configuration. • The register values will be sent to the corresponding I/O pins. • Remove the Observe configuration from the selected I/O pins. Because the Observe command is a product term, any input to the array can be used to set Observe (including I/O pins and registers). If I/O pins are used, the observe mode could cause a value change, which would cause the device to oscillate in and out of the Observe mode. Therefore, only dedicated input pins should be used for the Observe command. Parameter Description Min Rec. Max Unit to Delay Time 0.5 1.0 5.0 Ils tllO Valid Output 100 500 ns Input Pins t= tD =ise~e Mode ~tlOl VIH VIL VOH I/O Pins VOL - - - - VIH elK VIL Pin 1 (2) 08811G-44 Observability Waveform PALCE29MA 16H-25 2-421 ~ AMD POWER-UP RESET The registered devices in the AMD PAL Family have been designed with the capability to reset during system power-up. Following power-up, all registers will be reset to LOW. The output state will depend on the polarity of the output buffer. This feature provides extra flexibility to the designer and is especially valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the asynchronous operation of the power-up reset, and the wide Parameter Symbol range of ways Vee can rise to its steady state, two conditions are required to ensure a valid power-up reset. These conditions are: • The Vee rise must be monotonic. • Following reset, the clock input must not be driven from LOW to HIGH until all applicable input and feedback setup times are met. Parameter Description Min tPR Power-Up Reset Time ts Input or Feedback Setup Time tw Clock Width tR Vee Rise Time ~t~~-- Jls Vee tPR ----.1.1 ZI \t tw Power-Up Reset Waveform 2-422 Jls 500 tR Clock Unit 10 See Switching Characteristics Power Registered Active LOW Output Max PALCE29MA 16H-25 08811G-45 AMD~ TYPICAL THERMAL CHARACTERISTICS Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Sjc Sja Sjma Typ Parameter Description Thermal impedance, junction to case PLCC Unit 17 11 °elW 63 51 °elW 200 Ifpm air 60 43 400 Ifpm air 52 38 600 Ifpm air 43 34 800 Ifpm air 39 30 °elW °elW °elW °elW Thermal impedance, junction to ambient Thermal impedance, junction to ambient with air flow SKINNYDIP Plastic Sic Considerations The data listed for plastic Sjc are for reference only and are not recommended for use in calculating junction temperatures. The heat-flow paths in plastic-encapsulated devices are complex, making the Sjc measurement relatIve to a specific location on the package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom center of the package. Furthermore, Sjc tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a similar environment. PALCE29MA16H-25 2-423 _ COM'L: H-15/25 ~ MIL: H-20 • Advanced Micro Devices PALCE610 Family EE CMOS High Performance Programmable Array Logic DISTINCTIVE CHARACTERISTICS • AMD's Programmable Array Logic (PAL) architecture • Asynchronous clocking via product term or bank register clocking from external pins • Electrically-erasable CMOS technology providing half power (90 rnA Icc) at high speed • • Register preload for testability Power-up reset for Initialization • Space-saving 24-pln SKINNYOIP and 28-pln PLCC packages -15 = 15 ns tPD -25 = 25 ns tPD • Sixteen macrocells with conflgurable 1/0 architecture • Fully tested for 100% programming yield and high reliability • • Registered or combinatorial operation Registers programmable as 0, T, J-K, or S-R • Extensive third-party software and programmer support through FusionPLO partners GENERAL DESCRIPTION The PALCE610 is a general purpose PAL device and is functionally and fuse map equivalent to the EP610. It can accommodate logic functions with up to 20 inputs and 16 outputs. There are 16 I/O macrocells that can be individually configured to the user's specifications. The macrocelis can be configured as either registered or combinatorial. The registers can be configured as 0, T, J-K, or S-R flip-flops. The PALCE610 uses the familiar sum-of-products logic with programmable-AND and fixed-OR structure. Eight product terms are brought to each macroceli to provide logic implementations. The PALCE610 is manufactured using advanced CMOS EE technology providing high density and low power consumption. Moreover, it is a high-speed device having a worst-case tPD of 15 ns. Space-saving 24-pin SKINNYDIP and 28-pin PLCC packages are offered. This device can be quickly erased and reprogrammed providing for easy prototyping. Once a device is programmed the security bit can be used to provide protection from copying a proprietary design. BLOCK DIAGRAM 1/016 CLK2 2-424 II0a 1107 1/06 lIas 1/012 1/011 1/04 1/03 1/010 1/0g 1/02 1/01 Publication# 12950 Rev. F Issue Dale: June 1993 CLK1 12950F-1 AmendmenllO AMD~ CONNECTION DIAGRAMS Top View SKINNVDIP CLK1 PLCC/LCC Vee ;:: 1I0g ()O .... ...J 0 0 0 ~'" - 0 »-::. I 1/01 1/010 1/02 1/011 1/03 1/012 1104 1/013 1105 1/014 1/015 1/06 1/07 1/016 II0a GND CLK2 25 24 23 22 21 20 1/0 10 1/° 11 1/° 12 1/° 13 1/°14 1/°15 NC 1/°2 1/°3 1/°4 1/°5 1/°6 1/°7 NC ~-QQ~- o ::::. zz .... (!)(!)o co ~ 12950F-3 12950F-2 Nots: Pin 1 is marked for orientation PIN DESIGNATIONS ClK GND I = Clock = Ground = Input 1/0 = Input/Output NC = No Connect Vee = Supply Voltage PAlCE610 Family 2·425 ~ AMD ORDERING INFORMATION Commercial Products AMD programmable logic products for commercial applications are available with several ordering options. The order number (Valid Combination) is formed by a combination of: PAL CE 610 H -15 T..J FAMILYTYPE _ _ _ _ _ PAL = Programmable Array Logic TECHNOLOGY-------------..J CE = CMOS Electrically Erasable P C L OPERATING CONDITIONS C = Commercial (O°C to +75°C) PACKAGE TYPE P = 24-Pin 300 mil Plastic SKINNYDIP (PD3024) J = 28-Pin Plastic Leaded Chip Carrier (PL 028) DEVICE NUMBER 610= 600 Gates POWER _____________________-..J H = Half Power (90 mA Icc) SPEED -15 = 15 ns tPD -25 = 25 ns tPD Valid Combinations PALCE610H-15 PC,JC PALCE610H-25 2-426 Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMD sales office to confirm availability of specific valid combinations, and to check on newly released combinations. PALCE610H-15/25 (Com'l) AMD~ ORDERING INFORMATION APL Products AMO programmable logic products for Aerospace and Defense applications are available with several ordering options. APL (Approved Products List) products are fully compliant with MIL-STO-883 requirements. The order number (Valid Combination) is formed by a combination of: PAL CE610 H ·20/B L A t= FAMILYTYPE _ _ _ _ _T..J PAL. Programmable Array Logic TECHNOLOGY------------..J CE - CMOS Electrically Erasable DEVICE NUMBER 610 .. 600 Gates POWER ------------------~ LEAD FINISH A - Hot Solder ~ip, PACKAGE TYPE L - 24-Pin (300 mil) Ceramic SKINNYOIP (C03024) 3 - 28-Pin Ceramic Leadless Chip Carrier (CL 028) DEVICE CLASS H = Half Power (90 mA Icc) 18 = Class B SPEED -20 = ,20 ns tPD Valid Combinations PALCE610H-20 I I8LA,183A Valid Combinations Valid Combinations lists configurations planned to be supported in volume for this device. Consult the local AMO sales office to confirm availability of specific valid combinations, and to check on newly released combinations. Group A Tests Group A tests consist of Subgroups 1,2,3,~8,~ 1~ 11. Military Burn-In Military burn-in is in accordance with the current revision of MIL-STO-883, Test Method 1015, Conditions A through E. Test conditions are selected at AMO's option. PALCE610H·20 (Mil) 2·427 ~ AMD FUNCTIONAL DESCRIPTION The PAlCE610 is a general purpose programmable logic device. It has 16 independently-configurable macroce lis. Each macrocell can be configured as either combinatorial or registered. The registers can be 0, T, J-K, or S-R type flip-flops. The device has 4 dedicated input pins and 2 clock pins. Each clock pin controls 8 of the 16 macrocells. The programming matrix implements a programmable AND logic array which drives a fixed OR logic array. Buffers for device inputs have complementary outputs to provide user-programmable input polarity. Unused input pins should be tied to Vee or ground. The array uses AMD's electrically erasable technology. An unprogrammed bit is disconnected and a programmed bit is connected. Product terms with all bits unprogrammed assume the logical-HIGH state and product terms with both the TRUE and Complement bits programmed assume the logical-lOW state. The programmable functions in the PAlCE610 are automatically configured from the user's design specifications, which can be in a number of formats. The design specification is processed by development software to verify the design and create a programming file. This file, once downloaded to the programmer, configures the design according to the user's desired function. In The 0 and T configurations, feedback can be either from Q or the output pin. This allows 0 and T configurations to be either outputs or I/O. In the J-K and S-R configurations, feedback is only from Q: therefore, J-K and S-R configurations are strictly outputs. D Flip-Flop All 8 product terms are available to the OR gate. The 0 input polarity is controlled by an exclusive-OR gate. For the D flip-flop, the output level is the D-input level at the rising edge of the clock. an an+1 D 0 0 0 0 1 1 0 0 1 1 1 1 T Flip-Flop All 8 product terms are available to the OR gate. The T input polarity is controlled by an exclusive-OR gate. For the T register, the output level toggles when the T input is HIGH and remains the same when the T input is lOW. an an+1 T 0 0 0 Macrocell Configurations 0 1 1 The PAlCE610 macrocell can be configured as either combinatorial or registered. Both the combinatorial and registered configurations have output polarity control. The register can be configured as a 0, T, J-K, or S-R type flip-flop. Figure 1 shows the possible configurations. 1 0 1 1 1 0 Each macrocell can select as its clock either the corresponding clock pin or the ClK/OE product term. If the clock pin is selected, the output enable is controlled by the ClK/OE product term. If the ClK/OE product term is selected, the output is always enabled. J-K Flip-Flop The 8 product terms are divided between the J and K inputs. N product terms go to the J input and 8-N product terms go to the K input, where N can range from 0 to 8. Both the J and K inputs to the flip-flop have polarity control via exclusive-OR gates. The J-K flip-flop operation is shown below. Combinatorial I/O All 8 product terms are available to the OR gate. The output-enable function is performed by the elK/OE product term. Registered Configurations There are 4 flip-flop types available: 0, T, J-K and S-R. The registers can be configured as synchronous or asynchronous. In the synchronous configuration, the clock is controlled by the clock input pin. The output enable is controlled by the product term function. In the asynchronous configuration, the clock input is controlled by the product term. The output is always enabled. 2-428 PAlCE610 Family an+1 J K an 0 0 0 0 0 1 1 0 1 0 0 0 1 0 1 1 0 0 1 1 0 1 1 1 1 0 1 1 1 1 0 0 AMD~ Combinatorial Vee Vee CLK------1 CLK----t o Register T Register Vee ----------1 CLK------1 Vee --------~ CLK------1 J-K Reg ister S-R Register 12950F-4 Figure 1. Macrocell Configurations PALCE610 Family 2·429 ~ AMD S-R Flip-Flop The 8 product terms are divided between the Sand R inputs. N product terms go to the S input and 8-N product terms go to the R input, where N can range from 0 to 8. Both the Sand R inputs to the flip-flop have polarity control via exclusive-OR gates. The S-R flip-flop operation is shown below. S R an an+1 0 0 0 0 0 0 1 1 0 1 0 0 1 0 0 0 1 0 1 0 1 1 through long test vector sequences to reach a desired state. In addition, transitions from illegal states can be verified by loading illegal states and observing proper recovery. Security Bit After programming and verification, a PALCE610 design can be secured by programming the security bit. Once programmed, this bit defeats readback of the internal programmed pattern by a device programmer, securing proprietary designs from competitors. However, programming and verification are also defeated by the security bit. The bit can only be erased in conjunction with the array during the erase cycle. Preload is not affected by the security bit. Technology Not Allowed Asynchronous Reset All flip-flops have an asynchronous-reset product-term input. When the product term is true, the flip-flop will reset to a logic LOW, regardless of the clock and data inputs. Power-Up Reset All flip-flops powerupto a logic LOWforpredictable system initialization. Outputs of the PALCE61 0 depend on whether they are selected as registered or combinatorial. If registered is selected, the output will be LOW. If combinatorial is selected, the output will be a function of the logic. The Vcc rise must be monotonic and the reset delay time is 1000 ns maximum. The PALCE610 is manufactured using, AMD's advanced Electrically Erasable (EE) CMOS process. This technology uses an EE cell to replace the fuse link in bipolar parts, and allows AMD to offer lower-power parts of high complexity. In addition, since the EE cells can be erased and reprogrammed, these devices can be 100% factory tested before being shipped to the customer. Inputs and outputs are designed to be compatible with TTL devices. This technology provides strong input clamp diodes, output slew-rate control, and a grounded substrate for clear switching. Programming and Erasing Register Preload The PALCE610 can be programmed on standard logic programmers. It also may be erased to reset a previously configured device back to its virgin state. Bulk erase is automatically performed by the programming hardware. No special erase operation is required. The register on the PALCE610 can be preloaded from the output pins to facilitate functional testing of complex state machine designs. This feature allows direct loading of arbitrary states, making it unnecessary to cycle The PALCE610 has CMOS-compatible outputs. The output voltage (VOH) is 3.85 V at -2.0 rnA. 2-430 CMOS Compatibility PALCE610 Family AMD~ PALCE610 LOGIC DIAGRAM DIP (PLCC) Pinouts 49 ~ C LKI ~ IN PUT £[ I (3) v 11O.~ Macrocell NODEl -- (1,2111 r=I ~ 0 80 ~ Macrocell I.n AR OEICLK !ill .... ~ 89 I 90 v 9 R;: 11O,0W- Macrocell Macrocell NODE2 AR AR OEICLK 99 ~ I 110 2 NO DE 15 OEICLK T 100 20 Macrocell NODE3 ~ A V 11O"W- ~ 19 NODE16 A I 10 ~ INPUT 110, Macrocell AR AR OEICLK ~ NODE14 OEICLK 109 29 A I I 110 30 V W""" 110 12 Macrocell Macrocell NODE4 ~ NODE13 AR OEICLK 119 DFJ!' ~ IC 39 A I W- 110 13 I 40 120 V Macrocell NO DES Macrocell AR AR OEICLK 129 ~ 1105 NODE 12 OEICLK 49 A I W- 11O .. I 130 50 V Macrocell Macrocell NODE6 AR AR OEICLK 139 ~ NODEll OEICLK 59 A I v IIO,s~ I 60 1140 Macrocell NODE7 Macrocell AR AR OEICLK ~ NODE10 OEICLK 149 69 A I v 11O,.~ 1 150 70 Macrocell NODE8 f-ffij1 110, Macrocell NODE9 AR OEICLK 159 79 A INP UTmf GND v liD--. (14.T5) i:- 0 8 16 • 24 32 16 39 40 • 24 32 39 I '""'"' ~ ml INPUT CLK2 12950F-5 PALCE610 Family 2-431 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES - Storage Temperature .......... -65°C to + 150°C Commercial (C) Devices Ambient Temperature with Power Applied ............ -55°C to + 125°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V DC Input Voltage ........... -0.5 V to Vee + 0.5 V DC Output or I/O Pin Voltage ............ -0.5 V to Vee + 0.5 V Static Discharge Voltage Ambient Temperature (TA) Operating in Free Air. . . . . . . . . . .. O°C to +75°C Supply Voltage (Vee) with Respect to Ground ..... +4.75 V to +5.25 V Operating ranges define those limits between which the functionality of the device is guaranteed. ............... 2001 V Latehup Current (TA= O°Cto +75°C) .................... 100 rnA Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Max imum Ratings for extended periods may affect device reliabi lity. Programming conditions may differ. DC CHARACTERISTICS over COMMERCIAL operating ranges unless otherwise specified (Note 2) Parameter Symbol Test Conditions Output HIGH Voltage VIN = VIH or VIL Vee = Min IOH = -4.0 rnA IOH =-2.0 rnA VOL Output LOW Voltage VIN = VIH or VIL Vee = Min IOL = 8.0 rnA IOL = 4.0 rnA VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 1) VIL Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 1) 0.8 V hH Input HIGH Leakage Current VIN = 5.25 V, Vec = Max (Note 2) 10 ~ Input LOW Leakage Current VIN = 0 V, Vec= Max (Note 2) -10 ~ loZH Off-State Output Leakage Current HIGH VOUT = 5.25 V, Vee = Max VIN = VIH or VIL (Note 2) 10 ~ loZL Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or VIL (Note 2) -10 ~ Isc Output Short-Circuit Current VOUT = 0.5 V, Vec = Max (Note 3) -150 rnA lee Supply Current VIN = 0 V, Outputs Open (lOUT = 0 rnA) Vee = Max 90 rnA . IlL Min Max Unit Parameter Description VOH 2.4 3.84 V V 0.5 0.45 2.0 -30 V V V Notes: 1. These are absolute values with respect to device ground and all overshoots due to system and tester noise are included. 2. 1/0 pin leakage is the worst case of liL and 10ZL (or liH and 10zH). 3. Not more than one output should be tested at a time. Duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. 2-432 PALCE610H-15/25 (Com'l) AMD~ CAPACITANCE (Note 1) Parameter Symbol CIN GoUT Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance VOUT= 2.0 V Typ Unit 8 Vee = 5.0 V TA = +25°C f = 1 MHz 8 pF Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over COMMERCIAL operating ranges (Note 2) -15 Parameter Symbol Parameter Description Min tPD Input or Feedback to Combinatorial Output ts Setup Time from Input or Feedback to Clock tH Hold Time teo Clock to Output tWL tWH Clock Width fMAx Maximum Frequency (Note 3) -25 Max Min 15 12 15 0 0 Max Unit 25 ns ns ns 12 8 ns LOW 6 10 HIGH 6 10 ns 50 76.1 83.3 37 40 50 MHz MHz MHz I External Feedback 1/(ts + teo) Internal Feedback (feNT) No Feedback 1/(twH + tWL) I ns tEA Input to Output Enable Using Product Term Control 15 25 tER Input to Output Disable Using Product Term Control 15 25 ns tAR Asynchronous Reset to Registered Output 15 25 ns tARW Asynchronous Reset Width tARR Asynchronous Reset Recovery Time tSA Setup Time from Input or Feedback to Clock (Note 4) 5 8 tHA Hold Time (Note 4) 5 12 tCOA Clock to Output (Note 4) tWLA Clock Width LOW (Note 4) 6 10 HIGH (Note 4) p 10 ns Maximum Frequency (Notes 3 and 4) External Feedback 1/(tsA + teoA) Internal Feedback (feNT) No Feedback 1/(twLA + tWHA) 50 61.6 83.3 28.6 29.4 50 MHz MHz MHz tWHA fMAXA 10 15 ns 15 I ns ns ns 27 15 I ns 25 ns ns Notes: 2. See Switching Test Circuit for test conditions. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. 4. These parameters are measured using the asynchronous product-term clock. PALCE610H-15/25 (Com'l) 2-433 ~ AMD ABSOLUTE MAXIMUM RATINGS OPERATING RANGES Storage Temperature ........... -65°C to +150°C Military (M) Devices (Note 1) Operating Case Temperature (Te) ........... -55°C to +125°C Ambient Temperature with Power Applied ............. -55°C to + 125°C Supply Voltage with Respect to Ground ............. -0.5 V to +7.0 V Supply Voltage (Vee) with Respect to Ground ....... +4.5 V to +5.5 V DC Input Voltage ........... -0.5 V to Vee + 0.5 V Operating ranges define those limits between which the functionality of the device is guaranteed. DC Output or I/O Pin Voltage ............... -0.5 V to Vee + 0.5 V Note: 1. Military products are 100% tested at Tc and -5SOC, per MIL-STD-883. , Static Discharge Voltage ................. 2001 V Latchup Current (Te = -55°C to + 125°C) ................ 100 rnA = +2SOC, + 12SOC Stresses above those listed under Absolute Maximum Ratings may cause permanent device failure. Functionality at or above these limits is not implied. Exposure to Absolute Maximum Ratings for extended periods may affect device reliability. Programming conditions may differ. Absolute Maximum Ratings are for system design reference; parameters given are not tested. I DC CHARACTERISTICS over MILITARY operating ranges unless otherwise specified Parameter Symbol VOH Parameter Description Test Conditions Output HIGH Voltage VIN = VIH or Vil Vee = Min Min I IOH =-2 rnA IloH=-1 rnA Val Output LOW Voltage IOl = 4 rnA Vee = Min VIH Input HIGH Voltage Guaranteed Input Logical HIGH Voltage for all Inputs (Note 3) Vil Input LOW Voltage Guaranteed Input Logical LOW Voltage for all Inputs (Note 3) IIH Input HIGH Leakage Current VIN = 5.5 V, Vee = Max (Note 4) III Input LOW Leakage Current VIN = 0 V, Vee = Max (Note 4) Off-State Output Leakage Current HIGH VOUT =5.5 V, Vee = Max VIN = VIHor Vil (Note 4) lozl Off-State Output Leakage Current LOW VOUT = 0 V, Vee = Max VIN = VIH or Vil (Note 4) Ise Output Short-Circuit Current VOUT = 0.5 V, Vee = 5.0 V, TA = 25° C (NoteS) Icc Supply Current Outputs Open (louT= 0 A) Vee = Max loZH Max 0.5 VIN = VIH or Vil 2.0 V V 0.8 -30 Unit V 2.4 3.84 V 10 ~ -10 ~ 10 JlA -10 JlA -150 rnA 90 rnA Notes: 2. For APL products, Group A, Subgroups 1,2 and 3 are tested per MIL-STD-833, Method 5005, unless otherwise noted. 3. VIL and VIH are input conditions of output tests and are not themselves directly tested. VIL and VIH are absolute voltages with respect to device ground and include all overshoots due to system andlor tester noise. Do not attempt to test these values without suitable equipment. 4. 110 pin leakage is the worst case of hL and 10zL (or hH and 10ZH). 5. Not more than one output should be shorted at a time and duration of the short-circuit should not exceed one second. VOUT = 0.5 V has been chosen to avoid test problems caused by tester ground degradation. This parameter is not 100% tested, but is evaluated at initial characterization and at anytime the design is modified where los may be affected. 2-434 PALCE610H-20 (Mil) AMD l1 CAPACITANCE (Note 1) Parameter Symbol CIN CoUT Parameter Description Test Conditions Input Capacitance VIN = 2.0 V Output Capacitance VOUT = 2.0 V Typ. Vee TA = +25°C f = 1 MHz Unit 8 pF 8 Note: 1. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where capacitance may be affected. SWITCHING CHARACTERISTICS over MILITARY operating ranges (Note 2) -20 Parameter Symbol tPD Parameter Description Input or Feedback to Combinatorial Output ts Setup Time from Input or Feedback to Clock tH Hold Time 0 teo Clock to Output 10 tWL tWH fMAX Clock Width Maximum Frequency (Note 3) LOW Min Max 20 Unit ns 15 ns ns ns ns 8 HIGH I External Feedback 1t'{ts + teal Internal Feedback (feNT) No Feedback 1/(tWH + twd I 8 ns 40 50 62.5 MHz MHz MHz tEA Input to Output Enable Using Product Term Control (Note 3) 20 ns tER Input to Output Disable Using Product Term Control (Note 3) 20 ns tAR Asynchronous Reset to Registered Output 20 ns tARW Asynchronous Reset Width (Note 3) tARR Asynchronous Reset Recovery Time (Note 3) tSA Setup Time from Input or Feedback to Clock (Note 4) 8 tHA Hold Time (Note 4) 10 tCOA Clock to Output (Note 4) 20 tWLA tWHA fMAXA Clock Width Maximum Frequency (Notes 3 and 4) 20 ns 20 ns ns ns ns LOW (Note 4) 8 HIGH (Note 4) 8 ns 35.8 45 52.6 MHz MHz MHz I External Feedback 1/(tSA + tcOA) Internal Feedback (teNT) No Feedback 1/(tWLA + tWHA) I ns Notes: 2. See Switching Test Circuit for test conditions. For APL Products, Group A, Subgroups 9, 10, and 11 are tested per MIL-STD-883, Method 5005, unless otherwise noted. 3. These parameters are not 100% tested, but are evaluated at initial characterization and at any time the design is modified where frequency may be affected. 4. These parameters are measured using the asynchronous product-term clock. PALCE610H-20 (Mil) 2-435 ~ AMD SWITCHING WAVEFORMS Vr Input or Feedback _ _ _- - I Input or Feedback Combinatorial Output - -~-vr .....---.. tH CO -= V Vr : Ciock _ _ _ _ _ _ _ _..... Registered Outp,ut _ _ _ _ _ _ _ __ 12950F-6 12950F-7 Registered Output Combinatorial Output Input -----I tEA tER Clock Output ------'-.L..I.-1 12950F-9 12950F-8 Input to Output Disable/Enable Clock Width Input or Feedback _ _ _- - I Vr Product-Term Clock _ _ _ _ _ _ _ __ Registered Output _ _ _ _ _ _ _ __ 12950F-10 Clock Width Using Product-Term Clock Input Asserting Asynchronous Reset Registered Output _ _ _L...K~...x...~ tARR Vr Clock 12950F-12 Notes: 1. Vr Asynchronous Reset = 1.5 V 2. Input pulse amplitude 0 V to 3.0 V 3. Input rise and fall times 2 ns-5 ns typical. 2-436 12950F-11 Registered Output Using Product-Term Clock PALCE610 Family AMD 11 KEY TO SWITCHING WAVEFORMS WAVEFORM INPUTS OUTPUTS Must be Steady Will be Steady \\\\\ May Change from Hto L Will be Changing from H to L ///// May Change from L to H Will be Changing from L to H XXXXXX 1B EK Don't Care, Any Change Permitted Changing. State Unknown Does Not Apply Center Line is HighImpedance "Off" State KSOOOO10-PAL SWITCHING TEST CIRCUIT 5V Output D - -.....- - -....--te) Test Point 12950F-13 Commercial Specification tPD, teo tEA Sl Closed 35 pF Z~ H: Open L: Closed 35 pF H ~Z: Open L~ Z: Closed 5 pF Z~ tER CL R1 Military R2 R1 R2 Measured Output Value 1.5 V 1.5 V 855n 340n 855n 340n H ~ Z: VOH - 0.5 V L ~ Z: VOL + 0.5 V PALCE610 Family 2-437 ~ AMD ENDURANCE Symbol tOR N Parameter Description Test Conditions Min Unit Pattern Data Retention Time Max Storage Temperature Max Operating Temperature 10 20 Years Years Reprogramming Cycles Normal Programming Conditions 100 Cycles INPUT/OUTPUT EQUIVALENT SCHEMATICS Vee ESD ProgramNerify Protection Circuitry -= Typical Input Vee 1 Preload Circuitry Feedback Input Typical Output 12950F-15 2·438 PALCE610 Family AMD 11 Power-Up Reset The power-up reset feature ensures that all flip-flops will be reset to LOW after the device has been powered up. This feature is valuable in simplifying state machine initialization. A timing diagram and parameter table are shown below. Due to the synchronous operation of the power-up reset and wide range of ways Vcc can rise to Parameter Symbol its steady state, two conditions are required to insure a valid power-up reset. These conditions are: • The Vee rise must be monotonic. • Following reset, the clock input must not be driven from LOWto HIGH until all applicable input and feedback setup times are met. Parameter Description Max Unit tPR Power-up Reset Time 1000 ns ts Input or Feedback Setup Time tWL Clock Width LOW Power Registered Output Clock See Switching Characteristics Vee 4V~ '~~,- _ _- tPR ~ \\ ~ tWL 12950F-16 Power-Up Reset Waveform PALCE610 Family 2-439 ~ AMD TYPICAL THERMAL CHARACTERISTICS Measured at 25°C ambient. These parameters are not tested. Parameter Symbol Typ SKINNYDIP PLCC Parameter Description Sjc Thermal impedance, junction to case Sja Thermal impedance, junction to ambient Sjma Thermal impedance, junction to ambient with air flow Unit 21 20 °elW 72 57 °elW °elW °elW °elW °elW 200 Ifpm air 64 47 400 Ifpm air 60 44 600 Ifpm air 55 40 800 Ifpm air 49 36 Plastic Sjc Considerations The data listed for plastic Sjc are for reference only and are not recommended for use in calculating junction temperatures. The heat-flow paths in plastic-encapsulated devices are complex, making the Sjc measurement relative to a specific location on the package surface. Tests indicate this measurement reference point is directly below the die-attach area on the bottom center of the package. Furthermore, Sjc tests on packages are performed in a constant-temperature bath, keeping the package surface at a constant temperature. Therefore, the measurements can only be used in a similar environment. 2-440 PALCE610 Family 3 MACH DEVICES MACH 1 and 2 Device Families ............................................. 3-3 MACH110-12115/20 ........................................................ 3-7 MACH120-15/20 .......................................................... 3-9 MACH130-15/20 ........................ ~ ................................ 3-11 MACH210A-10, MACH210-12115/20, MACH210AQ-15/20 ......................... 3-13 MACHLV210A-15/20 ...................................................... 3-15 MACH220-12115/20 ....................................................... 3-17 MACH230-15/20 ......................................................... 3-19 MACH215-12/15/20 ....................................................... 3-21 MACH 3 and 4 Device Families ............................................ MACH435-15/20, 0-25 .................................................... MACH355-15/20 ......................................................... MACH445-15/20 ......................................................... MACH465-15/20 ..............................' ........................... MACH Devices 3-23 3-27 3-29 3-31 3-33 3-1 ~AMD 3-2 MACH Devices - Advanced Micro Devices MACH 1 and 2 Device Families High-Density EE CMOS Programmable Logic DISTINCTIVE CHARACTERISTICS • High-performance, high-density, electrically-erasable CMOS PLD families • • • 900 to 3600 PLD gates 44 to 84 pins In cost-effective PLCC and CQFP packages • 32 to 128 macrocells • 0.8 JlI11 CMOS provides predictable design-independent high speeds • - • • Commercial 10/12/15/20-ns tPD, 80/67/50/40-MHz fMAX - Military 20-ns tPD, 40-MHz fMAX Synchronous and asynchronous devices PAL blocks connected by switch matrix - • Provides optimized global connectivity Switch matrix integrates blocks into uniform device Configurable macrocells - Programmable polarity - Registered or combinatorial - Internal and I/O feedback - D-type or T-type flip-flops - Choice of clocks for each flip-flop - Input registers for MACH 2 family Extensive third-party software and programmer support through FusionPLD partners - Schematic capture and text entry - Compilation and JEDEC file generation - Design simulation - LogiC and timing models - Standard.PLD programmers Each MACH product has a factory programming option available for high-volume applications PRODUCT SELECTOR GUIDE Device MACH 1 Family Pins Macrocells PLD Gates Max Inputs Max Outputs Max Flip-Flops Speed (ns) MACH110 MACH120 MACH130 MACH 2 Family 44 68 84 32 48 64 900 1200 1800 38 56 70 32 48 64 32 48 64 12,15,20 15, 20 15, 20 MACH210 44 64 1800 38 96 128 2400 3600 56 70 32 48 64 64 96 128 10,12,15,20 12,15,20 15, 20 64 1500 38 32 64 12,15,20 MACH220 MACH230 68 84 Asynchronous MACH Device MACH215 44 GENERAL DESCRIPTION The MACH (Macro Array CMOS High-density) family provides a new way to implement large logic designs in a programmable logic device. AMD has combined an innovative architecture with advanced electricallyerasable CMOS technology to offer a device with several times the logic capability of the industry's most popular existing PAL device solutions at comparable speed and cost. Their unique architecture makes these devices ideal for replacing large amounts of TTL, PAL-device, glue, and gate-array logic. They are the first devices to provide such increased functionality with completely predictable, deterministic speed. The MACH devices consist of PAL blocks interconnected by a programmable switch matrix (Figure 1). Designs that consist of several interconnected functional modules can be efficiently implemented by placing the modules into PAL blocks. DeSigns that are not as modular can also be readily implemented since the switch matrix provides a high level of connectivity between PAL blocks. The internal arrangement of resources is managed automatically by the deSign software, so that the designer does not have to be concerned with the logic implementation details. The MACH family consists of the MACH 1 and MACH 2 series of synchronous devices and the MACH215, an MACH 1 and 2 Device Families ~AMD asynchronous device. The MACH 1 and 2 series are ideal for synchronous subsystems like memory controllers and peripheral controllers. The MACH215 is appropriate for applications having asynchronous inputs and for collecting random glue logic. of tools for each new device, but rather can use the tools with which he or she is already familiar. The MACH devices can be programmed on conventional PAL device programmers with appropriate personality and socket adapter modules. AMD's FusionPLD program allows MACH device designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide timely, accurate, quality support. This ensures that a designer does not have to buy a complete new set MACH devices are manufactured using AMD's state-ofthe-art advanced CMOS electrically-erasable process for high performance and logic density. CMOS EE technology provides 100% testability, reducing both prototype development costs and production costs. r-----~ Output: Buried Output: Buried Macro- I Macro- 1----+11> Macro- I Macro- •• ~e!l: .:••c~~s: _ •• :e~l~ .:••c~~s: • PAL Block PAL Block t I t Switch Matrix I 1 PAL Block PAL Block . O~tP~t·:· B~ried ·Output : ·B~ri;d· ........- -..... ~ Macrocells I I Macro- I - - -......~i...... ~ Macrocells· cells I I • •• I ••• Macrocells· 14051G-1 • Buried macrocell available on MACH 2 devices only. Figure 1. MACH 1 and 2 Block Diagram Design Methodology Design tools for MACH devices are widely available both from AMD and from third-party software vendors. AMD provides PALASM software as a low-cost baseline tool set and works with tools vendors to ensure broad MACH device support. This allows designers to do MACH device deSigns using the same tools that they would use to do PAL device designs, whether PALASM software or any of the other popular PAL device design packages. Design entry is the same as that used for PAL devices. The basic logic processing steps are the same steps that are needed to process and minimize logic for any PAL device. Simulation is available for verifying the correct behavior of the device. Functional (unit-delay) simulation of MACH devices is supported in all approved software packages, and other options for simulating the timing and board-level behavior of the MACH devices are available. The end result is a JEDEC file that can be downloaded to a programmer for device configuration., MACH device desian methodoloav differs somewhat from that of a PAL device due to the automatic design 3-4 fitting procedure that the software performs. DeSigns written by logic designers-whether by schematic capture, state machine equations, or Boolean equations-are partitioned and placed into the PAL blocks of the MACH device. While this procedure is handled automatically by the software, the software can also accept manual direction based upon the user's working knowledge of the design. The overall device utilization provided by the Fitterwill vary from deSign to design, but a utilization target of no greater than 70% is recommended. Since a design must be entered and fit in order to calculate actual utilization, it is best to be conservative when estimating utilization before starting a design. AMD recommends allowing the software to decide the best fit and pin· placement automatically for the first design iteration to provide the best chance of fitting. With this approach, large designs can be implemented incrementally, starting with low device utilization and building up by adding logic until the device is full. This generally means that designs are done without any specific pinout assignments. with the final pinout decided by the software. While it is possible to pre-place signals, it is not recommended in most cases. If done carefully, pre-placement can help the software fit MACH 1 and 2 Device Families AMD~ difficult designs; if not done carefully, it may make it harder for the design to fit. Guidelines on specifying the initial pinout are provided in the MACH Technical Briefs book. The design is partitioned and placed into the MACH device by the software so as not to affect the performance of the design. With deSigns that do not fit, it is possible to make some performance tradeoffs to aid in fitting (for example, by optimizing the flip-flop type or passing through the device more than once), but those tradeoffs must be specifically requested, and any additional delays are entirely predictable. Once an initial design fits, there may be subsequent changes to the deSign. This is important if board layout has already started based on the original pinout. Design changes make it necessary to refit the design, which may result in a different pinout. Some deSign changes may make it impossible to refit the deSign, regardless of the pinout. The stability of the design and the expected extent of any changes should therefore be considered before committing the design to layout. Careful designs that target about 70% utilization will make future changes much easier. Higher utilization will make deSign changes much more difficult to implement. Hints on designing for change can be found in the MACH Device Design Planning Guide near the end of this book, and in the article Designing for Change with MACH Devices in the Technical Briefs book. MACH 1 and 2 Device Families 3-5 ~AMD 3·6 MACH 1 and 2 Device Families CONDENSED _ COM'L: -12/15/20 MIL: -20 MACH110-12/15/20 High-Density EE CMOS Programmable Logic ~ Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • 44 Pins 32 Macrocells • 12 ns tPD Commercial 20 nstpD Military • 66.7 MHz fMAX external Commercial 40 MHz 'MAX external Military • 381nputs • • 32 Outputs 32 Flip-Flops; 2 clock choices • 2 "PAL22V16" Blocks • Pin-compatible with MACH210, MACH215 GENERAL DESCRIPTION The MACH110 is a memberof AMD's high-performance EE CMOS MACH 1 family. This device has approximately three times the logic macrocell capability of the popular PAL22V1 0 with no loss of speed. The MACH110 consists of two PAL blocks interconnected by a programmable switch matrix. The two PAL blocks are essentially "PAL22V16" structures complete with product-term arrays and programmable macrocells. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. Publication# C14127 Rev. G AmendmentlO Issue Date: June 1993 The MACH110 macrocell provides either registered or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type or T-type to help reduce the number of product terms. The register type decision can be made by the designer or by the software. All macrocells can be connected to an I/O cell. If a buried macrocell is desired, the internal feedback path from the macrocell can be used, which frees up the liD pin for use as an input. 3-7 ~ CONDENSED AMD BLOCK DIAGRAM 10- h. 13-14 1/00-1/015 16,;~ ... 16 , .... 1/0 Cells 16 16 , .... 1 2.... Macrocells DE 44 4,; ... x 70 AN D Logic Array and Logic Allocator 22 Switch Matrix 22 44 x 70 AND Logic Array and Logic Allocator 2 ... DE .2L Macrocells 16 L...t. I 16 , 1/0 -, 2~ Cells, 16L , 16/ 11016 - 1/031 3-8 MACH 110-12/15/20 CLK1/Is. CLKo/12 C14127G-1 CONDENSED _ ~ COM'L:-15/20 MACH120-15/20 High-Density EE CMOS Programmable Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • 68 Pins 48 Macrocells • 48 Outputs • 48 Flip-flops; 4 clock choices • 15 ns tPD • 4 PAL blocks • 50 MHz fMAX external • Pin-compatible with MACH220 • 56 Inputs GENERAL DESCRIPTION The MACH120 is a memberof AMO's high-performance EE CMOS MACH 1 family. This device has approximately five times the logic macrocell capability of the popular PAL22V10 with no loss of speed. The MACH120 consists of four PAL blocks interconnected by a programmable switch matrix. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. Publication# C14129 Rev. G AmendmentlO Issue Date: June 1993 The MACH120 macrocell provides either registered or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as O-type or T-type to help reduce the number of product terms. The register type decision can be made by the designer or by the software. All macrocells can be connected to an lID cell. If a buried macrocell is desired, the internal feedback path from the macrocell can be used, which frees up the lID pin for use as an input. 3-9 ~ CONDENSED AMD BLOCK DIAGRAM 12 -13. 16-17 52 x 54 52 x 54 AND Logic Array AND Logic Array and Logic Allocator and Logic Allocator 26 ~x~ ~x~ AND Logic Array AND Logic Array and Logic Allocator and Logic Allocator OEr-__~____, OEr-__~____~ ~~-r+--+~ '------a--c::::::"L.....J 4 <~--~~~_I,------' 12 12 1/036 - 1/047 3-10 1/024 -1/035 MACH120-15/20 CLKo/lo, CLK1/11. CLK2/14, CLK3/15 C14129G-1 CONDENSED _ , COM'L:-15/20 MIL: -20 MACH130-15/20 High-Density EE CMOS Programmable Logic ~ Advanced Micro De'vices DISTINCTIVE CHARACTERISTICS • • • 84 Pins 64 Macrocells 15 ns tpo Commercial 20 ns tpo Military • 50 MHz fMAx external Commercial 40 MHz fMAx external Military • 70 Inputs • • 64 Outputs 64 Flip-flops; 4 clock choices • 4 "PAL26V16" Blocks with buried Macrocells • Pin-compatible with MACH230, MACH435 GENERAL DESCRIPTION The MACH130 is a memberof AMO's high-performance EE CMOS MACH 1 family. This device has approximately six times the logic macrocell capability of the popular PAL22V10 at an equal speed with a lower cost per macrocell. The MACH130 consists of four PAL blocks interconnected by a programmable switch matrix. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. Publication# C14131 Rev. F Issue Date: June 1993 AmendmentlO The MACH130 macrocell provides either registered or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as O-type or T-type to help reduce the number of product terms. The register type decision can be made by the designer or by the software. All macrocells can be connected to an I/O cell. If a buried macrocell is desired, the internal feedback path from the macrocell can be used, which frees up the 1/0 pin for use as an input. 3-11 ~ AMD CONDENSED BLOCK DIAGRAM 12,15 1/016-11031 4 OE 52 x 70 52 x 70 AND Logic Array and Logic Allocator AND Logic Array and Logic Allocator 2 26 Switch Matrix 26 52 x 70 52 x 70 AND Logic Array and Logic Allocator AND Logic Array and Logic Allocator OE OE 1/048 - 1/063 4 4 1/032 -1/047 CLKo/lo, CLK1/11 CLK2/13, CLK3/14 C14131F-1 3-12 MACH130-15/20 CONDENSED COM'L: -10/12/15/20, Q-15/20 _ MIL: -20 MACH210A-10 MACH21 0-12/15/20 MACH210AQ-15/20 High-Density EE CMOS Programmable Logic ~ Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • 44 Pins 64 Macrocells • 38 Inputs; 210A Inputs have built-In pull-up resistors • 10 ns tPD Commercial 20 ns tPD Military • • 32 Outputs 64 Flip-flops; 2 clock choices • 80 MHz fMAX external Commercial 40 MHz fMAX external Military • 4 "PAL22V16" blocks with burled macrocells • Pin-compatible with MACH110, MACH215 GENERAL DESCRIPTION The MACH210 is a memberof AMD's high-performance EE CMOS MACH 2 device family. This device has approximately six times the logic macrocell capability of the popular PAL22V1 0 with no loss of speed. The MACH210 consists of four PAL blocks interconnected by a programmable switch matrix. The four PAL blocks are essentially "PAL22V16" structures complete with product-term arrays and programmable macrocells, including additional buried macrocells. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. tered, latched, or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type or T-type to help reduce the number of product terms. The register type deciSion can be made by the designer or by the software. All output macrocells can be connected to an liD cell. If a buried macrocell is desired, the internal feedback path from the macrocell can be used, which frees up the liD pin for use as an input. The MACH21 0 has dedicated buried macrocells which, in addition to the capabilities of the output macrocell, also provide input registers or latches for use in synchronizing signals and reducing setup time requirements. The MACH210 has two kinds of macrocell: output and buried. The MACH21 0 output macrocell provides regis- Publication# C14128 Rev. G AmendmentlO Issue Date: June 1993 3-13 ~ CONDENSED AMD BLOCK DIAGRAM 1100-1/07 10-11. 13-14 1/08-1/015 2 44 x 68 44 x 68 AND Logic Array and Logic Allocator AND Logic Array and Logic Allocator 4 Switch Matrix 44 x 68 AND Logic Array and Logic Allocator 44 x 68 AND Logic Array and Lo ic Allocator DE DE 1/024-1/031 1/016-1/023 CLKo/12. CLK1/15 14128G-1 3-14 MACH21 0-1 0/12/15/20, 0-15/20 i.j.N·Ul3JWieh'M·"'eUi CONDENSED COM'L: -15/20 ~ MACHLV210A-15/20 Advanced Micro Devices High Density EE CMOS Programmable Logic DISTINCTIVE CHARACTERISTICS • • • • • • Low-voltage operation, 3.3 V JEDEC compatible - Vee = +3.0 V to +3.6 V 10 rnA standby power 44 Pins 64 Macrocells 15 ns tPD 50 MHz fMAX external • • • • • • 38 Inputs with pull-up resistors 32 Outputs 64 Flip-flops; 2 clock choices 4 "PAL22V16" blocks with burled macrocells Pln-, functlon-, and JEDEC-compatlble with MACH210 Pin-compatible with MACH110, MACH215 GENERAL DESCRIPTION The MACHLV210A is a member of AMD's highperformance EE CMOS MACH 2device family. This device has approximately six times the logic macrocell capability of the popular PAL22V10 at an equal speed with a lower cost per macrocell. It is architecturally identical to the MACH210, with the addition of 1/0 pull-up resistors and low-voltage, low-power operation. The MACHLV21 OA provides 3.3 V operation with lowpower CMOS technology. At 10 rnA maximum standby current, the MACHLV21 OA is ideal for low-power applications. The MACHLV210A consists of four PAL blocks interconnected by a programmable switch matrix. The four PAL blocks are essentially "PAL22V16" structures complete with product-term arrays and programmable macrocells, including additional buried macrocells. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity Publication# C17908 Rev. A Issue Date: June 1993 Amendment/O between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. The MACHLV210A has two kinds of macrocell: output and buried. The MACHLV210A output macrocell provides registered, latched, or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type or T-type to help reduce the number of product terms. The register type decision can be made by the designer or by the software. All output macrocells can be connected to an I/O cell. If a buried macrocell is desired, the internal feedback path from the macrocell can be used, which frees up the I/O pin for use as an input. The MACHLV210A has dedicated buried which, in addition to the capabilities of macrocell, also provide input registers or use in synchronizing signals and reducing requireme nts. macrocells the output latches for setup time 3-15 ~ CONDENSED AMD BLOCK DIAGRAM 1100-1/07 10-11. 13-14, I/OS-1/01S 8 2 44 x 68 AND Logic Array and Logic Allocator 44 x 68 AND Logic Array and Logic Allocator 22 4 22 Switch Matrix 44 x 68 AND Logic Array and Logic Allocator 44 x 68 AND Logic Array and Lo ic Allocator 2 OE OE 2 1/024-11031 1/016-11023 CLKo!l2. CLK1/Is 1790SA-1 . 3-16 MACHLV210A-15/20 - CONDENSED ~ COM'L: -12/15/20 MACH220-12/15/20 High-Density EE CMOS Programmable Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • 68 Pins • 96 Macrocells • • • 12 ns tPD 66.7 MHz fMAX external 56 Inputs with pull-up resistors • 48 Outputs • 96 Flip-flops; 4 clock choices • 8 PAL blocks with burled macrocells • Pin-compatible with MACH120 GENERAL DESCRIPTION The MACH220 is a memberof AMD's high-performance EE CMOS MACH 2 device family. This device has approximately nine times the logic m,acrocell capability of the popular PAL22V1 0 with no loss of speed. The MACH220 consists of eight PAL blocks interconnected by a programmable switch matrix. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. The MACH220 has two kinds of macrocell: output and buried. The output macrocell provides registered, Publication# C14130 Rev. G Amendment/O Issue Date: June 1993 latched, or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type or T -type to help reduce the number of product terms. The registertype decision can be made by the designer or by the software. All output macrocells can be connected to an 1/0 cell. If a buried macrocell is desired, the internal feedback path from the macrocell can be used, which frees up the I/O pin for use as an input. The MACH220 has dedicated buried macrocells which, in addition to the capabilities of the output macrocell, also provide input registers for use in synchronizing signals and reducing setup time requirements. 3-17 m r- ~ o > o :s: ~ c c Cf ...... ClO 1100-1105 1106-11011 11012-11017 11018-11023 12-13, 16-17 :; G') ::xJ l> 3: 4 ~ s: (') (') O :t:- Z C ::I: I\) I\) Switch Matrix 9...... m z ~ tTl N 0 en m C 4 11042-11047 11036-11041 11030-11035 11024-11029 CLKO/IO, CLK1/11 CLK2I14 ; CLK3I15 C14130G-1 CONDENSED _ ~ COM'L:-15/20 MACH230-15/20 High-Density EE CMOS Programmable Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • 84 Pins 128 Macrocelis • • 15 ns tPD 50 MHz fMAX external • 70 Inputs • • • • 64 Outputs 128 Flip-flops; 4 clock choices 8 "PAL26V16" blocks with burled macrocelis Pin-compatible with MACH130, MACH435 GENERAL DESCRIPTION The MACH230 is a memberof AMD's high-performance EE CMOS MACH 2 device family. This device has approximately twelve times the logic macrocell capability of the popular PAL22V10 with no loss of speed. The MACH230 consists of eight PAL blocks interconnected by a programmable switch matrix. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. The MACH230 has two kinds of macrocell: output and buried. The output macrocell provides registered, Publication# C14132 Rev. G Amendment/O Issue Date: June 1993 latched, or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type or T-type to help reduce the number of product terms. The register type decision can be made by the designer or by the software. All output macrocells can be connected to an 1/0 cell. If a buried macrocell is desired, the internal feedback path from the macrocell can be used, which frees up the 1/0 pin for use as an input. The MACH230 has dedicated buried macrocells which, in addition to the capabilities of the output macrocell, also provide input registers for use in synchronizing signals and reducing setup time requirements. 3-19 m r Co) N Q 0 ~ 0 " c i> C') VOO - V07 (Block A) VOS - 0015 (Block B) V016 - VOl3 (Block C) V024 - V031 (Block 0) 12,15 :D l> :s:: (') 3: o l> z c (') ::t I\) w m 01 en m 9 ..... z i\) Q c VOS6 - V063 (Block H) V048 - VOSS (Block G) V040 - V047 (Block F) V031 - V039 (Block E) CLKO/IO, CLKlI11 CLK2I13, CLK3J14 C14132G-1 CONDENSED _ ~ COM'L: -12/15/20 MACH215-12/15/20 High-Density EE CMOS Programmable Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • 44 Pins 32 Output Macrocells • 67 MHz fMAX external • 3S Inputs with pull-up resistors • 32 Input Macrocells • 32 Outputs • Product terms for: 64 Flip-flops 4 "PAL22RAS" blocks with burled macrocells Pin-compatible with MACH110, MACH210 • - Individual flip-flop clock • • - Individual asynchronous reset, preset • - Individual output enable 12 ns tPD GENERAL DESCRIPTION The MACH215 is a memberof AM D's high-performance EE CMOS MACH device family. This device has approximately three times the capability of the popular PAL20RA10 with no loss of speed. The MACH215 consists of four PAL blocks interconnected by a programmable switch matrix. The four PAL blocks are essentially "PAL22RAS" structures complete with product-term arrays and programmable macrocells, individual register control product terms, and input registers. The switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. Publication# C16751 Rev. C Issue Date: June 1993 AmendmentiO The MACH215 has two kinds of macrocell: output and input. The MACH215 output macrocell provides registered, latched, or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type or T-type to help reduce the number of product terms. The register type decision can be made by the designer or by the software. Each macrocell has its own dedicated clock, asynchronous reset, and asynchronous preset control. The polarity of the clock signal is programmable. All output macrocells can be connected to an 1/0 cell. The MACH215 has dedicated input macrocells which provide input registers or latches for synchronizing input signals and reducing setup time requirements. 3-21 ~ CONDENSED AMD BLOCK DIAGRAM VOo-I/O; VOa-1/01S Q Q T ..j 8/ Output Macrocells --I 1 S1 DE T 8~ I 1/0 Cells VOCelis 8f I tnput Macrocells I J I I t ClK J 8 E~ / 1 8" J S! Output Macrocells OE 8" J 44x64 AND Logic Array and Logic Allocator I 10-11. 13-14 Input Macrocells ClK , 8" S 8" J 44x64 AND Logic Array and Logic Allocator 22 J I / J 4" J 22 Switch Matrix 22 22 44x64 AND Logic Array and Logic Allocator OE Output Macrocells 8 -.j 8" 1 ~T 1 ~ 1/024-1/031 I J / 81- ! 1 I I Output Macrocells 8v 1 I J 1 J I Input Macrocells I l I/OCell5 1 8,,8 11'2,. 8,1' ~ OE Input Macrocells ~ 1/0 Cells 8,,8, 44x64 AND Logic Array and Logic Allocator 1 8-I II' ~ 1/016-1/023 CLKo/12 CLK1/1s C16751C-1 3-22 MACH215-12/15/20 - ~ Advanced Micro Devices MACH 3 and 4 Device Families High-Density EE CMOS Programmable Logic DISTINCTIVE CHARACTERISTICS • • High-performance, high-density electrically-erasable CMOS PLD families Predictable design-Independent 15- and 20-ns speeds • High density, pin count - 3500-10,000 PLD Gates - 84-208 Pins - 96-384 Registers • Input and output switch matrices increase ability to hold a fixed pinout JTAG, 5-V In-circuit programmability on devices with more than 84 pins Synchronous and asynchronous modes available for each macrocell - Clock generator in each PAL block for programmable clocks, edges in either mode - Individual clock, initialization product terms in asynchronous mode • • • Central, Input, and output switch matrices - 100% Routability with 80% Utilization • • Up to 20 product terms per function 96-256 configurable macrocells - D/T/J-K/S-R Registers, latches - Synchronous or asynchronous mode - Programmable polarity - Reset/preset swapping • • • XOR gate available Registered/latched inputs on MACH 4 series Extensive third-party software and programmer support thro~gh FusionPLD partners PRODUCT SELECTOR GUIDE Device Pins Macrocells PLD Gates Max Inputs Max Outputs Max Flip-Flops JTAGI 5 V Prog Speed 132 96 3500 102 96 96 Y 15,20 84 128 5000 70 64 192 N 15,20, 0-25 MACH 3 Family MACH355 MACH 4 Family MACH435 MACH445 100 128 5000 70 64 192 Y 15,20 MACH465 208 256 10,000 146 128 384 Y 15,20 GENERAL DESCRIPTION The MACH (Macro Array CMOS High-speed/density) family provides a new way to implement large logic designs in a programmable logic device. AMD has combined an innovative architecture with advanced electrically-erasable CMOS technology to offer a device with many times the logic capability of the industry's most popular existing PAL device solutions at comparable speed and cost. The second-generation MACH devices provide approximately three times the density and register count, and two times the amount of I/O of the original MACH 1 and 2 families. By increasing the pin count, adding function- ality, and improving routing, the MACH 3 and 4 families build upon the strength of the MACH architecture without sacrificing predictable timing. Their unique architecture makes these devices ideal for replacing large amounts of TTL, PAL-device, glue, and gate-array logic. They are the first devices to provide such increased functionality with completely predictable, deterministic speed. The MACH devices consist of PAL blocks interconnected by a programmable central switch matrix (Figure 1). Designs that consist of several intercon- MACH 3 and 4 Device Families 3-23 ~ AMD nected functional modules can be efficiently implemented by placing the modules into PAL blocks. Designs that are not as modular can also be readily implemented since the central switch matrix provides a very high level of connectivity between PAL blocks. The use of input and output switch matrices allows logic to be implemented independent of pin connections. This allows greater flexibility when making initial pin assignments for PCB layout, or when trying to maintain the pinout through design changes. The internal arrangement of resources is managed automatically by the design software, so that the deSigner does not have to be concerned with the logic implementation details. AMD's FusionPLD program allows MACH device designs to be implemented using a wide variety of popular industry-standard design tools. By working closely with the FusionPLD partners, AMD certifies that the tools provide timely, accurate, quality support. This ensures that a designer does not have to buy a complete new set of tools for each new device, but rather can use the tools with which he or she is already familiar. The MACH devices can programmed on conventional PAL device programmers. Devices with pin counts greater than 84 have an additional 5-V programming algorithm option that can be implemented with the devices soldered onto the board. MACH devices are manufactured using AMD's state-ofthe-art advanced CMOS electrically-erasable process for high performance and logic density. CMOS EE technology provides 100% testability, reducing both prototype development costs and production costs. PAL Block PAL Block 110 Cells 0 ~ Q) 0 Macrocells C> Logic Allocator ~ 0 ..g ~ x ·c cn1il Array CD c: Q) C> '5~ ~ ~ C3 a. 0 C3 1il ~ c: Q) 110 Cells Output Switch Matrix ~ .B Logic Allocator .§ .~ cn1il '5~ ••• a. 0 I --... Macrocells 0 • • , • I Output Switch Matrix Array ~ h " Central Switch Matrix ., , 4 I I Array ~ CD c: Q) 0 0 C3 x ~·5 1\1 Logic Allocator C> ~ Array ~ ..g '5~ a. ~ Macrocells , 0 Q) 0 110 Cells Logic Allocator C> ~ 0 (,) Output Switch Matrix .B .§ ~ Q) c: ~ .~ cn~ '[~ ••• ~ Macrocells Output Switch Matrix 110 Cells 17466A-1 Figure 1. MACH 3 and 4 Block Diagram 3-24 MACH 3 and 4 Device Families AMD~ Design Methodology Design tools for all MACH devices are widely available both from AMD and from third-party software vendors. AMD provides MACHXLTM software as a low-cost baseline tool set and works with third-party vendors to ensure broad MACH device support. MACHXL software is based on the popular PALASM o V024-V031 "c ~ C j; G') :D » 3: 3: o o l> o :::J: C'lC'l z r-r- -'=" ;0::;0:: Co) ~~ ''(;S'0r) C{1 ...a. C11 ~fS Central Switch Matrix C'lC'l iii r-r- i\3 ;0::;0:: ~~ P 'i::: 66 X90 4 ....1 "".0:' U;, ~ AND Logic Array and Logic Allocator V056-I/063 m z en m c '? N C11 c V043-V055 V04O-V047 V031-V039 CONDENSED "'I"Z.'ij'3Jitii·';i~t.5i(.]tl ~ COM'L: -15/20 MACH355-15/20 High-Density EE CMOS Programmable Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • 132 Pins In PQFP • 96 Macrocells • • 15 ns tpo 50 MHz fMAX external • 102 Inputs with pull-up resistors • 96 Outputs • • 96 Flip-flops Up to 20 product terms per function, with XOR • Flexible clocking - Four global clock pins with selectable edges - Asynchronous mode available for each macrocell • JTAG, 5-V programmability • 6 "PAL33V16" blocks • Input and output switch matrices for high routablllty • Fixed, predictable, deterministic delays GENERAL DESCRIPTION The MACH355 is a memberof AMO's high-performance EE CMOS MACH 3 family. This device has approximately nine times the macrocell capability of the popular PAL22V10, with significant density and functional features that the PAL22V10 does not provide. The MACH355 consists of six PAL blocks interconnected by a programmable central switch matrix. The central switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. Routability is further enhanced by an input switch matrix and an output switch matrix. The input switch matrix provides input signals with alternative paths into the central switch matrix; the output switch matrix provides flexibility in assigning macrocells to lID pins. The MACH355 has macrocells that can be configured as synchronousor asynchronous. This allows designers to implement both synchronous and asynchronous logic together on the same device. The two types of design can be mixed in any proportion, since the selection on each macrocell affects only that macrocell. Up to 20 product terms per function can be assigned. It is possible to allocate some product terms away from a macrocell without losing the use of that macrocell for logic generation. The MACH355 macrocell provides either registered or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type, T-type, J-K, or S-R to help reduce the number of product terms used. The flip-flop can also be configured as a latch. The register type decision can be made by the designer or by the software. All macrocells can be connected to an I/O cell through the output switch matrix. The output switch matrix makes it possible to make significant design changes while minimizing the risk of pinout changes. Publication# C17467 Rev. A AmendmentiO This document contains information on a product under development at Advanced Micro Devices, Inc. The information is ~I_ss_u_eD_a_te_:_Ju_n..;..8..;..19;.;.9.;..3_ _ _ _ _..J :~t~~~~~~~i~~P you evaluate this product. AMO reserves the right to change or discontinue work on this proposed product 3-29 ~I vo vo vo OJ r 0 0 "> c ~ » ~ c G') :c :5: » (') ~ » (') o z c o ::J: r ~ Central Switch MatrIx S f{1 m ...a. 33 ~ I\) o 4 66X98 AND Logic Array and Logic Allocator 4 66X98 AND Logic Array and Logic Allocator .a ~5. e!.cn ~.! 4 66X98 AND Logic Array and Logic Allocator g. ~ ~ O'l ~ vo vo vo z en m c CONDENSED ','·';z.'ttgJi~i4·';i~t.ii[.'~1 ~ COM'L: -15/20 MACH445-15/20 High-Density EE CMOS Programmable Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • • 100 pins In PQFP MACH435 with JTAG, 5-V programmability • 128 Macrocells • • • 15 ns tPD 50 MHz fMAx external 70 Inputs with pull-up resistors • • 64 Outputs 192 Flip-flops - 128 Macrocell flip-flops 64 Input flip-flops • Up to 20 product terms per function, with XOR • Flexible clocking - Four global clock pins with selectable edges Asynchronous mode available for each macrocell • • 8 "PAL33V16" blocks Input and output switch matrices for high routability • • Fixed, predictable, deterministic delays JEDEC-file compatible with MACH435 GENERAL DESCRIPTION The MACH445 is a memberof AMD's high-performance EE CMOS MACH 4 family. This device has approximately twelve times the macrocell capability of the popular PAL22V10, with significant density and functional features that the PAL22V1 0 does not provide. It is architecturally identical to the MACH435, with the addition of JTAG and 5-V programmability capabilities. The MACH445 consists of eight PAL blocks interconnected by a programmable central switch matrix. The central switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows deSigns to be placed and routed efficiently. Routability is further enhanced by an input switch matrix and an output switch matrix. The input switch matrix provides input signals with alternative paths into the central switch matrix; the output switch matrix provides flexibility in assigning macrocells to I/O pins. The MACH445 has macrocells that can be configured as synchronous or asynchronous. This allows designers to implement both synchronous and asynchronous logic together on the same device. The two types of design can be mixed in any proportion, since the selection on each macrocell affects only that macrocell. Up to 20 product terms per function can be assigned. It is possible to allocate some product terms away from a macrocell without lOSing the use of that macroce/l for logic generation. The MACH445 macrocell provides either registered or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as D-type, T-type, J-K, or S-R to help reduce the number of product terms used. The flip-flop can also be configured as a latch. The register type decision can be made by the designer or by the software. All macrocells can be connected to an I/O cell through the output switch matrix. The output switch matrix makes it possible to make significant design changes while minimizing the risk of pinout changes. Publication# C174S8 Rev. A Amendment/O This document contains information on a product underdevelopment at Advanced Micro Devices, Inc. The information Issue Date: June 1993 is intended to help you evaluate this product. AMD reserves the right to change or discontinue work on this proposed ' - - - - - - - - - - - - - - - ' product without notice. 3-31 m ~ Co) N 1100-1/07 1/08-11015 11016-11023 r- ~ o » o 11024-11031 "; 3: c c C) :a » 3: 0 3: » o 00 r-r- 0 33 ,," ~g :J: ~ ~ J;5!=>r; , ~~ C{1 ..... en ,," ~~ .<<1 fJ Central Switch Matrix .1 u; i:.- N 66X90 4 ~ :~ z en AND Logic Array and Logic Allocator c 5" :!:S "0 ~CJ) ~~ )(§: ~:? ...... :~ m 1 m c . Z C 11056-11063 11048-11055 11040-11047 1/031-V039 CONDENSED ',',P"'fHJ'f'i·,;!!t·"t.1f' ~ COM'L: -15/20 MACH465-15/20 High-Density EE CMOS Programmable Logic Advanced Micro Devices DISTINCTIVE CHARACTERISTICS • 208 pins In PQFP • Up to 20 product terms per function, with XOR • 256 macrocells • Flexible clocking • • 15 ns tPD 50 MHz fMAX external • 146 Inputs with pull-up resistors • 128 Outputs • 384 Flip-flops - 256 Macrocell flip-flops . - 128 Input flip-flops - Four global clock pins with selectable edges - Asynchronous mode available for each macrocell • 16 "PAL34V16" blocks • JTAG, 5-V programmability • Input and output switch matrices for high routability • Fixed, predictable, deterministic delays GENERAL DESCRIPTION The MACH465 is a memberof AMO's high-performance EE CMOS MACH 4 family. This device has approximately 25 times the macrocell capability of the popular PAL22V10, with significant density and functional features that the PAL22V10 does not provide. The MACH465 consists of 16 PAL blocks interconnected by a programmable central switch matrix. The central switch matrix connects the PAL blocks to each other and to all input pins, providing a high degree of connectivity between the fully-connected PAL blocks. This allows designs to be placed and routed efficiently. Routability is further enhanced by an input switch matrix and an output switch matrix. The input switch matrix provides input signals with alternative paths into the central switch matrix; the output switch matrix provides flexibility in assigning macrocells to I/O pins. The MACH465 has macrocells that can be configured as synchronous or asynchronous. This allows designers to implement both synchronous and asynchronous logic Publication# C17470 Rev. A AmendmentlO ...1;.;;.;ss;.;;.ue;;..D;;;.;a;;.;;te;.;..:..;;.Ju;;.;.n;.;;.8..;.;19;.;;.9;;..3_ _ _ _ _--' together on the same device. The two types of design can be mixed in any proportion, since the selection on each macrocell affects only that macrocell. Up to 20 product terms per function can be assigned. It is possible to allocate some product terms away from a macrocell without losing the use of that macrocell for logic generation. The MACH465 macrocell provides either registered or combinatorial outputs with programmable polarity. If a registered configuration is chosen, the register can be configured as O-type, T-type, J-K, or S-R to help reduce the number of product terms used. The flip-flop can also be configured as a latch. The registertype decision can be made by the designer or by the software. All macrocells can be connected to an I/O cell through the output switch matrix. The output switch matrix makes it possible to make Significant design changes while minimizing the risk of pinout changes. This document contains Information on a product underdevelopment at Advanced Micro Devices. Inc. The information ~r~~~~d!1t~~~e~~:. evaluate this product. AMD reserves the right to change or discontinue work on this proposed 3-33 ~ AMD CONDENSED BLOCK DIAGRAM I/O 110 ClKII 110 L l - - - -..... 110 ~l-------------------~ 110 C J - - - - - - - - - - - - - - - - - - , I/O L l - - - - . . . , 1/0 110 ~---_C:l 1/0 ~--------------~ 110 r------------------~ 110 .------L::::l 110 110 1/0 1/0 C17470A-1 3-34 MACH465-15/20 4 GENERAL INFORMATION Inside AMD's CMOS PLD Technology ......................................... 4-3 Military PAL Devices ...................................................... 4-43 Military ProPAL Devices .................................... . . . . . . . . . . . . . .. 4-52 Electrical Characteristic Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 4-54 fMAX Parameters ......................................................... 4-57 Physical Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 4-58 General Information 4·1 ~ 4·2 AMD General Information t1 Inside AM D's CMOS PLD Technology Advanced Micro Devices Application Note by Bryon Moyer, Applications Manager, Advanced Micro Devices, Inc. TECHNOLOGY DESCRIPTION The EE5 process is similar, except that Left is 0.65 J..Lm. The EE CMOS technology used by AM D in programmable logic is a single-poly, double-metal n-well process. It has been optimized for high-speed programmable logic devices, which do not have the same density constraints that memories have. The basic characteristics of the EE4 process are: CMOS PLDs use standard CMOS logic internally, with the addition of a programmable array. The output buffers of most devices are designed to be compatible with TTL circuits, and therefore have n-channel enhancement pull-up transistors. Exceptions to this are the zeropower devices and the PALCE610H-15, which have rail-to-rail switching outputs provided by a p-channel pull-up in the output buffer. • CMOS, n-well • Grounded substrate • Single-poly, dual metal • 1.2 J..Lm minimum feature • 0.8 J..Lm gate length (Lett) • 180A gate oxide thickness AMD's CMOS process for programmable logic is simplified by the absence of standard depletion-mode transistors in the more advanced processes. Depletion mode transistors are a vestige of NMOS design, and are not really needed. This results in the elimination of a mask and implant step, reducing the process cost and simplifying the structure. • 90A tunnel oxide thickness • 1.4 J..Lm contact dimension on wafer • 3.2 J..Lm metal 1 pitch • 3.8 J..Lm metal 2 pitch Transistor Cross-Section Fi~ure 1 shows a cross-section of a basic inverter. This is a very straightforwa.rd structure. The gates consist of poly-silicon; the other connections are made with metal. Out In n-well p-5ubstrate 16507A-001A Figure 1. CMOS Inverter Cross-Section Substrate Voltage There are two basic substrate configurations for CMOS PLDs: grounded substrate and floating (or negative) substrate, as shown in Figure 2. In the first case, the substrate is connected to ground; no voltage on the chip Publication# 16507 Rev. A Issue Date: February 1993 Amendment/O is more negative than ground. The substrate is directly hard-wired to the ground pin. In the second case, the substrate is capacitively coupled to ground. A charge pump pumps the substrate to a negative voltage, typically -3 V. 4-3 ~ AMD Input pin Ground Thus the· negative substrate has no practical advantages over a grounded substrate, and lacks the advantages of a grounded substrate. For this reason, none of AMO's CMOS PLOs use a negative substrate. Erasable Technology 16507A-002A a. n in roundin Any erasable CMOS technology is based upon the concept of stored charge. The charge is stored on a transistor with a floating gate-that is, a gate that has no connection. The transistor actually has two gates: one that floats, and 0 ne that acts as a control gate. The control gate is used to establish the field across the floating gate (see Figure 3) . .. Flo-a-,ti-ng-::......"S-u.,....bs-tr-a--'teL........., 16507 A-003A ~I 0 b. Figure 2. Substrate Configurations: a. Grounded; b. Floating Control Gate Floating Gate / AMO's CMOS process uses a grounded substrate. This 16507A-004A means that no large charge pump is used to pump the substrate negative. This technology has several beneFigure 3. Floating-Gate MOS Transistor fits. Providing effective clamp diodes is easier if the substrate is grounded; this helps protect against negative overshoot. A grounded substrate permits In the programmed state, there is a net deficit of elecquieter operation on boards where the oscillation of a trons in the floating gate. The resulting positive charge charge pump can radiate and disturb sensitive circuits. turns the transistor ON. In the erased state, there are Also, this approach permits the design of zero-power enough electrons on the control gate so that the _ _ _ Rarts that w()LJ!d rlott?_~ p_~~It>.I_~_!t~~harge pump were negative charge turns the transistor OFF. constantly running. -- ---------=-----------------There are two basic ways of transferring the charge onto the floating gate: a) hot electron injection, and b) tunnelA negative substrate is sometimes used to get speed, ing. Electrically erasable devices rely on tunneling; howand also makes it more difficult to induce latch-up by making the substrate more negative than any board ever, it is useful to compare these two methods. voltage. However, latch-up is generally an issue only during power-up on standard boards with standard logic UV-Erasable Technology Hot electron injection is used for UV-erasable devices, drivers. While the device is powering up, the device is most vulnerable to latch-up because of the many transuch as EPROMs. With this scheme, a bias is set up between the source and drain of the transistor, and sients occurring. However, it takes time to pump down the substrate, so the substrate cannot immediately probetween the control gate and the substrate (see Figure 4). The channel is pinched off, and a strong curtect against latch-up. This means that the negative rent flows. Because of the high fields, the electrons are substrate approach provides no latch-up protection at hot. The two fields (source-to-drain, and substrate-tothe time when latch-up is most likely. control-gate) combine to form a field in a diagonal direcThe only signal excursions into negative territory during tion, but because of the oxide barrier, electrons cannot cross in that direction. Occasionally, electrons acquire normal operation will be from overshoot, and overshoot enough energy to cross the barrier in the shortest direccannot induce latch-up because there is not enough energy. Negative overshoot is discussed in more detail tion: from the channel to the floating gate. This is below. referred to as hot electron injection. 4-4 Inside AMO's CMOS PLO Technology AMD~ Oxide Barrier I Pinched-Off Channel Depletion Region p-substrate 16S07A-OOSA Figure 4. Programming by Hot-Electron Injection Figure 5 shows the energy band diagram for the gate/ channel interface. Because the fields give the electrons more energy, more electrons can cross the oxide barrier. The height of the barrier determines how easily charge can be transferred across. Once an electron is on the other side of the oxide, it is on the floating gate, with no path. It is therefore effectively trapped, and remains there. During programming, large fields are set up so that a significant number of electrons are injected. manufacturing, limiting the amount of extra quality that can be provided by the erasability feature. Electrica"y Erasable Technology Electrically-erasable devices use Fowler-Nordheim tunneling as the mechanism for getting charge onto the floating gate. This is defined roughly as tunneling that occurs as a result of a field placed across the barrier that the electrons tunnel through. Erasing these devices requires exposure to ultraviolet light. The energy from the ultraviolet light causes the electrons to cross back over the oxide barrier, erasing the device. Forthis to happen, the device package must have a window that lets the ultraviolet light pass through. Some amount of direct tunneling, or tunneling that occurs without an applied field, is always possible through any energy barrier. It may be extremely small or significant; the determining factor is the width of the barrier_ Since tunneling electrons are going through the barrier instead of over it, the height ofthe barrier does not affect the amount of tunneling. UV-erasable technology has a few distinct drawbacks. The fact that the parts require a window to be erased makes the devices much more expensive. Although they are usually available in plastic one-timeprogrammable (OTP) packages, they are then not erasable, and have no advantage over fuse technology. In addition, windowed devices take about 90 minutes to erase. This limits the number of times that the device can practically be reprogrammed and tested during For an electrically-erasable cell, the tunnel oxide is about one third the thickness of the oxide of a UVerasable part; therefore tunneling occurs at relatively low fields. Even so, the field used to cause tunneling is about five times the field used to cause hot-electron injection for UV parts. Note that tunneling is theoretically possible on a UV part, but a very high field is required, and the normal electron injection would swamp out any tunneling that would occur. Inside AMD's CMOS PLD Technology 4-5 ~AMD ;,,--------------. -Electron Energy Distribution Barrier Height ,/ ~\// Oxide Barrier >~ Substrate Floating Gate Q) \ c \ W \ \ \ \ \ \ \ \ \ \ \ \ ,, \ , 16507A-OOSA a. ! Electron Energy Distribution ~ ,~ I ~ ,/ /,/ " ' " \\"\\\\\\\ ---------~ --Substrate-- "Hot" ~ Electrons Oxide Barrier ---- - ~ ------------------------~ C \ W \ Floating Gate 16507A-007A b. Figure 5. Energy Band Diagrams: a. Neutral Floating Gate; b. Hot-Electron Injection Fowler-Nordheim tunneling involves placing a potential across the barrier, which distorts the band diagram as shown in Figure 6. The "angle" caused by the applied potential effectively thins part of the already-thin barrier, making tunneling easier. It is this tunneling under bias that is used to program electrically-erasable devices. Note that by reversing the bias, the tunneling can occur just as well in the opposite direction. This is what makes electrical erasure possible. 4-6 Electrical erasure has advantages over UV erasure both in cost and quality. Because the erasure is electrical, no expensive window is required in the package. This makes erasability cost-effective even in high-VOlume production quantities. In addition, the fast erasure allows AM D to reprogram the device many times, allowing many more paths to be tested than can be tested in a UV part. This provides much higher quality, especially in higher-density devices. Inside AMD's CMOS PLD Technology AMD~ Minimal Intrinsic ________ (Direct) Tunneling Tunnel Oxide Barrier >. ~ Electron Source Floating Gate Q) (Neutral) c::: W 16507A-008A a. Narrower Barrier; _________ More Tunneling eElectron Energy e- DiSlribUtiOn~!/ ." Tunnel Oxide Barrier >. ~ Electron Source Q) c::: W Floating Gate (Uncharged) 16507A-009A b. Electron Energy DiSlribUtiOn~("/ >. Reversible Tunneling Direction ,: Tunnel Oxide Barrier ~ Floating Gate (Charged) Q) c::: W Electron Source 16507A-Ol0A c. Figure 6. Energy Band Diagrams: a. Direct Tunneling; b. Fowler-Nordheim Charging; c. Fowler-Nordheim Discharging Inside AMD's CMOS PLD Technology 4-7 ~ AMD In addition to speed, there are a number of other benefits to this approach. At the most basic level, this eliminates a poly-silicon layer, simplifying the process. This reduces costs and improves reliability. Cell Configuration and Programming The programming cell is shown in Figure 7. To improve device speed, the programming cell has been divided into the programming portion and the data path portion. - - Program Line - - - - - Read Line WordLine _ _ - - - - - Read Transistor Electron Source - - - - - - - 1 Control Gate _ _ o--~~~~--------~ - - - - - Sense Transistor - - - - - Source Floating Gate 16507A-011A a. Program Word Line Program Transistor Word Tunne Electron Source 0sX~idleIF~lo~at~in~lgIG~a~te?;;;2D2D5i~I[I~:::::: Read Line Floating Gate Program Junction Control Gate Sense Transistor . Read Transistor 16507A-012A b. Figure 7. EE PLD Programming Cell: a. Circuit; b. Cross-Section The programming half requires long-channel transistors capable of sustaining high electrical fields; the data path requires short-channel transistors that are fast. Note that this does take more space, but in PLDs, the size of the cell is not a limiting factor as it is. in memories. In a PLD, the programming array can take up as little as 10% of the die area, while a memory typically uses more than 90% of the die area for the programming array. Programming and erasure are complementary procedures in EE technology. However, the sense of programming and the sense of erasing are perhaps opposite to what one might assume. A cell is considered to be programmed if there is a charge deficit on the float4-8 ing gate, providing a positive voltage; it is erased if there is excess charge on the floating gate, generating a negative voltage. This means that programming a device only requires turning ON those cells that are needed, ratherthan turning off all of the cells that are not needed. A cell fresh from wafer fabrication has no net positive or negative charge on the gate. To balance the threshold of the transistor for reliable turn-on and turn-off, a cell imp/antis used to center the threshold voltage near 0 V. Programming and erasing involve either removing electrons from the conduction bands of the poly-silicon gate or adding excess electrons, providing a net charge that Inside AM D's CMOS PLD Technology AMD~ will move the gate voltage solidly on one side or the other of the threshold voltage. When programming or erasing the device, a voltage is applied between the program and control gate nodes. The direction of the voltage determines whether the cell is erased or programmed. When erasing, the control gate is given a positive voltage, and the program node is grouncled. This attracts electrons from the program transistor across the tunnel oxide to the floating gate, turning the read transistor OFF (see Figure 8a). + ~: When programming the cell, the program node voltage is elevated, and the control gate is grounded, reversing the electron flow, as indicated in Figure 8b. Enough electrons flow off the floating gate to leave a net positive charge; this turns the transistor ON. AMO has modified the programming cell to increase programming efficiency and has a patent on the resulting circuit. On traditional devices, the source node is grounded during programming. On AMO's devices, the source node is raised to the same potential as the control gate, as shown in Figure 9. This increases the coupling ratio of the cell. The coupling ratio is the percentage of the applied field that appears across the tunnel oxide. When the source is grounded, the field across the tunnel oxide is reduced (since there is another capacitor in parallel with the tunnel oxide). By raising the source voltage, more of the field is available for programming. The coupling ratio can therefore be thought of as a measure of the programming efficiency; since the efficiency is higher, lower voltages are required for programming. ty_ @~ @\J @ 16507A-013A a. + + 16507A-01SA Figure 9. Source Is at Same Potential as Control Gate to Improve Coupling Ratio 16S07A-014A b. Figure 8. Cell Biasing: a. Charging; b. Discharging The split-cell configuration also allows a simpler programming algorithm, since the programmer can take advantage of the self-limiting nature of programming and erasure. The split cell places the read cell gate and the floating cell gate in "parallel" with each other. Therefore the floating cell can be either completely charged (with a net excess of electrons) or completely discharged (with a net deficit of electrons, or an excess of holes), as shown in the energy diagrams in Figure 10. This is simple to do, since the electrons that have crossed the barrier set up a field that opposes further tunneling. As more electrons cross the barrier the opposing field grows strong enough to block more Inside AMD's CMOS PLD Technology 4-9 ~ AMD electrons from tunneling (Figure 11). Regardless of the state of the floating cell, the select line will turn on or off the read transistor; the cell will only be read, however, if both the read transistor and the the floating transistor are ON. Built-in ~Field m..... --------EJ /Electrons Trapped o o_\J00 Tunnel Oxide Barrier >~ CD [f;\ 00 fo:\0 0 ~ 0 000 Floating Gate Electron Source c: UJ 16507A-016A a. Built-in /'-Field EJ-------·m Tunnel Oxide Barrier >~ Electron Source CD c: Floating Gate UJ 16507A-017A b. Figure 10. Stable EE Cell: a. Charged (Erased); b. Discharged (Programmed) 4·10 Inside AMD's CMOS PLD Technology AMD~ ~ Applied Programming Field E1--------~1±l ~ Built-in Field Growing I±l.------- E1 Tunneing Electrons __________ Tunnel Oxide Barrier >. ~ Trapped ~Electrons ~ Electron Source Q) 0 0 00 ® o ® ® ®® c W Floating Gate 16507A-018A a. I ~ / No Net Field; Programming Stops O~------;....+-- -----. Time Applied Field 16507A-019A b. Figure 11. Self-Limiting Programming and Erasure: a. Energy Band Diagram; b. Fields vs Time In standard one-transistor cells, the two gates are actually in "series". If the floating gate is charged, then the transistor is OFF, regardless of the state of the select line. In order to read the cell, the floating gate has to be neutralized so that the select line controls the transistor (Figure 12). If the floating gate were completely dis- charged, then the transistor would be ON regardless of the state of the select line. The programming algorithm is therefore more complicated, since the amount of charge removed must be monitored to ensure that just enough charge is removed to neutralize the floatinggate. Inside AMD's CMOS PLD Technology 4-11 ~ AMD Net Electron /SurpIUS Oxide Barrier o ®/ (OJ ® ®®-191 ® 0®® (!)0 Substrate Floating Gate 16507A-020A 8. /Charge No Net Oxide Barrier Substrate Floating Gate 16507A-021A b. Figure 12. UV Cell: a. Charged (Programmed) State; b. Neutral (Erased) State Array Configuration The discussion above focused on individual cells. These cells must be hooked together to form a complete array that is driven by input lines and drives product terms. There are two configurations used in AMD's EE CMOS devices (see Figure 13). The configuration in Figure 13b provides some benefit because the parasitic capacitor does not couple the input line and the product term, but both approaches are actively used in designs. Bit Line ~' i'---ll ~' Input Line ol---... i'---tl Input Line o - - -... Floating Gate - - - - Floating Gate - - - Bit Line 16507A-023A 16507A-022A a. . b. Figure 13. Two Array Configurations: a. Bit Line at the Drain; b. Bit Line at the Source 4·12 Inside AM D's CMOS PLD Technology AMD~ PROGRAM INTEGRITY Reliable programming of PLDs requires the use of wellcalibrated, quality programming equipment. To ensure that the device is correctly programmed, the correct voltages and times must be applied. As discussed above, it is impossible to over-charge or over-discharge the programming cell since the mechanism is self-limiting. This provides more leeway and makes the programming algorithms less sensitive to programmer variations. This ultimately provides higher, more consistent programming yields under real-life production programming conditions. However, if the cell is under-programmed or undererased, an insufficient amount of charge might transfer onto or off of the floating gate. When programming, this might not turn the read cell ON sufficiently, potentially slowing down the device. In the case of erasure, the read cell might be partially ON if it is not completely erased. This may cause "disconnected" inputs to appear partially connected. Thus it is important to ensure that the programming pulsewidths are long enough to provide adequate programming. If the programming voltages are slightly inaccurate, CMOS devices often can still be programmed correctly. However, excessive voltage might cause device damage if breakdown voltages are exceeded. Extremely low voltages might fail to engage the programming circuitry completely. date, we certify programmers that meet strict criteria for all products through our FusionPLDsM program. AMD guarantees the performance of any device when programmed on an approved programmer .. For a list of FusionPLD partners, please refer to the FusionPLD Catalog. Data Retention In an electrically erasable device, the floating cell is programmed by forcing electrons to tunnel through the tunnel oxide into the floating gate. Ideally, these trapped electrons mean that the device remains programmed indefinitely. Actually, the charge cannot remain indefinitely, but its lifetime is normally extremely long. The stability of the program charge is called data retention; that is, the ability of the device to retain its charge as programmed. There are two basic leakage mechanisms: direct tunneling and thermal leakage. These mechanisms occur independent of whether the cell was programmed by electron injection or tunneling. The amount of direct tunneling is a function of the potential across the tunnel oxide, and is generally very low. Leakage is normally dominated by thermal charge decay. On one side of the energy barrier, there are electrons with a distribution of energies (see Figure 14). Some have enough energy to escape over the top of the barrier. As the temperature is raised, more electrons achieve the energy required to overcome the barrier. Because of the need for accurate programming, and for ensuring that the programming algorithms are up-to- Thermal Leakage ________ . . ________ Direct Tunneling '0 ?a\~\e\\... Trapped Electron -------- 0 \\ ~EnergY Distribution Tunnel Oxide Barrier Electron Source \ \, , ,, ,, , , ,, ,, , ,, ,, ,,, , ,, ,, Floating Gate 16507A-024A Figure 14. Data Loss Mechanisms Inside AMD's CMOS PLD Technology 4-13 ~AMD The tendency of the gate to leak can be modelled as an Arrhenius function, which means the formula forthe programming "decay time" td has the form: where Ea is the intrinsic activation energy T is the temperature in Kelvin k is Boltzmann's constant K is a scaling constant. If we can measure the rate at two known temperatures, then: tdl KeEa/kTl td2 Ke E a/kT2 Note that the constant K drops out, so we need not be concerned with it's specific value. From this we find that Ea = T21ntdl - Tllntd2 kT1T2 This lets us measure Ea, which should be constant for a given process. The higherthe value of Ea, the longerthe decay time will be. This is because Ea roughly represents an energy "barrier" that must be overcome for an electron to leak away. The higher the barrier, the fewer electrons have the energy to overcome Ea. Charge leakage can be aggravated by poor quality tunnel oxide. Defects in the oxide provide a lower energy path for discharging, effectively lowering Ea. Baking a device accelerates this leakage, and identifies devices with weak oxide. AM D uses a bake for all EE products to ensure that the production devices have a high Ea and therefore good data retention. The average Ea for all devices, including those with weak oxide, is about 0.8 eV. After eliminating the weak devices by a 250°C 24-hr bake, the average Ea is about 1.8 eV. Data retention time depends on the temperature to which the devices are exposed. The higher the temperature, the shorter the decay time because the electrons have more energy, and more can leak off the gate. There are two temperatures that may be of concern for different reasons: the maximum device storage temperature (150°C) and the maximum operating temperature (125°C for military). In the first case, the idea is to know that if a programmed part sits on a shelf for some period of time before being used, that the program will remain intact forthat time. The second case is intended to give an idea of how long a device will remain operational in-system. Using the equation above to solve for the decay time at these temperatures, the result is several decades forthe 4-14 storage temperature, and even longer for the operating temperature. For room temperature, the exponential nature of the function makes the decay time increase to centuries. AMD specifies 10 years at the maximum storage temperature (an industry standard for EPROMs and EEPROMs), and 20 years in-system under worst-case military conditions. That the calculated numbers are so much higher builds confidence in the numbers specified. In general, the typical end-of-life failure mechanisms that affect all devices (and which are unrelated to the EE cells) will cause device wear-out before the program data is lost. The integrity of the charge in the electrically erasable cell also stands up to any electrical fields that exist in surrounding equipment. For charge to be transferred off, or onto, the floating gate, a field must be placed across the oxide. Such a field cannot be generated outside the programming mode; an external field, no matter how strong, cannot set up the programming mode. The charge might also be pulled through some other oxide if the field were large enough. However, to remove the charge through anything but the tunnel oxide requires an external field so high that the rest of the device would break down before any cell charge were ever lost. This would occur on any device, programmable or not. Therefore any external field strong enough to remove charge from a floating cell will destroy the rest of the device first. Cell Endurance Anotherfactorthat affects data retention in the long term is the cell endurance. The endurance is the number of times the device can been erased and reprogrammed. Over time, the oxide can wear out, resulting in a gradual reduction in Ea. This occurs as defects are created in the oxide. These defects trap electrons; these electrons then oppose the field that is required for programming. Given enough trapped charges, the established potentials will be insufficient for programming. This typically happens after hundreds of thousands of reprogramming cycles. The ability to charge up a cell with good data retention can be measured by the margin voltage. This is the voltage that must be applied to the control gate to counteract the charge on the floating gate. If the gate is highly charged, a larger margin voltage is needed to overcome the charge. Thus, put simplistically, a higher margin voltage indicates beUer cell charging. Figure 15 illustrates measurements of the margin voltage as the number of program/erase cycles is increased. By 100,000 cycles, the margin voltage still is greater than 4 V; for the cell to fail, the margin voltage must fall to below about 1 V. Inside AMO's CMOS PLD Technology AMD~ 5.0 4.5 ~ 4.0 .r= -> 3.5 3.0 10 1000 100 10000 # Cycles 100000 16507A-025A Figure 15. Cell Endurance: Margin Voltage Solid After 100,000 Program/Erase Cycles For EEPROMs, which often are reprogrammed in-system, it is important to know how many thousands of times the device can be reprogrammed. However, most EE PLDs are not intended to be programmed insystem, and probably are programmed very few times. Most production units are programmed only once by the user. Prototypes might be programmed tens of times at most. Therefore we specify a maximum number of 100 erase/reprogram cycles. This does not imply that the devices are weaker than EEPROMs; it is just that more extensive testing would have to be done to justify specifying a larger number. Since this larger number is not needed, a cost savings is realized because of the test simplification. Note that the devices are actually programmed hundreds of times in testing before they are shipped out, giving outstanding programming and functional yields; however, the number of erase/reprogram cycles specified refers only to programming done by the user. DEVICE CHARACTERISTICS Power Dissipation CMOS technology is associated with low power, and indeed, all CMOS PLDs provide lower power than their bipolar counterparts. However, most PLDs do not provide the zero-standby power that standard CMOS logic parts provide. The basic CMOS inverter lowers operating power because at any given time, only one of the two transistors can be fully ON. The other is OFF and blocks the flow of DC current. Thus, when the device is in a stable state, no current can flow. While the device is switching, both transistors are partially ON, allowing for a transient current spike. Thls means that power is consumed only when the device switches. Because a spike occurs for each transition, the average power consumption is affected by the frequency of operation (see Figure 16). This type of circuitry is found throughout most of a PLD circuit. However, one portion of the PLD circuit does not use a standard CMOS inverter: the programmable array. One of the necessary elements of zero-power operation is that the output of the inverter have a voltage swing from ground to Vcc, so-called rail-fa-rail operation. In the array, such a wide swing makes the propagation delays too long. To speed up the device, the sense amps that determine the state of a product term are designed to have a much more limited swing. This means the sense amps are constantly drawing power, even when not switching. These are the half- and quarter-power CMOS PLDs; their power consumption is still less than that of a bipolar PLD. Since most CMOS PLDs are used in TTL sockets, the CMOS PLDs work well. For designs that require the absolute lowest-power operation, the half- and quarter-power CMOS PLDs are inadequate. Zero-power PLDs have been designed to address this range of applications. These devices operate by turning off the sense amps in the array if no signals switch for a period of time. If the transition detectors at the inputs indicate that some Signal is changing, then the array is activated to process the incoming data. In this manner, the average operating power consumption can be reduced, especially at low frequencies, and the standby power consumption is negligible (less than 15 ~). The only penalty is a small wake-up delay of a few nanoseconds. Inside AMD's CMOS PLD Technology 4-15 ~AMD Vee + LC Vout= LOW - 16507A-026A The output load can have a dramatic effect on power dissipation, especially on devices that have many I/O pins. For an output driving a purely capacitive load, the power dissipation contributed by the load for one output is determined by the load capacitance, the frequency at which the output is switching, and the output voltage swing (Vs). The output stage will go through a process of repeatedly charging and discharging the capacitor. Although the direction of charge flow reverses itself every other transition, the relative voltage change does too, so that the power contribution is the same for a charge and a discharge. If we consider the case of charging the capacitor, we will be placing a charge QL on the capacitor that is determined by a. Vee OL = CLVO whereCL is the load capacitance and Vo is the output voltage. The current contribution from this is i Vout = HIGH dOL - dt dVo =CL- 16507A-027A dt In one half the output transition period tp, the change in output voltage will be equal to the output swing Vs. This means that b. i =CL~ tp 2 =2CL~ HIGH tp Vout =2CL Vsfo LOW where fo is the frequency at which the output is switching. Power Dissipation o 16507A-028A The power dissipation is the product of the current and the voltage. Since the voltage is changing during the time that the power is being dissipated, we can approximate by dividing the voltage swing by 2. This gives P =iv c. Figure 16. CMOS Inverter Power Dissipation: a. VOUT Static LOW; b. VOUT Static HIGH; c. Dynamic Power Dissipation = = Icc VS VI and Loading The greatest external contributors to Icc are the input HIGH level (VIH) and the output load. As the VIH drops from its ideal level of Vcc, the inverter starts to draw current. The worst case scenario would be a VIH at the minimum of 2.0 V, which could contribute some 5 rnA per input buffer to the power consumption. 4-16 Vs =2CL vsfoT 2 =2CL V sfo This means that for a 100-output device (PLD or any other device) with each output driving 35 pF loads, where the output swing is 3 V and the output frequency' is 50 MHz, the power dissipation contributed only by the load will be about 1.6 W regardless of the power dissipation of the chip itself. Icc vs Frequency The operating current increases with frequency for both standard and zero-power CMOS devices. The Inside AMD's CMOS PLD Technology AMD~ device typically draws less than 10 J.1A. Figure 17 shows typical curves for standard and zero-power devices. difference is the current at low frequencies. A standard device typically can draw 35 rnA at 0 MHz; a zero-power 100 50 o 4 8 12 20 16 24 Frequency (MHz) 16507A-029A a. 110 90 70 50 30 Unear 10 --------- ----- -- ------- --------- -- ---------------- ------ -- -----Log 0.1 Log 0.01 Unear -+----+----+-----+----+----+---+--+----1 0.1k 1k 10k 100k 1M 10M Frequency (Hz) 30M 50M 70M 16507A-030A b. Figure 17. Icc vs frequency: a. Half-Power Device; b. Zero-Power Device Inside AMD's CMOS PLD Technology 4-17 ~AMD All but the PALCE16V8 and PALCE20V8 have their Icc specified at 0 frequency (that is, DC). For compatibility with existing specifications, the 16V8 and 20V8 have their Icc specified at a frequency level: 15 MHz for devices with tpo of 15 ns or slower, and 25 MHz for 10 ns, 7.5 ns, and faster devices. Icc vs Number of Product-Terms The number of product terms switching can sometimes affect Icc. On standard devices, however, the design of the particular sense amp determines whether the Icc will increase or decrease with more product terms. therefore it cannot be predicted in general. From a practical standpoint, the change in Icc due to different numbers of product terms is negligible. AM D's zero-power devices have been designed with a product-term power-down feature that turns off those product terms not being used. The graph in Figure 18 shows the effects. Because these devices are intended for low-power and battery-operated use, the substantial extra power savings can significantly help extend the time between battery charges. 110 100%50%- 90 25%- 70 50 30 <' .s Linear 10 Log ~ 0.1 Log 0.01 Linear -;----+---+---+----+---+---+---1----1 0.1k 1k - Percent of product terms used 10k 100k 1M 10M 30M Frequency (Hz) SOM 70M 16507A-031A Figure 18. PrOduct-Term Power-Down on Zero-Power Devices Icc vs Temperature The amount of current drawn by a device depends on how much current can pass through the transistors. Simplistically speaking, the channel of a transistor can be modelled as a resistor. The resistance is affected by temperature, since temperature affects the mobility of electrons. The hotter the device is, the more the molecules are vibrating around, and the harder it is for electrons to pass through without a collision with a mole4-18 cule; that is, electrons are less mobile in a hot device. This means that the resistance of the channel is higher, which in tum means that the device conducts less current. Therefore Icc is greatest when the device is cold, and is minimized when the device is hot. A typical curve is shown in Figure 19. This curve has been generalized by normalizing the current to the room temperature current. Inside AMD's CMOS PLD Technology AMD~ +20% +10% ~ O~----------~--=-------~--------~~--------~ 80 20 -10% Temperature (OC) -20% 16507A-032A Figure 19. Icc vs Temperature, Normalized to Room Temperature 2 Icevs Vee P The variation of Icc with changes in Vee should come as no surprise; as Vee increases, so does Icc. This means that the power consumption actually increases roughly as the square of Vec, since power consumption can be expressed as = Vee Icc v cc = Reff . where Reft is defined as i~~ This is a simplification, of course, since Ref! is non-linear, and varies with Vee. A typical Icc vs Vee curve is shown in Figure 20. 30 20 10 0~--~~--~--~----4---~----4----+--~ o 2 4 6 Vcc(V) 8 16507A-033A Figure 20. lee vs Vee Inside AMD's CMOS PLD Technology 4-19 ~ AMD Input/Output Structures Newer devices have pull-up resistors as shown below. In these devices, there is also a transistor in series with the resistor. The basic input and input/output structures are shown in Figure 21. The ESD circuits and the programming voltage detection circuits will be discussed in more detail later. Vee Vee >50kn ESD Protection and Clamping Programming -=Pins Only ----- -- --- -- --- ---- -16507A-034A a. Vee Vee > 50 kn [~ Provides ESD Protection and Clamping Preload Circuitry Feedback Input 16507A·035A b. Figure 21. Equivalent Input/Output Schematics: a. Input with Pull-Up Resistor and Overshoot Filter; b. Output with Pull-Up Resistor 4-20 Inside AMD's CMOS PLD Technology AMD~. I-V curves Since the input is effectively a capacitor, the impedance has no real component; the imaginary portion falls with increasing frequency. A typical device has an input capacitance of 8 pF at 1 MHz. Assuming a capacitance around 8 pF at higher frequencies, this yields a capacitive reactance of 2.5 KQ at 50 MHz. Figure 22 shows a typical I-V curve for an input buffer. Within the range of normal input signals, the input buffer has extremely high impedance, with diodes and MOS transistors that turn on when the input is below ground. On higher speed devices, this has the effect of a highspeed diode capable of clamping negative overshoot on noisy signals. II (rnA) 20 2 3 4 5 -20 -40 -60 -80 -100 16507A-036A Figure 22. I-V Curve for an Input with No Pull-Up Resistor Inside AMD's CMOS PLD Technology 4-21 ~AMD Figure 23 shows typical 1-V curves for high and low TTLstyle outputs. The impedance of a low output is about 10 Q; a high output has an impedance of about 30 n. The fact that the impedances are somewhat more symmetric than those found on a bipolar device makes it a bit easier to terminate long traces accurately. IOL(mA) 80 VOL (V) -1.0 -0.8 -0.6 -0.4 .6 1.0 .8 16507A-037A 8. IOH (rnA) 25 2 3 4 5 VOH(V) -3 ·2 -1 16507A-038A b. Figure 23. I-V Curves for 8 TTL-Style Output with No Pull-Up Resistor: 8. Output LOW; b. Output HIGH 4-22 Inside AMD's CMOS PLD Technology AMD~ loaded to about 50 n when lightly loaded. The n-channel impedance is lower, at about 10 n. Figure 24 shows the curves for rail-to-rail switching outputs. The p-channel impedance, when the output is HIGH ranges from 200 n when extremely heavily IOL(mA) +100 VOL (V) -2 2 6 4 -100 16507A-039A a. 10H (mA) +10 VOH (V) -1 2 3 5 -10 -20 -30 -40 -50 16507A-040A b. Figure 24. I-V Curves for a CMOS-Style Output: a. Output LOW; b. Output HIGH Inside AMD's CMOS PLD Technology 4-23 ~ AMD Open Inputs Newer devices have input pull-up resistors with a minimum resistance of 50 kil. When unused, these inputs can be left unconnected. With older devices, an unused input should be pulled HIGH or LOW. A floating input may cause no trouble, but there are some potential concerns. First, if an input is floating with its voltage near threshold (1.5 V for a TTL-style device), the input buffer can conduct tens of mA of current. This does not damage the device, but must be calculated into the power budget. Of course, as the input moves away from the threshold, the current decreases. In a noise-free environment, any floating inputs will generally tend to drift to ground. The second concern is the fact that the environment is not usually noise-free. The unused input is not directly connected to any internal logic, so there should be no interference between the input and the other logiC. However, if there is noise on an unused input floating near threshold, internal noise could be generated should the input buffer start to oscillate. This could disturb some of the surrounding circuitry as well as the internal ground, compounding the problem. On a device without pull-up resistors, an unused 1/0 pin can be pulled HIGH or LOW by programming it as an output with constant value 1 orO. The output buffer itself then acts as the pull-up or pull-down. Output Drive vs Temperature and Vee The output drive varies with the temperature just as Icc does. As the temperature increases, electron mobility decreases, cutting the drive. Likewise, the drive increases as the temperature decreases. For example, at 75° IOL decreases by about 18% from its room temperature value; IOH decreases by about 7%. The drive also varies directly with Vee, although the effect is most pronounced on IOH; it increases by about 18% when taken from 5.0 V to 5.25 V. Because a low output transistor is already ON hard, the little extra bit of drive that its gate gets as Vee goes to 5.25 V only increases IOL by about 3%. There is no explicit current-limiting resistor on the pullup. The resistance of the pull-up channel limits the current. The fact that this resistance is smaller than what one might find in a bipolar device contributes the the more symmetric impedances, but also gives a higher short-circuit current Ise. The slew-rate-limiting circuit also limits the drive; slew rate limiting is discussed below. benchmark for confirming the guaranteed performance, but as the application changes the conditions, the actual system performance may change for the better or worse. AC Test Conditions AC test conditions are sometimes treated differently for CMOS than they are for bipolar. However, since most of the CMOS products are designed to work in a TTL environment, the test conditions that AMD uses generally are the same as those used for bipolar devices. The resistor network is chosen to match the output drive levels, and the load capacitor is normally 50 pF. JEDEC recently changed the load standards from what had to date been the industry de facto standard, but for TTL parts this only affects the resistor values; a 50 pF capacitance is still part of the standard. Note that in the JEDEC standard, the decision affecting which load to use depends only on the interface level, not the technology. Thus a" parts intended to operate at TTL levels are given the same load, whether bipolar or CMOS. AMD has made two exceptions to the 50-pF load. The first is forthe zero-power devices, which are designed to operate at true CMOS levels. The JEDEC load standard is different for these devices; it has a different resistor network, and uses a 30-pF capacitor. The second exception is the MACH family: in this density range, a precedent had been set at 35 pF, prior to the JEDEC standardization. To be compatible with existing devices, the MACH devices are measured with 35 pF loads. tpo vs Temperature Propagation delays decrease (that is, they speed up) at colder temperatures for the same reasons that Icc increases. In general, devices at O°C operate about 15% faster than those at 75°C. tpo vs Vee As Vee is increased, more power is available, and the device can operate faster. However, the effect is less pronounced than that of temperature. A device operating with a 5.25 V supply runs about 4% faster than one running with a 4.75 V supply. tpo vs Loading The tPD increases as the device load increases, although much of this results from the increase in rise and fall times of the outputs. For every 50 pF change in load, roughly a 2- to 5-ns change in the rise and fa" time can be expected. In addition, as the load increases, more transient current is switched,. creating more internal noise. This can slow the speed path inside the Chip. AC Parameters AC parameters vary with a number of conditions. The data sheet specs pick one set of conditions that act as a 4-24 Inside AMD's CMOS PLD Technology AMO~ Power-Up Reset Power-up reset is a feature that forces a device to power up into a known state. Without this feature, the power-up state is not known. Power-up reset helps make system initialization and testing simpler. The ramp rate of Vcc is not critical to the power-up reset function. However, there are two other requirements: the supply ramp must be monotonic, and the clock must be suppressed until power-up is complete. (Figure 25). The danger in such glitches is that if the timing and voltage are just right, the registers themselves may think that the device powered down temporarily, causing them to lose their state. If the glitch is fast enough, however, the power-up reset circuit may not notice the glitch, and may think that everything is proceeding just fine. At the end, the registers may be in a random state. Even if the power glitches low enough for long enough to shut down all circuits, the power-up timing must be restarted from the end of the glitch. The monotonicity requirement basically says that there should be no low-going glitches in the power-up ramp T , Power OFF ~: . Registers May Lose State Time , : _____ Power Back ON Reset Circuit May Not Restart If Duration Short, Reset Circuit May Not Notice 16507A-041A Figure 25. Non-Monotonic Power-Up Can Cause Power-Up Reset To Fail There is also a requirement that the clock not be running during power-up (Figure 26a). If the clock is running while the device is powered up, then, as different parts of the device-and, indeed, the whole circuit boardturn on, parts of a single device, or different devices, may be out of synchronization with each other (Figure 26b). At some point, a part of a device will be ON enough to start recognizing the clock. It will then start to sequence as per the inputs it sees. If the inputs are not stable, the sequence may not be correct. In addition, if not all parts of the circuit or board recognize the clock at exactly the same time, some parts will start cycling before others, and the whole system will be out of synchronization. The ot~~r potential (although remote) problem with clocking during power-up is metastability. If a register powers ON in time to see the clock edge, its setup time might have been violated, making the results at the output unpredictable. Inside AMO's CMOS PLO Technology 4-25 ~AMD 4V Power Vee IIr 7 / /I~ Registered ActiveLow Output ~ \\(, Clock -,L J 16507A-042A a. ~----Nominal PartA----, .. ' . .. ' •• " - - - - - Part B :......... " I Time Threshold ", Part A ON Part A : - Responds To Clock , Part B ON Part B i-Responds , To Clock , ~I~ If Too Short, Metastability Possible 16507A-043A b. Figure 26. Clocking During Power-Up Reset: a. Correct Operation; b. Free-Running Clock Places Part B One Clock Cycle Out of Sync with Part A 4-26 Inside AMD's CMOS PLD Technology AMD~ Powered-Down Characteristics Some applications place the CMOS PAL device in a situation where it is itself powered down, but it is driving or is driven by other devices that are still powered up. This is especially typical of devices that are talking directly to a bus (Figure 27). The characteristics of the device in such a condition depend on how the power was removed. There are two ways of removing power: • opening up the Vcc line (e.g., if Vcc is fused, and the fuse blows; Figure 28) , • grounding Vcc (Figure 29) Bus - Powered Up I ...--_ _-----' u u u u u I u u u u u'--_ _---, I I PLD Board - Powered Down 16507A-044A Figure 27. Powered-Down Device with Active Inputs and Outputs / Blown Fuse ~~-------------y ~------------~) les 16507A-045A Figure 28. Power Down with Vee Open Inside AMD's CMOS PLD Technology 4-27 ~AMD Vee Effectively Grounded I 1 16507A-046A Figure 29. Power Down with Vee Grounded 16507A-047A a. Internal Current External Current I' External Diode Only Blocks External Current IC On-Chip Off-Chip 16507A-04SA b. Figure 30. Powered-Down Current Paths with P-Channel Pull-Up: a. Vee Grounded; b. Vee Open 4-28 Inside AMD's CMOS PLD Technology AMD~ It is important to know whether, for a given device, there is some kind of path from the pin to Vcc when Vee is lower than the pin voltage. If any current can flow, it is not necessarily catastrophic, but there can be some effect. If Vce is grounded, then there is a direct path to ground for any current flowing from the pin to Vee (Figure 30a). If Vce is open, then the only path from Vee to ground is through the device itself, and through the Vee lines of any other devices on the same Vec line (Figure 30b). In the latter case, the pin is essentially powering up the device(s) itself; realistically, it cannot provide enough power to drive the chip, and this could result in the pin being loaded down. Most of AMD's CMOS PLDs have no such path when powered down. Figures 31 and 32 show the I-V curves of inputs and I/O pins while Vec is open and Vce is grounded. Figure 31 is for TTL-compatible devices, which have n-channel pull-ups on the outputs. Figure 32 is for the HC/HCT-compatible zero-power devices and the PALCE610H-15, which have p-channel pull-ups. 8 6 14 2 0 o 2 4 6 0 2 4 6 VI (V) 16507A-049A 16507A-053A a. b. Figure 31. Power-Down Characteristics of TTL-style CMOS Inputs and Outputs: a. Standard; b. Older ESD Structure 14 12 10 10 8 8 <' E 6 ::;6 14 4 -=2 2 0 0 0 2 4 6 0 2 4 6 VI (V) 16507A-055A a. 16507A-056A b. Figure 32. Power-Down Characteristics of CMOS-Style Output: a. Vcc Open; b. Vce Grounded Inside AMD's CMOS PLD Technology 4-29 ~ AMD Note that for most of the TTL-compatible devices, there is no leakage on the pins. This means that signals on a pin are not affected by the powered-down device. Therefore it can be safely connected to an active bus. It also allows for safe hot insertion, where the device (or the board that contains the device) is plugged into a socket that has Vcc applied. As a result of one of the ESO structures (which are discussed below), some devices do conduct some current when Vcc is powered down (Figure 31 b). Newer devices do not have this characteristic. With the HC/HCT-compatible devices, the input structures are the same as for TTL devices, but the outputs conduct because of the p-channel pull-up. There is a parasitic diode between the output and Vcc (Figure 33). This can cause latch-up if the output voltage is higher than Vcc. Thus it is not recommended that devices with p-channel outputs be directly connected to a bus if the device will be powered down while the bus is active. Hot-insertion of these devices should also be avoided. Output p-substrate 16507A-057A Figure 33. Parasitic Diode In CMOS-Style Outputs DEVICE INTEGRITY AND ROBUSTNESS The reliability of AMO's CMOS processes is documented in product and process qualification books. For EE4 process products generally, the extended life FIT rate is under 100 and declining. The rate for the devices with 2000 hours burn-in is around 30; similar devices with only 1000 hours burn-in have a FIT rate closer to 100. With more burn-in experience, the FIT rate will decline even further due to the statistics used to calculate FIT rates. The FIT rate calculation is such that with fewer burn-in hours, a lower confidence factor is applied, giving higher FIT rates on newer products even when there are no failures. ESO Every pin on the devices is protected against electrostatic discharge (ESO), a formal name for static electricity shocks. Output pins rely on the large output drivers as protection. Inputs normally do not have large drivers, so a circuit must be added for input protection. These input protection circuits also provide clamping against negative overshoot. All new devices make use of the structures in Figures 34a and 34c for ESO protection. Most input pins 4-30 use the circuit in Figure 34a. On pins requiring high voltages, the circuit has been modified as shown in Figure 34c. Some older devices have the configuration shown in Figure 34b. Because the active pull-downtransistor is not ON when Vcc is disconnected, it cannot necessarily hold off the ESO transistors; this causes the current seen in Figure 31 b. This circuit is no longer being used in new devices. Noise Generation and Sensitivity AMO's CMOS PLOs are designed with noise concerns in mind. This affects both the amount of noise generated by the devices and the way in which the devices react to externally-generated noise. As more is understood about the nature of system-level noise, new design techniques are being used to make the devices quieter and more robust. Ground Bounce Ground bounce occurs when many outputs simultaneously switch from HIGH to LOW. This occurs because of the fact that CMOS devices generally have outputs that switch very quickly. If left uncontrolled, ground bou nce can make a device with many outputs unusable. Inside AMD's CMOS PLD Technology AMO~ Vee Vee r r Input ~----.- - 16507A-058A Output - - Causes Ground Bounce a. 16507A-061A Vee Figure 35. Origins of Ground Bounce Input Vee y 16507A-059A Any output that is at a static LOW level maintains a VOL with respect to the chip ground. If the chip ground is bouncing with respect to the board ground, the LOW output will track the moving chip ground and will also appearto bounce (see Figure 36). This is sometimes seen as a glitch by the next device. Even if there is no output glitch, instances of high ground bounce can slow the performance of the internal circuits by temporarily starving them of power. In extreme cases, this can interrupt the internal circuits. b. LOW Output I VOL Vee (/) .g Chip Ground > I 0 Input Time 16507A-060A 16507A-062A c. Figure 34. ESO Protection: a. Standard; b. Older Version; c. Supervoltage Pins Ground bounce is generated by the natural parasitic inductance in the ground lead (see Figure 35). When a large current surge goes through the inductor, the high ~ induces a voltage that puts the ground level on the chip at a higher voltage than the ground level seen on the board. Figure 36. Symptoms of Ground Bounce Excess ground bounce can be handled in two ways: by limiting the amount of ground inductance and by reducing the ~ . Inductance can be reduced by improving the configuration of the ground pin. On AM D's 28-pin devices with many outputs (PALCE24V10 and PALCE26V12), the ground pin has been moved from Inside AMO's CMOS PLO Technology 4-31 ~AMD of the overshoot, and reduces or eliminates ringing. Figure 39 shows the ESD protection circuit used on most input pins. Parasitic p-n junction diodes exist between the substrate and the n-type source and drain, although these diodes are relatively slow. Faster reaction is provided by the n-channel devices themselves. When the input is too negative, the gate-to-drain voltage is positive. If the drain is more negative than the threshold voltage, the transistors turn on in the reverse direction, with the drains acting as a sources. This happens very quickly and acts as a clamp. This will also happen on an I/O pin, with the low output driver acting as the clamp. the corner to the center of the DIP package, effectively reducing the inductance by a factor of about four. Ground bounce is also controlled by limiting the slew rate of all the output drivers (see Figure 37). This slows down the fall time and reduces the rate of current change by as much as 25%. Overshoot Sensitivity Overshoot is a form of noise usually generated when signaltraces act as transmission lines but have not been adequately terminated. The resulting reflections can cause significant overshoot, with as much as double the intended swing applied to the input in the negative or positive direction. While it might appear that parts with negative substrate bias can "tolerate" more negative overshoot, it is really more accurate to say that these parts allow more negative overshoot, since there is no clamping. If there are effective input clamps, which are possible with a grounded substrate, then it will look like the part never gets as much negative overshoot. This does not mean it can't handle the overshoot; it means that it is clamping the overshoot. If you take the part out of the socket, you will see that when unclamped, the overshoot will increase dramatically, as illustrated in Figure 40. Since AMO's devices have a grounded substrate, they are inherently better equipped to handle negative overshoot. Negative Overshoot Negative overshoot (Figure 38) poses no problems for a device that has been carefully designed. There is no detrimental effect as long as no unexpected parasitic behavior occurs due to the fact that ground is no longer the most negative voltage. However, the ringing that usually follows overshoot can slow down system performance, since the system has to wait for the ringing to subside. Clamp diodes are useful for stealing the energy present in the ringing, and cutting the ringing short. A fast clamp reacts to the overshoot as it occurs, cuts the amplitude Vee Output Figure 37. Output Drivers with Slew-Rate Control Undershoot Overshoot 16507A-064A Figure 38. Definition of Negative Overshoot and Undershoot 4-32 Inside AMD's CMOS PLD Technology AMD~ Input tESD Circuit 16507A-065A a. -1 V ". ---------------------C>I---------------- --p-substrate 16507A-066A b. Figure 39. Negative Overshoot Clamping: a. Circuit Diagram; b. Cross-Section Inside AMD's CMOS PLD Technology 4-33 ~AMD ........................ ~ .......... ~ . . .. . . . . .l.,~f..~\ ·.. ·· .. ····,f.\.:·· .. ·.l·· ~ ~ ~ ~ J\ -=- ~~ . ...... } ......1......l· ".. ..!..f. . \., ," . . .l:'·.·· ······l,··:.;· l' t: . . . for':'" I ,. . ,.ll ' " ~: 1·:····'-ri" ¥ .... ······i .. · .. J \ I '..\ ...! . ~ ......:~f:~... ". ,i.,. . :.. . :.' t \,.i I ~1 r .. ······..;:···.:······· . ····V· . ·· rd: ~)l:··· ¥ i J J \ ~! ···········:\l·· .... i u.. . . . . .. := Ch.1 Timebase 2.000 volts/div 50.0 ns/div a. .;..~ l "':.:.;::i ~::=::. ""..:: .P ~\:.::4 ~F==,/l:\ "',;/ \ ~ I ~ ~j Ch. 1 Timebase . 2.000 volts/div 50.0 ns/div b. Figure 40. The Effect of Clamping: a. Signal Driving Empty Socket; b. Signal Driving Same Socket with CMOS PAL Device In It 4-34 Inside AMD's CMOS PLD Technology AMD~ Positive Overshoot Large amounts of positive overshoot (Figure 41) can be a problem on most PLDs, regardless of technology or vendor. This is because most PLDs are programmed using supeNoltages, and the pins therefore have supervoltage detectors that turn on the programming or test circuits, and potentially disable parts of the normal operating circuitry. If there is too much positive overshoot, the signal can travel into the programming voltage range, briefly activatingthe programming circuitry. This can result in functional interruptions, such as outputs momentarily starting to disable or going from HIGH to LOW. For earlier devices, the problem can only be avoided by revising the design to reduce the overshoot. A particular design in a particular device might work, but this might be because that device has no supervoltage function on that particular pin. But if you use an alternate source with different supervoltage pins, the design might not work. New AMD CMOS devices incorporate a filter, or delay circuit, that delays the reaction of the programming circuit for about 100 ns. This is enough to reject overshoot signals, which usually last for less than 30 ns. Positive overshoot wil not cause any functional interruptions on devices with this protection (see Figure 42). Overshoot Undershoot 16507A-067A Figure 41. Definition of Positive Overshoot and Undershoot Super Voltage Detect Circuit Programming or Test Input Data Path ':X)--_.>----l " ) 0 - - . Programming or Test Circuitry 16507A-068A Figure 42. Positive Overshoot Filter Inside AM D's CMOS PLD Technology 4-35 ~AMD Latch-up is normally triggered by an input or output at a voltage significantly above Vee or below ground, with enough current drawn to cause the SCR to turn on. This condition usually occurs when hot-socketing a vulnerable part; Le., plugging a part into a powered up board or inserting a board into a powered-up system. When this happens, the inputs and Vee power up uncontrolled, and there is a risk of latch-up. Latch-Up Latch-up occurs as'a result of parasitic bipolar transistors between the n-channel and p-channel devices (see Figure 43a). These transistors form a parasitic SCR (see Figure 43b), which turns ON when triggered, conducting large amounts of current. It is usually impossible to shut OFF without removing all power from the device. The amount of current drawn is so high that it can either overload a power supply or, if the power supply can supply huge amounts of current, destroy the device. R p-substrate p-substrate 16507A-069A a. Vee Rn-well Rp-substrate 16507A-070A b. Figure 43. Latchup Mechanism: a. Cross-Sectl,on; b. Equivalent Schematic TTL-compatible outputs are intrinsically less susceptible to latch-up, since they have no p-channel pull-up. This accounts for nearly all of AMD's CMOS PLDs; these devices can be used for hot-insertion. For true CMOS outputs, the SCR is an intrinsic part of the CMOS structure and cannot be eliminated. The SCR 4-36 must be made as difficult as possible to turn ON by using guard rings and very carefully laying out input and output circuits. All of AMD's CMOS devices are guaranteed to endu re a current pulse of 100 rnA into or out of the pin without inducing latch-up; most devices can actually withstand over 500 rnA. Since AM D's zero-power parts Inside AMD's CMOS PLD Technology AMD~ and the PALCE610H-15 have true CMOS outputs, hot insertion is not recommended. COMPATIBILITY WITH BIPOLAR Most of the CMOS PLDs are designed to be compatible with TIL circuits; indeed, many designers have replaced bipolar TTL devices with a CMOS equivalent. Often this can be done blindly without affecting system performance. The interface levels are compatible and should pose no problems. Even the zero-power parts have been designed with input buffers that can respond to TIL or CMOS signals. However, when making such a conversion, some details require attention, especially in cases where a straight conversion appears not to work. Ground Bounce Because CMOS devices generally have higher output slew rates, designs having many outputs switching at the same time (particularly if the outputs are heavily loaded) can cause more ground bounce than that generated by a comparable TTL device. It is important to use devices with output slew rate control. The slew-rate-limiting circuits help minimize the occurrence of conversion problems, but even when the output slew rate is limited, the signal still can switch more quickly than that from a TIL output. If a design cannot be modified to accommodate the faster edge rates, this ground bounce may make a conversion unfeasible. If deSign changes are possible, any of the following can be tried: • Limit the number of outputs that can switch at once. • • Reduce the loading on the outputs. Go to a lower-lead-inductance package (like a PLCC). • Ensure that the ground path on the circuit board has low inductance. pear on different pins for different devices, and the supervoltage functions vary. Thus, overshoot on one pin of a particular bipolar device might have had no effect. Once that device is changed (whether to CMOS or any other device that has no overshoot filter), the new device might react to the overshoot and cause problems. The solution is to ensure that all signals are clean and have minimal overshoot·, making them compatible with any device. Signal noise reduction can be accomplished most effectively by contrOlling the impedance of the signal traces and terminating correctly. As an alternative, if the driving device has extremely fast edge rates, it can be replaced with a device that has better controlled out' put slew rates. Direct JEDEC File Conversions[ from Bipolar to CMOS With some CMOS devices (most notably the PALCE16V8 and PALCE20V8), converting logic from a bipolar device is particularly simple once the noise issues have been addressed. This can be done in the programmer or by conversion software. It only affects the JEDEC file; the source file is not required. Generally, this is recommended only for designs whose source file is not available. If the source file is available, it is recommended that you change the device type in the source file, and then recompile to generate a new JEDEC file. This permits better documentation and revision control, since the source file is then consistent with the JEDEC file being used in production. SUMMARY By concentrating on the needs of CMOS PLD users, AM D has developed industry-leading CMOS technology that can provide cost-effective PLDs of unequalled quality, reliability, and performance. AMD provides value through: • AMD-owned fabs, for better control of quality, reliability, volume, and costs • electrical erasure, for higher quality and lower cost the highest performance available Overshoot The other possible problem when converting from bipolar to CMOS is reaction to signal overshoot in a noisy system. This is only an issue if the CMOS device has no overshoot protection. Overshoot sensitivity is not specifically related to CMOS, but results from programming algorithms being different between the technologies. This also can occur when changing between bipolar vendors, or when changing between CMOS vendors. If the noise on a signal can disturb supervoltage circuitry, this can be troublesome. Different devices have different sensitivities; this accounts for some of the apparent incompatibility. However, the culprit usually is the fact that supervoltages ap- a • • robust technology that is quiet and yet tolerant of noise an extremely broad offering of products; low and high density, low and zero power This application note has detailed many of the aspects of the technology that make it superior to any alternatives. This, together with the information in the individual data sheets, qualification books, and a crew of applications engineers, should provide answers to your questions as you make use of AMD's CMOS PLD technology. Inside AMD's CMOS PLD Technology 4·37 ~AMD INDEX charge pump latch-up protection negative substrate noise .................................................................. . compatibility with bipolar ...................................................... . ground bounce .......................................................... . input thresholds .......................................... '............... . JEDEC file conversion ............................................ : ....... . overshoot .............................................................. . contact dimension ........................................................... . control gate ................................................................ . coupling ratio ............................................................... . data retention .............................................................. . Arrhenius function ........................................................ . bake .................................................................. . effect of external fields .................................................... . thermal leakage ......................................................... . value of activation energy .................................................. . depletion mode transistor ..................................................... . endurance ................................................................. . ESD ...................................................................... . powered down .......................................................... . overshoot clamping ....................................................... . FIT rates .................................................................. . floating gate ............................................ '. ................... . FusionPLD program ......................................................... . gate length ................................................................ . gate oxide thickness : ........................................................ . ground bounce ................. ,............................................ . ground lead inductance ................................................... . hot electron injection ......................................................... . hot insertion ................................................................ . I-V curves ................................................................. . impedances ................................................................ . input pull-ups ............................................................... . inputs connection to array ........................................................ . ESD protection .....................' ..................................... . floating ................................................................ . hot insertion ............................................................ . I-V curves .............................................................. . impedance ............................................................. . overshoot clamping ....................................................... . overshoot filter .......................................................... . powered down .......................................................... . pull-up resistors ......................................................... . state during power-up ..................................................... . structure ............................................................... . transition detectors ....................................................... . TTL, CMOS compatibility .................................................. . voltage level ................ '............................................ . 4-38 Inside AM D's CMOS PLD Technology 4-4 4-3 4-4 4-37 4-37 4-37 4-37 4-37 4-3 4-4 et seq 4-9 4-13 4-14 4-14 4-14 4-13 4-14 4-3 4-14 4-30 4-30 4-32 4-30 4-4 et seq 4-13 4-3 4-3 4-31,4-37 4-31,4-37 4-4 4-30, 4-36, 4-37 4-21 4-21 4-24 4-12, 4~13 4-30 4-24 4-30 4-21 4-21 4-30,4-32 4-35 4-29 4-24 4-25 4-20 4-15 4-37 4-16 AMD~ INDEX (continued) inverter cross-section ............................................................ . power consumption ....................................................... . JEDEC file conversion ....................................................... . JEDEC load standard ........................................................ . latch-up ................................................................... . hot insertion ............................................................ . negative substrate ....................................................... . relationship to overshoot ................................................... . Left ••••.•...•••••....•••••....••••..••..•••..••..•.••..••••......•...•••••. MACH output load ........................................................... . margin voltage .............................................................. . metal interconnect ........................................................... . pitch .................................................................. . minimum feature ............................................................ . n-well structure .............................................................. . noise ..................................................................... . ground bounce .......................................................... . overshoot sensitivity ...................................................... . latch-up ................................................................ . outputs as ESD protection ........................................................ . as overshoot clamp ....................................................... . behavior due to overshoot ................................................. . compatibility with TTL, HC/HCT ............................................. . current limiting .......................................................... . drive .................................................................. . floating ................................................................ . ground bounce .......................................................... . HC/HCT-compatible ...................................................... . hot insertion ............................................................ . I-V curves .............................................................. . impedance .................... ; ........................................ . load ........................................................ .- .......... . power dissipation ......................................................... . powered down .......................................................... . pull-up transistors ........................................................ . rail-to-rail ............................................................... . rise and fall times ........................................................ . role in latch-up .......................................................... . short-circuit current ....................................................... . slew rate ................................................................' structure ............................................................... . TTL-compatible .......................................................... . overshoot causes ................................................................ . clamping ............................................................... . CMOS vs bipolar ........................................................ . filter ................................................................... . negative ............................................................... . positive ................................... : ............................. . Inside AMD's CMOS PLD Technology 4-3 4-15 4-37 4-24 4-36 4-30 4-4 4-4 4-3 4-24 4-14 4-3 4-3 4-3 4-3 4-30 4-31 4-32 4-36 4-30 4-32 4-35 4-3,4-37 4-24 4-24 4-24 4-31 4-30 4-30 4-22 4-22 4-24 4-15 4-27 4-3 4-3,4-23 4-24 4-30,4-21 4-36 4-24,4-37 4-20 4-30 4-32 4-3,4-21, 4-30,4-32 4-37 4-35 4-4,4-21, 4-30,4-32 4-35 4·39 ~AMD INDEX (continued) , relationship to latch-up .................................................... . overshoot clamping .......................................... '................ . role of ESD structure ..................................................... . poly-silicon interconnect ...................................................... . elimination of second layer ................................................. . power dissipation ........................................................... . frequency .............................................................. . input voltage .... . ................ " ...................................... . output loading ........................................................... . product terms ........................................................... . temperature ............................................................ . Vcc ................................................................... . power-up latch-up '.' .............................................................. . reset .................................................................. . power-up reset ............................................................. . powered-down characteristics .................................................. . programmable array ......................................................... . area .................................................................. . configuration ............................................................ . power consumption ....................................................... . programmers ............................................................... . programming cell ............................... ' ............................. . data retention ........................................................... . endurance .............................................................. . implant ................................................................ . programming and erasure mechanisms ....................................... . self-limiting programming .................................................. . underprogramming, undererasure ........................................... . propagation delay ........................................................... . rise and fall times ........................................................ . reliability .................................................................. . SeR ................................................................. ~ .... . substrate .................................................................. . as source of electrons for UV parts .......................................... . floating, negative ................................................ ,........ . grounded .............................................................. . overshoot clamping ....................................................... . supervoltages .............................................................. . transmission lines ........................................................... . tunnel oxide cell endurance .......................................................... . coupling ratio ........................................................... . data retention .......................................... '................. . field across ................................................. . ........... . thickness relative to UV ................................................... . tunneling ................................................................ " . direct .................................................................. . Fowler-Nordheim ........................................................ . UV-erasable devices ......................................................... . compared with EE ........................................................'. yields, programming and functional ............................................. . 4-40 Inside AMD's CMOS PLD Technology 4-4 4-5,4-21, 4-30,4-32 4-30 4-3 4-8 4-15 ' 4-16 4-16 4-18 4-18 4-18 4-19 4-4 4-25 4-25 4-27 4-3 4-8 4-12 4-15 4-13 4-8 et seq 4-13 4-14 4-8 4-8 et seq 4-9,4-13 4-13 4-24 4-24 4-30 4-36 4-3 4-4 4-3 4-3 4-33 4-35 4-32 4-14 4-9 4-13 4-6 4-6 4-6 4-6,4-13 4-6 4-4 4-6 4-13 AMOl1 INDEX (continued) testing to ensure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. zero-power devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. compatibility with TTL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. hot insertion ............................................................. I-V curves ..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . .. output load .............................................................. power dissipation ......................................................... product-term power down. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. rail-to-rail operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. role of grounded substrate .................................................. Inside AMO's CMOS PLO Technology 4-6,4-14 4-4, 4-15 4-37 4-37 4-23 4-24 4-15 4-18 4-3,4-15,4-23 4-3 4-41 ~AMD 4-42 ~ Military PAL Devices Advanced Micro Devices Advanced Micro Devices' Military Programmable Array Logic (PAL) devices provide state machine and combinato rial logic solutions processed to military criteria. We offer the largest number of Standard Military Drawing PAL products in the industry. sequence generators, decoders, multiplexers, adders, memory mapped I/O and much more. These designs go into radar systems, missile guidance, avionics, airport graphic terminals, parallel processors, military computer hardware, and product obsolescence solutions, just to name a few. Applications for our configurable PAL architectures include counters, shift registers, accumulators, control Military PAL Device Menu Product Terms/Output tPD 'MAl( ns MHz Icc mA 10 10 210 /BRA /B2A 12 47.6 210 20L, J, W 20 28.5 180 30 20 90 30 20 180 50 13.3 55 10 50 210 12 41.7 210 15 35.7 210 Family Part Number Package Technology Inputs 110 Outputs 16R8 PAL16L8-10 /BRA /B2A TTL 10 6 Comb 2 Comb 7 8 8 2 Comb 8 4 Comb 8 Reg 6 Reg 4 Reg 8 7,8 7,8 PAL16R8-10 PAL16R6-10 PAL16R4-10 PAL16L8-12 PAL16R8-12 PAL16R6-12 PAL16R4-12 PAL16L8B PAL16R8B PAL16R6B PAL16R4B PAL16L8B-2 PAL16R8B-2 PAL16R6B-2 PAL16R4B-2 PAL16L8A PAL16R8A PAL16R6A PAL16R4A PAL16L8B-4 PAL16R8B-4 PAL16R6B-4 PAL16R4B-4 20R8 PAL20L8-10 PAL20R8-10 PAL20R6-10 PAL20R4-10 PAL20L8-12 PAL20R8-12 PAL20R6-12 /BLA,/B3A TTL 14 12 12 12 6 Comb 2 Comb 4 Comb 2 Comb 8 Reg 6 Reg 4 Reg 7 8 7,8 7,8 PAL20R4-12 PAL20L8-15 PAL20R8-15 PAL20R6-15 PAL20R4-15 . 4-43 ~ AMD Military PAL Device Menu (Continued) Family Part Number Package Technology Inputs I/O Outputs 20R8 PAL20L8B PAL20R8B PAL20R6B PAL20R4B 24JS, W, 28L TIL 14 12 12 12 6 Comb 2 Comb 8 Reg 6 Reg 4 Reg 2 Comb 4 Comb Product Terms/Output 7 8 7,8 7,8 PAL20L8A PAL20R8A PAL20R6A PAL20R4A tPD ns MHz Icc mA 20 28.5 210 30 20 210 fMAX Universal PAL Devices Family Part Number PAL22Vl0-12 22Vl0 Package Technology Inputs 110 Product Terms/Output Features /BLA,/BKA TIL 12 10 Macro 8-16 Varied Term tPD ns 12 50 200 31.2 200 30 22 180 40 16.5 180 15 50 120 PALCE22Vl0H-20 20 33.3 120 PALCE22Vl0H-25 25 26.3 100 PALCE22Vl0H-30 30 25 100 15 41.6 130 AmPAL22Vl0A Distribution /B3A AmPAL22Vl0 PALCE22Vl0H-15 PALCE20V8H-15 EE CMOS EE CMOS /BLA,/B3A 12 8 Macro GAL~ 8 Device PALCE20V8H-20 Equivalent PALCE20V8H-25 16V8 Icc mA fMAX 20 PAL22Vl0-20 20V8 MHz PALCE16V8H-15 /BRA,/B2A EE CMOS 8 8 Macro GAL~ Device Equivalent 8 PALCE16V8H-20 PALCE16V8H-25 20 33.3 130 25 25 130 15 41.6 130 20 33.3 130 25 28.6 130 Asynchronous PAL Device Family 610 Part Number Package Technology Inputs Macro Product Terms/Output Clock Cells Other Features tPD ns fMAX Icc MHz mA PALCE610H-20 /BLA,/B3A EE CMOS 4 16 8 Programmabie J-K Flip-Flops 20 35.8 90 MACH Family (Macro Array CMOS High-density) Technology Inputs Output Macro Burled Macros Product Terms/Output tPD ns MHz 44COFP EE CMOS 6 32 20 40 170 EE CMOS 6 64 - 0-12 84COFP 0-12 20 40 220 44COFP EE CMOS 6 32 32 0-16 20 40 195 Family Part Number Package MACH 1 MACHll0-20 MACH 1 MACH130-20 MACH 2 MACH210-20 4-44 Military PAL Devices fMAX Icc mA AMD~ Standard Military Drawing Program Manufacturing and Screening Locations AMD is an active participant in the Standard Military Drawing (SMD) Program. The idea behind the SMD Program is to standardize MIL-STD-883, Class B microcircuits. The advantage to the user is that SMDs are a cost effective alternative to source control drawings and are offered as off-the-shelf stocking items by IC manufacturers participating in the program. MIL-STD-883 Class B products, and orders to source control drawings, are assembled at our Penang, Malaysia facility. This facility is qualified by AMD Quality Department, as well as by many of our customers, to manufacture non-MIL-STD-883 Class B products. Conformance to MIL-STD-883 requirements is routinely monitored through audits at the Penang facility. Standard Military Drawings should always be considered to improve availability over source control drawings. It is standard practice at AMD to convert our 883, Class B processing to SM Os for all products which we are approved to supply. AM D then dual marks these devices with both the SMD number and the Generic Part Number. DESC approved products can then be procured to either part number as standard product through both OEM and Distributor channels. Assembly location as well as fabrication and seal date codes are included in AMD's part marking. AMD will continue to work closely with DESC, generating new drawings, which will provide a steady flow of advanced technology products to standardize specifications. Example: 2A 92 1T 35 A P T Assermly 10ca. tion code Shift identity (1 st shift of week) Seal week (ww 35) ' - - - - - - - - Seal year (1992) Fab date code (1992, 1st qtr.) Product Introduction Procedures All new military products released by the Programmable Logic Products Division must successfully pass MIL-STD-883 Class B processing prior to new product announcement. This practice allows us to do checkout of bonding diagrams, electrical test tapes and burn circuits in a manufacturing environment. Programmability is checked when applicable. Our Engineering Department reviews electrical data to insure performance and yields to military data sheet limits are acceptable, prior to new product release. This procedure allows AMD to keep manufacturing start-up problems to a minimum on new product orders. Assembly Location Codes: Military PAL Devices Blank M P Sunnyvale Manila Penang 4-45 ~ AMD Standard Military Flow Chart Screening Class B Assembly Offshore assembly Internal visual 2010 condo B 100% Temperature cycling 1010 100% Constant acceleration 2001 test condo D or' E Y1 orientation only 100% Interim electrical parameter (1) Per applicable device (1) specification TA = 25°C only 100% *100 Cycles bulk program and erase (2) 100% *Bake 48 hr. 150°C 100% *Post bake electric TA = 25°C 100% Burn In 1015 Condo Cor D 100% Post electrical parameters Per applicable device specification TA = 25°C only 100% Percent defect allowable DC Parameters PDA = 5% or 1 device whichever is greater Final electrical parameters (hot and cold extremes) Per applicable device specification 100% 1014 Cond, A or B cond C 100% Group A lot 5005 Class B Sample every lot Group B inspection lot Group C 5005 Class B 5005 Class B Group D External visual 5005 Class B 2009 Every lot Every 4 qtrs. of fab date code Every 52 weeks 100% Seal A) Fine B) Gross *EE CMOS Devices (1) Programming and verification are performed at 25°C only (2) May be performed at wafer sort. 4-46 Requirements Military PAL Devices AMD~ Quality Programs Group B - Package Related Tests The Military Product Division quality system conforms to the following Mil-Standards: • QCI is performed in line on each inspection lot. • Purpose: To monitor assembly and device package integrity. Mil-M-38510, Appendix A, "Product Assurance Program" Mil-O-9858, "Quality Program Requirements" Group C - Product/Process Related Tests Mil-1-45208, "Inspection System Requirements" • Group C is performed based on fab date code, at least every four quarters. • Life test data may be used to qualify similar technologies. • Purpose: To monitor the reliability of the process and the parametric performance for each product technology. Quality Assurance The Programmable Logic Products Division ensures outgoing product quality and integrity by performing inspection Lot Group A's and B's per Mil-Std-883 Method 5005, conducting self audits in all areas involved in screening tests per Method 5004 of MIL-STD-883, gating all shipments to our customers, and maintaining a calibration control system in accordance with MilStd-45662. For products requiring programming prior to AC tests, testing is performed utilizing MIL-M-38510 slash sheet sample plans and approved SM 0 sample plans. Product Qualification/Quality Conformance Inspection (QCI) QCI is conducted every 52 weeks using devices which represent the same package construction and lead finish. • Any device type in the same package type may be used regardless of the specific part number. • Purpose: To monitor the reliability and integrity of various package materials and assembly processes. Process Audits AMD has a quality conformance testing program in accordance with MIL-STD-883, Method 5005. Quality Conformance Testing provides necessary feedback . and monitors several areas: Process Audits are performed in accordance with Mil-M-38510, Appendix A, (self audits) by the Quality Assurance Department. Electrostatic Discharge Control Procedures • • Reliability of Product/Processes Vendor Qualification for Raw Materials • Customer Quality Requirements Maintain Product Qualification Engineering Monitor on Products/Processes • • Group D - In-Depth Package Related Tests • Standard procedures for new product release specify that AMD, as a minimum, conduct qualification testing per Company Policy specification on Product Reliability Qualification (00-021). Once qualified, each package type (from each assembly line) and device (by technology group as delineated in MIL-M-3851 0) are incorporated into AMD Quality Conformance Inspection program which utilizes the requirements of MILSTD-883. When military programs do not require that QCI data be run on the specific lot shipped, AMD Quality Conformance program allows customers to obtain generic data on all product families manufactured by AM 0 Generic Qualification Data enables customers to eliminate costly qualification and destruct unit charges, and also improves delivery time by -a factor of eight to ten weeks. The following product data is available: AMD fully employs static control procedures throughout its facilities. All manufacturing areas where product is processed or handled including our Reliability Labs, Engineering Labs, etc., have full static control such as wrist straps, antistatic smocks, grounded stainless steel tables, conductive mats and ion generators wherever neces~ary. All product is moved throughout our facilities and shipped to customers in static shielded containers. In addition, AMD distributors must demonstrate that they meet the same stringent standards regarding ESD handling and control procedures as the factory. Individual distributor locations are audited and approved annually by AM D's Quality Assurance Department. An ESD identifier is marked on all products per MILSTD-883 1.2.1 b (30). All shipping containers are labeled with an ESD Caution Message. ESD procedures are continually reviewed, to ensure that our customers receive only the highest quality product from AMD. Military PAL Devices 4-47 ;t1 AMD MILITARY ORDERING INFORMATION APL Products AMD products for Aerospace and Defense applications are available in several packages and operating ranges. APL (Approved Products List) products are fully compliant with MIL-STD-883C requirements. The order number (Valid Combination) is formed by a combination of: 81035 01 M L t LEAD FINISH A = Hot Solder Dip X = Any Lead Finish PACKAGE TYPE K = 24 Lead 3/8" x 5/8" Cerpack L = 24 Lead 1/4" x 1 1/4" SKINNYDIP R = 20 Lead 1/4" x 1 1/6" DIP S = 20 Lead 114" x 1/2" Cerpack 2 = 20 Terminal .350 x .350 LCC 3 = 28 Terminal .450 x.450 LCC DEVICE CLASS M = Non-JAN level B product, processed to MIL-STD-883 DEVICE TYPE ~---- DRAWING NUMBER FOR PRODUCT FAMILY PART NUMBER INTERPRETATION: When ordering to Military Drawing numbers, the lead finish designator (last letter in part number) is commonly called out as "X." This is a way of stating that the customer will accept the standard manufacturer's lead finish for the package orders. "X" is not a lead finish designator in itself, therefore, when product is shipped, the actual lead finish designator will be marked on the devices. 4-48 Military PAL Devices AMD~ Military SMDs Military Drawing 8103501RA 81035012A 8103501 SA 8103502RA 81035022A 8103502SA 8103503RA 81035032A 8103503SA 8103504RA 81035042A 8103504SA 8103505RA 81035052A 8103505SA 8103506RA 81035062A 8103506SA 8103507RA 81035072A 8103507SA 8103508RA 81035082A 8103508SA 8103509RA 81035092A 8103509SA 8103607RA 81036072A 8103607SA 8103S08RA· 81036082A 8103608SA 8103609RA 81036092A 8103609SA 8103610RA 81 0361 02A 8103610SA 8103612RA 81036122A 8103612SA 8103613RA 81036132A 8103613SA 8103614RA 81036142A 8103614SA 8412901 LA 84129013A 8412901KA 8412902LA 84129023A 8412902KA 8412903LA AMD Part Number Military Drawing PAL 1OH8MJ/883B PAL 1OH8MU883B PAL 1OH8MW/883B PAL 12HSMJ/883B PAL 12HSMU883B PAL 12H6MW/883B PAL 14H4MJ/883B PAL14H4MU883B PAL14H4MW/883B PAL 16H2MJ/883B PAL16H2MU883B PAL 16H2MW/883B PAL 16C 1MJ/883B PAL16C1MU883B PAL 16C1 MW/883B PAL 1OL8MJ/883B PAL10L8MU883B PAL 1OL8MW/883B PAL 12LSMJ/883B PAL12L6MU883B PAL 12L6MW/883B PAL 14L4MJ/883B PAL 14L4M U883B PAL 14L4MW/883B PAL 1SL2MJ/883B PAL 16L2MU883B PAL 16L2MW/883B PAL 16L8AMJ/883B PAL 16L8AMU883B PAL 16L8AMW/883B PAL 1SR8AMJ/883B PAL 16R8AMU8838 PAL16R8AMW/8838 PAL 16R6AMJ/8838 PAL 16RSAMU8838 PAL 16R6AMW/8838 PAL 1SR4AMJ/883B PAL 16R4AMU883B PAL 1SR4AMW/883B PAL 16R8A-2MJ/883B PAL 16R8A-2MU883B PAL 16R8A-2MW/8838 PAL 1SRSA-2MJ/883B PAL 16RSA-2MU8838 PAL 16RSA-2MW/8838 PAL 16R4A-2MJ/8838 PAL 16R4A-2MU883B PAL 16R4A-2MW/8838 PAL20L8AMJS/8838 PAL20L8AMU8838 PAL20L8AMW/8838 PAL20R8AMJS/883B PAL20R8AMU8838 PAL20R8AMW/883B PAL20RSAMJS/8838 84129033A 8412903KA 8412904LA 84129043A 8412904KA 8412905LA 84129053A 8412905KA 8412906LA 84129063A 8412906KA 8412907LA 84129073A 8412907KA 8412908LA 84129083A 8412908KA 8412909LA 84129093A 8412909KA 8412910LA 84129103A 8412910KA 8412911LA 84129113A 8412911 KA 8412912LA 84129123A 8412912KA 850S501RA 85065012A 8506501 SA 8506502RA 85065022A 8506502SA 8506503RA 85065032A 8506503SA 8506504RA 85065042A 8506504SA 5962-8515501 RA 5962-85155012A 5962-8515501 SA 5962-8515502RA 5962-85155022A 59S2-8515502SA 5962-8515503RA 5962-85155032A 5962-8515503SA 5962-8515504RA 5962-85155042A 5962-8515504SA 5962-8515505RA 5962-85155052A Military PAL Devices AMD Part Number PAL20RSAMU883B PAL20RSAMW/883B PAL20R4AMJS/883B PAL20R4AMU883B PAL20R4AMW/883B PAL20L1OAMJS/883B PAL20L1OAMU883B PAL20L1OAMW/883B PAL20X8AMJS/883B PAL20X8AMU883B PAL20X8AMW/883B PAL20X 1OAMJS/883B PAL20X10AMU883B PAL20X10AMW/883B PAL20X4AMJS/883B PAL20X4AMU883B PAL20X4AMW/883B PAL20L8A-2MJS/883B PAL20L8A-2MU883B PAL20L8A-2MW/883B PAL20R8A-2MJS/883B PAL20R8A-2MU883B PAL20R8A-2MW/883B PAL20R6A-2MJS/883B PAL20R6A-2MU883B PAL20R6A-2MW/883B PAL20R4A-2MJS/883B PAL20R4A-2MU883B PAL20R4A-2MW/883B PAL 16L8A-4MJ/883B PAL 16L8A-4MU883B PAL 16L8A-4MW/8838 PAL 16R8A-4MJ/8838 PAL 16R8A-4MU8838 PAL 16R8A-4MW/8838 PAL 16R6A-4MJ/883B PAL 16R6A-4MU883B PAL 16R6A-4MW/8838 PAL 16R4A-4MJ/8838 PAL 16R4A-4MU8838 PAL 16R4A-4MW/8838 PAL 16L88MJ/883B PAL 16L88MU8838 PAL 16L88MW/8838 PAL 16R88MJ/8838 PAL16R8BMU8838 PAL 16R8BMW/883B PAL 16R68MJ/8838 PAL 16R6BMU8838 PAL 16R68MW/8838 PAL 16R48MJ/8838 PAL 16R48MU8838 PAL16R48MW/8838 PAL 16L88-2MJ/8838 PAL 16L88-2MU8838 4·49 ~ AMD Military SMDs (continued) Military Drawing AMD Part Number 5962-8515505SA 5962-8515506RA 5962-85155062A 5962-8515506SA 5962-8515507RA 5962-85155072A 5962-8515507SA 5962-8515508RA 5962-85155082A 5962-851550SSA 5962-8515509RA 5962-85155092A 5962-8515509SA 5962-8515510RA 5962-85155102A 5962-8515510SA 5962-8515511 RA 5962-85155112A 5962-8515511 SA 5962-8515512RA 5962-85155122A 5962-8515512SA 5962-85155132A 5962-8515513RA 5962-85155142A 5962-8515514RA 5962-85155152A 5962-8515515RA 5962-85155162A 5962-8515516RA 5962-85155172A 5962-8515517RA 5962-85155182A 5962-8515518RA 5962-85155192A 5962-8515519RA 5962-85155202A 5962-8515520RA 5962-8605301 lA 5962-86053013A 5962-8605301 KA 5962-8605302lA 5962-86053023A 5962-8605302KA 5962-8605304lA 5962-86053043A 5962-8605304KA 5962-8S063053A 5962-8605305KA 5962-8605305lA 5962-8680301 lA 5962-86803013A 5962-8680301 KA 5962-8680401 lA 5962-86804013A PAL 16l88-2MW/8838 PAL 16R88-2MJ/8838 PAL 16R88-2MU8838 PAL 16R88-2MW/8838 PAL 16R68-2MJ/8838 PAL 16R68-2MU8838 PAL 16R68-2MW/883B PAL 16R48-2MJ/8838 PAL 16R48-2MU8838 PAL 16R48-2MW/8838 PAL 16l8DMJ/8838 PAL 16l8DMU8838 PAL 16lSDMW/8838 PAL 16R8DMJ/8838 PAL 16R8DMU8838 PAL 16R8DMW/S83B PAL 16R6DMJ/8838 PAL 16R6DMU8838 PAL 16R6DMW/883B PAL 16R4DMJ/8838 PAl16R4DMU8S3B PAL 16R4DMW/8838 PAL 16l8-12182A PAL 16lS-1218RA PAL 16RS-12/82A PAL 16RS-12/8RA PAL 16R6-12/82A PAL 16R6-12/8RA PAL 16R4-12/82A PAl16R4-12/8RA PAL 16l8-1 0/82A PAL 16l8-1 0/8RA PAL 16R8-1 0/82A PAl16R8-10/8RA PAL16R6-10/82A PAL16R6-10/8RA PAL 16R4-1 0/82A PAL 16R4-1 0/8RA AmPAl22V10Al8lA AmPAl22V10Al83A AmPAl22V10Al8KA AmPAl22V10/8lA AmPAl22V10/83A AmPAl22V10/8KA PAl22V10-20/8lA PAl22V10-20/83A PAl22V10-20/8KA PAl22V10-12183A PAl22V10-1218KA PAl22V10-1218lA PAl20RA 1OMJS/S838 PAl20RA 1oMU8838 PAl20RA 1OMW/8838 PAL 18l4MJS/8838 PAL 18l4MU8838 4-50 Military Drawing 5962-8680401 KA 5962-8680402lA 5962-86804023A 5962-8680402KA 5962-8680403lA 5962-86804033A 5962-S680403KA 5962-8680404lA 5962-86804043A 5962-S680404KA 5962-8680405lA 5962-86804053A 5962-S680405KA 5962-8680406lA 5962-86804063A 5962-8680406KA 5962-8753001 lA 5962-87530013A 5962-8753001 KA 5962-8753002LA 5962-87530023A 5962-8753002KA 5962-S753003lA 5962-87530033A 5962-S753003KA 5962~S753004lA 5962-S7530043A 5962-8753004KA 5962-S7671 01 lA 5962-87671013A 5962-8767101 KA 5962-8767102lA 5962-87671023A 5962-8767102KA 5962-8767103lA 5962-87671033A 5962-8767103KA 5962-8767104lA 5962-87671043A 5962-8767104KA 5962-8767107LA 5962-87671073A 5962-8767108lA 5962-87671083A 5962-8767109lA 5962-87671093A 5962-8767110lA 5962-87671103A 5962-8767111 lA 5962-87671113A 5962-8767112LA 5962-87671123A 5962-8767113lA 5962-87671133A 5962-S767114lA Military PAL Devices AMD Part Number PAL18l4MW/883 PAL 12l1OMJS/8838 PAL12L10MU8838 PAL12l10MW/8838 PAL14l8MJS/8838 PAL14l8MU8838 PAL14l8MW/8838 PAL 16l6MJS/8838 PAL16l6MU8838 PAL 16l6MW/S838 PAL20L2MJS/8838 PAL20L2MU8838 PAl20L2MW/883B PAl20C1 MJS/8838 PAL20C1 MU8838 PAL20C1 MW/8838 PAl20S 1OMJS/8838 PAl20S10MU8838 PAl20S10MW/8838 PAl20RS10MJS/8838 PAl20RS10MU8838 PAL20RS 1OMW/8838 PAl20RSSMJS/8S38 PAl20RS8MU8838 PAL20RSSMW/8838 PAl20RS4MJS/8838 PAl20RS4MU8838 PAl20RS4MW/8838 PAl20l88MJS/8838 PAl20l88MU8838 PAL20l88MW/8838 PAL20R88MJS/8838 PAL20R88MU8838 PAl20R88MW/8838 PAl20R68MJS/8838 PAL20R68MU8838 PAL20R68W/8838 PAL20R48MJS/8838 PAl20R48MU8838 PAL20R48MW/8838 PAL20l8-15/8lA PAL20L8-15/83A PAL20R8-15/8lA PAl20R8-15/83A PAl20R6-15/8lA PAL20R6-15/83A PAL20R4-15/8LA PAL20R4-15/83A PAl20l8-1218lA PAl20lS-12183A PAl20R8-12/8LA PAl20R8-12/83A PAL20R6-12/8lA PAl20R6-12/83A PAl20R4-12/8lA AMD~ Military SMDs (continued) Military Drawing Military Drawing AMD Part Number 5962-87671143A 5962-8767115LA 5962-87671153A 5962-8767116LA 5962-87671163A 5962-8767117LA 5962-87671173A 5962-8767118LA 5962-87671183A 5962-8851501 RA 5962-88515012A 5962-8851501 SA 5962-8851502RA 5962-88515022A 5962-8851502SA 5962-8851503RA 5962-88515032A 5962-8851503SA 5962-8851504RA 5962-88515042A 5962-8851504SA 5962-89839012A 5962-8983901 RA 5962-89839022A PAL20R4-12/83A PAL20L8-10/8LA PAL20L8-10/83A PAL20R8-1 0/8LA PAL20R8-10/83A PAL20R6-10/BLA PAL20R6-1 0/83A PAL20R4-10/BLA PAL20R4-101B3A PAL 16L8B-4MJ/883B PAL 16L8B-4MU883B PAL 16L8B-4MW/883B . PAL 16R8-4MJ/883B PAL 16R8B-4MU883B PAL 16R8B-4MW/883B PAL 16R6B-4MJ/883B PAL 16R6B-4MU883B PAL 16R68-4MW/883B PAL 16R48-4MJ/8838 PAL 16R4B-4MU8838 PAL 16R4B-4MW/883B PALCE16V8H-25E41B2A PALCE16V8H-25E4IBRA PALCE 16V8H-20E4/B2A 5962-8983902RA 5962-89839032A 5962-8983903RA 5962-89840013A 5962-8984001 LA 5962-89840023A 5962-8984002LA 5962-89840033A 5962-8984003LA 5962-89841013A 5962-8984101 KA 5962-8984101 LA 5962-89841023A 5962-8984102KA 5962-8984102LA 5962-89841043A 5962-8984104KA 5962-8984104LA 5962-89841053A 5962-8984105KA 5962-8984105LA 5962-9169501 M3A 5962-9169501 MLA Military PAL Devices AMD Part Number PALCE16V8H-20E4IBRA PALCE 16V8H-15E41B2A PALCE 16V8H-15E41BRA PALCE20V8H-25E41B3A PALCE20V8H-25E4IBLA PALCE20V8H-20E41B3A PALCE20V8H-20E4IBLA PALCE20V8H-15E41B3A PALCE20V8H-15E4IBLA PALCE22V10H-301B3A PALCE22V10H-30IBKA PALCE22V10H-30IBLA PALCE22V10H-20E4IB3A PALCE22V10H-20E4IBKA PALCE22V10H-20E4IBLA PALCE22V10H-251B3A PALCE22V10H-25IBKA PALCE22V10H-25IBLA PALCE22V10H-15E4/B3A PAL.CE22V10H-15E4/BKA PALCE22V10H-15E4/BLA PALCE610H-201B3A PALCE610H-20/BLA 4·51 Military ProPAL™ Devices The Advantages An important service that AMD's Programmable Logic Division offers is ProPAL devices, the programming of a customer's PAL devices during the manufacturing cycle, saving aerospace customers significant time and money. We offer full 883, SMD, programmed products. Not only were we the first to offer this service, we by far have the most expertise in managing your programmed business. The pre-programming of PAL devices for the customer offers a number of advantages, most of which increase the parts' reliability. Unlike a blank generic device, a ProPAL device undergoes a more thorough testing process. Listed below are just a few of the many advantages. Although the price of the ProPAL device is higher than that of its unprogrammed counterpart, the added value provided to the user more than offsets the cost. In this era of cost-cutting and the need to get the most from each dollar spent, Pro PAL devices make it possible for the customer to save money over the life of the project. Below are just a few examples of where cost savings can be recognized. Cost Reduction Areas: • The more thorough testing significantly increases the long term reliability of the device. System failures and resulting rework costs, owing to a defective PAL device, can easily be 50 times the purchase price of the component. • The customer does not have to purchase and maintain costly programming and test facilities. He can concentrate on what he does best. .. design systems to aid in defense of our nation and allow AM D to do what we do best. .. produce and test Programmable LogiC Devices. ' • The parts are programmed prior to burn-in; weak parts are likely to fail to accept a program and be rejected on the spot. As volumes go up, the customer's production line won't be constrained by his programming and labeling capacities or throughput time. • Programmed devices then must pass 25°C DC, functional and AC electrical tests prior to burn-in and DC, functional and AC electrical tests after burn-in, at temperature extremes, per M5004. Inventories can be reduced. The customer no longer needs to store extra product for programming fall-out or human error. • Improve lead time of manufacturing. The customer will no longer need to plan on internal time to program or label devices. Enhanced Reliability • • • In the design environment, the pattern undergoes fault simulation with> 90% coverage, providing a high testability. (Some patterns do not allow 90% fault coverage. In those cases the customer will be notified.) 4-52 Military ProPAL™ Devices AMD~ ProPAL Device Flow for Products with Existing Mil Shell Programs Marketing/Sales Receives Masters (Programmed devices, Floppy disk or Mag tape) from customer and forwards them to Spec Writing. Military Programming Process Summary (per current Revisions of MiI-Std-BB3 and Mil-M-3B51 0) Assembly 100% program to customer bit pattern Initial Electrical Test (2S°C) 100% DC/FunctionaVAC at 2SoC, MS004 I Spec Writing Prepares and submits documentation packet to engineering. Packet contains PICS, master and blanks. Initiates Process Spec. I Electrical Test (2S°C) I Engineering Generates prototypes, fuse file and post functional tests on Auto VecHA • Method 101S, Condition CorD Burn-In PDA - 100% DC/FunctionaVAC at 2SoC per MS004 - Group A DC/Functional/ AC at 2SoC per MSOOS - 5% (DC only) - Resistance to solvents, solderability, bond pull - MSOOS Solder Dip Mark I Lot B Test Spec Writing Sends prototypes and Approval form to customer via Sales. Electrical Test (+ 12SoC &-SS°C) I 100% DC/FunctionaV AC, MS004 Group A DC/Functional/ AC, MSOOS Customer Verifies prototypes and returns Approval form to AMD. Hermiticity I External Visual Data Review/Pack/Ship - 100% Fine/Gross Leak per Method 1014 Method 2009 Spec Writing Checks Approval form for customer acceptance of prototypes. ( I NO Accept I YES Engineering Reviews and completes documentation packet. Release bit pattern and test programs to Production. AMD is the proven technology leader in PAL devices and has numerous years of experience in programming customer patterns. Currently several major customers are using ProPAL devices and several major programs are being converted to ProPAL devices. Factory programming of your PAL device products is another service of the Military Programmable Logic's long-term partnership with and commitment to the worldwide Military and Aerospace market. For more information on Military ProPAL devices contact your local AMD sales office. I Spec Wrltlng/QA Reviews and releases Process Spec to Production. Military ProPALTM Devices 4·53 Electrical Characteristic Definitions Parameter Symbol Parameter Name Parameter Definition tAPR Asynchronous Preset Recovery Time The minimum time after the asynchronous preset becomes inactive to the next input clock triggering edge. tAPW Asynchronous Preset Width The minimum pulse width required for the asynchronous preset signal. tH Hold Time The minimum time a valid data level is held after clock triggering edge. tHP Hold Time for Preload The minimum delay time for data to remain stable after the preload signal becomes inactive. This only applies to TTL-level preload. tSRR Synchronous Reset Recovery Time The minimum time between the synchronous reset going inactive and the next input clock triggering edge. ts Setup Time, Input or Feedback to Clock The minimum time a valid data level of input or feedback is stable before the next clock triggering edge. tsp Data Setup Time for Preload The minimum time for input data to be stable prior to the preload signal becoming inactive. This only applies to TTL-level preload. tWH Clock Width High The minimum width of the clock high from rising edge to the next falling edge. In some cases, simultaneous minimum clock widths (both high and low) will exceed the minimum period of the device. tWL Clock Width Low The minimum width of the clock low from falling edge to the next rising edge. In some cases, simultaneous minimum clock widths (both high and low) will exceed the minimum period of the device. twp Preload Pulse Width The minimum pulse width required to preload the registers. This only applies to TTL-level preload. tAP Asynchronous Preset to Output The maximum time required to preset the register output after the preset signal is asserted. tAR Asynchronous Reset to Output The maximum time required to reset the register output after the reset signal is asserted. Timing 4-54 AMD~ Parameter Symbol Parameter Definition Parameter Name Timing tco Clock to Register Output The maximum time it takes to obtain a valid data level on the output pin after an input clock triggering edge is applied. tCR Input or Feedback to Registered Output from Combinatorial Configuration; Output Mux Select 1 to 0 The minimum time from input or feedback to registered output as output mux selection changes from combinatorial to registered output 1 to 0). tEA Output Enable Time, Clock to Output The minimum delay between when an input is asserted and the output switches from a highimpedance state to HIGH or LOW logic state. tER Output Disable Time, Input to Output The minimum delay between when an input is asserted and the output switches from a HIGH or LOW logic state to a high-impedance state. tF Fall Time The minimum time for a signal to fall from 80% to 20% of its stabilized high value. tPD Propagation Delay, Input or Feedback to Combinatorial Output The time fora signal to propagate from input or feedback to output. tPR Power-up Reset Time The minimum time for a registered output signal to be reset after the power is applied. tpxz Output Disable Time, OE to Output The minimum delay between when a dedicated enable signal is asserted and the output switches from a HIGH or LOW logic state to be a highimpedance state. tpzx Output Enable Time, OE to Output The minimum delay between when a dedicated enable signal is asserted and the output switches from a high-impedance state to a HIGH or LOW logic state. tR Rise Time The minimum time for a signal to rise from 20% to 80% of its stabilized high value. tAC Input or Feedback to Combinatorial Output from Registered Configuration; Output Mux Select 0 to 1 The minimum time from input or feedback to combinatorial output mux selection changes from registered to combinatorial output (0 to 1). Supply Voltage, Positive Potential The voltage required across supply and ground terminals of a TTL or CMOS integrated circuit. VI Input Clamp Voltage The maximum input clamp voltage limit on every input pin. VIH High-Level Input Voltage The minimum high-level input voltage that is guaranteed to represent a high logic level. VIL Low-Level Input Voltage The maximum low-level input voltage that is guaranteed to represent a low logic level. VOH High-Level Output Voltage The minimum high logic level guaranteed for all outputs. VOL Low-Level Output Voltage The minimum low logic level guaranteed for all outputs. Voltage Vec Electrical Characteristic Definitions 4-55 ~AMD Parameter Symbol Parameter Name Parameter Definition Supply Current, Corresponding to Vcc The maximum current into the Vcc terminal of a TTL or CMOS integrated circuit. II Input Current with Maximum Input Voltage The maximum current into an input pin when the input voltage is applied to the input pin. IIH High-Level Input Current The maximum current into an input pin when a logichigh level is applied to the input pin. III Low-Level Input Current The maximum current into an input pin when a logiclow level is applied to the input pin. IOH High-Level Output Current The maximum current into an output pin to guarantee an output logic-high level. IOl Low-Level Output Current The maximum current into an output pin to guarantee an output logic-low level. Isc Output Short-Circuit Current The current into an output when that output is short-circuited to ground (0.5 V). lozH High-Level Leakage Current The maximum current into a high-impedance state output p!n when a high logic level is applied to the output pin. lozl Low-Level Leakage Current The maximum current into a high-impedance state output pin when a low logic level is applied to the output pin. , Input Capacitance The input pin capacitance at a specified voltage and frequency. Output CapaCitance The output or 1/0 pin capacitance at a specified voltage and frequency. TA Operating Free Air Temperature The ambient homogeneous temperature of the environment during operation. Tc Operating Case Temperature The maximum chassis temperature during operation. fMAX Maximum External Frequency The fMAX, External is the maximum clocking frequency with external feedback. It is the reciprocal of the clock period (ts +tco). fMAX Maximum Internal Frequency The fMAX, Internal is the maximum clocking frequency with internal feedback. An internal counter is used to determine'1cNT." fMAX Maximum Frequency without Feedback The fMAX, No Feedback is the maximum clocking frequency with no feedback. It is the reciprocal of the sum of the data setup time (ts) and the data hold time (th). Current Icc Miscellaneous CIN COUT 4·56 Electrical Characteristic Definitions fMAX Parameters The parameter fMAX is the maximum clock rate at which the device is guaranteed to operate. Because the flexibility inherent in programmable logic devices offers a choice of clocked flip-flop designs, fMAX is specified for three types of synchronous designs. The first type of design is a state machine with feedback signals sent off-chip. This external feedback could go back to the device inputs, or to a second device in a multi-chip state machine. The slowest path defining the period is the sum of the clock-to-output time and the input setup time for the external signals (ts + tco). The reciprocal, fMAX, is the maximum frequency with external feedback or in conjunction with an equivalent speed device. This fMAX is designated '1MAX external." The second type of design is a single-chip state machine with internal feedback only. In this case, flip-flop inputs are defined by the device inputs and flip-flop outputs. Under these conditions, the period is limited by the internal delay from the flip-flop outputs through the inter- nal feedback and logic to the flip-flop inputs. This fMAX is designated '1MAX internal". A simple internal counter is a good example of this type of design, therefore, this parameter is sometimes called '1cNT." The third type of design is a simple data path application. In this case, input data is presented to the flip-flop and clocked through; no feedback is employed. Under these conditions, the period is limited by the sum of the data setup time and the data hold time (ts + tH). However, a lower limit for the period of each fMAX type is the minimum clock period (tWH + tWl.). Usually, this minimum clock period determines the period for the third fMAX, designated '1MAX no feedback." fMAX external and fMAX no feedback are calculated parameters. fMAX external is calculated from ts and teo, and fMAX no feedback is calculated from tWL and tWH. fMAX internal is measured. ClK r------------- I I -----~ I I I--l-....I lOGIC REGISTER I I I I I ~-------------------~ I. . 1 - - - - - ts (SECOND CHIP) ••-- teo ---t"~lioIa.r- ts -.t ---"'~"'I fMAX External; 1/(ts + teo) ClK ClK r------------- I I lOGIC -----~ REGISTER I I I I I I I ~-------------------~ fMAX Internal (feNT) r------------- I I I I lOGIC -----~ I I I REGISTER J.----il~...... I I I I I I ~-------------------~ I~.~---ts----~~~I fMAX No Feedback; 1/(ts + tH) or 1/(tWH + tWL) fMAX Parameters 4-57 PHYSICAL DIMENSIONS* CD 020 20·Pin Ceramic DIP (measured in inches) I" .045 .065 ..'1 ..975 935. .100 Bse .005 MIN .300 TOP VIEW :~60~~. sc 1.015 ~.060 .125 .200 ~ .008 .Q18 --+l~ 94 0 1050 L·400 .014 .026 MAX -.l g~~~1o 613193 as END VIEW SIDE VIEW CD3024 24·Pin 300 mil Ceramic SKINNYDIP (measured in inches) ,.. 1.235 ... , 1.280 ~~:~~ ::: :~~ :]J!~~ -'I~ ..045 .065 .100 BSe TOP VIEW .005 MIN .300 BSe :~6g~~.,f· .008 .018 1.015 060 .125 .200 .014 .026 I SIDE VIEW "For reference only. BSC is an ANSI standard for Basic Space Centering. 4-58 Physical Dimensions I .400 I t---MAX~ 06850F 613193 CD3024 COO9ae END VIEW AMD~ CFL024 24·Pin Ceramic Flatpack (measured in inches) .050 sse + I f i i 1 I .015 .022 :~~: .58 o .63 o I 005 I I ·MIN ~ T I f j.-- MAX --..J .440 +L I I~ .060 .090 ===::I .250 .320 j I[ . - .250 .320 Physical Dimensions .026 .045 096750 CJ16 614/93 ae 4-59 ~ AMD CL020 20-Pin Ceramic Leadless Chip Carrier (measured in inches) .~::J .100-11 INDEX CORNER .020 X 45° REF. (OPTIONAL) T 1 .320 MAX .320 MAX J PLANE2~1 PLANE 1 TOP VIEW 073180 CJ46 CL 020 614193 ae BOTTOM VIEW 4-60 Physical Dimensions AMD~ CL028 28-Pin Ceramic Leadless Chip Carrier (measured in inches) '054~ .040 X 45 0 REF. INDEX CORNER .020 X 45 0 REF. (OPTIONAL) .088 (OPTIONAL) :~~6-1 .442 --. tl MAX ~:~~ .458 PLANE2~ PLANE 1 TOP VIEW .022L 028 . r 065951 CJ45 CL 028 6/7/93 ae .045L BonOMVIEW Physical Dimensions 4·61 ~ AMD PD 020 20·Pin Plastic DIP (measured in inches) I" 1.010 1.040 --I ~~~::~~]j - . I I..005 ' MIN. -.I I+.045 .065 .090 .no TOP VIEW :~60~mNMW·l.015 ~.060 .120 .160 I I Q:'--.. --.. 7° .014 .022 SIDE VIEW ~.330---.J .430 07522C CM73 PD020 12118/91 e de END VIEW PD3024 24·Pin 300 mil Plastic SKINNYDIP (measured in inches) I" --I 1.150 1.270 ~~~~:~:~::~ll~~~ .030 MIN .045 .065 .090 .110 .005 MIN TOP VIEW :~6~0~1.015 .125 .160 - . .060 I --.J~ .014 .022 SIDE VIEW 4·62 .008 .015 o· T ~ .330 ____ .430 END VIEW Physical Dimensions I 07089F·OO2 CM74 PD3024 617193 ae AMD l1 PD3028 28·Pin 300 mil Plastic SKINNYDIP (measured in inches) I_ .030 .065 -I 1.375 1.400 .045 .065 .090 .005 MIN .1'10 TOP VIEW :~6~0~~1.015 .120 .160 . --. .060 I -.j~ O' T .014 .022 SIDE VIEW END VIEW PL020 20·Pin Plastic Leaded Chip Carrier (measured in inches) .042 .020 MIN .050 ~Ioo.IooI.I.RE.I.I.F~ 1. .042 .056 n 1 r " 8.......... T· 350 .356 . !o4-_ _ . 385 .395 026 032 3 06970D Bve PL 020 112192e de 1 TOP VIEW SIDE VIEW Physical Dimensions 4-63 ~ AMD PL028 28·Pin Plastic Leaded Chip Carrier (measured in inches) .042 .020 MIN .050 m '04_~ ____~__R_E_F__~ :~~~ 1. T· 026 032 .485 .450 .495 .456 14-"450~ 06751F BV8PL028 12131191 e de .456 ...._ _ _ .485 .495 TOP VIEW SIDE VIEW SO 020 20·Pin Plastic Gull.Wing Small Outline Package (measured in inches) .045 .055 TOP VIEW t. Jtr-ll-----I~b ~.0091 I I ---r .0125 "1 JL8° -.I k- T .016 - ::: .035 END VIEW 4·64 Physical Dimensions 15683A BJ10 SO 020 214/92 c de AMD~ 50024 24·Pin Plastic Gull·Wing Small Outline Package (measured in inches) .045 .055 ===~ ~ ---ll-.0138 .0192 h ---r 0926 .1043 .0040 .0118 END VIEW SIDE VIEW Physical Dimensions 093109 10117/91 9J10 e de 4-65 ~ 4-66 AMD IIilt+iAi3,1 5 DESIGN AND TESTABILITY Basic Design with PLDs .................................................... 5-3 PLD Design Basics ................................................ 5-8 PLD Design Methodology .......................................... 5-16 Combinatorial Logic Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .. 5-27 Registered Logic Design ........................................... 5-40 State Machine Design ............................................. 5-60 Testability .............................................................. 5-73 Ground Bounce . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-92 Metastability ............................................................ 5-95 Latchup .......................................... ~ . . . . . . . . . . . . . . . . . . . .. 5-97 Converting Bipolar to CMOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-100 High-Speed-Board Design Techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5-107 Minimizing Power Consumption with Zero-Power PLDs .......................... 5-135 Designing with the PALCE16V8HD .......................................... 5-138 Design and Testability 5-1 ~ 5-2 AMD Design and Testability .~ Basic Design with PLDs Advanced Micro Devices INTRODUCTION The Programmable Array Logic device, commonly known as the PAL device, was invented at Monolithic Memories in 1978. The concept for this revolutionary type of device sprang forth as a simple solution to the short comings of discrete TTL logic. WHAT OTHER IMPLEMENTATIONS ARE POSSIBLE? There are essentially four alternatives to programmable logic: • Discrete Logic The successfully proven PROM technology which allowed the end user to ''write on silicon" provided the technological basis which made this kind of device not only possible, but very popular as well. • Gate Arrays • Standard Cell Circuits • Full Custom Circuits The availability of design software made it much easier to design with programmable logic. As designers were freed from the drudgery of low-level implementation issues, new complex designs were easier to implement, and could be completed more quickly. Discrete Logic This chapter outlines some basic information essential to those who are unfamiliar with Programmable Logic devices (PLDs). The information may also be useful to those who are current users of programmable logic. The specific issues which need to be addressed are: • What is a PLD? • What other implementations are possible? • What advantages do PLDs have over other implementations? Discrete logic, or conventional TIL logic, has the advantage of familiarity; hence its popularity. It is also quite inexpensive when only unit cost is considered. The drawback is that the implementation of even a simple portion of a system may require many units of discrete logic. There are "hidden" costs associated with each unit that goes into a system, which can render the overall system more expensive. Designing with discrete chips can also be very tedious. Each design decision directly affects the layout of the board. Changes are difficult to make. The deSign is also more difficult to document, making it harder to debug and maintain later. These items all contribute to a long design cycle when discrete chips are used extensively. WHAT IS A PLD? Gate Arrays In general, a programmable logic device is a circuit which can be configured by the user to perform a logic function. Most "standard" PLDs consist of an AND array followed by an OR array, either (or both) of which is programmable. Inputs are fed into the AND array, which performs the desired AND functions and generates product terms. The product terms are then fed into the OR array. In the OR array, the outputs of the various product terms are combined to produce the desired outputs. Gate arrays have been increasing in popularity. The attractiveness of this solution lies in the device's flexibility. By packing the functions into the device, a great majority of the available silicon is actually used. Since such a device is customized for an application, it would seem to be the optimum device for that application. PAL Devices The PAL device has a programmable AND array followed by a fixed OR array (Figure 1). The fact that the AND array is programmable makes it possible for the devices to have many inputs. The fact that the OR array is fixed makes the devices small (which means less expensive) and fast. However, one also pays substantial development costs, especially in the case of a design which needs changes after silicon has already been processed. Even though the unit costs are generally quite low for gate arrays, the volumes required to make their use worthwhile excludes them as a solution for many designers. This fact, added to the long design cycle and high risk involved, make this solution practical for only a limited number of designers. 5-3 , ~AMD 15 14 13 12 11 10 1~ 1~ 1~ 1~ 1~ 1~ "-- 'V' Fixed OR Array ~ ~ Programmable AND Array 03 02 01 00 -$- + Indicates Programmable Connection 101730-1 Indicates Fixed Connection Figure 1. PAL Device Array Structure Standard Cell Circuits Standard cell circuits are quite similar to gate arrays, their main advantage being that they consist of a collection of different parts of circuits which have already been debugged. These circuits are then assembled and collected to perform the desired functions. This can ideally lead to reduced turn around from conception to implementation, and a much more efficient circuit. The drawback is that even though the individual components of the circuit have been laid out, a complete layout must still be performed to arrange the cells. Instead of just customizing the metal interconnections, as is done in a gate array, the circuit must be developed from the bottom up. Development costs can be even higherthan for gate arrays, and despite the standard cell 5·4 concept, turn around time often tends to be longer than planned. Again, the volume must be sufficiently high to warrant the development costs. Full Custom Circuits Full custom designs require that a specific chip be designed from scratch to perform the needed functions. The intent is to provide a solution which gives the designer exactly what is needed for the application in question; no more and no less. Ideally, not a square micron of silicon is wasted. This normally results in the smallest piece of silicon possible to fit the needs of the design, which in turn reduces the system cost. Understandably, though, development costs and risks for such a design are extremely high, and volumes must be commensurately high in order for such a solution to be of value. Basic Design with PLDs AMD~ WHAT ADVANTAGES DO PLDS HAVE OVER OTHER IMPLEMENTATIONS? As user-programmable semicustom circuits, PLDs provide a valuable compromise which combines many of the benefits of discrete logic with many of the benefits of other semicustom circuits. The overall advantages can be found in several areas: • Ease of design • Performance • Reliability • Cost savings Ease of Design The support tools available for use in designing with PLDs greatly simplify the design process by making the lower-level implementation details transparent. In a matter of one or two hours, a first time PLD user can leam to design with a PAL device, program it, and implement the design in a system. The design support tools consist of deSign software and a programmer. The design software is used in generating the design; the programmer is used to configure the device. The software provides the link between the higher-level design and the low-level programming details. All of the available deSign software packages perform essentially the same tasks. The deSign is specified with relatively high-level constructs; the software takes the design and converts it into a form which the programmer uses to configure the PLD. Most software packages provide logic simulation, which allows one to debug the design before actually programming a device. The high-level design file also serves as documentation of the design. This documentation can be even easier to understand than traditional schematics. Many PLD users do not find it necessary to purchase a programmer; it is often quite cost effective and convenient to have either the manufacturer or an outside distributor do the programming for them. For deSign and prototyping, though, it is very helpful to have a programmer; this allows one to implement designs immediately. The convenience of programmable logic lies in the ability to customize a standard, off-the-shelf product. PLDs can be found in stock to suit a wide range of speed and power requirements. The variety of architectures available also allows a choice of the proper functionality for the application at hand. Thus, a design can be implemented using a standard device, with the end result essentially being a custom device. If a design change is needed, it is a simple matterto edit the original deSign and then program a new device, or, in the case of reprogrammable CMOS devices, erase and reprogram the old device. Board layout is vastly simplified with the use of programmable logic. PLDs offer great flexibility in the location of inputs and outputs on the device. Since larger functions are implemented inside the PLD, board layout can begin once the inputs and outputs are known. The details of what will actually be inside the PLD can be worked out independently of the layout. In any cases, any needed design changes can be taken care of entirely within the PLD, and will not affect the PC board. Performance Speed is one of the main reasons that deSigners use PAL devices. The PAL devices can provide equal or better performance than the fastest discrete logic available. Today's fastest PAL devices are being developed on the newest technologies to gain every extra nanosecond of performance. Performance cannot come strictly at the expense of power consumption. Since PLDs can be used to replace several discrete circuits, the power consumption of a PLD may well be less than that of the combined discrete devices. As more PLDs are developed in CMOS technology, the option for even lower power becomes available, including zero standby power devices for systems which can tolerate only minute standby power consumption. Reliability Reliability is an area of increasing concern. As systems get larger and more complex, the increase in the amount of circuitry tends to reduce the reliability of the system; there are "more things to go wrong." Thus, a solution which inherently reduces the number of chips in the system will contribute to higher reliability. A programmable logic approach can provide a more reliable solution due to the smaller number of devices required. With the reduction in units and board space, PC boards can be laid out less densely, which greatly improves the reliability of the board itself. This also reduces crosstalk and other potential sources of noise, making the operation of the system cleaner and more reliable. Basic Design with PLDs 5-5 ~ AMD Cost For any design approach to be practical, it must be cost effective. Cost is almost always a factor in considering a new design or a design change. But, the calculation of total system cost can be misleading if not all aspects are considered. Many of the costs can be elusive or difficult to measure. For example, it is difficult to quantify the cost of market share lost due to late product introduction. The greatest savings over discrete design are derived from the fact that a single PLD can replace several discrete chips. Board space requirements can drop 25% or more when PLDs are used. The relationship between the various alternatives is summarized in Figure 2. Cost Weeks Months Years Development Time 10173D-2 Figure 2. Development Cost vs. Time for Alternative Logic Implementations 5-6 Basic Design with PLDs AMD~ Another economic benefit of the use of PLDs is that when one PAL device is used in several different designs, as is often the case, the user has not committed that device to anyone of the particular designs until the device has been programmed. This means that inventory can be stocked for several different designs in the form of one device. As requirements change, the parts can be programmed to fit the need. And in the case of reprogrammable CMOS devices, one is not committed even after programming. One final subtle cost issue is derived from the ease with which a competitor can copy a deSign. PLDs have a unique feature called a security bit, whose purpose is to protect a design from being copied. By using secured PLDs extensively in a system, one can safely avoid having one's system easily deciphered. The added deSign security provided by this feature can buy extra market time, forcing competitors to do their own original design work rather that copying the designs of others. By assuming some of the attributes of gate arrays, programmable logic provides the cost savings of any other semicustom device, without the extra engineering costs, risks, and design delays. Reliability is also enhanced as quality increases and board complexity decreases. The design tasks are greatly simplified due to the design tools which are now available. Design software and device programmers allow top-down high-level designs with a minimum of time spent on actual implementation issues. Simulation allows some design debug before a device is programmed. For all of these reasons, programmable logic has become, and will continue to be, the deSign methodology of choice among digital systems designers. Summary Programmable logic provides the means of creating semi-custom designs with readily available standard components. There is a wide variety of PLDs; PAL devices are most widely used, and perform well for basic logic and some sequencing functions. Basic Design with PLDs 5-7 PLD Design Basics INTRODUCTION This section is intended as a beginner's introduction to PLD design, although experienced users may find it a good review. We will take a step-by-step approach through two very simple designs to demonstrate the basic PLD design implementation process. Through this . effort, you will be introduced to the concept of device programming. A C ---t------1 D F --t-----\ G ---+-----1 M N As can be seen from the figure, there will be six separate functions involving a total of twelve inputs. It is important to bear in mind that programmable logic provides a convenient means of implementing designs. With a real design, some work would be required before this pointto conceptualize the deSign, but due to the simplicity of these circuits, we are already in a position to start the implementation. 5-8 E L ---t----~ ---t-----I D------+-- 0 P R 10173D-3 Figure 1. The Basic Logic Gates Building the Equations We will start by generating Boolean equations. The first function to be generated is an inverter. This is specified according to Figure 1 as: B The first example we will try is a very simple combinatorial circuit consisting of all of the basic logic gates, as shown in Figure 1. This will be helpful forthose designs where you are integrating random logic into a PAL device to save space and money. B l-------+-- H Q --t-----I We will take no significant shortcuts for these examples, even though there may be times when we could. In this way, you can gain a better understanding of exactly what is happening as you implement your design. Constructing a Combinatorial Design-Basic Gates l---~- I J K By "beginner," we mean a logic designer who is just beginning to use programmable logic. You may have a lot of experience with discrete digital logic, or you may have just graduated from college. We assume a basic understanding of digital logic. Some computer experience is helpful, but not essential. We will talk about device programming, describing all of the steps that are necessary to program a PLD. However, due to the wide variety of programmers available, we will not get down to the level of detail that , tells you exactly which buttons to push. Although we will get as close as we can, we must defer the details to your programmer manual. --~------~ ~o-------+-- = fA Here the "equal" sign (=) is used to assign a function to output B. The slash (/) is used to indicate negation. Thus, this equation may be read: B is TRUE if NOT A is TRUE The next function is a simple AND gate. As shown in Figure 1, we can write: E = C*D Here we use the "equal" sign again, but this time we have introduced the asterisk (*) to indicate the AND operation. This equation may be read: E is TRUE if C AND D are TRUE PLD Design Basics AMO;t1 The third function is an OR gate, which may be written: H = F + G The "plus" sign (+) is used to specify the OR operation here. Because of the sum-of-products nature of logic as implemented in PLDs, it is often easy to place product terms on separate lines, which improves the readability. We may rewrite this equation as: H = I(x * I(X + Y) Y) Ix + IY = Ix * IY = We may generate our NAND function by writing: L = I (I * J * K) or, if preferred, II L + IJ F + + G IK This equation may be read: Likewise the NOR function may be specified as: H is TRUE if F OR G is TRUE o I + For the moment, we will assume that we have active-HIGH outputs on our device. The functions we have generated so far have essentially been active-HIGH functions. At times we wish to generate active-LOW functions; the next two functions are active-LOW functions that we wish to implement in an active-HIGH device. When we talk in terms of an active-HIGH or an active-LOW device, the real question is whether there is an extra inverter at the output. An active-HIGH device has an AND-OR structure; an active-LOW device has an AND-OR-INVERT structure which inverts the function at the output (see Figure 2). (M N) or o = 1M * IN Finally, an exclusive-OR (XOR) gate may be specified either as: R = P :+: Q where :+: represents the XOR operation, or more explicitly as: R = P * IQ + Ip * Q We have now specified all of the functions in terms of their Boolean equations. The equations are summarized in Figure 3. a. AND-OR Structure B IA E C H inverter * D F AND gate OR gate + G L II NAND gate + IJ + IK 0 b. ANO-OR-INVERT Structure 10173D-4 Figure 2. Active HIGH vs. Active LOW R 1M * IN P * IQ + Ip * Q NOR gate XOR gate Figure 3. BaSic Gates Equations NAND and NOR gates could be generated very simply in an active-LOW device, because we would just have to generate AND and OR functions, and let the output inverter generate their complements. However, given that we wish to implement these functions in an active-HIGH device, we must invoke DeMorgan's theorem, as follows: PLD Design Basics 5-9 ~ AMD very wide gates, which can be cumbersome to draw. To save space, the product terms are drawn as horizontal lines with a small AND gate symbol at one end to indicate the function being performed. Understanding the Logic Diagram A portion of a logic diagram is shown in Figure 4. The logic diagram shows all of the logic resources available in a particular device. In each device, inputs are provided in true and complement versions, as shown in Figure 4. These drive what are often called "input lines," which are the vertical lines in the logic diagram. These input lines can then be connected to product terms. The name "product term" is really just a fancy name for an AND gate. However, PLDs provide )}::.:.:.,.:.:.: ,.:.:. '.:.' ..... ::.:. .:.:.:.:.:.:.:....:.: I.:.:.',:,:,',:,',:,' l 48 -t-\::::::.:.:t.I:.t:.:I:.-r:.: . :.:.:.t:.:-.:.:.:.:.:.:.:.t:.:i.: .. -... ":':"1":"':"':':"':':' Although you really do not need to be concerned with the actual implementation of these functions inside the PAL device, you may be curious. Figure 5 shows how the inverter and the AND gate are implemented. An 'X' indicates a connection. A product term that is not used is indicated by an 'X' in the small AND gate. .... , .. .:.:..... :.:.:.:.: ...... ,.:.:.:.:... :.....:.: ........... ,.:.:.: ~:~==~;~==:t1=::j-. ~ I.:.:.:.:.:"':':':" . . I~:.:;~t~:~=.:.:.:W.-+-·.I-..-I. . .:.o=t--, 4. . :.l-.:.:.-:.:.:.:.:.:.:.+: .... P+oduct -D-J 49 arm 50-rrt-t--t--t-t-t--t+--+-f--H-t-+---J-+-I-J--U-ll-lQ--I1D-13 51---rtTt--t~Hr--rt--~--+4--~--~~+--+~~-I~ < .... :::.. <..... \ \ \ \ 9 \ o 123 456 7 8 9 1213 1617 2021 24252627 28293031 12 11 \ \ Input Lines True and Complement Inputs 10173D-5 Figure 4. A Portion of a Logic Diagram 5-10 PLD Design Basics AMD~ ... ~ A Y" E c -cr: .... B ... ~. D Y"" 10173D-6 Figure 5. Implementation of NOT, AND Gates Building the Design File Simulating the Gates Once the design has been conceptualized, the design file must be generated. After you have verified that your design file is correct, it is time to verify that the design itself is correct. This is done by simulating the design. Simulation provides a way for you to see whether your design is working as you expect it to. You provide a series of commands, or events, which are then simulated by the software. If requested, the software can tell you if the simulation matches what you expect, and, if not, where the problems are. We now know exactly what our functions are going to be. We have twelve inputs, six outputs, and the NAND function requires three product terms. Note that if we had specified: L = I (I * J * K) instead of: L II + IJ + IK forthe NAND gate, it would not be as obvious how many product terms would be needed. We are now in a position to create the design file. The design entry varies with the software package used. You must consult the manuals supplied with the software for design entry format. Generating a JEDEC File Once the design file has been entered, you can assemble the design to get a JEDEC file. We have two purposes here: to make sure there are no basic mistakes in the file, and to generate a JEDEC file for programming. Again, how this is done is determined by the software. The simulation section is the last part of the design file. It is not required, but is invariably helpful both in debugging the design, and in generating what can eventually be used as a portion of a test vector sequence. The simulator also converts the simulation results into test vectors, and appends the vectors to the JEDEC file. This file can be used with programmers that provide functional tests. Constructing a Registered DesignBasic Flip-Flops Next we will do a very simple registered design: we will be designing all of the basic flip-flop types (Figure 6). We will conceptualize the design by reviewing briefly the behavior of the D-type flip-flop. We will then present the results for T, J-K, and S-R flip-flops. PLD Design Basics 5-11 ~ AMD The devices we will be using in the examples only have D-type flip-flops. Thus, we will be emulating the other flip-flops with D-type flip-flops. D - I I - - - - - - - - + _ DT ClR b. DT= 0 I ClK o 0 ~t> Co OT po DC o I I Co IT .-1> po TC T T D c. DT:= 0 101730-8 I J Figure 7. Registered vs. Combinatorial Equations Co JKT po JKC .-t> K " K I S Co SRT po SRC -t> R R I PR 101730-7 Figure 6. Basic Flip-Flops Building the 0-Type Flip-Flop Equations A D-type flip-flop merely presents the input data at the output after being clocked. Its basic transfer function can be expressed as: DT : = We can also generate the complement signal (named DC) with the statement: DC : = ID I S D where we have used pins DT (D True) and D as shown in Figure 6. As shown in Figure 6, we want to add synchronous preset and clear functions to the flip-flops. This can be done with two input pins, called PR and ClR. To add these functions to the true flip-flop signal, we add IClR to every product term and add one product term consisting only of PR. Likewise, for the complement functions, we add IPR to every product term, and add one product term consisting only of ClR. With these changes, the equations now looks like: DT '= D * ICLR + PR DC '= + ID * IpR CLR In this way, when clearing the flip-flops, the active-HIGH flip-flops have no product terms true, and go lOW; the active-lOW flip-flops have the last product term true, and will therefore go HIGH. The reverse will occur for the preset function. Note the use of ':=' here instead of '='. This indicates that the output is registered forthis equation. The difference is illustrated in Figure 7. 5-12 OT ClK I J 0 PlD Design Basics AMD~ There is still one hole in this design: what happens if we preset and clear at the same time? As it is right now, both outputs will go HIGH. This makes no sense since one signal is supposed to be the inverse of the other. To rectify this, we can give the clear function priority over the preset function. We can do this by placing lelR on every product term for the true flip-flop signal. The results are shown as follows: DT D * /CLR + PR * /CLR DC /D * /PR + CLR The same basic procedure can be applied to all of the other flip-flops. The equations are shown in Figure 8. EQUATIONS ;emulating all flip-flops with D-type flip-flops DT := D * /CLR + PR * /CLR DC := /D + CLR TT := T * + /T * + PR * TC := T * + /T * + CLR * /PR ;output is D if not clear ;or 1 if preset and not clear at the same time ;output is /D if not preset iOr 1 if clear /TT * /CLR TT * /CLR /CLR igo HI if toggle and not clear istay HI if not toggle and not clear igo HI if preset and not clear at the same time /TC * /PR TC * /PR igo HI if toggle and not preset istay HI if not toggle and not preset go HI if clearing JKT:= J * /JKT * /CLR + /K * JKT * /CLR + PR * /CLR ;go HI if J and not clear ;stay HI if not K and not clear ;go HI if preset and not clear at the same time JKC:= /J * /JXC * /PR + K * /JKC * /PR + CLR ;go HI if not J and not preset ;stay HI if not K and not preset ;go HI if clear SRT:= S * /CLR + /R * SRT * /CLR + PR * /CLR SRC:= R * /PR + /S * SRC * /PR + CLR ;go HI if set and not clear ;stay HI if not reset and not clear ;go HI if preset and not clear at the same time igo HI if reset and not preset ;stay HI if not set and not preset ;go HI if clear Figure 8. Flip-Flop Equation Section Building the Remaining Equations and Completing the DeSign File Notice that in some Of the equations above, the output signal itself shows up in the equations. This is the way in which feedback from the flip-flop can be used to determine the next state of the flip-flop. An equivalent logic drawing of the TT equation is shown in Figure 9. PLD Design Basics 5-13 ~ AMD ClK ----------------------------~ ClR T l--+------I D Qt-tI--- IT PR - - - - - - - - - 1 - - - - 1 IT!1:= T* [Z!I] * IClR + T* [fIT) * IClR 10173D-9 + PR*/ClR Figure 9. Feedback In the Equation for TT We are now in a position to complete the design file. You must follow the instructions included with your software . package to complete the file. Simulating the Flip-Flops After processing the design and correcting any mistakes, we can run the simulation. The file can now be simulated in the same manner as the basic gates design. 10173D-10 Figure 10. A Connector Must Be Provided Between the Computer and the Programmer Programming a Device ~ft~r simulating the design, and verifying that it works, it time to program a device. There are several steps to programming, but the exact operation of the programmer naturally depends on the type of programmer being used. We will be as explicit as we can here, but you will need to refer to your programmer manual for the specifics. IS The first thing that must be done after turning the programmeron is to select the device type. This tells the programmer what kind of programming data to expect. The device type is usually selected either from a menu or by entering a device code. Your programmer manual will have the details. Next a JED EC file must be downloaded. To transfer the JEDEC file from the computer to your programmer, you will need to provide a connection, as shown in Figure 10. 5-14 If your programmer can perform functional tests, and you wish for those tests to be performed, you should download the JEDEC file containing the vectors; otherwise, you should download the JEDEC file without vectors. To download data, the programmer must first be set up to receive data. The programmer manual will tell you how to do this. Communication must be set up between the computer and the programmer. Whichever communication program is installed must be invoked. This is used to transmit the JEDEC file to the programmer. Follow the instructions for your program to accomplish the next steps. PLD Design Basics AMD~ Before actually sending the data, you must verify the correct communication protocol. Check to make sure you know what protocol the programmer is expecting; then set up the baud rate, data bits, stop bits, and parity, to match the protocol. Once the protocol has been set up the JEDEC file must be downloaded. . Enter the name of the JEDEC file you wish to use. The computer will then announce that it is sending the data, and tell you when it is finished. Note that just because it says it has finished sending data does not mean that the data was received. Your programmer will indicate whether or not data was received correctly. Once the data has been received, the programmer is ready to program a device. Place a device in the appropriate socket, and follow the instructions for your programmer to program the device. This procedure programs and verifies the connections in the device, and, if a JEDEC file containing vectors was used, will perform a functional test. The programmer will announce when the programming procedure has been completed. You may then take the device and plug it into your application. If you have actually programmed one of the examples that we created above, you naturally don't have a board into which you can plug the device. If you do have a lab setup, you may wish to play with the devices to verify for yourself that the devices perform just as you expected them to. You will find much more detail on many issues that were not discussed in this section in the remaining sections of this handbook. This section should have provided you with the basic knowledge you need to understand the remaining design examples in this book, and to start your own designs. PLD Design Basics 5-15 PLD Design Methodology Programmable logic devices (PLDs) are used in digital systems design for implementing a wide variety of logic functions. These logic functions range from simple random logic replacement to complex control sequencers. Programmable logic devices offer the multiple advantages of low cost, high integration, ease of use, and easier design debugging capability not available in other systems design options. In the following discussion we will detail the PlD design process. Most PlDs have an AND-OR array structure with programmable connections in either or both of the arrays. A programmable array implies that the connections can be programmed by the user. The . popular PAL (Programmable Array Logic) devices have a programmable AND array and a fixed OR array. PAL devices are used for a wide variety of combinatorial and registered logic functions. In this discussion we will also examine the various design constraints to be considered when selecting the correct architecture for a given application. tables. Occasionally Karnaugh maps and state diagrams are also used. The design file is then assembled to produce the "JEDEC"file. The JEDEC file gets its name from the fact that it is an approved JEDEC standard for specifying the state of every connection on the device. Simulation can then be performed. If the design is correct, the JEDEC file is downloaded into a device programmer for programming the connections on the device. The device can then be plugged into the PC board where it will function. The entire procedure can often be performed with the designer never having to leave the desk. Most programmers interface to personal computers, so that the design file can be edited, assembled, simulated, and downloaded, and the device programmed, all in one place. All digital logic can be efficiently reduced to two fundamental gates, AND and OR, provided both true and complement versions of all input signals are available. Such logic is generally built around what is known as the sum-ot-products (AND-OR) torm. Programmable logic devices are ideal for implementing such two-stage logic in the AND and OR arrays. Various process technologies offer many design options for PlDs. The connections in the programmable arrays can be fuse-based, commonly used in both ECl and TTL bipolar technologies, E/EEPROM cell based in UV-EPROM and EEPROM CMOS technologies, and RAM cell-based in CMOS RAM technology. The selection of technology is mostly dependent upon the system speed and power constraints. Most design engineers are familiar with these constraints, which not only dictate the technology of PlDs but also all of the other logic used in a system. Designing with PlDs involves the use of design software and a device programmer (Figure 1). The design software eliminates the need to identify every connection to be programmed for implementing the desired sum-of-products logic. The design process begins with the creation of a design file which specifies the desired function. The function is typically represented by its sum-of-products form and can be derived directly from the timing diagram and/or truth 5-16 JEOEC FLE 101730-11 Figure 1. PLDs are Designed Using Software and a Device Prog rammer The first stage in a PlD design process (Figure 2) is the conceptualization of a design problem; the second is the selection of the correct device; the third is the implementation of the deSign, which also includes simulating the deSign with test vectors; and finally, the PLD Design Methodology AMD~ actual programming and testing on a system board. We will take a simple design example and go through the various stages of this design process. Conceptualize A Design Problem ! Select Device ! Implement Design ! We will take the example of a simple address decoder circuit required for a 68000 microprocessor. The microprocessor has 24 address lines along with separate read and write signals. It requires some ROM to store the boot-up code as well as some RAM for storing and executing programs. The purpose of the address decoder circu itry is to select one of the memory addresses at a time. The RAMs and ROMs are assigned addresses on the 68000 microprocessor address space. The Address decoder circuit has to select one of the RAMs or ROMs for a specific range of addresses, called the address space. This selection is accomplished by asserting the specific chip-select signal for the RAM or ROM when the microprocessor accesses one of the addresses in the address space. There is additional circuitry in a typical microprocessor system for addressing 1/0 devices (such as disk controllers). These devices also require that chip-select signals be asserted when the microprocessor addresses them. Figure 3 shows an example address map for a 68000 microprocessor. Program PLD ! Test PLD PROM 1 OOOOOO-OFFFFF PROM 2 100000-1FFFFF DRAM 1 200000-2FFFFF DRAM 2 300000-3FFFFF DRAM 3 400000-4FFFFF DRAM 4 500000-5FFFFF ----------- ! ----------- ----------- Plug PLD Into Board 101730-12 600000-6FFFFF Figure 2. Programmable Logic Device Design Process 101730-13 Figure 3. Memory Address Map Conceptualizing a Design The first step in the PLD design process is also required for any SSI/MSI design. An advantage of PLDs is that at this stage the designer needs to be concerned only with the required logic function. With SSI or MSI, various device logic limitations must be accounted for before the design can be started. Clearly a designer needs to develop a brief and complete functional description, based upon the system design requirements. PLD Design Methodology 5-17 ~ AMD ± Top Reset Button • Bottom _INIT Interface Logic RW ~ RESET ROMCS1 .J I ROMCS2 AS I I RAMCS A21,A22,A23 DRAM Controller / M68000 Microprocessor /2 PROM1 A22,A23 PROM2 H DRAM1 ~ DRAM2 H rl DRAM3 DRAM4 rr- Read } Only ~ r----. r----. r----. Read & Write DO-16 101730-14 Figure 4. Microprocessor to Memory Interface Figure 4 show the circuit diagram. The address signals from the 68000 microprocessor are inputs to the interface logic block. The outputs generated are ROMCS1, ROMCS2 and RAMCS. The generation of signals for selecting device lias is similar and is not shown here for the sake of simplicity. Other system inputs to the interface are the address strobe signal generated by the 68000 microprocessor as well as the read/write signal. The truth table for generating the outputs is shown in Table 1. This truth table is derived from the memory address map and the functional description of the design. Table 1. Truth Table for Chip-Select Signals Addresses Hex Size A23 A22 A21 Signal OOOOOO-OFFFFF 1 MB a a a ROMCS1 1 ROMCS2 1 a RAMCS 1 1 RAMCS depends upon the functional nature of the problem. Sometimes timing and logic considerations can also dictate the use of registers; this will be discussed later. Registers are usually not required for such simple combinatorial functions such as encoders, decoders, multiplexers, demultiplexers, adders, and comparators. However, registers are required for functions such as counters, timers, control signal generation, and state machines. No registers are required for this simple address decoding example. The best choice for our combinatorial design would be a PAL device. The task now is to select a PAL device for implementing the desired function. General device selection considerations are listed below. These items are applicable to most designs. 100000-1FFFFF 1 MB 200000-2FFFFF 1 MB 300000-3FFFFF 1 MB a a a a 400000-4FFFFF 1 MB 1 0 0 RAMCS • Number of I/O pins 500000-5FFFFF 1 MB 1 0 1 RAMCS • Device speed • Number of output pins • Device power requirements Device Selection Considerations The first task for the designer is to identify the design problem and classify it as a combinatorial function or a registered function, depending upon whether or not registers are required. In most cases, this decision 5-18 • Number of input pins • Number of registers (if any) • Number of product terms • Output polarity control P.LD Design Methodology AMD~ - Address Decoding Time Memory Access Time Address Decoding - - Time :1::~~~_-_1_0_n_s~~~~~~_""'_".I_"~~~~~~~===-- ::: :: ==========:":14=====~_1o_ns_-_-_-_-_-_::~I ReadlWrite Cycle Time 101730-15 Figure 5. System Timing Requirements The first resource that must be provided in a PLD is the numberof pins needed forthe basic logic function. This consists of the number .of input and output pins. Many PLDs have internal feedback, which allows the generated output signal to be reused as an input. The same feedback also allows the pin to be used as a dedicated input, if required. This is especially useful for fitting various designs with different input/output requirements on the same device. The I/O pin capability of certain PLDs can also be very useful for certain bus applications. hold a signal stable forthe long durations required by the addressed peripheral or memory. However, this slows the initial response or access time of the device since the chip select must wait for the setup time before the rising edge of the clock cycle. Devices without registers provide fast access time but hold the signal valid only as long as the input conditions are valid. In most address decoders, the address signals are- kept asserted by the microprocessor until the read/write cycle is completed. lnthis case, the registers are not required for holding the signals asserted. The task is as simple as counting the number of input, output and I/O pins required by the design and picking a PLD which has the requisite number of pins. The remaining two general design considerations are the number of product terms and output polarity. We will discuss these two as we implement the design in the next section. The next selection issue is the device speed. The most important timing consideration for combinatorial PLDs is the propagation delay (tPD) of signals from the input to the output of the device. For registered PLDs, the important timing consideration is the device clocking frequency. This clocking frequency is in turn determined by sum of the register setup time (ts), and clock-to-output propagation delay (teo). Most systems impose some timing restrictions on the internal logic functions. These restrictions will determine the necessary tPD (for combinatorial devices) or fMAX (for registered devices). In our design example, the PLD will primarily perform address decoding. The critical system timing constraint is determined by the read/write cycle time of the microprocessor and the memory access time available (Figure 5). Most microprocessors allow anywhere from 10 to 35 ns for address decoding. That is, 10 ns - 35 ns after the address is available, the correct memory chip-select signal should be asserted. In our design example, the available cycle time of 240 ns and memory access time of 220 ns leaves barely 10 ns for address decode time. We can check the propagation delay and select the appropriate speed device for our design, which is tPD = 10 ns. . We have already briefly discussed the types of applications where registers are needed. Sometimes the consideration of system timing can affect whether or not registers are needed. Devices with registers can Implementing a Design Implementing a design (Figure 6) requires the creation of a design file. The design file contains three types of ' information. • Basic bookkeeping information • Design syntax • Simulation syntax Once the design file is complete, it is then assembled and simulated. Once it passes assembly and Simulation, the resultant JEDEC file is downloaded to a device programmer for configuring the device. Design Syntax In this example, as shown in Figure 6, there are two options available to the designer for expressing the design. The first is through traditional Boolean logic equations; the second is through a state machine syntax. The Boolean logic equations are the only option for combinatorial designs and can also be efficient for some registered deSigns. The Boolean equations can be derived from a combination of the functional description, the truth table and/or the timing diagrams (Figure 7). The state machine approach is ideal for large registered control designs, and can be derived from the functional description, state table, state diagram and/or the timing diagram (Figure 8). PLD Design Methodology 5·19 ~ AMD Functional Description Assemble Design File 101730-17 Figure 7. Writing Boolean Logic Equations 10173D-16 Figure 6. Implementing a Design 5-20 PLD Design Methodology AMD~ .------------------------~ Functional Description State Table Assemble Design File 10173D-18 Figure 8. State Machine Description Boolean Logic Equations Boolean equations are used to represent the ~um-of-pr?ducts logic form. The Boolean equations are Ideally sUited for representing the two-level AND-OR logic available in most PLDs. A conventional approach to the design is to convert the design problem to its discrete logic implementation. ~uch random SSI and MSI logic can be easily Implemented in PLDs. This usually involves converting to sum-of-products Boolean logic form. This approach ?an be a chore, and much effort can be saved by Imple':Tlenting a design with PLDs in a sum-of-products form right from the start. This essentially means that the designer does not have to design around the limitations of fixed SSI and MSI functions. A direct implementation of a design in sum-of-products form in a PLD can also yield a faster circuit. Boolean equations can be directly derived from the truth table or timing diagram (Figure 7). The truth table is used more often in simple combinatorial designs. The timing diagram method is used more often in registered control 'deSigns. We will first discuss the truth table method and then discuss the details of the timing diagram method. In addition to specifying the logic function, the Boolean in the design file help document the design. There IS no need to draw out an equivalent schematic. This allows design modularity; the schematic can just show a bloc~ for a particular PLD. Separate supporting documentation (the design file) provides the details without cluttering the drawing. equati~ns PLD Design Methodology 5-21 ~ AMD truth table is based upon the functional description of the design, and is derived from the address map (Figure 3) and the truth table (Table 1). Truth-Table-Based Design The requirements for our particular design example can be easily converted to a truth table format (Table 2). This Table 2. Truth Table for the Address Decoder Output Generated A23 A22 A21 0 0 0 0 0 1 0 1 1 0 0 0 0 1 0 1 1 1 IN IT 1 1 1 1 1 1 There are three additional input signals in this design example. The first, RW, is generated by the microprocessor, and distinguishes between read and write cycles. Since the ROM data is only for reading, the ROMCS1 and ROMCS2 signals are asserted only when RW is high (when the microprocessor attempts to read the ROM) and are not asserted for the write cycle. On the other hand, RAMCS is generated for both read and write cycles and the state of signal RW is "don't care." The second additional signal, AS, is the address strobe signal generated by the microprocessor, and is asserted only when the address lines carry a valid address. All of the chip select signals need to be gated with the AS Signal to ensure that they are only generated for valid addresses, and no spurious chip selects are generated. The last signal is the INIT Signal, which is a system initialization signal. This signal is used to initialize the microprocessor for a ''warm boot," and none of the chip selects is allowed when this INIT signal is asserted. Writing Boolean equations from the above logic is very straight forward. The output signal names, along with their polarity, are assigned to sum-of-product equations, which are based upon inputs and their polarities. AS 0 0 0 0 0 0 RW 1 1 X X X X ROMCSl 0 ROMCS2 1 RAMCS 1 1 1 1 1 1 0 1 1 1 0 1 1 0 0 0 The equations are derived directly from the truth tables. Each one of the AND equations uses up one product term of the device as shown in Figure 9. One device selection consideration is to ensure that all the outputs have sufficient product terms to accommodate the desired function. This brings us to the issue of output polarity. Suppose we had to generate active-HIGH outputs. In that case the output equations for the ROMCS1 signal would be: ROMCSl = /A23 + /A22 * /A21 * INIT */AS * RW If the device has active-LOW outputs only, this equation's output polarity needs to be inverted to be able to fit the device. Using DeMorgan's theorem for Boolean logic we get: /RCMCSl = A23 + A22 + A21 + /INIT + AS + /RW This equation requires a large number of product terms (six). Some signals are efficient and use fewer product terms in their true form, while others are more efficient in their inverted form. The device selection issues of product terms and output polarity also apply to registered designs. Timing-Diagram-Based Design IA23 IROMCS2 = IA23 lRAMCS = IA23 + IA23 + A23 + A23 IROMCSl = * IA22 * * IA22 * * A22 * * A22 * * IA22 * * IA22 * IA21 * INIT A21 * INIT IA21 * INIT A21 * INIT IA21 * INIT A21 * INIT * lAS * RW * lAS * RW * lAS * lAS * lAS * lAS Figure 9. The Implementation In Boolean Equations 5-22 Until now, we have discussed a PLD design using truth tables as the primary design vehicle. In this section we will attempt a design using a timing diagram as a design vehicle. Earlier in the address decoder design we mentioned the INIT Signal. This IN IT signal essentially an initialization signal for the entire system. The INIT signal is used internally (via feedback) for disabling the chip selects during initialization. Externally it can be used to initialize PLD Design Methodology AMO~ other system signals. This INIT signal is generated from a RESET switch connected to the inputs of the device as shown in Figure 10. Most experienced designers understand the tradeoffs for device selection. They implicitly go through the steps of design conceptualization and device selection, explained earlier. They typically draw a block around the logic being designed, with the previous knowledge that it would fit a PLD which has sufficient inputs, outputs, las and product terms. RESET Switch Top Bottom Vee :> .~ RESET:±: f "> .~ TOP Bottom PAL Device Debounce Circuit 10173D-20 SODOM Figure 11. Timing Diagram for the Oebounce Switch 10173D-19 Figure 10. RESET Switch for System Initialization To avoid unwanted initialization, the RESET switch must be debounced. That is, we want the INIT signal to remain HIGH until the switch actually contacts the bottom side. Once the bottom side is hit, INIT should be asserted active LOW. Once asserted, it should stay LOW and not change until the top side is hit again. The timing requirements of the debounce circuitry are shown in Figure 11. Signals TOP and BOnOM are inputs to the programmable logic device. These signals are activated when the RESET switch touches the top and the bottom contacts, respectively. We can formulate the equations by looking at the timing requirements of the debounce circuitry shown in Figure 11. The idea is to identify the key elements of this timing diagram. The arrows in Figure 11 showthe critical events. The first arrow shows the normal state of all the pins when the RESET switch is not asserted. Subsequent arrows show each event in the timing of the INIT signal, depending upon the movement of the switch. The logic level of the signals at each critical event carries useful logic information for deriving Boolean equations. This logic information for each event is converted into direct Boolean equations as shown in below. For example, at instant 1 the INIT signal remains HIGH as long as the TOP signal remains LOW; this is converted to INIT = ITOP * BOnOM. 1. Normal state INIT 2. Switch travels from TOP to BOTTOM INIT INIT 3. Switch contacts BOTTOM /INIT 4. Switch travels from BOnOM to TOP /INIT BOTTOM = /TOP = TOP * BOTTOM * /BOTTOM * /INIT TOP * 5. Normal State Again We can combine the two active-LOW events into one equation: /INIT /BOTTOM + /INIT * BOTTOM PLD Design Methodology * TOP 5-23 ~AMD Minimizing, this becomes: /INIT = /BOTTOM + / INIT * TOP This can also be done by way of a truth table and Karnaugh map. Table 3. Truth Table of INIT Logic TOP BOTTOM 1 1 1 1 1 1 a a a a INIT1 INIT+ 1 a a a a a a a 1 1 a a a a 1 1 1 1 X X 1 Here TOP or BOTTOM will be LOW if contacted. Note that both TOP and BOTTOM can not be contacted at the same time. The truth table of Table 3 yields the Karnaugh map shown in Figure 12. Grouping the zeros (because we are using active-LOW outputs) yields the Boolean equation identical to the one derived from the timing diagram. Most experienced deSigners understand the tradeoffs for device selection. They implicitly go through the steps of design conceptualization and device selection, explained earlier. They typically draw a block around the logic being designed, with the previous knowledge that it would fit a PLD which has sufficient inputs, outputs, lOs and product terms. Simulation Design simulation is an integral part of the design process, as shown in Figure 13. The purpose is to exercise all of the inputs and test the response of outputs to verify that they will work as desired in the system. These are essentially test vectors which designate the state of every input on the device; the outputs are then checked for an appropriate response. The simulation test vectors identify any flaws in the design equations which could affect the logical operation of the devices programmed. Thus, the simulation vectors serve as a design debugging tool. (fOP IBOTTOM IRESET 00 01 11 Simulate the Design with Simulation Vectors 10 o 10173D-21 Fix Errors No -----e: Figure 12. Karnaugh Map of INIT Signal Logic There is essentially no difference between the truth table and timing diagram techniques for writing Boolean logic. Also, a careful analysis will indicate that we implicitly assumed a truth table in the timing diagram example. Some designers prefer to make a separate truth table (at least in the first few PLD designs), while others prefer to design directly from timing diagrams. While the truth table method allows a more optimal utilization of product terms, the timing diagram method is easier to visualize as it retains the design perspective. In both cases the logic should be minimized by the design software to ensure that the deSign is testable. Download JEDEC File to the Device Programmer Program the PLD 10173D-22 Figure 13. Device Simulation and Programming 5-24 PLD Design Methodology AMD~ Simulation test vectors will eventually make up part of a larger set of test vectors called '1unctional test vectors". These functional test vectors are used to exercise a real device after programming to identify any individual devices which are defective. Other means of identifying defective devices, such as signature analysis, are also available. In this section we will strictly focus on simulation vectors. Simulation is included in the design file along with the logic equations. There is little standardization in these simulation expressions among various PLD design software packages, although most of them rely on test vectors to exercise the logic. The simulation vectors or events can be directly derived from the truth table and the timing diagram of the design. The logic level and functions of all Signals can be expanded and rewritten in a test vector form by the software. For example, the truth table for the address decoder example discussed earlier can be easily rewritten as shown in Table 4. Table 4. Truth Table Used to Derive Simulation Vectors A23 A22 A21 TOP BOTTOM AS RW ROMCS1 ROMCS2 RAMCS INIT 0 0 0 0 0 0 0 0 1 H H H H H H 0 0 1 1 0 0 H H H 0 1 1 1 1 H 0 0 1 1 1 1 H H H H 0 0 1 1 0 0 0 0 1 1 H H H H H 0 H H 0 0 1 1 1 1 0 0 1 1 0 H H H H H 1 1 1 1 1 0 0 0 0 0 0 H H H H H 0 0 1 1 0 0 H H H H H L H H 1 1 1 1 1 0 0 1 1 1 1 1 0 H H H H H H H H H H H H H H H H H L L 0 0 0 1 1 1 0 1 1 1 1 0 1 1 1 0 1 0 1 1 X X 0 1 1 1 1 1 These are essentially the simulation vectors which will allow us to define the inputs to the device and check the outputs of the device. The simulator then interprets the design file and generates the output logic levels and/or waveforms, which can be checked by the deSigner. X X X X X X X X X X X X X L L L L L H H H H H Once the simulation is complete, the design file can be assembled to generate the JEDEC file. In the proceeding discussions we have assumed prior knowledge of the design file assembly. The procedure for assembly varies with different software packages. PLD Design Methodology 5-25 ~ AMD Device Programming and Testing Once the design simulation is completed, the final step is device programming and testing (Figure 14). Programmers are available from a variety of vendors. It is important to note that Advanced Micro Devices, Inc., qualifies programmers upon verifying that the algorithms used by the programmers are correct and that other basic criteria are met. When purchasing a programmer, check that the programmer is qualified for the devices you intend to use. There are two types of programmers available: menu-driven or device code based. The menu-driven programmer directly indicates the part type being programmed, whereas the latter type requires the user to enter the device code before programming. Once the JEDEC fuse file has been downloaded, the programmer can program the device; the PLD is then ready for use. The programmer also verifies the connections after the programming cycle. Programmers also provide the capability of reading a previously programmed device and creating duplicates of that device. Testing PLDs Download JEDEC File to Device Programmer The testing of PLDs can be performed by the device programmer or by other test equipment. For a manufacturing environment, where high yields are required, device testing is critical. After testing is complete, the device security bit may be programmed, if desired, to secure the design from copying. Program OPLD~ Performed by Device / Test PLD with Simulation and Other Test Vectors I I Programmers Test PLD 10173D-23 Figure 14. Device Programming and Testing 5-26 PLD Desig n Methodology Combinatorial Logic Design INTRODUCTION In this section we will take a detailed look at several aspects of combinatorial logic design. Most combinatorial design applications can be easily segmented into five major fields. • • • • • co} Encoder C1 Encoded Outputs Encoders and Decoders Multiplexers 10173D-24 Comparators Figure LA Block Diagram of an Encoder Adders and Arithmetic Logic Latches Table 1. Truth Table of a Typical Encoder We will not only focus on the design methodology for these functions, but will also explore further functionspecific PLD selection requirements. Generalized designs will be developed, which can be customized later to suit specific system applications. Ways of optimizing the design will also be discussed. Encoders and Decoders Two of the most important functions required in digital design are encoding and decoding. The encoding and decoding of data are used extensively in digital communications as well as in peripherals. Both these areas use various complex encoding and decoding techniques. Most of these techniques are extensions of the simple encoding and decoding techniques often used in other digital designs. In this discussion we will focus on simple encoding and decoding techniques. More complex techniques will be discussed later. Encoders A binary code of n bits can be used to represent 2n distinct pieces of coded data. A simple combinatorial encoder is a circuit which generates n bits of output information based upon one of the 2n unique pieces of input data information. This encoding of information is controlled by other independent control signals in a typical digital circuit. An illustration of a typical encoder is shown in Figure 1. The deSign methodology typically followed is based on truth tables (Table 1), from which the Boolean equations are directly derived forthe design. The same generic device selection considerations discussed in the section on PAL device design methodology apply for encoder and decoder deSigns. Inputs Outputs A 8 C D CO C1 1 a 0 0 0 1 0 0 0 a 1 0 0 0 0 1 L L H H L H L H The Boolean equations can then be optimized using Karnaugh maps or the software minimizer. The resulting Boolean equations are: el /A + /A eo /A + /A * * B /B * * /B /B * /e * /e e * /e * * /D D * * /D D A Priority Encoder Let us take another look at the encoder example of Table 1. In this example it is assumed that only one of the inputs A, B, Cor D is asserted HIGH at anyone time. If two of the inputs are asserted HIGH simultaneously, a conflict would be created. To resolve this, a priority needs to be assigned to each of the inputs. Such a priority assignment is used to select a particular element when several inputs are asserted simultaneously. Each input is assigned a priority with respect to the other inputs. The output code generated is the code assigned to the highest priority input asserted. Thus, a priority encoder is a combinatorial circuit block similar to a general encoder, except that the inputs are assigned a priority. Such priority encoders are used often in state machine applications, where they detect Combinatorial Logic Design 5·27 ~ AMD the occurrence of the highest priority event. They are also used for microprocessor interrupt controllers, where they detect the highest priority interrupt. Another use for priority encoders is in bus control, where they are used in arbitration schemes for allowing selective access to the bus. The model of a priority encoder is shown in Figure 2. The four input signals are A, B, C and D. These are to be encoded as LL, LH, HL and HH outputs. Let us assign priority to Dover C, Cover B, and B over A. The next design step would be to modify the truth table (Table 2) to reflect these priorities. The Boolean equations, directly derived from the truth table, are: Cl /A + /A + + co /A + /A + + * * * * B /B B * /e * /B /B * C * /C * * * * /C /C C * * * /D D /D D /D D /D D These equations can be further optimized by the design software to the following: Cl Cl D D + /C + C * B A B C Although a priority encoder is a purely combinatorial function, output registers are frequently used to hold the output signal stable for longer durations. CO Priority -Encoder C1 Decoders D 10173D-25 Figure 2. A Four-Input Priority Encoder Block Diagram Table 2. Priority Encoder Truth Table Inputs Outputs A 8 C 0 CO 1 0 1 a 1 0 0 0 L 0 0 0 0 0 0 0 1 X X X 1 X X 0 0 0 5-28 1 X 1 C1 L L H H H H L H H H Priority L Assignments H L A decoder performs the reverse function of an encoder. It converts an n-bit code to one of its 2n unique items. It is a combinatorial circuit designed such that at most one of its several outputs will be asserted based upon the unique input codes. A decoder may have as many outputs as there are possible binary input selection combinations. As shown in the truth table (Table 3), only one output may be asserted at any time. When a new combination is applied, another output is asserted and the original output is returned to its non-asserted state. Combinatorial Logic Design AMD~ Table 3. The Truth Table of an Active-LOW 4-to-16 Decoder Input Select Lines Output Lines A B C D 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 1 1 1 0 0 0 1 1 0 0 0 0 1 0 0 0 1 0 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 02 1 1 0 1 1 1 1 01 1 0 03 1 1 1 04 1 1 1 1 1 1 1 1 0 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 The Boolean logic equations can be directly derived from the truth table shown in Figure 5. The procedure is the same as explained in the previous section on PLD design methodology. The Boolean equations derived are shown in Figure 3. /QO /Ql /Q2 /Q3 /Q4 /Q5 /Q6 /Q7 /Q8 /Q9 /Ql0 /Qll /Q12 /Q13 /Q14 /Q15 /D /D /D /D /D /D /D /D D D D D D D D D * * * * * * * * * * * * * * * * /e /e /e /e .e e e e /e /e /e /e e e e e * * * * * * * * * * * * * * * * /B /B B B /B /B B B /B /B B B /B /B B B * * * * * * * * * * * * * * * * 05 1 1 1 1 1 /A A /A A /A A /A A /A A fA A fA A /A A Figure 3. Decoder Boolean Logic Equations Probably the most commonly used decoders are the address decoders required by most microprocessors and bus interfaces. These also constitute the most common application of PLDs in digital designs. The design considerations for address decoders have been covered earlier in the PLD Design Methodology section. Laterwe will develop a general Boolean equation for an address decoder circuit when we discuss range decoders. OS 1 1 1 1 1 1 1 0 1 1 1 1 1 1 07 1 1 1 1 1 1 1 OS 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 09 010 011 012 013 014 015 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 a 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 1 Encoder/Decoder Device Selection Considerations The general device selection considerations are listed below. Based upon the number of inputs and outputs required, a device can be selected. • • • • • • • • Number of Input Pins Number of Output Pins Number of I/O Pins Device Speed Device Power Requirements Number of Registers Number of Product Terms Output Polarity Control Encoders typically require a large number of inputs and fewer outputs, whereas decoders typically require a large number of outputs and fewer inputs. Notice from the truth table that there is no combination of inputs that will send all the outputs to their non-asserted state. Many designs actually need to be able to make all outputs inactive. This can be done simply by putting enable lines in all of the output AND gates. Many such design modifications can be easily added once the basic Boolean equations have been derived, instead of redoing the truth table. Another important device selection consideration for encoders and decoders is the number of product terms required for a design. A careful selection of code values Combinatorial Logic Design 5-29 ~ AMD (and priority assignments in priority encoders) can often reduce the required number of product terms. This can sometimes determine whether or not a design fits a device successfully. Figure 4 shows the truth tables of two simple partial 3-to-2 encoders. The product terms required for the two designs are different due to the different assignment of encoded bits. Inputs Outputs Inputs Outputs A B C X1 XO A B C X1 XO ~ 0 0 0 1 0 0 1 0 0 1 0 1 0 1 0 0 0 1 0 0 0 1 0 1 1 1 0 1 Xl XO 0 /A * /B * C /A * /B * Ic Xl XO /A * B * + IA */B * A * IB * + /A * IB * Table 4. Truth Table for a Three-to-One Multiplexer Select Inputs A 11CO 11C1 11C2 01Y 0 0 0 0 1 1 0 0 1 1 0 0 0 1 X X X X X X 0 1 X X X X 0 1 0 1 0 1 0 1 C IC C 10173D-26 X X Deriving the Boolean equation from this truth table is a straight forward task. In this case no further minimization is possible. The Boolean equation is: IOlY IC Output B + + IB * /A * /B B * A /A * * IIlCO /IlCl IIlC2 * The equations derived in the above example can be easily generalized forother multiplexers. The symbol for a general 2n-inputs-to-one-output multiplexer is shown in Figure 5 where n select lines are used. Figure 4. Two Encoders with Different Product Term Requirements Anotherway of looking at a decoder is as a logic function which, depending upon the select code applied, connects one data input to the selected outputs. Also known as a demultiplexer, a decoder essentially connects an input to one of 2n outputs based upon n select code bits. The reverse logic function, which combines data from multiple sources to an output signal, is called a multiplexer and is discussed next. 10 11 Inputs . Multiplexer Y Output Multiplexers A multiplexer (sometimes referred to as a data selector) is a special combinatorial circuit, widely used in digital design. It is designed to gate one of several inputs to a single output. The input selected for connection to the output is controlled by a separate set of select inputs. The traditional use of a multiplexer is for ''time division multiplexing" in data communication, when gating several data lines to a single data transmission line for short intervals of time. The data received is then demultiplexed by using a demultiplexer. The design methodology employed for multiplexer design is the truth-table approach. As an example, we can look at a three in put-to-one-output (3:1) multiplexer, which uses two select Signals A and B. Based on these two select bits, the data on one of the three inputs is sent to the output. The truth table is shown in Table 4. 5-30 SO S1 Sn-1 10173D-27 Input Select Lines Figure 5. General Model of a 2n-to-1 Multiplexer Combinatorial Logic Design AMD~ The Boolean equations are: Comparators n=2 A comparator is a combinatorial circuit designed primarily to compare the relative magnitude of two binary numbers. Table 5 shows the truth table for a two-bit comparator. Y /Sl + /Sl + Sl + Sl, * * * * IsO so ISO so * (IO) * * (Il) (I2) * Table 5. Truth Table for a Comparator (I3) Inputs n=3 Y = /S2 + /S2 + /S2 + /S2 + S2 + S2 + S2 + S2 * * * * * * * * /Sl /Sl Sl Sl /Sl /Sl Sl Sl * * * * * * * * ISO so ISO so ISO so ISO SO * * * * * * * * (IO) (Il) (I2) (I3) (I4) (IS) (I6) (I7) Multiplexer Device Selection Considerations Multiplexers typically require more inputs than outputs, so the devices with a large number of inputs and VOs are usually more useful. Careful consideration must also be given to the number of product terms available on each output. Several multiplexers are often used simultaneously to route multiple address and data bits, under the control of the same select lines. In such cases, multiple devices can be cascaded when the number of inputs and outputs exceeds device limits. Cascading is also possible for large multiplexers that do not fit in a single device. In such cases, the select bits should also be judiciously selected for each PLD, to minimize the number of product terms. Outputs EQl lES 8 A GTR A>B A2 A1 82 81 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 0 0 0 0 0 0 0 1 0 1 1 1 0 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 1 1 0 0 0 1 1 0 1 1 0 0 1 1 0 1 0 0 1 0 1 1 0 0 0 0 0 1 1 1 1 1 1 0 1 A=B AB). Based on this truth table, the equations for the three output signals EOL, LES and GTR can be easily derived. These equations can then be optimized by using Boolean algebra, Kamaugh maps, or the minimization routine available with the software. Another common trick for designing a multiplexer is to connect a number of outputs together and control the output enables using the select bits to multiplex data. Timing considerations for such designs include the output enable and disable times, which should be carefully selected to avoid output contentions. Combinatorial Logic Design 5·31 ~AMD + (An: +: IBn) An-2 * IBn-2 The final Boolean equations are: EQL LES * IB2 * IBI IA2* IAI + IA2 * Al* IB2 * Bl * B2 * + A2* IAI Al * B2 * + A2 * IA2 * IAI * + IA2 * IAI * + IA2 * IAI * + IA2 * A * +/A2 Al * * + A2 */AI * B2 = = IAI + IA2 + IA2 GTR * * * = IA2 + + + + + A2 A2 A2 A2 A2 * * * * * * IAI + IA2 + IA2 * * * = IB2 IAI IB2 B2 B2 B2 B2 * * * Bl Bl * * * * * IB2 IB2 IB2 IB2 IB2 * B2 * * * * * Bl Bl + + + Bl + (An :+: IBn) IBI (A2 :+: /B2) IBI Bl IBI Bl B2 Al IAI Al IAI Al Al IB2 Al * * * * * * IBI IBI IBI Bl Bl IBI * IBI * IBI IB2 Comparator Device Selection Considerations The number of product terms needed is directly related to the number of bits compared. For LES (less than) and GTR (greater than) functions, the number of product terms required depends upon the number of bits in the two operands compared, as well as their value. The LES and GTR equations can be written as follows: LES GTR B2 * IA2 + (B2 :+: IA2) * Bl*/Al A2 * IB2 + (A2 :+: /B2) * LES Bn * *Bn-l * IAn-l * (Bn -1 : + : IAn -1 ) *Bn-2 * IAn-2 + + + GTR 5-32 The values of the comparands themselves affect the number of product terms used. When the comparison is made with comparands which are power-of-two numbers, the number of product terms required can be reduced drastically. This essentially relies on the fact that when the lower bits of a comparand are all zeros only the highest bit needs to be compared, requiring only one product term. For example, in a two-bit comparator, if A1 is zero and A2 is one, the equation for the greater-than function becomes very simple and requires only one product term: + (Bn :+: IAn) (B2 :+: IA2) * (Bn -1 : + : I An -1) ... * Bl * I Al An * IBn + (An: +: IBn) *An-l * = GTR /B2 The general equation for the GTR signal can also be simplified when comparing a number B to a fixed powerof-two comparand A with p least significant zeros. GTR /An + (Bn : +: I An) + (Bn :+: IAn) *(An-l:+:/Bn-l) ... * Al * IBI As is obvious from these equations, comparators require exclusive-OR functions. They can be efficiently implemented in devices that offers exclusive-OR functions but, can still be implemented in those devices that do not. 000010000... p = n These equations can then be extended for a general comparison of n-bit comparands as follows: * The total number of product terms required for an n-bit comparison is 2n-1. Comparators required a large numberof product terms so, devices that offer many product terms can be used very effectively. A Al*/Bl * (An-I: +: IBn-I) 00 1 = IBn */Bn-l ... */Bp+l */Bp This general GTR equation can also be considered as an equation for comparing a number to a range of numbers extending from zero to number A. In fact, this trick is used very often by many system designers for address decoder functions. In the PLD design methodology section the ROMCS1 signal is one such signal that is generated for the address range from (000000) hexto (OFFFFF) hex. Forthis design n=23, the comparand A=(OFFFFF + 1)=1 00000, and p=21. Substituting in the general equation we get the same address decoder Boolean logic equation. ROMCSI = IBn-l Combinatorial Logic Design IA23 * IA22* IA21 AMD~ As such designs require few product terms and no XOR gates, they are efficiently implemented on standard combinatorial PlDs. A general form of range comparators with two boundary comparands will be discussed later. The third output signal is the EQl signal. The EQl Boolean equation tells us whether the two numbers are identical. Such information is useful not only in address decoders, but also in digital signal processing designs. This equation requires a large number of product terms. A closer examination reveals that it is essentially an exclusive-OR function. EQL EQL /A2 * /B2 * (/A1 * /B1 + Al * B1) + A2 * B2 * (/A1 * /B1 + Al * B1) (A1:*:B1)* (A2:*:B2);Exclusive-NOR ; function Inverting this: /EQL = (A1:+:B1) + (A2:+:B2); Exclusive-OR ; function This equation can be extended to give a general equation for equal-to comparison for two n-bit comparands. /EQL (An :+: Bn) + (An-1 :+: Bn-1) + (An-2 :+: Bn-2) + (An-3 :+: Bn-3) let us analyze these equations further. The lES and GTR outputs indicate whether one number is greater than or less that another. In fact, these equations can also be judiciously combined to get a comparison of a range of numbers such as A>X>B. Such range comparisons are very useful for address decoder circuits. Range Decoders Range decoders implemented as address decoders are one of the most commonly used applications of PlDs in digital systems. A good example is the address decoder illustrated earlier. Range decoders compare a number (address) to a given range of comparands (addresses). One way to arrive at the range decoder Boolean equations is to use the. traditional truth table approach. Another way is to use the Boolean equations generated earlier in the comparator section for greater-than and less-than functions. To decode a range of three-bit numbers from B to A, we must compare another number X such that A>X>B. The Boolean equations for the GTR (A>X) and lES (B 1 1 -> 2 2 -> 3 3 ->4 4 -> 5 5 -> 6 6 -> 7 7 -> 8 8 -> 9 9 -> 0 0 0 0 0 0 0 0 1 1 0 02 01 0 0 0 1 1 1 1 0 0 0 0 1 1 0 0 1 1 0 0 0 00 1 0 1 0 1 0 1 0 1 0 AMD~ A good example of a modulo counter is a BCD counter. Such a counter is useful in applications where the computer's outputs are generated using a decimal system. While a four-bit binary counter can count to sixteen, the BCD counter terminates the count at the modulus of 10. Modulo counters can be designed in a variety of ways. One direct way is to use the truth table to implement a count to a modulus and directly derive the equations from it. The truth table for a BCD count (from zero to nine) is shown in Table 4. A state diagram for the BCD counter is shown in Figure 8. For active-LOW outputs, the Boolean equations can be derived directly from the truth table and optimized using Karnaugh maps or the software minimizer. The Boolean equation for 03 is: /Q3 := + + + + + + + + + + + + + Now let us consider what happens if the device accidentally powers up in one of the count values from ten to fifteen. These are illegal counts (states) and, for a good design, a mechanism must be built into the equations to allow it to recover back into a legal state. What we actually need is to consider the truth table in Table 5 in conjunction with the one in Table 4 for deriving the Boolean equations. Table 5. Truth Table for Illegal State Recovery to Count Zero Present State Next State 03 02 01 00 1 1 1 1 1 1 0 0 1 1 1 1 1 1 0 0 1 1 0 1 0 1 0 1 10 -> 11 -> 12 -> 13 -> 14 -> 15 -> 0 0 0 0 0 0 03 02 01 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 /Q3 /Q3 /Q3 /Q3 /Q3 /Q3 /Q3 Q3 Q3 Q3 Q3 Q3 Q3 Q3 * * * * * * * * * * * * * * /Q2 /Q2 /Q2 /Q2 Q2 Q2 Q2 /Q2 /Q2 /Q2 Q2 Q2 Q2 Q2 * * * * * * * * * * * * * * /Ql /Ql Ql Ql /Ql /Ql Ql /Ql Ql Ql /Ql /Ql Ql Ql * * * * * * * * * * * * * * /QO QO /QO 60 /QO QO /QO QO /QO QO /QO QO /QO QO The equation can be reduced to the following: /Q3 := /Q3 * /Q2 + /Q3 * /Ql + /Q2 * QO + Q3 * Ql + Q3 * Q2 Similar Boolean equations can be generated for 02, 012 and QO. Figure 9 shows the circuit diagram of a loadable dual BCD counter. Figure 8. State Sequence of a BCD Counter Showing Illegal State Recovery Registered Logic Design 5-47 ~ AMD CEB LDB ---- Four-Bit BCD Count '-- DO Input Buffers 01 02 - 00 - 04 05 06 07 01 02 03 03 Four-Bit BCD Count CEA LDA A 101 73D-43 Clock Figure 9. Circuit of a Dual BCD Counter Modulo Counter Device Selection Considerations We have illustrated a counter that counts from zero to a fixed modulus. The same technique can be applied for a counter which counts down from a maximum power-oftwo numberto a fixed modulus, or even counter which counts from one modulus to another. The important considerations will be the number of product terms used. a The registered PLDs used for modulo counters are similar to the ones selected for other counters. Since the counts used are binary, devices with J-K, T-type flipflops, or XOR gates will help optimize the number of product terms used. The product term usage also depends upon the modulus selected. Generally, a powerof two or a multiple-of-two modulus will require fewer product terms. Another factor for flip-flop selection is the illegal states. D-type flip-flops are generally belter suited for illegal state recovery than the J-K or T-type flip-flops. This is because when no product term is asserted, the D-type flip-flops reset to zero. Designers using J-K or T-type flip-flops must design-in illegal state recovery. Certain devices allow the use of a synchronous RESET product term for modulo counters. The idea is to use the minimal number of product terms to build a binary counter that counts up to a power-of-two number. However, this counter is RESET to zero using the synchronous RESET product term when the desired modulus is reached. It then begins counting afresh from zero, and the procedure is repeated. Similar operation can also be achieved with a synchronous PRESET product term for a down counter. 5-48 Using synchronous RESET and PRESET product terms allows the counter to recover from illegal states. Notice that the logic product terms in the counter are designed for a complete binary count. If the counter powers up in any illegal state (as shown in Figure 10), it will continue the count until the terminal count and then, return to zero, where the correct modulo count will begin. This illegal state recovery will take an unpredictable numberof clock cycles, and you may wish to design a more systematic recovery system. Cascading Modulo Counters For large modulo counters, the technique of generating Boolean equations from the truth tables is very tedious and time consuming. Another approach for designing modulo counters is to divide it into two smaller modulo counters. In addition to simplifying the design, this approach usually helps optimize the number of product terms. As an example, a modu10-360 counter can be directly implemented with nine register bits. However, instead of implementing this as a straight 9-bit counter, we can implement this as two counters: one four-bit counter (counting from zero to 14) and another five-bit counter (counting from zero to 23). Together, the two counters count up to 360. The terminal count output, MOUT, is asserted when the count reaches 360, as shown in Figure 11. The design requires nine inputs, nine outputs, one clock pin, one LOAD pin, one RESET and one MOUT (module output Signal) pin. Note that no extra flip-flops or pins were ne'eded. Obviously, the count values of this counter are not the same as a straight modu10-360 counter. Actually, this is what contributes to the optimization of the number of product terms used. Registered Logic Design AMD~ Synchronous Reset Product Term Asserted Figure 10. A BCD Counter Using Synchronous RESET Product Term 024 (4) (3) (2) (1) (0) 015 (3) (2) (1) (0) D ~ 5, A (8:4) load RST :~:~:~ A (3:0) Modulo 24 5 Bits 4, , ~ elK - Modulo 15 4 Bits r--~ AmPAl22V10 10173D-45 Figure 11. A Modulo-360 Counter Registered Logic Design 5-49 ~ AMD decoding circuitry. This combinatorial circuit modifies the counter bits and generates output signals in the manne~ required for peripheral timing and control. Since these circuits require extra combinatorial logic, they are not very efficient. They are also more susceptible to hazards and output glitches. Counters with Encoding Until now, we have discussed counters that generate binary output sequences. Most peripherals require a predetermined sequence of control signals. Custom control sequences can be generated by decoding the binary sequence with combinatorial logic. Figure 12 shows a general model of a counter with combinatorial output combinatorial~ Logic 1 Control Inputs •• I ,-......",. ~ _ Next State Decoder Clr k Output Decoder v •• • FlipFlops •• • • • Outputs 1 10173046 Figure 12. Counter with an Output Decoder It is possible to have a different output coding for a fourbit counter, as shown in Table 6. This code, called Gray code, allows only one output bit to toggle for each new count value. This code can be easily derived from a fourbit binary counter code (also shown in Table 6) using an output decoder. Table 6. Generating Gray Code from a Binary Code Binary Code X3 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 5-50 X2 0 0 O. 0 1 1 1 1 0 0 0 0 1 1 1 1 Gray Code X1 XO G3 G2 G1 GO 0 0 0 0 0 1 1 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 1 1 0 0 1 1 1 1 0 1 0 1 0 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 0 0 1 1 1 1 We can derive the Boolean equations forthe combinatorial output decoder from the truth table. The equations are: G3 G2 GI X3 X3 :+: X2 X2 :+: Xl GO Xl :+: Xo A more efficient and easier technique for generating control signals is to implement the decode circuitry before the registers. This alternative is shown in Figure 13. This essentially generates a non-standard counter with state values that are not a binary progression. It can be considered as a counter where the product terms for a binary count and encoding the outputs have been combined. Many different codes can be generated using such tech- : niques. We will limit ourselves to the ones that are most commonly used: Gray-code counters and Johnson counters. 0 0 1 1 0 Registered Logic Design AMD~ Combinatorial logic ClK .r...c;;--r---,._____r-1r--, Control _ _ _ _ Inputs Next State Decoder I I I Output "PRE" : Decoder I--'."'----.....-t-l~ FlipFlops •• • Outputs 10173D-47 Figure 13. Counter with Combined Next State Generation and Output Encoding Circuit Gray-Code Counters Gray-code counters are often used in digital designs for control timing functions. The primary advantage of Gray-code counters stems from the characteristic that only one output bit changes value for every clock cycle. These output signals can be easily decoded using a combinatorial decoder without any risk of hazards. Gray-code counters are used extensively as system clocks, since the different output bits provide different clock pulses, without the risks of hazards. Gray-code is also used in high-speed data communication applications, where data is transmitted from one part of the system to another, and where the erro~ susceptibility increases with the number of bit changes between adjacent numbers in a sequence. These are also used for such specialized applications as shaft encoders and real-time process control. Table 7. Truth Table for a Four-Bit Gray-Code Counter Present State Next State X3 X2 X1 XO X3 X2 X1 XO 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 The implementation of a Gray-code counter is very simple. A truth table can be derived from the transition table as is done for a binary counter. The Boolean equations can then be directly derived from the truth table. The truth table for the Gray-code counter is shown in Table 7. Registered Logic Design 5-51 ~AMD The Boolean logic equations for a Gray-code counter are: X3 + + + + + + + /X3 X3 X3 x3 X3 x3 X3 x3 /X3 /x3 /X3 /X3 /X3 X3 X3 x3 + + + + + + + /X3 /X3 /X3 /X3 X3 X3 X3 X3 '= + + + + + + + X2 Xl ,= * * * * * * * * * * * * * * * * * * X2 X2 X2 X2 X2 /X2 /X2 /X2 /X2 X2 X2 X2 X2 X2 X2 X2 * /XI * /XI * /XI * Xl * Xl /X2 /X2 /X2 X2 X2 X2 x2 /X2 /XI Xl Xl Xl /XI Xl Xl Xl * * * * * * * * * * * * * * * * * * * * * * * * * XO /X3 * /X2 * + /X3 * /X2 * + /X3 * X2 * + /X3 * X2 * + x3 * X2 * + X3 * x2 * + X3 * /X2 * + X3 * /X2 * Johnson Counters Xl Xl /XI Xl Xl Xl /XI /XI /XI /XI Xl /XI /XI Xl Xl /XI /XI Xl Xl * /XO * /XO * * * * * * * * * * * * * * * * * * * * * * * XO XO /XO /xO XO XO /XO /XO XO XO /XO /xO XO XO XO XO /XO /XO XO XO /XO /XO /XO * XO * /XO * XO * /XO * XO * /XO * xo A Johnson counter is part of a family of counters known as "ring counters." These counters are used for special applications where code symmetry is desired. Ring counters are also often used for timing purposes, since all the outputs are essentially a 'series of pulses. This code symmetry also allows use of the fewest possible product terms with a D-type register. Devices that provide a small amount of logic per cell, can implement Johnson counters very easily. Johnson counters are also known as circular-shift counters. The sequence for a five-stage Johnson counter is shown in Table 8. As can be seen in the truth table, the counter first fills up with 1's from left to right and then it fills up with zeros again. Note from the output sequence that only one of the Johnson counter bits changes for every clock period, like the Gray-code counter. One major advantage of the Johnson counter is that it can be readily decoded with small two-input NAND gates and hence is suitable for high-speed applications. 5-52 Note that the five-stage sequence has a table of 10 legal states and 22 illegal states (Table 9).ln general, an n-bit Johnson counter will produce a modulus of 2n. Figure 14 shows the state diagram of the five-bit counter. Table 8. Five-Bit Johnson Counter Truth Table Legal States Next State Present State 04 03 02 01 00 0 0 0 0 1 1 1 1 1 0 0 0 0 1 1 1 1 1 0 0 0 0 1 1 1 1 1 0 0 0 0 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 04 03 1 0 0 0 0 0 1 1 1 1 0 0 0 0 0 1 1 1 1 1 02 01 00 0 0 0 0 1 1 1 0 0 0 1 1 1 1 1 0 0 0 0 1 1 1 1 0 1 1 1 0 0 0 The implementation of a Johnson counter is relatively straight-forward, and is the same regardless of the number of stages. When D-type flip-flops are used, the Q output of each flip-flop is connected to the D input of the following stage. The single exception is the Q output of the last stage, which is complemented and connected to the D input of the first stage. Table 9, Illegal States for a Five-Bit Johnson Counter Illegal States Next State Present State 04 03 02 01 00 0 0 0 0 1 1 0 1 1 1 0 0 0 0 1 1 1 0 0 0 1 1 1 1 0 0 0 1 1 0 0 1 0 0 1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 1 1 1 1 Registered Logic Design 0 0 1 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 1 1 0 1 0 1 0 1 1 0 1 1 04 03 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 02 01 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 AMD~ Shift Registers A Shift Register is a special digital circuit often used as a primary building block in digital computer systems. It is closely related to a ring counter. Its fundamental usage is for temporary data storage and bit-wise data manipulation for advanced arithmetic and multiplication operations. Shift registers are also frequently used in communications, for converting parallel byte-wide data from the microprocessor to a serial data bit-stream for transmission. Shift registers are also used in graphics systems for serializing parallel data for use by the display mqnitor. A number of examples of video shift registers are included in the graphics section. The fundamental purpose of a shift register (Figure 16) is to shift data from one flip-flop to another. There are several types of shift registers. They are classified by the way in which incoming data is received (parallel or serial), and how outgoing data is transmitted (parallel or serial). Figure 14. State Diagram of a Five-Bit Johnson Counter One disadvantage of the counter is the number of invalid (or illegal) states. The invalid states increase exponentially with the length of the counter. The bigger the counter becomes, the greater are its chances of entering an illegal state. Johnson counters are very susceptible to illegal states, and can "hang up" very easily. Noise or improper use can cause this counter to end up in an illegal state. Therefore, a design with illegal state recovery circuitry is always recommended. Figure 15 shows a nine-bit Johnson counter that can be derived by directly extending the design of a five-bit Johnson counter. In the following example, we will discuss a simple universal shifter that provides both serial and parallel input and output functions. Depending upon the control signals 10 and 11, the data is shifted from one flip-flop to another in the left or the right direction. These inputs also control when the new parallel data is loaded onto the registers. When shifting left or right, serial data can be received and transmitted on serial pins LlRO and RILO. Since the flip-flop outputs appear on the output pins at all times, the parallel output data is always available. The truth table is shown in Table 10. The Boolean logic equations can be directly derived from the truth table, and are shown Figure 17. Shift registers can be modified to suit various system design requirements. This universal shift register can be used for serial in/serial out, parallel in/parallel out, serial in/parallel out and parallel in/serial out functions. CLK---.------~------~------~------*-------.-----~~----~~----~ 10173D-49 Figure 15. Block Diagram of a Nine-Bit Johnson Counter Registered Logic Design 5-53 ~AMD Clock Parallel Data In Control Signals Left Signal Data In and Out (L1RO) Right Serial Data Out and In (RILO) Shift Register 00 01 02 03 04 05 06 07 10173D-50 Parallel Data Out Figure 16. A Shift Register Block Diagram Table 10. The Truth Table for a Universal Shift Register 07 06 05 04 03 02 01 00 11 10 07 RILO 06 D7 06 07 05 D6 05 06 04 D5 04 05 03 D4 03 04 02 D3 02 03 01 D2 01 02 00 D1 00 01 L1RO DO 0 0 1 0 1 5-54 Registered Logic Design a 1 1 ;Retain Data ;Shift Right ;Shift Left ;Load Data AMD~ Equations /QO .= + :+: + /Ql .= + :+: + /Q2 .= + :+: + /Q3 .= + :+: + /Q4 .= + :+: + /Q5 .= + :+: + /Q6 .= + :+: + /Q7 .= + :+: + /Il*/IO*/QO /Il*IO*Q1 Il*/IO*/LIRO Il*IO*/DO HOLD QO SHIFT RIGHT SHIFT LEFT LOAD DO /Il*/IO*/Ql /Il *10* /Q2 I1*/IO*/QO Il*IO*/Dl ;HOLD Q1 ;SHIFT RIGHT ;SHIFT LEFT ;LOAD D1 /Il * /10* /Q2 /11 *10* /Q3 11*/IO*/Q1 11 *10* /D2 ;HOLD Q2 ;SHIFT RIGHT ;SHIFT LEFT ;LOAD D2 /Il*/IO*/Q3 /11 *10* /Q4 11* /10* /Q2 I1*IO*/D3 ;HOLD Q3 ;SHIFT RIGHT ;SHIFT LEFT ;LOAD D3 /I1*/IO*/Q4 /11*IO*/Q5 Il* /10* /Q3 11*IO*/D4 ;HOLD Q4 ;SHIFT RIGHT ;SHIFT LEFT ;LOAD D4 /11 * /10* /Q5 /11* IO*/Q6 11*/IO*/Q4 11* IO*/D5 ;HOLD Q5 ;SHIFT RIGHT ;SHIFT LEFT ;LOAD D5 /11*/IO*/Q6 /11 *10* /Q7 11 * /10* /Q5 I1*IO*/D6 ;HOLD Q6 ;SHIFT RIGHT ;SHIFT LEFT ;LOAD D6 /11*/IO*/Q7 /11 *10 * /RILO Il*/IO*/Q6 I1*IO*/D7 ;HOLD Q7 ;SHIFT RIGHT ;SHIFT LEFT ; LOAD D7 /LIRO = /QO LIRO.TRST = /11*10 ;LEFT IN RIGHT OUT /RILO = /Q7 RILO.TRST = Il*/IO ;RIGHT IN LEFT OUT Figure 17. Boolean Logic Equations for an Octal Shift Register Registered Logic Design 5-55 ~AMD Barrel Shifters In most data processing systems, some form of data shifting or rotation is necessary. In typical computer systems, the shifter is located at the output of the ALU, and usually requires a single-cycle shift and add function (Figure 18). For such applications as floating-point arithmetic or string manipulation, ordinary shift registers are inefficient, since they require n clock cycles for an n-bitshift. and the Boolean equations are then written based upon the truth tables. An eight-bit barrel shifter requires at least eight data inputs, eight registered data outputs. three control lines to specify the shift distance, a clock input and an output enable that controls the three-state buffer on the register output. Figure 19 shows the block diagram for an eight-bit registered barrel shifter, while Table 11 shows the truth table. The registered barrel shifter requires a total of 14 inputs and 8 outputs. Input Oata Bus 07 06 05 04 03 02 01 00 Register RST SO S1 S2 ClK File OE 07 06 05 04 03 02 01 00 10173D-52 Figure 19. Block Diagram of an Eight-Bit Barrel Shifter Shift Oistance--...... Table 11. Truth Table for an Eight-Bit Barrel Shifter S2 0 0 0 0 1 1 1 1 ClK---~ OE----"'-A Output Oata Bus SO 0 1 0 1 0 1 0 1 07 07 06 05 04 03 02 01 00 06 06 05 04 03 02 01 00 07 05 05 04 03 02 01 00 07 06 04 03 04 03 03 02 02 01 01 00 00 07 07 06 06 05 05 04 02 02 01 00 07 06 05 04 03 01 01 00 07 06 05 04 03 02 CO 00 07 06 05 04 03 02 01 Gray-Code, Johnson Counter and Shift Register Device Selection Considerations 10173D-51 Figure 18. Typical ALU Architecture A specialized shift register, called a "barrel shifter," is used to shift (or rotate) data by any number of bits in a single clock cycle. The name "barrel shifter" is used because of the circular nature of the shift operation. The storage registers on the output of the shifter are used in this architecture to pipeline the data operation, increasing throughput. The three-state buffer on the output registers is also useful for providing an interface to the data bus. The design of a barrel shifter proceeds in the same manner as a regular shift register. The truth table is drawn, 5-56 S1 0 0 1 1 0 0 1 1 Gray-code counters, Johnson counters and shift registers are not very logic-intensive; the number of product terms required is minimal. The D-type flip-flops provide the most efficient implementations, allowing these designs to be easily implemented in most PAL devices. Since Gray-code counters are often used as system clocks, very high speed PAL devices provide the highest resolution clocks. Barrel shifters are very logic-intensive and require many product terms, since data from all the inputs needs to be accessible at any output. Registered PLDs with a large number of product terms are ideal for barrel shifters. Large barrel shifters can also be partitioned into a number of PLDs. Registered Logic Design AMD~ Asynchronous Registered Designs Until now, we have discussed strictly synchronous registered designs, where a common system clock is used. In asynchronous registered designs, a common clock is not used. The register clock may be generated by the output of another register, or by a logical combination of various other signals. Such designs are usually slow for such applications as timing generation, because when the output of one register is used to clock another, multiple delays are encountered before all the register outputs stabilize. On the other hand, designs can be very fast for asynchronous applications such as bus arbitration and control, where a fast response to a bus signal can be provided without waiting for a common systemclock. Combinatorial hazard conditions can cause false clocking of registers, destroying the logic intended by the designer. The designer also needs to worry about race conditions when clocking a number of register simultaneously. Careful design analysis is strongly recommended before implementing any asynchronous design. Ripple counters are probably the easiest examples of such asynchronous designs. Figure 31 shows the logic diagram of a five-bit binary ripple counter. These counters clearly have the advantage of design simplicity. The output from one stage is fed as the clock to the next stage. However, this results in a slower counting rate, since the clock signals need to propagate through all five registers before the next count is reached. Although asynchronous designs are easierto visualize, they present larger problems in implementation. CK RESET r-------- -1 I I I I I I I I I T III T III hiV RO S Q a r-- ROY S Q a LJ LJ r- T I III RO v RD Y S S a Q Q a 11 RD S Q a LJ LJ LJ SET r- L___ -<= }:==-=-~_~ ___ ~ __~_____ J 02 I I Extra Circuit Required For Modulo 20 Counter 01 00 10173D-53 Figure 20. A Five-Bit Ripple Counter Figure 20 shows the implementation of a modul0-20 counter that is RESET when output bits 04 and 02 are both HIGH. Since the RESET is implemented with a product term, the extra AND gate shown can be implemented directly within the PAL device. Asynchronous Designs Device Selection Considerations The device selection for asynchronous designs is easy. As the clock signals require logic, only PLDs that allow implementations of Boolean logic on the clock signals are useful. Registered Logic Design 5-57 ~ AMD OTHER APPLICATIONS OF REGISTERED PLDs Registered PLDs are used for a number of miscellaneous applications that are not covered by the synchronous and asynchronous design applications discussed up to now. One such application is as a frequency divider. • Frequency dividers • Addressable Registers Frequency Dividers Standard synchronous counters provide the basic capability of dividing an input frequency. A single register of a PAL device will let us divide by two. If we stack these registers, a binary counter provides symmetrical division by 2, 4, 8, 16, etc. This divider has been a standard for years, and the PAL device has always been on excellent choice for such applications. One unique application of PAL devices is for dividing inputfrequencies by odd numbers. This has been done historically by designing a counter that cycles an odd number modulo, and decoding the specific states of the counter. The disadvantage of this approach is that the output is not symmetrical and the duty cycle is not 50%. Let us examine a simple divide-by-five counter. This counter can be implemented using three flip-flops that start at zero and reset at four, resulting in a five-state counter. Table 12 shows the outputs of the three individual flip-flops. Table 12. Truth Table for a Five-Bit Counter Present State Next State 02 Q1' 00 0 0 0 0 0 0 1 1 0 0 1 02 01 00 1 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 State zero to one. State one to two. State two to three. State three to four. State four to zero. TheBoo~anequ~~nsare: Q2 Ql QO 5-58 := /Q2 := /Ql := /Q2 * Ql * QO * /QO * QO + Ql iMSB bit * /QO iLSB bit The waveforms for this divider are shown in Figure 21. Notice that the 02 output goes HIGH for one state and that this output is one fifth of the input frequency, but it is a 20% duty cycle. 01 is active for two states; it provides the same frequency, but with a 40% duty cycle. If we want a 50% duty cycle, we are going to have to divide a state in half. To provide the 50% duty cycle, the two edges should be evenly spaced in the count sequence, one edge in the middle of state two and one at the beginning of state zero. The first edge can be formed by logically "ANDing" state_2 with the falling edge of the clock. The second edge can be formed by decoding state zero. edge_l =/ clock * /Q2 * Ql * /QO iedge between istates two and ithree edge_2= /Q2 */Ql */Ql iedge at state izera The logical "OR" of these two equations will provide the needed rising edges. To provide a clean output, this signal should clock another output register. The next step in the design is to pick the appropriate PAL device to fit this design. Our biggest concern is that we need the capability of clocking the counter at one speed and the output flip-flop at another. To do this, we cannot use a PAL device that has a dedicated clock pin; we need an architecture that allows programmable clocks. The clock signal requires two product terms (one for each edge). Another technique is to use the independent asynchronous SET and asynchronous RESET product terms of the output register. A HIGH on the SET product term asserts the register output, and a HIGH on the RESET product term unasserts the register output. Due to the asynchronous nature of the product terms some adjustment in timing is required. The SET product term is asserted when instate 0 (02=0, 01 =0 and 00=0), and the RESET product term is asserted when between states two and three. OUTPUT.SET = /clock * /Q2 * Ql * /QO iset between istates 2 & 3 OUTPUT.RESET = /Q2 */Ql * /QO ireset at istate zero Registered Logic Design AMD~ I_ State 0 _I . 1 State 3 1 State 4 1 State 0 1 State 1 1 ,. State 1 -1,. State 2 -,. -,. -- - 00 01 02 Output Set Reset 101730-54 Figure 21. Waveform for a Frequency Divider Addressable Registers Addressable registers are commonly-used MSI functions, often implemented in PAL devices. Addressable registers are used as building blocks for digital computers. Depending upon the address input one of the many flip-flops in the register retain their previous values. Registered Logic Design 5·59 State Machine Design INTRODUCTION What Is a State Machine? State machine designs are widely used for sequential control logic, which forms the core of many digital systems. State machines are required in a variety of applications covering a broad range of performance and complexity; low-level controls of microprocessor-toVLSI-peripheral interfaces, bus arbitration and timing generation in conventional microprocessors, custom bit-slice microprocessors, data encryption and decryption, and transmission protocols are but a few examples. A state machine is a digital device that traverses through a predetermined sequence of states in an orderly fashion. A state is a set of values measured at different parts of the circuit. A simple state machine can consist of PALdevice based combinatorial logic, output registers, and buried (state) registers. The state in such a sequencer is determined by the values stored in the buried and/or output registers. Typically, the details of control logic are the last to be settled in the design cycle, since they are continuously affected by changing system requirements and feature enhancements. Programmable logic is a forgiving solution for control logic design because it allows easy modifications to be made without disturbing PC board layout. Its flexibility provides an escape valve that permits design changes without impacting time-to-market. A majority of registered PAL device applications are sequential control designs where state machine deSign techniques are employed. As technology advances, new high-speed and high-functionality devices are being introduced which simplify the task of state machine deSign. A broad range of different functionality-and-performance solutions are available for state machine design. In this discussion we will examine the functions performed by state machines, their implementation on various devices, and their selection. A general form of a state machine can be depicted as a device shown in Figure 1. In addition to the device inputs and outputs, a state machine consists of two essential elements: combinatorial logic and memory (registers). This is similar to the registered counter designs discussed previously, which are essentially simple state machines. The memory is used to store the state of the machine. The combinatorial logic can be viewed as two distinct functional blocks: the next state decoder and the output decoder (Figure 2). The next state decoder determines the next state of the state machine while the output decoder generates the actual outputs. Although they perform two distinct functions, these are usually combined into one combinatorial logic array as in Figure 1. Combinatorial Logic , Next State Decode Device Inputs S~ r Output, Decode r---r----•• • r----- Outputs Memory - •• • 1 101730-55 Figure 1. Block Diagram of a Simple State Machine 5-60 State Machine Design AMDll Combinatorial Logic Combinatorial Logic Inputs -.. r Next State Decoder (Transition Function) ~ •• • Memory (Registers) ~ ••• Output Decoder (Output Decode Function) •• • - Outputs 1 10173D 56 Figure 2. State Machine, with Separate Output and Next State Decoders The basic operation of a state machine is twofold: State Machine Applications 1. It traverses through a sequence of states, where the next state is determined by next state decoder, depending upon the present state and input conditions. State machines are used in a number of system control applications. A sampling of a few of the applications, and how state machines are applied, is described below. 2. It provides sequences of output signals based upon state transitions. The outputs are generated by the output decoder based upon present state and input conditions. As sequencers for digital signal processing (DSP) applications, state machines offer speed and sufficient functionality without the overkill of complex microprocessors. For simple algorithms, such as those involved in performing a Fast Fourier Transform (FFT), a state machine can control the set of vectors that are multiplied and added in the process. Forcomplex DSP operations, a programmable DSP may be better. On the other hand, the programmable DSP solution is not likely to be as fast as the dedicated hardware approach. Using input signals for deciding the next state is also known as branching. In addition to branching, complex sequencers provide the capability of repeating sequences (looping) and subroutines. The transitions from one state to another are called control sequencing and the logic required for deciding the next states is called the transition function (Figure 2). The use of input signals in the decision-making process for output generation determines the type of a state machine. There are two widely known types of state machines: Mealy and Moore (Figure 3). Moore state machine outputs are a function of the present state only. In the more general Mealy-type state machines, the outputs are functions of both the state and the input signals. The logic required is known as the output function. For either type, the control sequencing depends upon both states and input signals. Most practical state machines are synchronous sequential circuits that rely on clock signals to trigger the state transitions. A single clock is connected to all of the state and output edge-triggered flip-flops, which allows a state change to occur on the riSing edge of the clock. Asynchronous state machines are also possible, which utilize the propagation delay in combinatorial logic for the memory function of the state machine. Such machines are highly susceptible to hazards, hard to design and are seldom used. In our discussion we will focus solely on sequential state machines. Consider the case of a video controller. It generates addresses for scanning purposes, using counters with various sequences and lengths. Instead of implementing these as actual counters, the sequences involved can be "unlocked" and implemented, instead, as state machine transitions. There is an advantage beyond mere economy of parts. A count can be set or initiated, then left to take care of itself, freeing the microprocessor for other operations. In peripheral control the simple state machine approach can be very efficient. Consider the case of run-Iengthlimited (RLL) code. Both encoding and decoding can be translated into state machines, which examine the serial data stream as it is read, and generate the output data. Industrial control and robotics offer further areas where simple control functions are requ ired. Such tasks as mechanical positioning of a robot arm, simple decision making, and calculation of a trigonometric function, usually does not require the high-power solution of microprocessors with stacks and pointers. Rather, what is required is a device that is capable of storing a limited number of states and allows simple branching upon conditions. State Machine Design 5-61 ~AMD Outputs Inputs St~ Next State Decoder r --•• • Output Decode Registers ~ ••• 1 •• • Output s are Functions of Stat e and Inputs a. Mealy State Machines Outputs Inputs ,....... r Next State Decoder --- Output Decode ~ •• • Registers ~ ••• 1 b. Moore State Machines •• • Output s are Functions of Stat eOnly 10173D-57 Figure 3. The Two Standard State Machine Models Data encryption and decryption present similar problems to those encountered in encoding and decoding for mass media, only here it is desirable to make the scheme not so obvious. A programmable state machine device with a security Bit is ideal for this because memory is internally programmed and cannot be accessed by someone tampering with the system. Functions Performed All the system design functions performed by controllers can be categorized as one of the following state machine functions: • Arbitration I!I Event monitoring • Multiple condition testing • Timing delays • Control signal generation State Machine Theory Let us take a brief look at the underlying theory for all sequentiallogic systems, the finite state machine (FSM), or simply state machine. Those parts of digital systems whose outputs depend on their past inputs as well as their current ones can be modeled as finite state machines. The "history" of the machine is summed up in the value of its internal state. When a new input is presented to the FSM, an output is generated which depends on this input and the present state of the FSM, and the machine is caused to move into new state, referred to as the next state. This new state also depends on both the input and present state. The structure of an FSM is shown pictorially in Figure 2. The internal state is stored in a block labeled "memory." As discussed earlier, two combinatorial functions are required: the transition function, which generates the value of the next state, and the output function, which generates the state machine output. Later we will take a design example and illustrate how these functions can be used when designing a state machine. 5-62 State Machine Design AMD~ State Diagram Representation The behavior of an FSM may be specified in graphical form as shown in Figure 4. This is called a state diagram, or state transition diagram. Each bubble represents a state, and each arrow represents a transition between states. Inputs that cause the transitions are shown next to each transition arrow. a. Moore Machine b. Mealy Machine 101730-60 Figure 6. Output Generation For this notation, there is a specification uncertainty as to which signals are outputs orinputs, as they both occur on the drawing next to the arrow in which they are active. This is usually resolved by separating the input and output Signals names with a line (Figures 4 and 6). Sometimes an auxiliary pin list detailing the logic polarity and input or output designations is also used. 101730-58 Figure 4. State Machine Representation Control sequencing is represented in the state transition diagram as shown in Figure 5. Direct control sequencing requires an unconditional transition from state A to state B. Similarly conditional control sequencing shows a conditional transition from state C to either state D or state E, depending upon input signal 11. a. Direct Control Sequencing State transition diagrams can be made more compact by writing on the transitions not the input values that cause the transition, as in Figure 4, but a Boolean expression defining the input combination or combinations that cause this transition. For example, in Figure 7, some transitions have been shown for a machine with inputs START, X1. and X2. In the transition between states 1 and 2, the inputs X1 and X2 are ignored (that is, they are "don't cares") and thus do not appear on the diagram. This saves space and makes the function more obvious. b. Conditional Control Sequencing 101730-61 101730-59 Figure 5. Control Sequencing Figure 7. State Transition Diagram with Mnemonics For Moore machines the output generation is represented by assigning outputs with states (bubbles) as shown in Figure 6. Similarly, for Mealy machines conditional output generation is represented by assigning outputs to transitions (arrows). as was shown in Figure 4. More detail on Mealy and Moore output generation is given later. There can be a problem with this method if one is careless. The state transitions in figu re 8 show what can happen. There are three input combinations, (10, 11, 12, 13) = {1 011}. {11 01} and {1111}, which make both (110 * 112 + 13) and (10 * 11 + 10 * 12) true. Since a transition to two next states is impossible, this is an error in the State Machine Design 5-63 ~AMD specification. It must either be guaranteed that these input combinations never occur, or the transition conditions must be modified. In this example, changing (10· 11 + 10· 12) to (10 • 11 + 10· 12) • 113 would solve the problem. 10173D-62 Figure 8. State Diagram with Conflicting Branch Conditions State Transition Table Representation A second method for state machine representation is the tabular form known as the state transition table, which has the format shown in Table 1. Listed along the' top are all the possible input bit combinations and internal states. Each row gives the next state and the next output; thus, the table specifies the transition and output functions. However, this type of table is not suitable for specifying practical machines in which there is a large number of inputs, since each input combination defines a row of the table. For example, with 10 inputs, 1024 rows would be required! Table 1. A State Transition Table Present State SO-Sn Inputs 10-lm Next State SO-Sn Outputs Generated OO-Op Flowcharts Another popular notation is based on flowcharts. In this notation, states are represented by rectangular boxes, and alternative state transitions are determined by strings of diamond-shaped boxes. The elements may have multiple entry points, but in general have only one exit. The state name is written as the first entry in the rectangular state box. Any Moore outputs present are written next in the state box, with a caret (A) following those that are unregistered. The state code assignment, if it is known, is written next to the upper right corner of the state box. Decision boxes are diamond or hexagonal shaped boxes containing either an input Signal or a logic expression. Two exits labeled "0" and "1" lead to either another decision box, a state box, or a Mealy output. 5-64 The rounded oval is used for Mealy machine outputs. Again, a caret follows those outputs that are unregistered. All the boxes may need to be expanded to accommodate a number of output signals or a larger expression. The use of these symbols is shown in Figure 9. Each path, through the decision boxes from one state to another defines a particular combination or set of combinations of the input variables. A path does not have to include all input variables; thus, it accommodates "don't cares." These decision trees take more space than the expressions WOUld, but in many practical cases, state machine controllers only test a small subset of the input variables in each state and the trees are quite manageable. Also, the chain of decisions often mirrors the designer's way of thinking about the actions of the controller. It is important to note that these tests are not performed sequentially in the FSM; all are performed in parallel by the FSM's state transition logic. A benefit of this method of specifying transitions is that the problem of Figure 8 can be avoided. Such a conflict would be impossible as one path cannot diverge to'define paths to two states. This flowchart notation can be compacted by allowing more complex decisions, when there is no danger of conflicts due to multiple next states being defined, Expressions can be tested, as shown in Figure 1Oa, or multiple branches can extend from a decoding box, as in Figure 1Ob. In the second case, it is convenient to group the set of binary inputs into a vector, and branch on different values of this vector. The three methods of state machine representation state diagrams, state tables, and flowcharts are all equivalent and interchangeable, since they all describe the same hardware structure. Each style has its own particular advantages. Although most popular, the state transition diagrams are more complex for problems where state transitions depend on many inputs, since the transition conditions are written directly on the transition arrows. Although cumbersome, the state tables allow the designer tight control over signal logic. Flowcharts are convenient for small problems where there are not more than about ten states and where up to two or three inputs or input expressions are tested in each state. For larger problems, they can become ungainly. Once a state machine is defined, it must be implemented on a device. Software packages are then used to implement the design on a device. The task is to convert the state machine description into transition and output functions. Software packages also account for device-specific architectural variations and limitations, to provide a uniform user interface. State Machine Design AMDl1 Some software packages accept all three different state machine representations directly as design inputs. However, the most prevalent design methodology is to convert the three state machine design representations to a simple textual representation. Textual representations are accepted by most software packages although the syntax varies. Since the most common of all state machine representations is the state transition diagram representation, we will use it in all subsequent discussions. Transition table and flowchart representation implementations will be very similar. ,....._....&._.......NN.., - (X". y".Z"', ...) (X, y,Z .... ) State Name State Code Asyn S ync Moore Output 10173D-63 Figure 9. Flowchart Notation 10173D-64 a. Testing Expressions b. Multiway Branch Figure 10. Using Flowcharts State Machine Design 5-65 ~AMD State Machine Types: Mealy & Moore With the state machine representation clarified, we can no~ return to the generic sequencer model of Figure 1, which has been labeled (Figure 11) to show the present st~t~ (PS), next state (NS), and output (OB, OA). This wlllillustr~te how Mealy and Moore machines are implemented with most sequencer devices that provide a single combinatorial logic array for both next state and output decode functions. There are four ways of using the sequencer, two of which implement Moore machines and two Mealy. First, let us look at the Mealy forms. The ~tandard Mealy form is shown in Figure 12, where the signals are labeled as in Figure 11 to indicate which registers and outputs are used. The register outputs PS are fed back into the array and define the present state. The combinatorial logic implements the transition function, which produces the next state flip-flop inputs NS, and the output function, which produces the machine output OB. This is the asynchronous Mealy form. Combinatorial Logic ~ Device Inputs Next State Decode Output Decode ~ r--•• • - ~ r ~ Memory (Registers) •• • I 08 Outputs OA Prese nt State PS 10173D-65 Figure 11. Generic Model of an FSM Inputs ~ Transition Function Next State NS Register (State Memory) Output Function -- 08 Outputs A Present State PS Clock Figure 12. Asynchronous Mealy Form An alternative Mealy form is shown in Figure 13. Here the outputs are passed through an extra output register (OA) and thus, do not respond immediately to input changes. This is the synchronous Mealy form. 5-66 State Machine Design 10173D-66 AMD~ Output Function In puts Transition Function Next State NS r----+ Register r----- f---+ Register OA Outputs /' A Present State PS Clock 101730-67 Figure 13. Synchronous Mealy Form Transition Function In puts .--- Next State NS Output Function Register (State Memory) ~ 08 Outputs /' Present State PS 10173D-68 Clock Figure 14. Asynchronous Moore Form In puts r----+ Transition Function Next State NS Register OA Outputs A Pre sent State PS Clock 101730-69 Figure 15. Synchronous Moore Form The standard Moore form is given in Figu re 14. Here the outputs 08 depend only on the present state PS. This is the asynchronous Moore form. The synchronous Moore form is shown in Figure 15. In this case the combinatorial logic can be assumed to be the unity function. The outputs (08) can be generated directly along with the present state (PS). Although these forms have been described separately, a single sequencer is able to realize a machine that combines them, provided that the required paths exist in the device. State Machine Design 5-67 ~AMD In the synchronous Moore form, the outputs occur in thestate in which they are named in the state transition diagram. Similarly, in the asynchronous Mealy and Moore forms the outputs occur in the state in which they are named, although delayed a little by the propagation delay of the output decoder. This is because they are combinatorial functions of the state (and inputs in the Mealy case). However, the synchronous Mealy machine is different. Here an output does not appear in the state in which it is named, since it goes into another register first. It appears when the machine is in the next state, and is thus delayed by one clock cycle. The state diagram in Figure 16 illustrates all the possibilities on a state transition diagram. • • • Number of inputs/outputs - I/O flexibility - Number of output registers Speed Intelligence/functionality - Number of product terms Type of flip-flops Number of state registers Number of II0s The number of inputs, outputs and I/O pins determine the Signals that can be sampled or generated by a state machine. Synchronous Mealy Output Available Timing and Speed The timing considerations for sequencer design are similar to those for registered logic design. A system clock cycle forms the basic kernel for evaluating control function behavior. Forthe most part, all input and output functions are specified in relationship to the positive edge. Registered outputs are available after a period of time teo, the clock-to-output propagation delay. Asynchronous outputs require an additional propagation delay (tPD) before they are valid. Asynchronous Moore Output Available 101730-70 Figure 16. State Diagram Labelling for Different Output Types As a matter of notation, Moore outputs are often placed within the state bubble and Mealy outputs are placed next to the path or arrow that activates them. The relationship of Mealy and Moore, synchronous and asynchronous outputs to the states is shown in Figure 17. 5-68 Device Selection Considerations There are three major criteria for selecting the correct state machine device for a design: For the circuit to operate reliably, all the flip-flop inputs must be stable at the flip-flop by the minimum set-up time (ts) of the flip-flops before the next active clock edge. If one of the inputs changes after this threshold, then the next state or synchronous output could be stored incorrectly; the circuit may even malfunction. To avoid this, the clock period (tp) must be greater than the sum of the set-up time of the flip-flops and the clock to output time (ts + teo). This determines the minimum clock period and hence the maximum clock frequency, fMAX, of the circuit. Metastability and erroneous system operation may occur if these specifications are violated. The timing relationships are shown in Figure 18. In each cycle there are two regions: the stable region, when all signals are steady, and the transition region, when the machine is changing state and Signals are unstable. The active clock edge causes the flip-flops to load the value of the new state that has been set up at their inputs. State Machine DeSign -t AMD~ Clock State Input Registered Moore Input Registered Mealy Output Asynchronous Moore Output Asynchronous Mealy Output On+2 __~____-J)(~_________ OT!n__________ State n J)(~____~r-o__ n_+_1________ ~)(~ __T-0 __ n_+_2____ State n + 2 State n + 1 101730-71 Figure 17. State Machine Timing Diagram Clock ~------tp ------~~---- Input Registered Output Asynchronous Output State X n ~D~ X X n+ 1 X n+2 )C X n+3 101730-72 Figure 18. Timing Diagram for Maximum Operating Frequency State Machine Design 5-69 ~AMD At a time after this, the present state and output flip-flop outputs will start to change to their new values. After a time has elapsed, the slowest flip-flop output will be stable at its new value. Ignoring input changes for the moment, the changes in the state register cause the combinatorial logic to start generating new values for the asynchronous outputs and the inputs to the flipflops. If the propagation delay of the logic is tpo, then the stable period will start at a time equal to the sum of the maximum values of teo, and tpo. Asynchronous Inputs The timing of the inputs to an asynchronous state machine is often beyond the control of the designer and may be random, such as sensor or keyboard inputs, or they may come from another synchronous system that has an unrelated clock. In either case no assumptions can be made about the times when inputs can or cannot 00 Clock arrive. This fact causes reliability problems that cannot be completely eliminated, but only reduced to acceptable levels. Figure 19 shows two possible transitions from state "51" (code 00) either back to itself, orto state "52" (code 11). Which transition is taken depends on input variable "A" which is asynchronous to the clock. The transition function logic for both state bits B1 and B2 include this input. The input A can appear in any part of the clock cycle. For the flip-flops to function correctly, the logic for B1 and B2 must stabilize correctly before the clock. The input should be stable in a window ts (setup time) before the clock and tH (hold time) after the clock. If the input changes within this window, both the flip-flops may not switch, causing the sequence to jump to states 01 or 10 which are both undefined transitions. This type of erro~ neous behavior is called an input race. _/ Input 11 81 82 101730-73 Figure 19. Asynchronous Input Cascading Race 5-70 State Machine Design AMD~ A solution to this problem is to change the state assignment so that only one state variable depends on the asynchronous input. Thus, the 11 code must be changed to 01 or 10. Now, with only one unsynchronized flip-flop input, either the input occurs in time to cause the transition, or it does not, in which case no transition occurs. In the case of a late input, the machine will respond to it one cycle later, provided that the input is of sufficient duration. There is still the possibility of an input change violating the setup time of the internal flip-flop, driving it into a metastable state. This can produce system failures that can be minimized, but never eliminated. The same problem arises when outputs depend on an asynchronous input. Very little can be done to handle asynchronous inputs without severely constraining the design of the state machine. The only way to have complete freedom in the use of inputs is to convert them into synchronous inputs. This can be done by allocating a flip-flop to each input as shown in Figure 20. These synchronizing flip-flops are clocked by the sequencer clock, and may even be the sequencer's own internal flip-flops. This method is not foolproof, but significantly reduces the chance of metastability occurring. Functionality The functionality of different devices is difficult to compare since different device architectures are available. The number of registers in a device det~rmines the number of state combinations possible. However, all the possible state combinations are not necessarily usable, since other device constraints may be reached. The numberof registers does give an idea of the functionality achievable in a device. Other functionality measures include the number of product terms and type of flip-flop. One device may be stronger than another in one of these measures, but overall may be less useful due to other shortcomings. Choosing the best device involves both skill and experience. In order to give an idea of device functionality, we will consider each of the architecture options available to the designer and evaluate its functionality. Ao - I Input Register A f-- Combinatorial Logic rI' L..-- Clock ~ --- Output Register So r-. A State Register ~ /'\.. 1 10173D-74 Figure 20. Input Synchronizing Register PAL Devices as Sequencers A vast majority of state machine deSigns are implemented with PAL devices. Early versions of software required the user to manually write the sum-of-products Boolean equations for using PAL devices. Second generation software allows one to specify the design in "state machine syntax," and handles the translation to sum-of-products logic automatically. PAL devices implement the output and transition functions in sum-ofproducts form through a user-programmable AND array and a fixed OR array. PAL devices deliver the fastest speed of any sequencer and are ideally suited for simple control applications characterized by few input and output signals interacting within a dedicated controller in a sequential manner. The number of flip-flops in a typical PAL device range from 8 to 12, which offer potentially more than one thousand state values. Since some of the flip-flops are used for outputs, and the number of product terms is limited, the usable numberof states is reduced drastically. Generally, up to about 35 states can be utilized. State Machine Design 5·71 ~AMD PAL Device Flip-Flops PAL device based sequencers implement small state machine designs, which have a relatively large number of output transitions. Since the output registers change with most state transitions, they can be used simultaneously as state registers, once the state values are carefully selected. Most PAL devices are used for sm.all state machines, and efficiently share the same register for output and state functions. High-functionality PAL device based sequencers provide dedicated buried state registers when sharing is difficult. As a state machine traverses from one state to another, every output either makes a transition (changes logic level) or holds (stays at the same logic level). Small state machine designs require relatively more transitions and fewer holds. As designs get larger, state machines statistically require relatively fewer transitions and more holds. 5-72 Most PAL devices provide D-type output registers. Dtype flip-flops use up product terms only for active transitions from logiC LOW to HIGH level, and for holds for logic HIGH level only. J-K, S-R, andT-type flip-flops use up product terms for both LOW-to-HIGH and HIGH-toLOW transitions, but eliminate hold terms. Thus, D-type flip-flops are more efficient for small state. ~achine ~e signs. Some PAL devices offer the capability of .conflguring the flip-flops as J-K, S-R or T-types, which are more efficient for large state machine designs since they require no hold terms. Many examples of PAL-device-based sequencers can be found in system time base functions, special counters, interrupt controllers, and certain types of video display hardware. PAL devices are produced in a variety of technologies for multiple applications, and provide a broad range of speed-power options. State Machine Design ~ Testability Advanced Micro Devices INTRODUCTION With digital logic design, it is all too easy to design a circuit which merely implements a specified function. When production starts it is suddenly found that the circuit cannot be tested, or perhaps that tests cannot be performed economically. Dealing with this situation can, at the very least, have a negative impact on the introduction of the system into the marketplace. Potential headache can be avoided by taking test issues into consideration during the initial design. Instead of just designing a circuit which implements a specified function, which is the bare minimum that must be accomplished, that function needs to be implemented in a manner which can be tested. The purpose of this section is to establish the notion of testability and its importance, and then to provide ways of avoiding the most common untestable circuits. The issues will be discussed primarily in the context of logic design in PLD's, although they are also relevant for general logic design. In addition, test vectors will be reviewed. Various kinds of vectors are mentioned, and the general tools available for vector generation will be summarized. Defining Testability - These are, of course, the age-old issues of controllability and observability, which are as important for digital logic circuits as they are for so many other kinds of systems. If any portion of a circuit is uncontrollable or unobservable, then the testability of the entire circuit is compromised. Figure 1 shows a couple of completely untestable circuits. The integrity of the top input in Figure 1 a can never be verified. No matter whether it is shorted to ground, to V ,or whether it is functioning correctly, the output will be the sa~e. That is to say, any faults on the top input cannot be observed at the output. The circuit in Figure 1 a would appear pretty useless as is. It is possible, however, that instead of being directly grounded, the second input may be driven by some distant signal, possibly on a different PC board, which happens to be a a logic low. If you cannot bring this line to a logic high, then it might as well be grounded. The circuit in Figure 1b essentially has no input. This circuit can be thought of as a latch, but there is no way to change its logic state. Therefore, it is completely uncontrollable. A Qualitative Look A completely testable design is one in which any and all device faults can be systematically detected. a. Unobservable First note that the issue is one of devices, not designs. The design itself must work as specified; that is the main job of the design engineer. Once the design is implemented in a device, the issue is how to test the device to make sure that the design has been correctly implemented. Throughout this paper, then, it will be assumed that a particular design works as is; we will just be addressing its testability. The easiest and most effective means of testing a circuit is through a systematic series of tests. A random set of tests may also do well, but does not yield much information regarding the testability of a circuit itself. No number of random (or systematic) vectors can test an inherently untestable circuit. In order to be able to perform a systematic test sequence, every part of the circuit under test must be accessible, so that it can be controlled. Only then can each node be forced high or low as needed. This is essentially a requirement of complete control/ability of the circuit. In order to be able to detect faults every part of the circuit must also be visible to the outside world, so that the results of each test can be observed. In this manner, each node can be inspected to determine its logic level. This requires complete observability. Publication# 14099 Rev. A Issue Date: February 1992 'Amendment/O b. Uncontrollable 14099-001A Figure 1_ Untestable Circuits Quantifying Testability In theory, if we want to quantify the testability of agiven circuit, we might first attempt to make a list of all possible things that could go wrong with a circuit (no matter how unlikely), and then verify that all such "faults" can be tested, in all combinations and permutations. But for a circuit of any significance whatsoever, it will rapidly become apparent that this is not a practical solution. What we need instead is a measure which can give an empirically reliable indication of the testability of a circuit, or of the quality of a given set of tests. There are several different such measures but the most popular of these is the single stuck-at faults model: Testability 5-73 ~AMD There are several ways of analyzing circuits for single stuck-at faults. For very large circuits, various testability analysis schemes have been developed. However, for smaller circuits, especially of the size that would be put into a PLD, the more common method uses simulation. Simulating Single Stuck-At Faults A given circuit is first simulated. The quality of the simulation is important; the more complete the simulation the better. A thorough simulation can then serve as a benchmark test sequence later. In this way, the fault simulation procedure also allows us to measure the quality of a given simulation, or set of tests, in addition to the testability of the circuit. The results of the simulation are recorded. Next, one node in the circuit is modeled with a "stuck-at" fault - either stuck-at-one (SA 1) or stuck-at-zero (SAO), as shown in Figure 2. The circuit is now resimulated. H the simulation results of the modified circuit are different from the simulation results of the good circuit, then the fault was detected. If not, then we have a faulty circuit which appears to operate correctly. Needless to say, this process of analyzing the testability of a circuit is not done all by hand; software aids are used. There are many different kinds of programs that run on many different kinds 'Of systems, ranging from pes to workstations to mainframes. Some of them are standalone programs; others are integrated into larger overall environments. Their specific capabilities also vary, but in general, they can simulate a given circuit with a given set of vectors; analyze the test coverage that the vectors provide for the circuit; and generate new tests, either from scratch or by improving on the coverage of a few manually generated "seed" vectors. Most can also point out potential problems areas of a circuit, such as race conditions and logic hazards. STUCK-AT-ONE (SAl) Finally, one frequently asked question is "So what if there is a fault that can never be detected. Who cares?" Theoretically, this question is not unreasonable. However, most companies will not feel comfortable telling a customer "We only tested haH of the system, but if anything goes wrong with the otherhaH, you'll never notice it." In addition, as will be seen, many untestable circuits occur as a result of poor design practices. STUCK-AT-ZERO (SAO) Testability issues for sequential circuits have implications far beyond the test bed. Indeed, failure to take these issues into account can greatly affect the normal performance of a system. The key for state machines is controllability. The challenge is to make all elements of the circuit controllable, both for testing and for general functionality. 14099-002A Designing Testable Combinatorial Circuits Figure 2. Single "Stuck-At" Faults This procedure is repeated for each node, one node at a time (hence the name "single" stuck-at faults). The nodes are modeled with both SA1 and SAO faults, so that for N nodes, we will have 2N simulations. If of those 2N simulations, D of them produced simulation results different from those of the original circuit, then we say that this simulation tested this circuit with a test coverage of D/(2N)*100%. Whereas this specifically tests only for single faults, experience shows that it is also a good test for mUltiple stuck-at faults. Undetected Faults Why are some of the faults not detected? For simple combinatorial logic, there are two basic reasons: either the simulation was not complete enough to find the fault, or the circuit itself cannot be tested for the fault. So when an undetected fault is located, the first step taken is to add vectors to the simulations which will exercise the node being tested. By doing this, we gradually improve the quality of the simulation, and thus the quality of the test sequence that we can use in production. 5-74 It is possible that certain nodes will have undetectable faults for which no new vectors can be added. These are the result of an untestable design. It is the joint job of the test and design engineers to generate a test sequence that is as complete as possible. It is the design engineer's responsibility to provide a circuit which is testable. If both of these responsibilities are carried out, the result will be a testable circuit which can be tested with an exhaustive test sequence. This will yield the highest quality system. Note, however, that the overall responsibility is shared between the design and test engineers. All of the previous procedures dealt mostly with the ways in which existing circuits are treated. However, if a finished circuit is found to be untestable, then it must be redesigned for testability. An easier approach is to design for testability from the beginning. Unfortunately there is no direct recipe for a testable design. There are, however, many common ways of making a circuit untestable. Most of this section is devoted to pointing out such problems. The simplest kind of problem is redundant logic. Figure 3a shows one such circuit. It has a purely redundant product term. H the output of either of the product terms is stuck low, for any reason, then as long as the other product term is good, the fault will never be visible at the output. This may initially look like a benefit, since we have what we could call a "primary" circuit with a "backup." One can cover up some of the failures of the other (but not all failures). If this kind redundancy is truly desired, this is not the way to achieve it. When you ship out thi$ circuit, you do not know if you really have a working primary and backup. The primary may already be malfunctioning; since it was never tested, you will never know. If you want useful, reliable redundancy, test circuitry must be added, as in Figure 3b, so that each part of the circuit can be independently tested. Testability of AMol1 Figure 4 shows another redundant circuit. Although the product terms are not identical, the larger AND gate is really redundant. Any stuck-low faults at the output of this gate are not detectable. ~~"B.A'B'A" FANS OUT RECONVERGES 14099-OO5A Figure 5" Reconvergent Fanout a. A Purely Redundant Circuit A one condition. In other words, there is no way that we can guarantee that,that node is operating correctly. BACKUP It is worth analyzing this circuit a bit more closely. This will give some insight into the kinds of analyses that are necessary when evaluating circuits and generating tests, and into the ways in which untestable nodes are created. PRIMAR~~ = A"B"PRIMARY + A"B"BACKUP A b. Testable Redundant Circuit 8---+--I 14099-003A C Figure 3. Making Redundancy Testable 14099-00SA ReconvergentFanout Figure 6. A Reconvergent Circuit with an Untestable Node Redundant logic is a special case of what is called reconvergent fanout. This is a term that refers to circuits that have inputs If we wish to prove that the node in question is not stuck high, then we must force it low and prove that we were successful in doing so. Thus we have two requirements: forcing the node low, and seeing the logic low on the output -controlling and observing the node. C ~~ C'D'E. D'E. D'E splitting up, going through independent logic paths, and then reconverging to form a single output, as shown in Figure 5. When this happens, it is very easy to introduce untestable nodes. It may not be easy to identify where such nodes are. First we raise input C high to force the node to a logic low condition, as in Figure la. This satisfies our controllability requirement. Next we need to provide a way to propagate this logic low to the output (Figure lb). This is referred to as sensitizing a path to the output. The first step is to get the logic low past the AND gate. But if either input A or B is low, then the output of the AND gate will be low regardless of the node being tested. Thus we must force both A and B to a logic high, so that if there is a low on the output of the AND gate, we will know for sure that it came from the node we are testing. This is shown in Figure lc. Figure 6 is an example of a reconvergent circuit. The inputs are shared between two different product terms, which are eventually summed. This circuit appears harmless enough, but it turns out that the node indicated by "SA 1" cannot be tested for a stuck-at- Next we wish to get the logic low through the OR gate to the output. To do this, we must insure that the second OR input is always low; if it is high, then the output of the OR gate will be high regardless of the node being tested. If we can keepthe lower OR 14099-004A Figure 4. Circuit with a Redundant 3-input AND Gate A =1 B = 1---+_-I C =1 a. Controllability: Forcing the Node Low c. Propagating Past the AND Gate =1 8 = 1---+--1 A C = 1 b. Observability: Sensitizing a Path to the Output d. Propagating Past the OR Gate Sets Up an Impossible Condition 14099-007A Figure 7. Analyzing Testability Testability 5-75 l1AMD A input low, then if the node we are testing was sucessfully forced into a low condition, then the output will be low. Otherwise the output will be high. This can be seen in Figure 7d. -4-------, B - t.......- - - - t c How do we keep the lower OR input low? Sy making the output of the lower AND gate low, which can be done by setting one of its inputs low. However, we have already required that all of the inputs be high. Thus we have required a set of conditions that cannot be met. One of three things will result: 1. The lower AND gate has both inputs high, and therefore keeps the lower OR input high. In this case, we may have been successful in forcing the node under test low, but we cannot see it at the output. 2. We bring input S low, allowing the lower OR input to go low. However, now the output ofthe upper AND gate will always be low. So we will see a low at the output, but we cannot be sure exactly where the low came from. 3. We bring input C low, allowing the lower OR input to go low. However, now we are no longer forcing the node under test low. So we can either force the node low, but cannot see the low at the output; or, we can see a low at the output but cannot be sure of its source; or, we cannot force the node itself low. In any case, we will never be able to guarantee that the node under test is not stuck high. Note that the two "independent logic blocks" which generate the signals that eventually reconverge are testable by themselves; they are just AND gates. It is only when we hook them together via the OR gate that the overall circuit becomes untestable. Thus Figure 9* A Messy Reconvergent Circuit addition to this, there is again a stuck-at-one fault that cannot be tested. Circuits like this can result from the design iteration process, as a designer tries to debug a circuit. Sy adding this and that, eventually the circuit works. Sut it is a mess, has poor timing characteristics, and is untestable. A little analysis of the logic itseH shows that: the bottom output is (A+ B) = A*S thus the middle output is (A"B) =A+ B which makes the top output the testability of individual portions of a circuit does not guarantee that the entire circuit will be testable when the testable pieces are all connected. (A*S*C + C*(A + B)) = (A*S*C + A*S*C) = (A"B) =A+B We can minimize this circuit using the following steps: A*S"c + S*C = A*Soc + S*C + A*S*S (by consensus) = A*Soc + S*C + A*S = A*S + S*C _Thus the node we were trying to test is really not needed in the logic. The resultant circuit is shown in Figure 8, and is completely testable. Not all reconvergent circuits are so simple. Figure 9 shows a more complicated reconvergent circuit. Here some signals have to travel through several levels of logic to reach their final destination. This introduces considerable skew into the circuit, and will produce glitches on the outputs during certain transitions. In A S ~. =A*S + S*C That is, the top two outputs are actually the same, and the third output is just the inverse of the top two. As convoluted as the original circuit looks, the logic itseH is actually trivial. So if three outputs are really needed for some reason, we can generatethem independently, as in Figure 10a. If only two outputs are needed, it is even easier. Figures 1Db and 1Dc show two possibilities. These circuits are much easier to understand, their timing characteristics are better, and they are completely testable. The Importance of Minimization The common factor behind all of the untestable circuits we have examined is the fact that all of them were not minimal. Sy minimizing the logic, we made the circuits testable. This is true in general: UNMINIMIZED LOGIC CANNOT BE FULL Y TESTED. C 14099-00BA Figure 8. The Minimized Circuit is Testable 5-76 Testability AMOl1 A_----t""""\ B -+-+----L..-/ = I(A'B) x= B A'B + IB'A = I(A'B) = I(/A +/8) a. A Glitchy Circuit A a. A Cleaner 3·0utput Version B :I-~_/(A.BJ ~_/(IA'I8J C b. A Clean. Fast 2-0utput Version u x b. Waveform for the Glitchy Circuit A C B 00 01 11 10 c. A Slower 2-0utput Version. 14099-010A Figure 10. Simplifying the Circuit of Figure 9. c. "Gap' in the Karnaugh Map Indicates a Logic Hazard 'ery often, especially when designing with PLDs, an attempt is lade to minimize logic only to the point where it fits into a articular PLD. Any further minimization is considered an acaemic waste of time. This is a grave misconception. Getting rid of II extra product terms, and eliminating all extra literals on the 3maining product terms has real value. Failing to do so will result 1 untestable nodes in the circuit. ~inimizing is not always enjoyable, since hand techniques are sually too tedious, and Karnaugh maps are essentially useless )r more than four or five inputs. However, computers have long ,een used to minimize logic. In particular, PALASM® software lersion 2.22 and later) has a minimization routine which can linimize logic automatically before assembly. .ogic Hazards )ne occasional side effect of minimization can be the introduction If glitches into a circuit. Figure 11 a shows such a "glitchy" circuit. -he waveform in Figure 11 b shows that under steady-state onditions, as long as inputs A and C are high. the output is high egardless of B. However, as Bchanges from high to low, causing 1e top product term to shut off and the bottom one to turn on. the werter adds a bit of delay to the path that will turn on the lower Iroduct term. Thus the top term may shut off before the bottom Ine gets a chance to turn on. In this case, we have two logic low ignals going into the OR gate, giving a Iowan the output. As soon lS the lower product term turns on, the output goes back high, but 14099-011A Figure 11. Examining a Glitchy Circuit not before the appearance of the high-low-high glitch. Figure 11 c shows the Karnaugh map for this circuit. It is minimal, but there are two product terms which do not overlap; they are "adjacent" in one location. These represent the two AND gates in the circuit diagram. The arrows indicate the troublesome transition: when A and C are high, and when B changes from high to low or the reverse. We can intuitively think of this as a "gap" between the two adjacent product terms, in which a glitch may occur. Note that glitching is not a certainty. It is called a hazard because in certain situation, given certain timing situations, there is a chance that a glitch will occur . Note also that the glitch is not really caused by the minimization process itself, but is caused by these "gaps" in the Karnaugh map. Unminimized logic with such gaps may also be glitchy. A PROM is a good example of such a circuit. PROMs can be used to implement any logic function of their inputs. However, regardless of the function, it is implemented in acompletely unminimized fashion, using complete minterms. So even a function as simple as the one in Figure 12 (which could be implemented using a single product term, grouping all 1's into a single cell) is implemented with each 1 in its own cell. Thus there is a gap between every cell, meaning that every transition is a potential glitch. Testability s-n ~AMD PROMs are notoriously glitchy, and it is for this reason that the output of a PROM is actually undefined until its access time has elapsed. If we go back tothe Karnaugh mapin Figure 11c, we see that we can eliminate the gap-and the glitch-by adding aproductterm which overlaps both existing product terms and covers the gap. This is shown in Figure 13a, with the resultant circuit shown in Figure 13b. input. But if we isolate just that portion which corresponds to th test input being high, which is the normal operating mode (se Figure 14c), it looks exactly like the map of Figure 13a. Of cours we should expect this, since we do not want the addition of ate! circuit to affect the basic function. Thus, in general, these types of glitches can be eliminated first b adding some redundant logic to get rid of the gaps in th Karnaugh map, and then by adding a test input to make the circu testable. This circuit is no longer glitchy. Unfortunately, it is also no longer testable, since we hxave added in a redundant product term that Z Yoo W 00 0 01 0 11 0 10 0 01 11 CD CD CD CD CD Q) CD CD 10 0 0 ~mP-- • 0 0 14099-012A TEST ~ = A'B'ITEST : !!;oITEST a. A Testable. Glitch-Free Circuit Figure 12. In a PROM, Every Transition Can Glitch cannot be tested (try it yourself). In order to have a circuit that is both testable and glitch·free, we must add a test input to the circuit which we can use to shut off the outside gates, isolating the middle gate for testing (Figure 14a). When the circuit is operating normally, the extra input is kept at a logic high condition, where it does not interfere with the basic logic function. A C B 00 A C TE ST Boo 00 0 01 01 0 0 0 11 0 0 V1 1" 10 l' 0 1 ~.L J 0 11 1\.1....1 10 0 0 b. Karnaugh Map a. A Redundant Product Term Can Eliminate the Glitch '~l'~X_AO. . . C + IB'C + A'C b. A Glitch·Free. but Untestable Circuit c. Karnaugh Map Showing Non-Test-Mode Portion 14099-013A Figure 13. Eliminating Glitches Figure 14. Making a Glltch·Free Circuit Testable The Karnaugh map for this circuit is shown in Figure 14b. Note that all product terms overlap, but now the circuit is minimal. The size of the Karnaugh map has doubled, since we added another 5-78 14099-014, Testability AMD~ Using Output Enable Most state machine PLDs are equipped with an enable pin for disabling the outputs. This is a key feature when the circuit board is to be tested in a bed-of-nails tester. When the devices driven by by the PLD are tested, it is recommended that the PLD be disabled so that there is no output level contention. Since the enable pin is usually grounded to keep outputs permanently enabled, it can instead be made available for use during testing. Note that for combinatorial devices, there is generally no output enable pin. The disabling feature is instead implemented through a product term. This feature is called programmable three-state. Designing the part such that the outputs can be disabled during bed-of-nails testing is also encouraged for these combinatorial designs. The user must be especially aware of the observability of outputs with programmable output three-state. In Figure 15, input B controls both the basic circuit logic and the three-state control logic. Therefore, any function which involves B in a LOW state will not be observable, since the output will not be on. Figure 16a.is a simplified representation of a register whose output cannot be observed because the three-state buffer is disabled when the output is LOW. Likewise, the circuitry in Figure 16b cannot be observed when the flip-flop output is HIGH. The user must make sure that an output will not be disabled when the results of a test are to be observed. A------------r-~ B J-----------~ a. LOW state observable c---...---t~ D b. HIGH state unobservable 10483A-171 14099-015A Figure 16. Untestable registered output with programmable three-state Figure 15. Untestable combinatorial circuit with programmable three-state Testability 5-79 ~AMD Designing Testable Sequential Circuits Latches The design of sequential circuits involves considerations above and beyond those required for simple combinatorial circuits. Latches and oscillators are circuits which appear combinatorial, but which use feedback to introduce sequential properties. State machines use flip-flops and feedback to generate what can be complex sequential circuits. A combinatorial logic circuit which uses positive feedback is a latch. The simplest possible latch is shown in Figure 19a. The_ output is fed back as an input in its TRUE form. This means, of course, that the output will stay at its present level; hence the name "latch." Feedback Whereas combinatorial circuits depend only on the conditions of present inputs, sequential circuits depend on both present conditions and past behavior to determine future behavior. This is made possible primarily by feedback. Feedback takes an output signal and routes it back for use as an input to the same circuit, as shown in Figure 17. We now have a situation where an output depends on itself; this can introduce new testability problems. a. Completely Uncontrollable b. Cannot Set Output HIGH 14099-016A Figure 17. Logic with Feedback Most sequential circuits (under varying circumstances also called state machines, finite state machines, and sequencers) make use of flip-flops as memory elements. These memory elements serveto remembera past condition (called a state) sothat afuture decision can be made based on it. This state is then fed back as input. With PLDs, the flip-flops and combinatorial logic are contained within a single device, as shown in Figure 18. c. Cannot Reset Output LOW 14099-018A Figure 19. Uncontrollable Latches The circuit as shown is clearly not useful, since it will always remain in its power-up state. If another input is added, as in Figure 19b, a HIGH output could be made to go LOW by setting the RESET input LOW. However, once the output goes LOW, there is no way to make it go HIGH again. Likewise, the circuit could be modified as in Figure 19c. Now a LOW output can be made HIGH by setting the SET input HIGH. However, once HIGH, the output can never be made to go back LOW. FLlPFLOPS 14099-017A Figure 18. Structure of a Sequential PLD Of course, the effects of feedback may have to be considered even when there are no flip-flops. The circuit in Figure 17 has feedback, but has no flip-flops. Such a circuit will either function as a latch or as an oscillator, as will be seen. Controllable latches For a latch to be useful, it must be completely controllable. The previous latches cannot be completely controlled. In order for a latch to be controllable, it must have both SET and RESET controls, as shown in Figure 20. Before we look into the special needs of circuits with feedback, bear in mind that all of the testability criteria discussed for combinatorial logic still hold. The blocks of combinatorial logic shown in Figures 17 and 18 must be testable by themselves. What we will discuss here are issues which must be considered in addition to the issues involving combinatorial logic. 5-80 Testability SET~ RESET~ 14099-019A Figure 20. A Controllable Latch AMD~ In PLDs, a latch can be detected by simplifying the logic for each function. If an output is a function of itself in TRUE form, then it is a latch. To be controllable, A B • productterms containing the feedback should have at least one other direct input in the product (providing RESET control). C • there should be at least one product term with no feedback (providing SET control). o x The circuit in Figure 21a provides an example. At first it is not immediately obvious that the circuit is a latch, but when the logic is simplified, we see that indeed it is. It is controllable since it has both SET and RESET controls. H the logic were as shown in Figures 21 b or21 c, the latch would be uncontrollable under some circumstances. = A + B'Y =A Latch hazards + B'(C + O'X) .. A + B'C The circuit of Figure 20 can be generalized to have several inputs on both the set and reset controls. Such a circuit is shown in Figure 22. In this case, we have two inputs on the set AND gate. If the two set inputs A and B change from 0 and 1 to 1 and 0, respectively, then there will be a glitch or a false latch at the output if both inputs were 1 at some time during the transition (Figure 22). For this transition, it is important to make sure that the 1-0 transition be made before the 0-1 transition to avoid anomalous output behavior. Merely delaying one input will not help, since it will delay both rising and falling transitions. ] SET + B'O'X LJ RESET 8. latch with SET and RESET B The simplest solution to this problem is the use of an edge- o x = B'Y a. Circuit + B'O'X L.J RESET A b. latch with RESET Only B A C B C X x= L....._ _ _ _~_ - - - ____ ... EXPECTED b. Glitch and False Latch A + Y = A + S'C + X 14099-021A ] SET Figure 22. A Latch with More Complex SET logic c. Latch with SET Only 14099-020A Figure 21. More Complex Latches triggered flip-flop to synchronize the signals. This will eliminate any such glitches. If a flip-flop cannot be used, it is possible to delay reaction to a "11 n condition to make sure that such a condition is not transitory. A circuit that accomplishes this is This delay circuit will delay the effect of an "11 n input by an extra Testability 5-81 Ir1 AMD shown in Figure 23a. This is relatively efficient in that only one delay circuit is required regardless of the number of inputs used on the set con~rol (within the limits of the size of the AND gate). It will require an extra output on a PAL device. DELAY GATE A B Because we have introduced redundancy, the circuit must be modified to be testable. If the circuit is implemented in a combinatorial PAL device, then programmable three-state can be used to test the circuit, as shown in Figure 23b. By enabling output X, the redundant circuit can be observed without regard to V. Then, to test V, output X is disabled and then the pin is used as an input to drive the circuitry for V directly. This provides a simple means of testing the circuit, but it only works if pin X can be measured and driven. The complete circuit is shown in Figure 24a. Hnode X is not so accessible, then additional circuitry and test inputs must be added. In the worst case, if node X is completely --+-f--4--~_ --+-~--~ N - ......- - - - 1 TEST - - - - - - I SET CONTROL AND GATE a. Circuit Which Delays "11 ... 1" signals TEST ~ ==±:!===:J ~ ------I C-----r-~--_/.:..-- Y x a. Complete Latch Circuit ~ =::t::!::=:=t j J - - - - - Y To control X _TEST1 - - - - - - - - - - , independently ,. of A and B. ~TEST2 _-;::====~ b. Testable Delay Circuit 14099-022A Figure 23. Delay Circuit TEST3 --+--l-I--~ propagation delay. However, it also provides a window of one propagation delay which will screen out any transitory "11" conditions that occur within that window. This allows up to one propagation delay's worth of skew between inputs during a transition from "01" to "10." /'--jr.==:t=~==f-- To MUX X to output" directly '" A B - .........+--~ TEST4 C Y ----~ b. Circuit if Node X is Completely Inaccessible A B F C Y c. Latch Circuit Behavior Figure 24. A Testable Glitch-Free Latch 5-82 Testability 14099-023A AMD~ inaccessible, the resulting testable circuit is shown in Figure 24b. Note that although the three-state capability is not needed, the circuit requires two extra gates, and, worst of all, four test inputs. an oscillator out of standard logic or PLDs will not yield a very predictable, accurate oscillator; where these circuits occur, it is usually by accident. Figure 24c shows the behavior of either of the testable glitch-free latches. An oscillatory circuit may not always be obvious. It also may not oscillate all of the time. The oscillator shown in Figure 26 is uncontrollable; it always oscillates. However, just as we can design controllable latches, we can also design controllable oscillators (on purpose or by accident). This means that there may be an oscillator hidden in the circuit which will sometimes oscillate and sometimes be stable. Such a circuit is shown in Figure 27a. Transparent latches Many designers like to use PLDs to design standard D-type ''transparent" latches. A D-type latch is a very simple circuit, shown in basic form in Figure 25a. As it turns out, however, this is a glitchy circuit of the type discussed in the combinatorial section. The problem is compounded in this case, since, given the right timing, the glitch can actually be latched; the glitching problem is no longer transitory. If this type of circuit is desired, it must be designed to be both glitch-free and testable; the resultant circuit is shown in Figure 25b. ---;===[)------=r- x == A"S"Z A"S"D"E + A*S"C"/x )------+--if--- Y ---""':::;--L../ = C"/X + + + DATA---------r~ GATE OUT Z IA·C IS"C C"/O"IV C"E·IV = D"E + Y = a. Complete Circuit D"E +/A*C + IS"C + C"/Z OUT = GATE"DATA + IGATE"OUT x= A*S"O"E + A*S"C"/X a. Glitchy TERM 1 TERM 1 b. The Equation for X 14099-026A Figure 27. A Conditional Oscillator Detecting oscillators The oscillator in the circuit is not obvious. But if we simplify the logic completely, we can see that output X depends on IX; output Y depends on N; and output Z depends on IZ. Since the outputs are fed back to themselves in COMPLEMENT form, the circuit constitutes an oscillator. OUT = GATE"DATA*ITEST + IGA TE"OUT*/TEST + DATA"OUT b. Glitch·Free and Testable 14099-024A Figure 25. 0-Type Transparent Latches Oscillators Circuits whose outputs are fed back in TRUE form are latches. If the outputs are fed back in COMPLEMENT form, then the circuit is an oscillator. A simple oscillator circuit is shown in Figure 26. 14099-02SA Figure 26. A Simple Oscillator Latches are very often useful in circuits; oscillators rarely are. Crystals and other specialized oscillators are useful when it is necessary to generate a clock signal, for example. Trying to build This circuit will sometimes be stable. If we examine the logic function determining X, we see that it has two product terms, shown in Figure 27b. Term 1 is independent of IX; term 2 is dependent on IX. If inputs A, B, D, and E are all TRUE, then term 1 becomes TRUE, and the output stays HIGH regardless of the status of the rest of the circuit. It is thus stable. However, if signals D andlor E are LOW, then term 1 will be FALSE. If, at the same time, input C is HIGH, then, as long as the output X is LOW, term 2 will be TRUE, making the output HIGH (which makes the product term FALSE, which makes the output LOW, etc.). That is, the circuit oscillates. In this manner, we can identify the conditions under which a conditional oscillator will oscillate. The mere presence of an oscillator is usually an indication that the circuit needs to be changed. It may be that the circuit only oscillates underconditions that could never possibly exist. One must be very certain of the impossibility of such a condition, however, if a conditional oscillator is to be tolerated. In addition, a thorough test sequence will usually expose a circuit to conditions that it may never encounter in a real system. Thus oscillators may interfere with the test process even if they do not disrupt the system. Testability 5-83 ~AMD Using a Programmable Clock When using the programmable clock on an asynchronous device, caution must be exercised with data setup. Refer to Figure 28a, where A and B are primary inputs. One setup time (t s) after signal A goes active, signal B goes active, clocking signal A into the register. In Figure 28b, B is a primary input but signal A is fed back from another register. In this case it may be harder to ensure that the proper setup time is allowed before signal B is asserted, possibly causing improper information to be clocked into the register. This is a simplified scenario. It does not take into account the product term on the clock, which can be programmed with a combination of any of the array inputs. A complex clock term can be a hidden source of frequently-violated setup time when feedback terms are used. Always be aware of which input or combination of inputs and feedbacks will clock each register, and calculate setup time backwards from the last input which will assert the clock term. This is the best and probably the easiest method for determining when data must be made available at the input of the register. o This is an important testability issue because with a programmable clock, the tester may no longer be in control of the clock timing. Automatic test equipment is capable of handling the timing for dedicated clock pins, but the programmable clock feature does not allow the tester the lUXUry of a single controlled clock pulse. D a A a c B A _ _ _ _ _--' B A----+-+------I ... ----------1 B-----~- c--------...J c--------+-+----b. Using feedback to drive the clock a. Using an input to drive the clock 14099-027A Figure 28. Using a programmable clock 5-84 Testability AMD~ Designing Testable State Machines input pin and potentially one product term on some outputs; this can affect the choice of device for the design. State machines have their own set of controllability issues. These essentially boil down to the concepts of initialization and illegal states. To provide initialization in an otherwise complete design when Boolean equations are being used: State machine Initialization The nature of a state machine is that there is a well-defined sequence of states through which the machine will traverse as it operates. This implies the existenc~ of a '1irst~ state. Of co~rse, these initial states vary from design to design. One obvIous problem is the fact that many flip-flops - especially older varieties - do not power up in a predictable state. Power-up Initialization Flip-flops that truly power up into a random state must be initialized explicitly. Lately, however, flip-flops have become available which have "power-up reset". This allows the flip-flops to power up into a predictable state every time. This is helpful when the power-up state also happens to be the initial state. But even if it is not the initial state, a predictable initialization sequence can bring the state machine into its start-up state. • determine the start-up state. • assign each bit as being initialized active or inactive, based on the desired start-up state. • if a bit is to be initialized inactive, add "'IN IT" to every product term for that bit. • if a bit is to be initialized active, add one product term consisting solely of "INIT." Here we have assumed that the initialization pin has been called "IN IT. " "Active" would mean HIGH for an active high device; LOW for an active low device. "Inactive" is just the reverse. The equation in Figure 29a can be. init~alized inacti~~ ~s s~0v.:n in Figure 29b, or active as shown In Figure 29c. Initialization IS accomplished by asserting the INIT pin and clocking once. This "cookbook" approach is very reliable. ao:= 01'02 + 02"03 Unfortunately, such initialization schemes rely on th~ ability of the device to initialize itself when being powered up. If the system needs to be re-initialized, it will have to be completely turned off and then turned on again. Anyone who has had to turn off a computer in order to reboot will know that this is not an elegant way of re-initializing. By building initialization into the design, a means of performing a "warm boot" is provided. It is for this reason that initialization must be considered along with all other aspects of the design. a. Uninitializable 00: = 01'02"INIT + 02"03"INIT b. Initialized Inactive 00:= 01'02 + 02"03 + INIT Some devices have mechanisms specifically designed for initializing a state machine. These are usually in the form of global preset and reset product terms. By programming the conditions for initialization onto such terms, the device can be re-initialized at any time. c. Initialized Active 14099-028A Figure 29. Designing In Initialization Including Initialization In a design Some of the simpler devices do not have specific provisions for initialization. However, the need is still present in these devices; here the initialization should be included in the design. This is a very simple process; it can be added in after all of the other design details have been worked out. Adding initialization will use up one PALASM software also makes it possible to design state machines with a special syntax which essentially allows the state diagram to be transferred directly into a design file. For devices which have no dedicated initialization features, the initialization branches should be explicitly built into the state diagram. The software then performs the remainder of the processing needed. Testability 5-85 ~AMD Illegal states A state machine is formed by using a set of flip-flops to remember states, and assigning a code to each state. Since there are 2" different codes that can be assigned to a group of n flip-flops, there is a good chance that some codes may not be used. For example, if a state machine is to have 6 states, 2 flip-flops will not be sufficient; 3 are needed. But 3 flip-flops allow 8 states, which will result in 2 unused states (see Figure 30). 14099-030A Figure 31. Using Initialization to Recover 14099-029A Figure 30. Illegal States Assuming that the state machine has been designed correctly, there is no reason why these extra states should ever be entered; therefore they are called "illegal" states. Unfortunately, situations do occur, thanks to noise and other unpredictable occurrences, which result in the state machine being in an illegal state. When this happens, the immediate need is to return to a normal sequence of states: there must be a predictable means of getting from any illegal states into a legal state. Illegal state recovery is a controllability issue which actually affects functionality more than it affects testability. But the concepts used for functionality and testing are so closely related that it is worth treating here. 14099-031A Figure 32. Cycling Back to a Legal State which eventually leads back to a legal state. In these cases, merely clocking enough times will cause the machine to recover. Recovering from Illegal states The drawback here is that one does not know ahead of time how many clock cycles will be needed. This necessitates some builtin way of knowing just when a legal state has been re-entered. And once that state has been reached, further cycling may be needed to get to a point where operation can resume. There are three basic ways to get out of an illegal state: Designing-In one-step recovery • re-initialize • make sure that one can continue clocking until the machine recovers • design the machine such that the start-up state is reached from any illegal state in one clock cycle, independent of any conditional inputs The most predictable way of dealing with illegal states is to provide a one-step path back to a legal state. Depending on the state desired, more or less work may be involved to do this. For PAL devices, we can consider three cases: Of course, re-initializing will take the machine back into its startup state from any state, legal or illegal (Figure 31). The disadvantage here is that outside control is needed to force initialization. Very often, a path will exist which eventually takes the state machine back into a normal sequence (Figure 32). These paths are not usually designed in; they just happen to be there. In fact, if D-type flip-flops are used, it is surprisingly difficult to get a "closed" set of illegal states (that is, a set such that once one of the illegal states is entered, the machine will forever remain in illegal states) by accident. In most cases, there will be a path 5-86 • all illegal states go to state 00 ... 0 • all illegal states go to one state other than 00 ... 0 • each illegal state goes to some legal state The cause of poor illegal state recovery can be illustrated conceptually with Karnaugh maps (although realistically, Karnaugh maps are often not used). When calculating the equations for a particular bit, it is tempting to use Don't Care cells from the Karnaugh map (Figure 33) to simplify the logic. The success of illegal state recovery depends on how these Don't Care cells are treated. Recovering Into state 00 ...0 Testability AMD~ 0102 00 01 Q3 Q4 11 10 00 X 0 X X 01 X 0 0 X - 11 0 1 X X 10 0 X X X J oJ J trated conceptually with a Karnaugh map. It must first be decided which legal state will be entered, and the resultant value of each state bit. The Don't Care cells for each bit are then filled with the corresponding next state bit value; if the next state for a bit is to be1, then Don't Care cells are filled w~h 1's for that bit's Karnaugh DONT CARE CELLS CORRESPOND TO ILLEGAL STATES 01 14099-032A Figure 33. Illegal State his is the simplest case; it is illustrated in Figure 34. It is ccomplished by not using any illegal states to generate the logic )r any of the b~s. Since most PAL devices have only Ootype flipops, a bit will go HIGH only as a result of legal states. Any illegal tates will cause all bits to be LOW. his prOCedure does not work when J-K or T-type flip-flops are sed. In fact, it is deadly. Whereas a Ootype flip-flop defaults to OW, J-K and T-type flip-flops hold their present state as a a. State Diagram 1 X X X 0 ~ X X 1 1 ~ X 0 0 0 1 b. Karnaugh Map for Bit On 18 a. State Diagram 0 0 0 ~ 0 0 Ir;- l:J' lo 0 0 0 0 CD 0 X X ~ 0) r-..! X 0 I\... X X 0 X X 0 C. oeD b. Karnaugh Map 14099-033A Bit On Recovers to, 0 (1 0 Figure 34. Recovering to State 0 ... 0 efault. Thus if illegal states are not considered in the transfer mctions, an illegal state will cause the state machine to be locked p in that state. CD 1" 1 1 1 3 1 Ie 1 1 II 1 0 0 0 1 ~ d. Bit On Recovers to 1 tecoverlng Into one fixed state 14099-034A Figure 35. Recovering to a State Other Than 0 ... 0 his case is shown in Figure 35a. The procedure can be iIIus- Testability 5-87 ~AMD map; the procedure for a O-b~ is analogous. The equations are now taken by including either all Don't Care cells if filled with 1's, or none of them if filled with O's. This procedure is illustrated in Figures 35b, c, and d. When Karnaugh maps are not used, the same result can be obtained by explic~ly considering all illegal states. When calculating the Boolean equations for: • a bit that will be 0 after recovery, included. no illegal states should be • a bit that will be 1 after recovery, all illegal states should be included. When J-Kflip-flops are used, then the transfer function for either J or K - but not both - will include all illegal states . • H a bit is to be HIGH after recovery, J should account for all illegal states; K should account for none. a. State Diagram • H a bit is to be LOW after recovery, K should account for all illegal states; J should account for none; This must be done explicitly for J-K flip-flops even if state 0... 0 is the recovery state. 1(1 1 1 1 0 0 0 1 1 "",1 1 1(1 When T-type flip-flops are used, there is no easy way out; any recovery must be explicitly designed-in as part of the original function. 0 0 ~ b. Karnaugh Map Recovering Into Any Legal State The third case allows one to fill in the Don't Care cells of a Karnaugh map in such a way that some legal next state is always reached in one clock cycle, but such that the 1's and O's are placed to keep the logic functions simple. This is shown in Figure 36. The disadvantage here is that since different illegal states result in a different legal state, some additional cycling may be required to allow operation to resume. 14099-035J Figure 36. Recovery Such That Logic Functions Are As Simple As Possible When Karnaugh maps are not used, this can be implemented more simply by explicitly including the illegal states as part of the complete state diagram. 5-88 0 Testability AMD~ ·esting Illegal state recovery )ne of the difficulties of designing illegal state recovery into a ircuit is the fact that it is difficult to test. Because the state is legal, it is impossible to force the circuit into such a state. The use If register preload circumvents this problem. With preload, any tate - legal or illegal - can be loaded into the register. If an legal state is loaded, then the circuit can be tested to verify that orrect recovery does indeed occur. o The use of preload must be considered carefully with devices having programmable asynchronous preset and reset features. If these are driven by feedback from an output, then situations can occur where preloading one state immediately causes a preset or reset to the opposite state (Figure 37). There are two alternatives: either avoid pre loading such states, or include a control input in the preset and/or reset product terms which can disable the feature when testing. o P Q P Q R Cannot Preload 1 Stable R R Cannot Preload 0 Stable Stable Case: Can preload any state Other Cases: Preloading any state will cause PRESET or RESET to opposite state. 14099-036A Figure 37. Preloading Registers with PRESET and RESET Testability 5-89 ~AMD Using Test Vectors Digital systems are generally tested by applying a sequence of test vectors. A test vector is a group of signals which are applied (forced) and measured (sensed) on a device or a board. The vector thus defines all inputs and expected outputs for a given test. As we have noted, the sequence of tests performed greatly affects the quality of the overall tests, as measured by the fault coverage. In general, we can talk in terms of three kinds of vectors. Simulation (or application) vectors, functional test vectors and signature test vectors. Simulation vectors are generated during the design process. Their main purpose is to help the designer verify that the design has been correctly implemented. They represent the way in which the circuit was intended to operate. When PALASM software (or almost any other PLD design software package) is used, simulation may be performed prior to programming a device. The software simulates the operation of the circuit, and then generates vectors from the simulation, adding the vectors to the JEDEC file. These vectors can then be used for testing by programmers that have the capability of performing functional tests. While simulation vectors may be adequate for verifying that the design is operating as expected, they generally do not provide very extensive test coverage. For this reason, we distinguish functional test vectors from simulation vectors. It is very difficult to generate a complete set of functional test vectors by hand; computer programs are generally used instead. The simulation vectors are often used as a basis for generating a more comprehensive set of functional test vectors; in this capacity, the simulation vectors serve as seed vectors. There are many programs which perform this function although many of the programs require larger computers and take a long time to run. AMD also generates functional test vectors for patterns that are used in ProPAL devices. quality. If complex internal feedback is used in a particular design, then some manual test generation may still be needed to improve the test coverage. 80th of these programs support the use of register preload for initializing states. While functional vectors provide more extensive tests, they may not exercise the circuit in the manner in which it was meant to be used. Thus, for example, a conditional oscillator in a circuit (as discussed previously) may not be a problem during simulation, since the conditions causing oscillation are not thought to be possible by the designer. However, the functional vectors will take all situations (some of which may not be physically possible) into account in the tests. Thus more subtle design problems may become apparent when functional test vectors are generated. Signature vectors are random vectors which are first applied to a device which is known to be good in order to generate a "signa· ture". This same set of vectors is then applied to a device 01 unknown quality; if the same signature results, the device is said to be good; if a different signature results, then the device is assumed to be faulty. Signature vectors can vary greatly in the quality ottesting they can provide. Since they are generated with no knowledge of the circu~ being tested, many more vectors must be used to perform a gooC test. The quality of the test depends on the circuit being tested, the number of vectors used, the speed with which the tests are applied, and the algorithm used to generate the vectors. The tester must also be able to apply a preload sequence to device~ that have registers; otherwise two devices may power up into twe different states. In that case, both devices will generate differen signatures even if both are good devices. Quality signature testing can be very cost effective, since ne advance knowledge of a device pattern is needed. This reduce! the amount of resources that must be dedicated to test vecto generation. The different types of vectors are summarized in Table 1 below Programs have been developed to generate vectors for use in testing PLDs. These programs use the programming information in the JEDEC file to generate tests. On most patterns, they can generate test sequences of high TYPE OF VECTOR PURPOSE GENERATED BY: Simulation (Application) Used for verifying whether or not a design will operate as expected when implemented. Sequence defined by the design engineer, usually by hand. Actual vectors generated by design software, placed in the JEDEC file. Functional Used for verifying that a device is operating correctly. Usually generated by a computer program. The simulation vectors can be used as seed vectors Signature Used for verifying that a device is operating correctly without functional vectors. The tester generates the test sequence during the test. Table 1. Test vectors 5-90 Testability AMD~ UMMARY This can be done by simplifying each combinatorial logic block to see whether any signal ultimately depends on itself. Ie time to start considering ways of testing a circuit is before the 'cuit has been designed. The key to testability lies in the way the 'cuit is implemented. lsic combinatorial logic can be made completely testable sim( by minimizing logic. It is not even necessary to analyze the 'cuit for redundancy or reconvergent fanout; automatically inimizing all logic will eliminate any occurrences. here a sequential circuit is generated from simple feedback lths in the logic, the circuit must be analyzed as a combinatorial ·cuit. All combinatorial logic must be included to determine lether the circuit is a latch or an oscillator. If a latch is desired, ,hould be completely controllable. If an oscillator is found, it is obably not desired, and will generally indicate a mistake in the Isign. If a conditional oscillator is to be tolerated, one must be re that the oscillation conditions can never occur, and that the st procedure will not cause oscillation. general, combinatorial circuits should be analyzed completely . the presence of latches and oscillators (wanted or unwanted). When the sequential nature of a circuit is derived through the use of flip-flops to generate a state machine, the two key issues are initialization and illegal state recovery. A combination of device features and careful circuit design will yield circuits that can behave predictably even in unexpected situations. It is important to analyze the testability of a circuit before committing it too far. Thus any changes can be made early on. In particular, if the test analysis software points out any logic hazards in your circuit, you can easily remedy them by modifying the design. These simple steps, taken early in the design phase, can help avoid later redesigns, and ultimately provide a higher quality system. Finally, the ultimate test quality depends also on the quality ofthe test sequence used for production, functional test vectors and high quality signature tests will provide you with the highest confidence in the quality of your system . Testability 5·91 ~ Ground Bounce Advanced Micro Devices INTRODUCTION Mechanism The development of fast PAL devices has increased the importance of analog considerations the digital designer has been able to overlook in the past. One of these is ground bounce. Ground bounce refers to the ringing on an output signal when one or more outputs on the same device are being switched from HIGH to LOW. This ringing can be in excess of 3 V. The system cannot consider the data valid until the ringing settles to below the VIL of the receiving devices. The ringing in a fast device can last so long that a slower device with less ground bounce could actually be a faster solution. Figure 1 shows a schematic of an output driver and load including parasitic elements. The load capacitor is charged to the HIGH-level voltage. When the transistor turns on, the capacitor discharges into the transistor and lead impedance. The resultant RLe circuit will have a damped ringing (Figure 2). The peak amplitude depends on the edge rate of the switch and the RLC values, while the frequency of the ringing and the rate of decay depend only on the RLC values. The phenomenon of ground bounce is associated with the inductance and resistance of the ground connection in the integrated circuit. As there is always some inductance and resistance, ground bounce cannot be totally eliminated; however, it can be reduced to a level tolerable to the system. This article will discuss the mechanism of ground bounce in CMOS circuitry and the utilization of slew-rate control used by AMD to keep ground bounce down to reasonable limits. The ringing caused by a single output switching is normally below the LOW-threshold voltage. However, the voltage at the ground pad of the device is proportional to the number of outputs switching simultaneously. In addition, the voltage at the ground pad is coupled to any LOW output through its output transistor. Therefore, if enough outputs switch, ringing on the ground pad will be coupled to LOW outputs, causing the detection of false HIGHs. Most PAL devices used today have relatively low output drive current: 16 mA or 24 mA. It is tempting to think that the low current level will somehow limit the switching en- Vee 2.0V 1.0 V 0.5V 1 2.0V 1 Quiescent Output 1.0 V 0.5V 13090-001A Figure 1. Simplified Schematic of an Output Driver 5-92 13090-002A Figure 2. Ground Bounce AmendmenllO AMD~ ergy and therefore ground bounce. Actually, even a lowpower transistor can pass a relatively large current. The transistor I-V curve in Figure 3a shows that a MOS transistor designed for 16 rnA at 0.5 V will pass 90 rnA at 3.0 V. Figure 3b shows the VII path when the output transistor switches between HIGH and LOW. Notice that the transistor switches from 3.5 V at 0 rnA to 3.0 V at 90 rnA. If eight outputs were to switch simultaneously, 90 rnA X 8, or 720 rnA, would flow through the ground lead. This sudden current surge is actually self-limiting. As the ground-pad voltage rises due to the high current change, the internal Vos and the available gate bias voltage are reduced, lowering the drive current. However, the ringing can still exceed 3 V. Controlled Edge Rate The parameters that influence ground bounce are the inductances and resistances of the device, the capacitance of the load, and the edge rate. Of these, the only one that the chip manufacturer can directly control is the edge rate. Turning on the output-driver transistor is equivalent to switching the charged load capaCitor to ground. This can be represented by a step-voltage source in series with the capacitor (Figure 4a). Slowing down the rate that the output transistor can tum on changes the voltage source from a step to a ramp (Figure 4b). With a 90 90 10 (rnA) 10 (rnA) 16 16 0.5 3.0 Vos (Volts) 0.5 5.0 13090-003A 3.0 5.0 Vos (Volts) 13090-OO4A 3b. The Path Followed as the Transistor Switches between the HIGH and LOW Levels 3a. The DC Curve of an Output Driver Transistor Figure 3. 13090-OO5A 4a. Equivalent Circuit of an Output Driver Transistor with a Capacitive Load 13000-OOSA 4b. Output Driver Circuit with Slew-Rate Limiting Figure 4. Ground Bounce 5-93 ~AMD shallower slope, less energy is available for ringing and the ground-bounce amplitude is reduced. A Spice simulation (Figure 5) illustrates the effect. The device without risetime control will have a very high charging current with a large di/dt: 2.1 X 107 Ns. Risetime control reduces the di/dt about 25%. This will result in a corresponding reduction in the voltage that can develop across the ground inductance. AMD has a proprietary technique that slows the edge rate of the output transistor, thereby reducing the amplitude of the ringing. Slowing down the fall time will add about a nanosecond to the output delay, but the system speed will still be greatly increased. On a high-capacitance load, a non-edge-rate-controlled device could ring for more than 25 ns. The additional delay required to allow for the ringing would be intolerable. System Ground Bounce Solutions There are some things that the system designer can do to reduce the ground bounce to a tolerable level. 1) Use AMD PAL devices that incorporate edge rate control. This the first line of defense against ground- bounce-related problems, and the most effective. 2) Use shorter lead packages. The bonding wires in a PLCC are 1/4 the length of the ground bonding wire in a DIP. The inductance is reduced proportionally. Any reduction in inductance will reduce the amplitude of the ringing. Some devices have center power and ground pins. The ground pin will be substantially shorter and have a proportionately reduced inductance. This will reduce the coupling between outputs. A good example is the PALCE26V12. 3) Reduce capacitive loading. Capacitive loading in any system should be reduced as much as possible. This may involve consideration of the transmission line characteristics of the layout. 4) Limit the number of outputs switching simultaneously. If the load naturally has high-capacitance such as a bus or memory board WOUld, ground bounce can be reduced by limiting the number of outputs that can switch simUltaneously in a single device. Many system designers consider 4 to be an acceptable upper limit. 4 60 3 2 VOUT (Volts) 1 40 I GND (rnA) 20 ·1 o ·2 1 2 3 4 5 6 7 8 9 1011 12 131415 Time (ns) Sa. Without Risetime Control 4 13090-007A 60 3 2 VOUT (Volts) 1 40 I GND (rnA) 20 -1 o ·2 Time (ns) Sb. With Risetime Control Figure S. Effect of Risetime Control S-94 Ground Bounce 13090-008A ~ Metastablility Advanced Micro Devices INTRODUCTION A significant number of digital systems must deal with inputs not synchronized to their own internal clocks. These asynchronous signals can arise from any of the various asynchronous protocols, such as are often used in bus designs; they can be the result of trying to share signals from systems with different clocks; or they may be the response of a system user, who is of course not synchronized with the system. The result can be metastability, a problem which can plague unwary designers. It is not a newly discovered phenomenon, but is normally dealt with somewhat qualitatively, and, unfortunately, is usually ignored as much as possible. Data ts Clock ...JX'--___ Causes of Metastability Outputs_ _ _ _ _ _ _ _ _ _ _ The flip-flop setup time is the parameter that is most often at the root of metastability. The setup time is basically a requirement that data be made available at the . input to the flip-flop before the clock signal arrives. The data must not only be there, but must also be stable. In a PAL device, the use of an array for the data adds to the setup time. The data passes through the array on its way to the flip-flop (Figure 1). The clock signal, on the other hand, goes directly from the clock pin to the flipflop. Its path is much shorter than the data path. The setup time is therefore essentially a requirement that the data signal must be given more time to get to the flip-flop before the clock signal. If the published setup time is satisfied, the data arrives at the flip-flop well before the clock, and the output to the flip-flop will change as desired (Figure 2). If the setup time is violated, then no guarantee can be made about what the output will do. The output may be normal, since the published setup time is a worst-case number. However, if the timing between the clock and data is just right, the output will be unstable for some time before it Figure 2. Output response when the setup time Is satisfied Data Clock _~_ _----I(VVV Outputs_ _ _ _ _ _ _ _,1\V1\vv\o... 1\ n Figure 3. Possible output response when the setup time Is violated D Q settles into some state. Neither the time the output remains unstable nor the final state is predictable (Figure 3). This condition is metastability. Ways of Dealing with Metastability Buffer Figure 1. The clock and data paths In a PAL device Amendment/O The most common way of dealing with this problem is to synchronize the inputs with an extra flip-flop (Figure 4). If the first flip-flop goes metastable, hopefully the delay between clock pulses will allow the ringing to die down 5-95 arbitration schemes, use synchronization not because synchronization itself is necessary, but because it provides the only convenient way to store data. This unfortunately takes a system that is inherently asynchronous and adds some synchronizing elements in the middle. Extra Flip-Flop for Synchronization 0 Q a D ~ Q a Summary Figure 4. Dual synchronizer before clocking into the next flip-flop. This improves the chances of having good data in the second flip-flop. This method is not without its costs. Each extra stage of flip-flop means an extra clock delay of the data which m!Jst be absorbed by the system. Moreover it is not foolproof. The possibility of metastability is reduced, but not eliminated. A flip-flop can go metastable if the preceding stage does not recover quickly enough. The best way to avoid metastability is to avoid synchronization when possible. Many applications, such as bus 5·96 Metastability can occur in a number of different kinds of asynchronous systems, usually due to the inability to guarantee that the setup time of the flip-flops will be satisfied. In standard synchronous systems, where the setup time (along with all other timing requirements) is specifically designed in, metastability will never be a problem. In some situations, metastability is caused by the need to interface systems with different clocks. In this case, it will never be possible to completely eliminate the possibilityof metastability. Instead, the designer must take steps to reduce the probability of a system failure due to metastability. Metastability ~ Latchup Advanced Micro Devices _atchup Circuit --------, .atchup is caused by an SCR (Silicon Controlled Rectifier) circuit. ~abrication of CMOS integrated circuits with bulk silicon process~g creates a parasitic SCR structure. The behavior of this SCR ~ similar in principle to a true SCA. These structures result from he multiple diffusions needed for the formation of complemenary MOS transistors in CMOS processing. The SCR structure :onsists of a four layer device formed by diffused PNPN regions. -hese four layers create parasitic bipolar transistors illustrated in :igure 1. &ih NWELL Vcc RpSUB GND 14105-001A ill 8 .. Figure 1 0 igure 2a shows a typical CMOS inverter layout with the schelatic of the parasitic bipolar SCR structure. Figure 2b is a cross ectional representation of the CMOS inverter, again with the chematic of the bipolar SCR structure. • 14105-OO2A Figure 28 Vcc Vss RpSUB 0 ACTIVE DIFFUSION - PTYPE ACTIVE DIFFUSION - N TYPE N-WELL POLYSILICON GATE METAL INTERCONNECT CONTACT PSUB 14105-OO3A Figure 2b Publication# 14105 Rev. A Issue Date: January 1990 Amendment/O Latchup 5-97 ~AMD Any CMOS diffusion can become part of the parasitic SCR structure, since all of these parts are interconnected through the bulk silicon substrate resistance. Other parasitic resistors shown result from doped regions of the semiconductor. The magnitude to which the resistors resist current flow depends upon geometric size and doping level. As illustrated in Figure 1, the complementary PNP and NPN transistors are cross-coupled, having common base-collector regions. The vertical PNP device, M1, has its base composed of the N-well diffusion while the emitter and collector are formed from P-type source-drain and substrate regions, respectively. The lateral bipolar transistor, M2, base is the P substrate with emitter and collector junctions formed from N-type source-drain and N-well diffusions, respectively. Latchup Conditions Under normal bias conditions the SCR conducts only leakage current and the SCR structure is in the blocking state. However, as current flows across any of the parasitic resistors, a voltage drop is developed, turning on the parasitic bipolar base-emitter junction. The forward bias condition of this junction allows collector current to flow in the bipolar transistor. This collector current flows across the base-emitter resistor of the complementary bipolar transistor, creating a voltage sufficient to turn on the transistor. A regenerative loop is now created between the complementary bipolar transistors such that current conduction becomes selfsustaining. Even after removal of the stimulus that triggered this action, the current conduction can continue. This region of operation is a high-current, low-resistance condition characteristic of a four layer PNPN structure. This is referred to as latch up. Once initiated, the excessive latchup current can permanently damage an integrated circuit by fusing metal lines or destroying junctions. Causes Of Latchup Latchup may be initiated in numerous ways. Just the critica causes frequently encountered in a system environment will bE discussed. These include power up, supply overvoltage, anc overshoot/undershoot at device pins. Power-Up Caution must be exercised when powering up CMOS ICs to avoic driving device pins before the supply voltage has been applied te the circuit. Placing a device or board in a "hot socket" will creatE this situation. When subjected to hot socket insertion, voltagE conditions at the device pins are uncertain such that the inpu diodes may be forward biased. Forward biasing the input diode! with a delayed or uncontrolled application of Vee could cause thE device to latch up. Advanced Micro Devices' CMOS circuits haVE substantial immunity to hot-socket power up, but since thi! condition is uncertain, and difficult to characterize, test, an< guarantee, it should be avoided. Supply Overvoltage Supply levels exceeding the absolute maximum rating can causl a CMOS circuit to latch up. Elevated supply voltage may causl internal junctions to break down, producing substrate curren capable of triggering latch up. Latchup is just one of the reason~ overvoltage should be avoided; other undesirable effects ma; result from this. Overshoot/Undershoot Generally the I/O pins experience the noisiest electrical environ ment. Fast switching signals with a large capacitive load ma' overshoot, creating a transient forward bias condition at the 1/( junction. These junction diodes are illustrated in Figures 3 and ~ Typically this is where latchup is most likely to be induced. Prope design of the input and output buffers is essential to minimize th risk of latchup due to overshoot. Vee TO INPUT GATE P + NWELL DIODE N + PSUB DIODE PSUB Vss PSUB 14105-OO4A Figure 3 5-98 14105-005 Figure 4 Latchup AMD resting For Latchup Idvanced Micro Devices characterizes the latchup sensitivity of s devices before they are released to the market. Testing is done , such a way as to completely cover every possible latchup ondition, including Vcc overvoltage, pin overcurrent, and pin vervoltage. rcc Overvoltage Test ·he Vee overvoltage test is applied to all power (Vee) pins. The 3st is performed at the highest guaranteed operating temperaJre of the device. All inputs and I/Os acting as inputs are tied to round or Vcc depending on the device logic, and outputs and V )s acting as outputs are floating (open). ree max is applied to the Vee pin. A positive high voltage pulse ; then applied to the Vee pin and returned to Vee max. The ccurence of latchup is detected if the voltage across the device ; less than Vcc max, and the current through the device is greater lan the normal DC operating current. ~ Vce depending on the device logic, and outputs and I/Os acting as outputs should be floating (open). Vee max is applied to the Vee pin. One pin is tested at a time. A three-state output under test should be disabled. A non-three-state output type under test should be a logic High when applying a positive current and a logic Low when applying a negative current. An VO pin should be placed into the input mode. A high current pulse is then applied to the pin under test. The magnitude of the pulse is stepped untillatchup is induced. Both positive and negative currents are tested. Latchup is observed as described previously. The sensitivity of the device is the worst case sensitivity found on any pin of the device. Pin Overvoltage Test The pin overvoltage test is performed on current-limited inputs. Current-limited inputs are inputs which present a resistor-like (or otherwise "limited") current characteristic for input voltages in the range (GND - 5 V) < Yin < (Vec + 5 V). 'In Overcurrent Test ·he pin overcurrent test is performed on every output, I/O pin, and on-current-limited input pin. Non-current-limited inputs are ,puts which present a diode-like (or otherwise "infinite") current haracteristic for input voltages in the range (GND - 5 V) < Yin < IIcc +5V). ·he pin overcurrent test is performed at the highest guaranteed perating temperature of the device. Input pins and I/O pins .cting as inputs (which are not under test) are tied to ground or The pin overvoltage test is performed at the highest guaranteed operating temperature of the device. Input pins and I/O pins acting as inputs (which are not under test) are tied to ground or Vcc depending on the device logic, and outputs and VOs acting as outputs are floating (open). Vee max is applied to the Vee pin. One pin is tested at a time. Both positive and negative voltage pulses are applied to the pin under test. latchup is observed as described previously. The sensitivity of the device is the worstcase sensitivity found on any pin of the device . Latchup 5-99 ~ Converting Bipolar PLD Designs to CMOS Advanced Micro Devices Application Note by Bryon Moyer The world learned about programmable logic through the use of bipolar PLDs. As PAL device designs proliferated, bipolar fuse technology was the only productionworthy vehicle for implementing the programming feature. As CMOS floating-gate technology was adapted to programmable logic, CMOS has increasingly become the technology of choice for new system designs. By using CMOS, bipolar speeds can be attained with lower power consumption. Electrical erasability joins a number of other reasons why designers now prefer CMOS. Today bipolar PLDs are being purchased largely to supply those designs that were done before CMOS was viable. Many systems makers are now looking for ways to convert their production systems to CMOS. The intent is always to have a seamless, "engineeringless" transition, or as close to that as possible. Few companies have access to the engineers that designed a system five years earlier. The purpose of this application note is to discuss issues that may arise when converting a socket from a bipolar PLD to its CMOS equivalent. • Overshoot • Ground bounce In addition, the issue of checksum consistency will be addressed where there is a chance of the checksum changing as a result of a conversion. Table 1. Bipolar/CMOS Direct Equivalents Bipolar Device The conversion issues discussed apply to devices from any manufacturer; the specific solutions apply primarily to AMD PAL devices. In particular, this article focuses on conversion where there is a direct CMOS architecture counterpart to the bipolar device, as shown in Table 1. For bipolar architectures where an exact CMOS equivalent is not available, or when using other CMOS architectures, additional logic design work may be needed. For more information on the CMOS architectures available, please refer to the application note Selecting the Correct CMOS PLD. There are a number of specific areas that need to be discussed for those designs that may have an easy architectural conversion, but more difficult electrical conversion. They are: • Floating unused input pins • Edge rates, termination, and layout 5-100 CMOS Bipolar Device PAL16R8 PAL20R8 PAL16L8 PAL20L8 PAL16R4 PAL20R4 PAL16R6 PAL20R6 Equivalent PAL10L8 PAL12L10 PAL10H8 PAL 14L8 PAL12L6 PALCE16V8 PALCE20V8 PAL16L6 PAL12H6 PAL18L4 PAL14L4 PAL14H4 PAL20L2 PAL22V10 PAL16L2 In theory, when converting from a bipolar device to a CMOS device, either of the two should work in the socket. In practice this is true for most designs, especially those that have been well adapted for high-speed signals. Most designs will not require that you take any special actions. It is only with more sensitive designs that one may have some applications issues to deal with. CMOS Equivalent PALCE22V10 PAL20RA10 PALCE20RA10 PAL16H2 Floating Unused Input Pins No input to a digital device likes to see its value kept at the threshold voltage for any length of time. These in-· puts expect a high or low signal level; the input signal should switch quickly and smoothly from one state to another, passing cleanly through the threshold voltage. If an input lingers too long at threshold, the input transistors will be in the active region, and, essentially being high-gain amplifiers, may oscillate as they decide whether they should be high or low. On a PAL device, this oscillation will not typically cause any first-order problems on unused pins. However, since this oscillation involves the rapid switching of a lot of current, it could generate internal ground noise, and affect other internal circuits. Bipolar devices tend to pull unused input pins to a high state. Many designers count on this to give their unused pins a default level, although it is not a recommended practice. Despite pull-up capability (typically 50 kn 100 kn effective), a standard TTL input will only pull up to at most a diode drop above threshold, as shown in Figure 1. In the presence of noise, the input can start to move across threshold and cause some disturbances. Publication# 1n64 Rev. A Issue Date: April 1993 Amendment/O AMD~ It is therefore always better to tie an unused pin high or low on the board. ill 17764A-3 b. No Current If Pull-Up Used 1.4 V --> 2.1 V S.ov 1n64A-1 Figure 1. A Typical Bipolar Input. All Voltages Are Nominal. 1n64A-4 When tying a pin high or low, a resistor is not needed. However, many designers feel safer putting in a currentlimiting resistor to protect the system if there is an accidental short on the pin. In addition, the resistor allows the pin t() be used later without needing to break the pu 11up or pull-down connection. It often makes no difference whether an unused input is tied high or low. However, when using a device with internal pull-up capability, tying high can save some power. If an input has a pull-up and is tied low, then up to 100 J..lA will be expended through the input. In addition, care must be taken to make sure that VIL is not exceeded when tying an unused pin low through a resistor. The external resistor will form a voltage divider with any on-chip pull-up. With a 50-kn. internal pull-up, a 10-kn. pull-down will bias the input at 0.8 V (maximum VIL) if Vee is 5.0 V. This is the maximum pull-down that should be left intact if the input has a built-in pull-up. c. Pull-Down Forms a Voltage Divider Figure 2. External Pull-Up/ Pull-Down Configurations Traditional CMOS devices have absolutely no pull-up or pull-down, so the pin is truly floating. Therefore it is completely at the mercy of leakage and noise as it seeks out some default level. Recent CMOS PAL devices from AMD have pull-up resistors that provide a default high level. The input will be pulled to about a diode drop below Vee, giving sufficient noise margin. The minimum effective pull-up resistance is 50 kQ, so it is still a high impedance input. Thus it is still a good idea to tie such an unused pin high for maximum noise immunity. The equivalent input schematic on the data sheet will indicate whether or not the product has built-in pull-up resistors. 1n64A-2 a. Current Flows if Pull-Down Used Converting Bipolar PLD Designs to CMOS 5-101 ~ AMD Typical Input Typical Output 1n64A-5 Figure 3. Input Pull-Up Resistors as Shown In the Datasheet This discussion applies primarily to unused input pins. There actually are three basic kinds of pin on a PLD: input, output, and 110. If an output pin is unused and has no three-state capability, then the output will be high or low, and does not need any external pull-up or pulldown. If the pin is an 110 pin, or an output with threestate, then the action taken depends on the default state of the unused 110 pin. Most software packages configure unused 110 pins as inputs; in this case, the I/O pin should be treated just like an input. If the I/O pin is configured like an output, then no extra action is needed. Note that PALASM software normally configures unused I/O pins as inputs; in the case of MACH devices, however, the software gives the option of configuring unused 110 pins as outputs instead. • If the original design has unused pins tied directly to Vee or ground, then the new device can simply be dropped into the old socket. • If the original design has unused pins tied high through a resistor, then the new device can simply be dropped into the old socket. • If the original design has unused input pins tied low through a pull-down resistor: - if the replacement part has no internal pull-up resistors, drop the CMOS part into the socket if the replacement part has internal pull-up resistors (as indicated in the data sheet), and the board's pull-down is less than about 10 kn, then drop the CMOS part into the socket What To Do What action you take when converting from bipolar to CMOS depends on the original design. 5-102 Converting Bipolar PLD Designs to CMOS AMD~ - if the replacement part has pull-up resistors (as indicated in the data sheet), and the board's pulldown resistor is greater than 10 kO, then either remove the pull-down resistor when using the new CMOS part or replace it with a smaller resistor. • . If the original design has unused pins that are floating, be sure to use one of AMD's newer CMOS devices that have pull-up resistors built in. single line may severely impact the DC loading on the drivers. • Too many vias (feedthroughs): each of these is a discontinuity. While some vias may be necessary for routing, use as few as possible. Do not route between an outside layer and an inside layer on multilayer boards. • Headers, sockets, and other components that act as discontinuities: these should be used as sparingly as pOSSible, since they can cause reflections, excessive EMI, and add to the inductance of the line. • Poorly decoupled Vee and ground: this can sabotage the best attempts at termination. Good termination relies on a solid ground system, and any noise being carried on the Vee or ground can make its way onto signals. It can also make an otherwise perfect termination ineffective, causing reflections. Edge Rates, Termination, and Layout Sensitivity Edge rates are important when converting from bipolar to CMOS. While there are exceptions, CMOS devices tend to have faster edge rates than their bipolar equivalents. The edge rates determine whether or not a signal needs termination. The faster the edge rate is, the more a PCB trace looks like a transmission line, which can generate reflections that impact system performance. More information on these issues is available in the Application Note High-Speed Board Design Techniques. With slow edge rates, long traces can be fabricated with no need for termination. As edge rates speed up, even short traces will need termination. There is no standard cutoff, and the need for termination will depend on many variables. These include PC board materials, layout, the number and kind of other components on the line, and the amount of noise that the design can tolerate. If a design is marginal with respect to edge rate sensitivity, then any changes in the edge rate of the component in that socket can affect the behavior of the system. As an example, the maximum unterminated line length for a particular trace may be 2"-4" with a 1.5-ns rise-time device, but 4"-7" with a 3-ns rise-time device. If a particular trace is 5" long, it is within the window for the slower device, and might work. But this definitely qualifies as marginal. If a replacement device has a 1.5-ns rise time, then this 5" line is now outside the allowable window for the faster device, and will require termination. Design sensitivity to edge rates is also affected by layout. The following items can contribute to increased noise in a system, and therefore may make it harder to replace one device with another that has a faster edge rate. • • 90 0 corners: these are impedance discontinuities. Use two 45 0 corners instead. The ideal is to have a rounded corner. Long unterminated stubs: these can introduce reflections onto a line that may be otherwise terminated at the end. Keep the stubs shorter than the maximum critical line length. If they must be longer, then they will have to be individually terminated. Note that multiple DC terminations on a What To Do When converting from bipolar devices to CMOS devices in a given deSign, the following considerations apply, depending on how the original PAL device output is terminated. • If the original design has parallel-terminated lines, then conversion should pose no problem. • If the original design has series-terminated lines, then the conversion will likely pose no problems. Series-terminated lines tend to be a bit more delicate to design; parallel-terminated designs are generally more robust, but cause greater power . dissipation. .• If the original design has no termination, but the CMOS device has a slower edge rate than the bipolar device (as in the case of the 1O-ns 22V10, for example), then the conversion should pose no problem. • If the original design has no termination and the CMOS device has a faster edge rate than the bipolar device, then your action depends on the length of the trace. No absolute rule can be given, but the following rules of thumb should generally be safe: - if the line is longer than about 5 inches, add termination. - if the line is shorter than 2 inches, the direct conversion will likely work. - if the line is between 2 and 5 inches, try the direct conversion, but be prepared to add termination if noise proves to be excessive. In general, terminate if you feel that it will save you future headaches. Converting Bipolar PLD Designs to CMOS 5-103 ~AMD AMO's CMOS PLO Technology. Even in the presence of large amounts of negative overshoot, no anomalous behavior has been observed in AMD's CMOS products. Negative overshoot should pose no conversion problem. Overshoot A design that is well terminated will likely have very little overshoot. As reflections grow, overshoot increases in both the positive and negative direction (note that "negative overshoot" is sometimes called "undershoot", although this is technically a misnomer). The effects of overshoot on inputs to a PAL device differ depending on whether it is positive or negative overshoot. • Negative overshoot: AMD's CMOS devices on EE4 technology or later (comprising all of the bipolar-equivalent devices being manufactured today) are able to clamp negative overshoot effectively. Details can be found in the Application Note, Inside • Positive overshoot: this can be an issue if the CMOS device used in the replacement has no positive overshoot filter. Not all PLD manufacturers use overshoot filters, but all of AMD's bipolarequivalent CMOS devices are being given overshoot filters; with these filters, positive overshoot will pose no conversion problems. The equivalent input schematic in the data sheet will indicate whether or not a device has overshoot filters. Typical Input Typical Output 17764A-5 Figure 4. Overshoot Clamping and Filters as Shown in the Datasheet 5-104 Converting Bipolar PLD Designs to CMOS AMD~ try it, but observe ground and signal conditions closely to see if any design modifications will be needed. Ground Bounce CMOS devices of any kind have a reputation of causing more ground bounce than their bipolar counterparts. While this is changing on AMD's fastest devices, it is generally true on many devices. The actual amount of ground bounce encountered will depend on the output loading and the number of outputs switching. The more current being switched, whether due to more outputs or heavier loads, the greater the ground bounce may be. AMD's newer CMOS devices have a special design that uses a split leadframe, as conceptualized in Figure 5. This technique allows for CMOS devices with less ground bounce than their bipolar equivalents. This means that faster CMOS devices will be possible without ground bounce making them unusable. Vee or Ground Pin In the last case, design changes may be difficult to accomplish, depending on the amount of ground bounce, the design flexibility, and the availability of engineering resources. Such changes will likely require more substantial board changes than a simple conversion would require, and may not be feasible. If deSign changes are possible, however, the following are some things to try. • Reduce the loading on the outputs. • If using DIP packages, switching to PLCC packages instead may help. PLCC packages have shorter, more uniform, lower-inductance leads, and offer lower ground bounce. • Keep the board-level ground inductance as low as possible to make sure that board-level ground bounce does not exacerbate any internal chip ground bounce. • Try to reduce the numberof outputs switching. This may be difficult in many designs, but a good candidate for this would be a state machine. If the existing state bit assignment has transitions that switch many outputs at once, try redoing the state bit assignment so that fewer simultaneous transitions occur. Please refer to the Application Note, Basic Design with PLDs for more information on tailoring state machines. ••• Inside Package Keeping Consistent Fuse Checksums Noisy Line 17764A-6 Figure 5. Conceptualization of Split Vee and Ground Leads What To Do When converting from bipolar to CMOS, the following ground bounce considerations apply. • If the CMOS replacement part has a split leadframe design (as indicated in the data sheet), then there will likely be no conversion issue (the PALCE22V1 OH-7 is presently the best example of this). • If the outputs in the design are lightly loaded, then there will likely be no conversion issue. While the amount of loading that can be tolerated will vary from design to design, outputs with 75-100 pF or greater should be considered heavily loaded. • If the deSign has few outputs switching at a time, then there will likely be no conversion issue. • If many outputs switch at a time, or if the loads are heavy, the conversion may work just fine anyway; As shown in Table 1 ,direct bipolar conversions can only be made between 16XX families and the 16V8; 20XX families and the 20V8; the bipolar and CMOS 22V10s; and the bipolar and CMOS 20RA10s. In the last two cases, the architectures are identical, and there will be no fuse checksum inconsistencies between the bipolar and CMOS versions. Note that there are actually two checksums in a JEDEC file: the fuse checksum and the transmission checksum. Only the fuse checksum reflects the array contents specifically, and is usually the only one of interest. Transmission checksum changes will occur if anything at all in the JEDEC file changes-even a comment; this may not reflect any functional change to the pattern, and can generally be ignored. For more information on checksums, please refer to JEDEC Standard 3. In the case of the 16R8 families and the 20R8 families, individual bipolar architectures are converted to a single universal CMOS architecture: the 16V8 and 20V8, respectively. The CMOS devices have extra architecture bits that determine the macrocell configuration. For example, the 16V8 can be configured differently to emulate a 16R4 or a 16L8 (as well as other arChitectures). Converting Bipolar PLD Designs to CMOS 5-105 ~AMD Because of this, the CMOS equivalent will generally have a different fuse checksum from the bipolar original. If all outputs on a device are used, then converting from bipolar to CMOS will give a consistent CMOS checksum, regardless of which software performs the conversions (of course, this checksum will be different from the original bipolar one). However, if some outputs are not used, then software defaults generally determine how the unused architecture bits are set. Different software programs may have different defaults; therefore they may generate different checksums forthe same design. Note that it is also possible to perform these conversions directly through cross-programming with no apparent change in checksum. However, this is not recommended as a long-term conversion mechanism, since it makes the production file being used inconsistent with the original source file. Cross-programming does not change the source file or the original JEDEC file. What To Do To convert designs with unused outputs, the following procedure is recommended if a consistent checksum is needed. The solution shown below uses PALASM syntax; similar equations can be specified in any PLD compiler. 5-106 1. Obtain the original source file. If the original source file is unavailable, then disassemble the JEDEC file to obtain a sou rce file. 2. Change the device type in the source file from the original bipolar device type to the 16V8 or 20V8, as appropriate. 3. For each unused output, add a "dummy" equation as follows: for a combinatorial output, add = GND = GND OUTPUT.TRST /OUTPUT for a registered output, add / OUTPUT : = GND 4. Recompile the design. Summary Converting from bipolar to CMOS PLDs is generally a painless process. Most designs can convert easily. The more robust the original design, the easier the conversion will be. For those designs that do not convert as easily, there are steps that can be taken to make the conversion successful. These steps can often result in a design that is cleaner and more robust than the original. Converting Bipolar PLD Designs to CMOS High-Speed-Board Design Techniques Advanced Micro Devices Application Note INTRODUCTION The most important factor in the design of many systems today is speed. 25-MHz processors are common; 40- and 50-MHz processors are becoming readily available. The demand for high speed results from: a) the requirement that systems perform complex tasks in a time frame considered comfortable by humans; and b) the ability of component manufacturers to produce highspeed devices. An example of a) is the large amount of information that must be processed to perform even the most rudimentary computer animation. Currently, Programmable Array Logic (PAL) devices are available with propagationdelaysof 4.5ns. Whilethismig htseemf ast, itis notthepropagationdelaythatcreatesthepotentialforproblems,butratherthefastedgeratesneededtoobtainthefast propagationdelays.lnthefuture, muchfasterdeviceswill becomeavailable,withcorrespondinglyfasteredgerates. Designing high-speed systems requires not only fast components, but also intelligent and careful design. The analog aspect of the devices is as important as the digital. In high-speed systems, noise generation is a prime concern. The high frequencies can radiate and cause interference. The corresponding fast edge rates can result in ringing, reflections, and crosstalk. If unchecked, this noise can seriously degrade system performance. This application note presents an overview of the deSign of high-speed systems using a PC-board layout. It covers: • the power distribution system and its effect on boardnoise generation, • transmission lines and their associated design rules, • crosstalk and its elimination, and II electromagnetic interference. 1. POWER DISTRIBUTION The most important consideration in high-speed board design is the power distribution network. For a noisefree board, it is necessary to have a noise-free power Publication# 16356 Rev. A Issue Date: February 1993 ~ AmendmentiO distribution network. Note that it is just as important to develop a clean Vcc as it is to get a clean ground. For AC purposes, which is what this application note mainly discusses, Vcc is ground. The power distribution network also must provide a return path for all signals generated or received on the board. This is often overlooked because the effect of the return path is less apparent at lower frequencies. Many designs work even when the nature of the return path is ignored. 1.1 Power Distribution Network as a Power Source 1.1.1 The Effect of Impedance Consider a 5" x 5" board with digital ICs and a power supply of +5.0 V. The goal is to deliver exactly +5.0 V to the power pins of every device on the board, regardless of its position relative to the power source. Furthermore, the voltage at the pins should be free of line noise .. A power source with these characteristics would be schematically represented as an ideal voltage source (Figure 1a), which has zero impedance. Zero impedance would ensure that the load and source voltages would be the same. It also would mean that noise signals would be absorbed because the noise generators have finite source impedance. Unfortunately, this is only an ideal. Figure 1b illustrates a real power source with associated impedances in the form of resistance, inductance, and capacitance. These are distributed over the power distribution network. Because of the network's impedance, noise signals can add to the voltage. The design goal is to reduce the power distribution network impedances as much as possible. There are two approaches: power buses and power planes. Power planes generally have better impedance characteristics than power buses; however, practical considerations might favor buses. 5-107 ~ AMD Vee r--------~--_u V+ .......................................................................................... Load a) Vee V+ 16356A-001A b) Figure 1. The Power Source. a) Ideal Representation; b) More Realistic Representation ::::::::::::::{~ t{~ff:' :::::;::::;::::::.: I :::;: t~~f ":;::::" tmm ::i::::::: !!::::'l ;:. :.; :.~:; !:.:.. .•.:.:.:.::..:.:.:.:.:.:..:..:......:..:.::' ... :....... :... :... 41 ::u.: . . .:..: ..... _.7.: ..:..:_ . .:. :. . ::'.:......:..:. : U8: .... ~~~:~~~::~~~:~~:~~f~~:~?r . :. . :.I:. . ._:. :.:.u .....:. :.:.:..... 11:,"," :9.:.:.:.'.:.••.:.•:.:.:.::.:••:.::.:. II b) a) 16356A-002A Figure 2. 'Power Distributions System. a) Power Buses; b) Power Planes 1.1.2 Power Buses vs Power Planes Two power-distribution schemes are shown in Figure 2. A bus system (Figure 2a) is composed of a group of traces with the various voltage levels required by the system devices. For logic, these are typically +5 V and ground. The number of traces required for each voltage level varies from system to system. A power-plane sys- 5-108 tern (Figure 2b) is composed of entire layers (or sections of layers) covered with metal. Each voltage level requires a separate layer. The only gaps in the metal are those needed for placing pins and signal feed-throughs. Early designs favored buses because of the expense of devoting entire levels to power distribution. The power High-Speed-Board Design Techniques AMD~ A capacitor, which is ideally represented in Figure 3a, is more realistically represented by Figure 3b. Resistance and inductance are the result of the construction of the plates and the leads necessary to build the capacitor. Because the parasitic components are effectively in series with the capacitance, they are called equivalent-series resistance (ESR) and equivalent-series inductance (ESL). bus shares layers with the signal lines. The bus must supply power to all devices, while leaving room for the signal traces; therefore, buses tend to be long, narrow ribbons. This results in a relatively small cross-sectional area with a small resistance. Although the resistance is small, it is significant. Even a small board can have 20 to 30 devices on it. If each device on a 20-device board sinks 200 rnA, the total current would be 4 A. A bus resistance of only 0.125.Q has a 0.5 V drop. With a 5 V power supply, the last device on the bus might receive only 4.5 V. Thus the capacitor is a series resonant circuit for which fR = As an example, the 10 JlF capacitors used forthe boardpower connections are typically made with rolls of metal foils separated by an insulating material (Figure 5). This results in large ESLs and ESRs. Because of the large ESLs, fR is generally less than 1 MHz. They are good filters for 60-Hz noise, but not good for the expected 1OO-MHz and higher switching noise. On a bus, currents are restricted to paths defined by the bus. Any line noise generated by a high-speed device is introduced to other devices on that power bus. On the board in Figure 2a, noise generated by U9 is sent to U7 by the bus. On the power plane, the noise currents are distributed because the current path is not restricted. This, along with lower impedance, makes power planes quieter than power buses. The ESL and ESR result from the construction of the capacitor and dielectric material used, ratherthan from capacitance value. The high-frequency reject capabilities cannot be improved by replacing a capacitor with a larger one of the same type. The impedance of a large capacitor is smaller than that of a small capaCitor at frequencies below the fR of the small capacitor. But at frequencies above fR, the ESL dominates and there is no difference between the impedance of the two capacitors (Figure 4b). This is because only the capacitance has changed; unless the construction is changed, the ESL remains essentially unchanged. To improve high-frequency filtering, one must replace the capacitor with a type that has a lower ESL. 1.1.3 Line Noise Filtering The power plane alone does not eliminate line noise. Since all systems gener~te enough noise to cause problems, regardless of the power distribution scheme, extra filtering is required. This is done with bypass capacitors. Generally, a 1 JlFto 10 JlF capacitor is placed across the power input to the board, and 0.01 JlF to 0.1 JlF capacitors are placed across the power and ground pins of every active device on the board. Since the g<;>al is to filter out any AC component on the power supply, it might seem initially that the largest possible capacitor is the best, minimizing the impedance as much as possible. However, this does not take into account that real capacitors do not have ideal characteristics. . nc As shown in Figure 4a, it is capacitive at frequencies below fR, and inductive at frequencies above fRo As as result, the capacitor is more a band-reject filter than a high-frequency-reject filter. Because the power plane fills an entire layer, the only area constraints are the dimensions of the board. The resistance of a power plane is a small fraction of that of a power bus supplying the same numberof devices. Thus a power plane is more likely than a bus to supply full power to all the devices. The bypass capacitor acts as a filter. The larger capac itor (== 10 JlF) is placed across the power input of the board to filter lower frequencies (like the 60-Hz line frequency) that usually are generated off the board. Noise generated on the board by the active devices have harmonics in the range of 100 MHz and higher. A bypass capacitor is placed across each chip and generally is much smaller (== 0.1 JlF) than the capacitor across the board. _I_ I Various types of capacitors are available for specific frequencies and applications. Table 1 gives a small overview of some available device types. a) 16356A-003A Figure 3. a) Ideal Representation of a Capacitor b) Parasitic Components added to Emulate Real Conditions High-Speed-Board Design Techniques 5-109 ~ AMD §: §: u u N N FR FR (Freq.) a} b) (Freq.) 16356A-004A . Figure 4. a} Capacitor Impedance Versus Frequency; b) the Effect of Lowering Capacitance While Using the Same Type of Construction (Constant ESL) Metal Foil Insulation Metal Foil 16356A-005A Figure 5. Internal Construction of a Large (> IlF) Capacitor The lowest ESL capacitors often are made with non-ferromagnetic materials, which have a low voltage-capacitance product. Thus it is difficult to make large capacitors with practical breakdown voltages to prevent board failure. However, because of better filtering characteristics, larger values might not be needed. Figure 6 compares a 0.01 IlF capacitor of type COG (non-ferromagnetic) to a 0.1 IlF capacitor of another type. Note that the 0.01 IlF capacitor gives better filtering at higher frequencies. The capacitor graphs imply that anyone capacitor has a limited effective frequency operating range. Because systems have both high- and low-frequency noise, it is desirable to extend this range. This can be done by putting a high-capacitance, low-ESL device in parallel with a lower-capacitance, very-Iow-ESL device. Figure 7 shows that this can significantly increase the effective filtering frequency range. 1.1.4 Bypass Capacitor Placement After the filter capacitors have been chosen, they must be placed on the board. Figu re 8a shows the standard placement for boards with slow device speeds. The capacitor is placed near the top of the device to help ensure its accessibility. While simple for layout, this does not give the best high-speed performance. Note that the Vcc capacitor connection is quite close to the chip's Vec connection, but the ground connection is far away. Because noise is not uniform on a power plane, the capacitor is not filtering noise at the chip leads; it is only filtering noise near the chip. Table 1. Bypass Capacitor Groups Type Range of Interest Application Electrolytic 1 IlF to> 20 IlF Commonly used at power-supply connection on board. Glass-Encapsulated Ceramic 0.01 IlF to 0.1 IlF Used as bypass capacitor at the chip. Also often placed in parallel with electrolytic to widen the filter bandwidth and increase the rejection band. Ceramic-Chip 0.01 IlF to 0.1 IlF COG < 0.1 IlF Primarily used at the Chip. Also useful where low profile is important. Bypass for noise-sensitive devices. Often used in parallel with another ceramic chip to increase rejection band. 5-110 High-Speed-Board Design Techniques AMD~ 100.000 10.000 9: 0.01 J.LF COG 1.000 N 0.100 0.010 0.001 0.1 1.0 10.0 100.0 1K Frequency (MHz) 16356A-006A Figure 6. Frequency Response of X7R and COG Type Construction ,, I I I , I , I , o I , x I , I , I I I I'" , , Frequency 16356A-007A Figure 7. Frequency Response of Two Capacitors in Parallel a) b) 16356A-008A Figure 8. a) Typical Placement of Bypass Capacitors; b) Preferred Placement of Bypass Capacitors High-Speed-Board Design Techniques 5-111 ~ AMD Better performance can be obtained by ensuring that the chip and the capacitor contact the Vee and ground planes at the same point. Because the capacitor size is different from that of the chip, it is necessary to run two traces from the Vee and ground plane contact pOints to the capacitor, as shown in Figure 8b. These "lead extensions" are placed on a non-power plane and should be kept as short as possible. It is generally best to place the capacitor on the opposite side of the board, directly under the Chip. A surface-mount chip capacitor works well here. Note that the "lead extension" traces from the capaCitor to the power pins take up space that could have been used for signal-line routing. However, putting extra effort into routing the signal lines now could prevent much noise-reduction work later on. For devices with multiple Vee and ground pins, how best to bypass depends on the device. In particular, it depends on whether the power pins are connected internally. On some devices, such as the PAL 16R8-4 series, the ground pins are connected by a common ground bus. On these devices, it is only necessary to bypass one ground pin to one Vee pin. If the power is kept separated internally, the separate Vee pins must be decoupled individually. In general, it is best to contact the device's manufacturer for specifiC recommendations. . 1.2 Power Distribution Network as a Signal Return Path One of the more surprising functions of the power network is the provision of a return path for all signals in the system, whether generated on or off the board. Designs that accommodate this aspect of the power distribution system eliminate many high-speed noise problems. 1.2.1 The Natural Path of the Signal-Return Line Of greatest concern in high-speed design is the energy generated at the signal switching edges. Each time a signal switches, AC current is generated. Current requires a closed loop. As illustrated schematically in Figures 9a and 9b, the return path needed to complete the loop can be supplied by the ground or Vee. The loop can be represented by Figure 9c. Vee Signal Current Loop GND a) Vee Signal Current Loop GND b) Signal Current Loop ACGND ACGND c) 16356A-009A Figure 9. Current Loop of a Signal on the Board. a) Through Vee; b) Through Ground; c) The Equivalent AC Path 5-112 High-Speed-Board Design Techniques AMD~ of least impedance is the path bringing the signal-return line closest to the signal line. If it can, the signal return follows the signal line as closely as possible, resulting in the smallest loop. In multiple layer boards, "as closely as possible" usually means in a ground or Vcc plane above or below the signal trace. In a two-layer board, this means the closest ground or Vcc trace. 1.2.2 Bus vs Planes for a Signal Return Path 16356A-O 1OA Figure 10. Inductance Increases as the Signal and Return Path are Separated Current ioops have inductance and can be thought of as single-turn coils. They can aggravate ringing, crosstalk, and radiation. The current-loop inductance and associated problems increase with loop size. Minimizing the size of the loop minimizes these problems. AC return signals have an entire plane in which to choose a path, but they take the path of least impedance (not necessarily least resistance) to the current. Impedance also includes inductance and capacitance. Metal has very little resistance; therefore, the impedance is primarily inductive. Because impedance increases with inductance, the path of least impedance is the path with the smallest inductance. Ifthe signal line goes from A to B on a random path, the natural return path is not necessarily a straight line, as would be dictated for least resistance. As noted in Figure 10, the inductance of a signal line and its return line increases with the separation of the two paths. The path Return Signal \. Plane '" Figure 2a shows that a power bus has a fixed path. The return signal must follow this path, whether optimal or not. Unless the signal lines are purposely laid out near the power buses and oriented to minimize loop size, there will probably be large loops. If the layout of a board using buses for power distribution is not thought out carefully, it can result in a configuration that generates much noise. The power plane imposes no natural restrictions on current flow. Thus the return signal can follow the path of least impedance, which is the path closest to the signal line. This results in the smallest possible current loops, which makes it the preferred solution for high-speed systems. Although power planes have an advantage over buses, the benefits they provide can be defeated by the designer. Any break in the natural path of the return signal forces it to go around the break, increasing the loop size (Figure 11). Be careful about cuts in the ground and power planes. 1.3 Layout Rules With Power Distribution Considerations The following layout rules will help you take advantage of power planes and avoid pitfalls. Break In Return Signal Plane Current forced around break, increasin~ L 16356A-011A Figure 11. The Increase in Loop Size Due to a Break in the Power Plane High-Speed-Board Design Techniques 5-113 ~AMD a. Be Careful with Feedthroughs Cuts in the power plane tend to show up at feedthroughs or vias., These are necessary for traces to cross sides of the board and to connect components and connectors to the board. They are surrounded by small gaps where the power planes are etched away to avoid shorts in the signal lines. If the vias are close and the etchings wide, they might touch and form a barrier to any return path. This can occur with backplane connectors and device sockets. For example, this can occur on the connectors on VME backplanes. The 104-pin connector has vias that can block the signal return. All the return signals are forced to the edge of the board. Not only are the loops longer, but the edge is shared by all the return signals; as we will see, this can result in crosstalk (Figure 12). b. Ground cables Sufficiently Current loop considerations are also applicable for cables going off the board. Every Signal should be a twowire pair: one for the signal, and one for the return. The two lines should be kept next to each other to minimize the loop size. Figures 13a and 13b illustrate poorer configurations. Figure 13c illustrates the proper configuration. c. Separate Analog and Digital Power Planes High-speed analog devices tend to be sensitive to digital noise. For example, amplifiers can amplify switching noise, making it appear as spikes. Thus on boards with analog and digital functions, the power planes are commonly separated; the planes are tied together at the power source. This causes a problem for devices using both types of signals (such as DACs or voltage comparators). The signal lines must cross the plane boundaries. These boundaries force the return path to the power source before returning to the driver. The solution is to place jumpers across the ground planes where signals cross (Figure 14). The jumper provides a bridge across the break for the return signal; this helps minimize the current loop. nal 'A' Return Path 'A" nal'S' Common Return Path for A and S 16356A-012A Figure 12. Common Paths of Signal Return Due to Vias 5-114 High-Speed-Board Design Techniques AMDl1 ...:.r-- ~ ,.,. -0 ~ ,.,. - " ,.,. ~ -0 ~ ,.,. -0 ,.,. ~ -0 ~ ,.,. ~ -0 ,.,. ~ ~ ~ a) 1 ~ b) - L..- 16356A-O 13A c) Figure 13. Connector Configuration. a) Insufficient Grounds; b) Enough Grounds but Grounds lumped Together Resulting In Larger Current Loops; c) Grounds.Evenly Distributed Among Signal Lines Digital Ground Analog Ground ~__---=====--"-/_-IA; Signal W Ground Bridge to Complete Current Loop 16356A-014A Figure 14. Jumper Between Analog- and Digital-Power Planes for Signal·Return Path d. Avoid Overlapping Separated Planes When separate power-planes are used, do not overlap the power plane of the digital circuitry and the power plane of the analog circuitry. The analog and digital power planes are separated to isolate the currents from each other. If the planes overlap, there is capacitive coupling, which defeats isolation. High-Speed-Board Design Techniques 5·115 ~AMD To ensure separation, take a board and cut between the separated planes. Then inspect the newly-exposed edges of the board. No metal should be showing, except where traces or connections are specifically designed to cross the boundary. e. Isolate Sensitive Components Certain devices, such as phase-locked loops, are particularly sensitive to noise interference. They require a higher degree of isolation. Good isolation can be achieved by etching a horseshoe .in the power planes around the device (Figure 15). All signals used by the device enter and leave through the narrow gap at the end of the horseshoe. Noise currents on the power plane must go around the gap and do not come close to the sensitive part. When using this technique, ensure that all other signals are routed away from the isolated section. The noise signals generated by these lines can cause the interference this technique was designed to avoid. f. Place Power Buses Near Signal Lines Sometimes, the designer must use two-layer boards and is forced to use power buses instead of planes. Even then it is possible to control loop size by placing the buses as close as possible to the signal lines. The ground bus could follow the most sensitive signals on the other side ofthe board (Figure 16). The loop for that signal is the same as it would be if the load used power planes. 2. Signal Lines as Transmission Lines Controlling the relationship between the signal line and AC ground takes advantage of the return signal's tendency to take the path of least impedance. Another advantage is the constant impedance along the signal line . Such signal lines are called controlled-impedance lines, and they provide the best medium for signal transmission on the board. GS.-----.. ,/ ~ NoiseSensitive Corresponding { Signal Lines ==~~~~~q~~========9 Gap in Power Planes Ground Plane Device Isolated Ground Plane 16356A-015A Figure 15. Isolation of Noise Sensitive Components '-------"~. D dl~_gn. D""@D""'~~' D"":~ :.II!::: ::: D Signal Line :{ DDD 16356A-016A Figure 16. Providing the Optimum Signal-Return Path with a Bus-Power Distribution System 5-116 High-Speed-Board Design Techniques AMDl1 However, when the signal delay is greater than a significant portion of the transition time, the signal line must be treated as a transmission !ine. An improperly terminated transmission line is subject to reflections, which distort the signal. The signal at the load end of the line can resemble ringing (Figure 17), slowing down the system. It can also cause false clocking, destroying system functionality. A controlled-impedance signal line can be modeled as shown in Figure 18. Inductance and capacitance are evenly distributed along the length of the line. Their units are henrys per unit length and farads per unit length, respectively. lossless signal line, Zo is an AC resistance; i.e., Zo appears to the driver as a pure resistor. Its units are ohms (il), and it is equal to where Lo = Signal Line Inductance in henrys per unit length Co = Signal Line Capacitance in farads per unit length The propagation delay also depends on Lo and Co. It has units of time per unit length, and it is equal to tpD = YLOCo From the model, we can derive two important parameters: impedance (Zo), and propagation delay (tpo). On a o ---- -- -- -- ----- _\ ---- --- --- -- -- --- ---- -- --- --- ----------- a) 16356A-017A b) Figure 17. Reflections on a Signal Line. a) at the Driver; b) at the Load 16356A-O 18A Figure 18. Transmission Line Transmission Line Categories Given that the designs discussed in this paper are for printed circuit boards, the possible types of signal lines fall into one of two categories: stripline and microstrip (Figure 19). The stripline has the signal line sandwiched between two power planes. This technique theoretically offers the cleanest signals because the signal line is shielded on both sides. However, the lines are hidden; there is no easy access to the signal lines. Microstrip has the signal line on an outer layer. The ground plane is to one side of the signal line. This technique allows easy access to the signal line. High-Speed-Board Design Techniques 5-117 ~AMD Example The dimensions of the trace and board are restricted by certain rules. Generally, the vendor sells the board with 1 oz of copper, so the metal thickness is about 1 mil. The trace width should be between 8 and 15 mils. Signal lines thinner than 8 mils tend to be harder to control. Signal lines thicker than 15 mils tend to have excess capaCitance. A typical value is 10 mils. The layer separation is determined by the required board thickness and the number of layers to be used. For this example, 30 mils is adequate. T h ~ a) Based on these assumptions, it is possible to calculate the parameters for a typical signal line: width = 10 mils, thickness = 1 mil, separation = 30 mils, and eR =5. h ~ f5 --rb) 87 In 5 98 * 03 + 1.41 0.8 * .001 + .01 67.0 16356A-019A tpD Figure 19. Signal Line Construction on a Circuit Board. a) Stripline; b) Microstrip = 1.017 VO.456 * 5 + 0.67 1000 ~ tpD = 1. Co 4h 0.67nw (0.8 + ~) n w o17Un ns / It 1000 tpD pF/ft Zo LO = 2 ZoCo pH/ft *~ 67.05 pF/ f t 26.1 pF/ft LO For Stripline ns/ft 1.75 n The parameters Co, Lo, Zo, and tPD can be determined from the physical dimensions of the signal line and the dielectric properties of the board material. They are discussed below. Zo = ~ In 67.05 2 * 26.1 pH/ft 117 nH/ft Distributed Load Calculations The calculations above are for a signal line with a lumped load at the end of the trace (Figure 20). If the load is distributed along the signal line (Figure 21), the capacitance of the load devices is also distributed along the line and adds to the line capacitance. This changes the signal-line parameters Zo and tPD. The new parameters are derived from the original values based on the added capacitance, el, in farads per unit length: For Microstrip 5.98h In 0.8w + tpD = 1.017 VO.457eR+ 0.67 ~1 t ns/ft + cL Co Co 1000 tpD pF/ft Zo 2 LO = Zo . Co pHI ft eR is the relative dielectric constant of the board material. A common material is epoxy-laminated fiberglass, which has an average eR of 5. 5-118 n High-Speed-Board Design Techniques ns/ft AMD~ 16356A·020A Figure 20. Transmission Line with a Lumped Load Figure 21. Transmission Line with a Distributed Load Distributed loading is common in memory banks. The input capacitance on these devices can range from 4 pF to 12 pF. The following example uses 5 pF. The physical size of memory devices usually permits placing two of them per inch. The distributed added capacitance is then: C = _ _--...:.5--=-p_F _ _ L 0.5 in *~ 12 in 120 pF/ft The new values of Zo and tPD, based on distributed loading, are: 67 05 n _ / 1 + 12 0 pF / f t 26.1 pF/ft 'V 28.34 n tpD = 1.75 ns/ft * .v 1 + 120 pF/ft 26.1 pF/ ft 4.14 ns/ft With this distributed load, the impedance has been greatly reduced, and the signal is now much slower. Reflections The source generates a signal with an energy content determined by Zo n. Even though the line is seen as a resistance, the signal line does not dissipate energy. The energy in the Signal must be dissipated by the load impedance (ZL), as shown in Figure 20. The maximum transfer of energy from source to load requires that the load impedance equal the source impedance. For the entire Signal to be transferred to ZL, ZL must equal Zoo If they are not equal, some of the signal energy is dissipated, and the rest is reflected back toward the source. The source generator output then adjusts to compensate for the "new" load. The waveform of the signal at the load can be thought of as the sum of the originally generated signal and the reflection from the load. The appearance of the waveform depends on the mismatch of the load and line impedances and the ratio of the signal-transition time (tR) to the propagation delay of the line (t), tR/t. If the transition time is significantly longer than the propagation delay of the line, the reflection reaches the source when the original signal has changed only a small amount. The generator compensates for the "new" load and transmits the corrected signal with little signal disturbance. The signal at the load then has a small overshoot. If the propagation delay of the line is long enough for the reflection to reach the source after the signal has changed a significant percentage, the generator must change significantly to compensate for the load. The load reflects the new transition, which results in the ringing shown in Figure 17. The amount of overshoot usually varies proportionally with the signal-line length until the signal-line delay is equal to the transition time. At this point, the overshoot can be as much as the original transition, effectively doubling the swing of the transition. A signal line long enough to produce significant reflections acts like a transmission line. The point at which the signal line is considered a transmission line depends on the amount of tolerable distortion. A liberal rule of thumb is to consider a signal line a transmission line when the transition time of the original signal is less than four times the propagation delay of the signal (Figure 22); that is, when tAlt ~ 4. Hlgh-Speed-Board Design Techniques 5-119 ~ AMD Original Signal at Source t I+--+t-- t < --E 4 Signal at Load 16356A-022A Figure 22. Minimum Delay Between Original and Reflected Signal Which Results in a Transmission LIne A more conservative rule is to consider the signal line a transmission line when tN't is less than eight times the propagation delay. Generally, the larger the transition time is in relation to the propagation delay of the signal line, the cleaner the resultant signal. For older devices with 5 ns transition times, signal lines shorter than 8.6" do not have to be treated as transmission lines. For newer, high-speed devices, even a twoinch line is a transmission line. Practically all signal lines are transmission lines on boards with high-speed devices. From this it is possible to determine what length of the microstrip line discussed above must be treated as a transmission line. On available devices, tR ranges from 5 ns (especially those using bipolar technology) to 1 ns (newer bipolar and CMOS devices). The rise times and corresponding signal-line lengths are shown in Table 2 for the example given above. If the transmission line has the distributed load in the example above, then the minimum transmission-line length must be reconsidered. As shown in Table 3, a four-inch line is a transmission line when tR = 5 ns. If tR =1 ns, a signal line smaller than one inch is a transmission line. Table 2 Example:tR and Corresponding Transmission-Line Length for tR = 4 Table 3 Example: tR and Corresponding Transmission-Line Length with Lumped and Distributed Loads for 't' 5-120 tR (ns) 5 Line Length (inch) 8.6 . 4 6.9 5.1 3 2 3.4 1 1.7 ..!B = 4 l' tR (ns) 5 3 2 1 Line Length (inch) Lumped Load Distributed Load 8.6 3.6 5.1 2.17 3.4 1.4 0.75 1.7 High-Speed-Board Design Techniques AMDl1 Quantifying Reflections Given that the signal line is long enough to be considered a transmission line, the size of the reflected signal depends on the difference between Zo and ZL The numerical indicator of the percentage, orthe original signal that is reflected, is called the reflection coefficient (KR). KR is equal to: The input impedance of the load is greater than 100 kn. This is so much greater than Zo (67 0), that KR at the load is practically equal to one. KR at the source is: K 8.3 - 67 8.3 + 67 R - = -0.78 The percentage of the original signal reflected back is 100 * KR. Referring back to the open load: The driver generates a signal switching from 3.5 V to 0.2 V. Since the driver-output impedance and ZO make up a voltage divider, the generated signal is: Zo 00 - 00 + Zo Ll v = KR = - - - Zo 1 = = oo+ 3. 5 V) + Zo --- = Zo 2.84 Zo Zs (0.2 V - 3.5 V) 50 + 8 For an shorted load: KR (0. 2 V - 50 v The resultant Signal at the source is: Vs = .-1 Foropen and shorted loads, the entire signal is reflected without attenuation .. KR is negative for the shorted load. This indicates that the reflected signal is inverted from the original. With a printed-circuit board, it is possible to estimate the expected type of mismatch. Zo typically ranges from 30 n to 150 o. Input impedances range from 10 kO (for bipolar devices) to over 100 kO (for CMOS devices). Output impedances can be very low. A CMOS PAL device, such as the PALCE16V8, has a typical-output LOW voltage of 0.2 Vat 24 rnA for about 8 o. The output-HIGH impedance is about 50 0, which is closer to the expected Zoo Consider the microstrip line derived earlier, with a CMOS device as its load. The following discussion shows what happens on the HIGH to LOW transition. The driver's output impedance (Zs) is: 3.5 V - Llv = 3.5 V - 2.84 V = .066 v When the signal reaches the load, VL changes by -2.84 V from the original transmission and a further -2.84 V from the reflection. Since VL originally was 3.5 V, it is now -2.19 V. At the start, Vs = 0.66 V. The reflected signal returns to the source. Some of it is reflected perthe source KR. VS is equal to the sum of the original signal, the reflected signal, and the second reflected signal. The second reflection is equal to: vR = -.78 * -2.84 2.21 V vs= 0.66 v+ -2.84 v+ 2.21 -0.035 V v The second reflection goes to the load. When it arrives, vL = -2.19 + 2.21 + 2.21 = 2.24 =~"'8.3Q. 24 rnA A more accurate numbercan be obtained from an actual IN curve of the output. The signal continues like this, bouncing back and forth, getting smaller each time. This is illustrated in the lattice diagram in Figure 23. The lines at the left and right are the voltage at the source and load, respectively. The angled lines show the value of the transmitted signal and the reflections. Hlgh-Speed-Board Design Techniques 5-121 ~AMD Vs O.66V -2.19V -O.04V Time I 2.24V O.59V -1.21 V O.26V 1.5 16356A-023A Figure 23. Lattice Diagram Representation of a Reflected Signal The same information in the time domain is shown in Figure 24. The top part of the Figure shows the source; the bottom shows the load signal. Note that it takes five complete cycles forthe signal strength to drop below the input threshold. Propagation delays are typically frolll 2 nslft to 5 nslft. With tPD = 3 nslft and a 6-inch line, the delay across the line is about 1.5 ns. The signal can be safely considered valid at about 13.5 ns after the original transition. To understand this, look at the nature of the input and output impedances of the PAL devices. As noted above, input impedances tend to be high. Bipolar is in the 10 kn range, while CMOS is in the 100 kQ range. Output drivers tend to have low impedance. would be too much for most systems. A technique is There are two schemes for termination: reduce ZL to Zo to eliminate load reflections, or increase Zs to Zo to eliminate secondary reflections at the source. ZL can be reduced by placing a resistor in parallel with the load-parallel termination; Zs can be increased by placing a resistor in series with the source and the line-series termination. needed to eliminated, or at least reduce, the reflections. Since the reflections are eliminated when ZL = Zo, it is necessary to change ZL to equal Zoo Parallel termination is shown in Figure 25a. Because of the extremely high input resistance of most devices, RL can be made equal to Zoo TERMINATION The amount of reflections shown in the last example 5-122 High-Speed-Board Design Techniques AMD~ 3 ~ n; 2 c: .!2l C/) ~ :J o o C/) -1 Time (Unit Delay) a) - 3 ~ n; 2· Input ---- -- ---- ------- --- ----- ------ ------- ---- ------ -----r------,---- -.,------- --------- -- Threshold c: CI U5 "0 n:l .9 0 -1 . I I I I I I I 2t 4t 6t 8t lOt J I II I I 12t -2 16356A-024A Time (Unit Delay) b) Figure 24_ Time Representation of a Reflected Signal; a) at the Source b) at the Load Vee a) b) Zs RT ~ Dnver I Zo c) d) e) 16356A-025A Figure 25. a) Parallel Termination; b) Thevenin Equivalent; c) Active Termination; d) Series Capacitor; e) Series Termination High-Speed-Board Design Techniques 5-123 ~ AMD This scheme has one disadvantage: the current drain is high forthe HIGH-output state. For a 50-0 termination, it can be as much as 48 rnA. Most drivers are rated for an IOH of 3.2 rnA. This is clearly above the level that the device can support and still maintain an adequate VOH. Terminating to Vcc can help, since IOL is usually higher than IOH. However, most CMOS devices designed for board-level applications have drivers rated for an IOL of 24 rnA or less. This is still below the level that can support and maintain an adequate VOL for a low-impedance transmission line. RL = Zoo This half-transition tracks down the transmission line until it is reflected at the load, which is unterminated. Since the reflection causes the original half-transition to double, it brings the signal at the load to its final value (Figure 27a). The reflection then travels back up the line, completing the transition all along the line (Figure 27b). The current can be reduced considerably by using two resisters, as shown in Figure 25b. The resistors form a voltage divider with the Theveninvoltage equal to: Vs a) _ ~"'------"'vee * R2 Rl + R2 V TH - The Thevenin resistance is equal to: _ Rl * R2 Rl + R2 RTH - --=.._----!e. Although this is a good solution, there is higher powersupply current because the resistors are between Vee and ground. Another approach to reducing load current is to reference the resistor to a positive voltage between VOH and VOL (Figure 25c). The current flow from 3 V to 2.5 V through a 50-0 resistor is considerably less than the flow from 3 V to ground through the same resistor. This does not present any signal problems, because the DC voltage reference is AC ground. However, it is difficult to find a terminating voltage source that can switch from sinking current to sourcing current fast enough to respond to the transitions. Another technique is to replace the original terminating resistor with a resistor and capacitor series-RC network (Figure 25d). The resistor is equal to Zoo The capacitor can be on the order of 100 pF; the exact value is not important. At these frequencies, the capacitor is an AC short but it blocks DC. Thus the driver does not see the DC loading effect of RL. This technique is referred to as AC termination. ,- b) Figure 26. a) Series Termination; b) Voltage Divider formed by Series Termination This can be illustrated by putting a series terminating resistor on the unterminated microstrip example considered earlier. A 59-0 resistor (6S0 - 90) is placed in series with the driver. For a LOW to HIGH transition, the signal at the source is: ~V = (0.2 v - 3.5 V) Zs + Zo + 59 0 (0 . 2 8 Q v + 3. 5 V) 67 Q + Zo * 67 Q 59 Q -1. 65 V Vs = 3.5 v + 3.5 V - ~v 1. 65 V Techniques that terminate at the load are designed to eliminate the first reflection. An alternate approach is to increase Zs to equal Zo by placing a resistor in series with the source (Figure 25e). When added to Zs, this resistor makes the new source impedance look like Zoo 1. 85 V If the load is effectively an open circuit, then a -1.65 V reflection returns. When the reflected signal reaches the source, no new reflections occur because Zs is matched to Zo by RT. Vs is 1.85 V - 1.65 V = 0.2 V. This type of termination works best with a lumped load because the voltage divider formed by the Zs and Zo attenuates the signal (Figure 26 a and b). The original transition is cut in half by this voltage divider, since Zs + The reflection at the load causes VL to equal 0.2 V when the original signal arrives. Vs does not equal 0.2 V until the reflected signal returns, in this example, 3 ns later (Figure 27). 5-124 High-Speed-Board Design Techniques AMD~ 3 2 o o 1'[ Time (Unit Delay) a) 3 2 o I I o 1'[ Time (Unit Delay) . 2'[ 16356A-027A b) Figure 27. a) Signal at Source; b) Signal at Load end This can be a risky approach if the load is distributed along the line, since those loads not at the end of the line will see some intermediate voltage until the reflection cleans them up on its return to the source. In addition, this technique adds the delay of the return trip because the signal cannot be considered valid until the device closest to the driver has a valid input. The input to the device closest to the driver becomes valid upon the re- , turn of the reflection. The delay is longer than indicated in the last example because the added capacitance of the distributed load reduces Zo and increases tPD. Despite this drawback, series termination is successfully used with DRAM drivers, even when the DRAMs are distributed along the signal line. The risk of the signal spending time near threshold and the extra delay are reduced by choosing RT so that the resultant Zs is slightly less than Zoo The voltage swing at the line is larger, and the voltage level is closer to VOL, below the input threshold. If the line is terminated with 20 n, Vs becomes: Vs = = 3.5 V (0.2 v - 3.5 V)' Zo + --------~ Zs + Zo + 20 3.5 V + (0.2 V - 3.5 V) * 67 8 + 67 + 20 n = n n n Because the termination is not an exact match, some ringing occurs. However, if the ringing is below a tolerable level, it can be used successfully. The designer must decide on the compromise. Furthermore, the high capacitance of memory lines often swamps out the ringing. Often, an exact match is not possible because of the differences between the HIGH- and LOW-output impedances. The output impedance of TTL-compatible devices is different for HIGH and LOW levels. For example, the PALCE16V8 is 8 n when LOW, and about 50 n when HIGH. This complicates the choice of a terminating resistor because no single value is ideal for both cases. A compromise value must be chosen that results in acceptable results in both transition directions. Layout Rules for Transmission Lines The controlled impedance signal line is the best practical medium for signal transfer on a board, and proper termination helps ensure proper noise-free operation. However, it is still possible to generate noise with an inefficient layout. The following layout rules further enhance board operation. n 1.17 V High-Speed-Board Design Techniques 5-125 ~ AMD 1. Avoid Discontinuities Discontinuities are pOints where the impedance of the signal line changes abruptly; they cause reflections. The formula for KR is as valid here as it as at the end of the line. Because they cause reflections, they should be avoided. Discontinuities can be at sharp bends on the trace or at vias through the board. At bends on the trace, the cross-sectional area increases, and Zo decreases. It is possible to compensate for the bend by cutting the trace as shown in Figure 28. The cut is chosen so that the resulting diagonal is equal to the trace width. This minimizes the delta in cross-sec- tional area, as well as the discontinuity. Using two 45°. bends makes use of the same concept and is a common way of smoothing out bends. A smooth circular arc would be ideal but is harderto generate with many tools. Vias take signals through the board to the other side (Figure 29). The vertical run of metal between layers is an uncontrolled impedance, and the more of these there are, the greater is the overall amount of uncontrolled impedance in the line. This contributes to reflections. Also, the 90° bend from horizontal to vertical is a discontinuity that generates reflections. If vias cannot be avoided, use as few as possible. w a) b) w w w w c) d} 16356A-028A Figure 28. Reducing Discontinuity. a) Corner on PC Board Trace which Causes Discontinuity; Solved: b) by Shaving the Edge; c) by 45° Corner; d) by Using Curves 5-126 High-Speed-Board Design Techniques AMD~ Uncontrolled Impedance I.-------------------x--------------------·~I a) T 1 16356A-029A b) Figure 29. a) Excessive Number of Vias; b) Preferred Solution Note that changing from an outer layer to an inside layer (or vice versa) generates an impedance change, since the design effectively is changed from stripline to microstrip (or vice versa). While it is theoretically possible to change geometries to compensate and keep impedances the same, it is very difficult to do so in production. The best results are obtained if outside signals remain outside, and inside signals remain inside. 2. Do Not Use Stubs or Ts When laying out the signal lines, it is often convenient to run stubs or Ts to the devices, similar to Figure 30a. Stubs and Ts can be noise sources. If long enough, they are transmission lines with the main line as the source and are subject to the same type of reflections. The signal lines should avoid long stubs and Ts. As long as the stubs are very short, a single line can be used with a single termination at the end, although Zo must then be derated to account forthe distributed load. Given the example in Figure 30a, if the stubs are too long, the Signal line could be made into two signal lines, as shown in Figure 30b. Both are transmission lines and require terminating; however, this is preferable to terminating each long stub individually. High-Speed-Board Design Techniques 5-127 ~AMD y 1 Y1 Y1 Y11 I 1 Y1 Y a) < 0 0 0 0 0 0 0 0 0 0 0 ,0 b) : 16356A-030A Figure 30. a) Stubs off of Transmission Line; b) Preferred Solution Noise Source [>~~--"I---[>- C Tracc Noise Receiver a) Noise Source Trace ~ Noise Receiver t----O Trace c) b) 16356A-031A Figure 31. a) Capacitive Crosstalk; b) Equivalent Circuit; c) Solution 3. Crosstalk Crosstalk is the unwanted coupling of signals between traces. It is either capacitive or inductive. Crosstalk can be handled effectively by following a few simple rules. 3.1 Capacitive Crosstalk -Capacitive crosstalk refers to the capacitive coupling of signals between Signal lines. It occurs when the lines are close to each other for some distance. 5-128 The circuit representation in Figure 31 shows two Signal lines, called the noise source and the noise receiver. Because of capacitance between the lines, noise on the source can be coupled onto the receiving line. This occurs in the form of current injected into the receiving line. In a transmission line, the current sees Zo in both directions, and propagates both ways, until it can be dissipated across the source and load. The voltage spike this causes on the line is determined by Zoo When the current High·Speed-Board Design Techniques AMD~ pulse gets to Zs and ZL, it dissipates across these resistors with a voltage proportional to the impedance. If there is an impedance mismatch at the source or load, reflections occur. In the case of an unterminated load, the voltage spike across ZL can be very large. Terminating the load can significantly reduce the voltage noise seen at the input of the next device. Capacitive crosstalk can also be reduced by separating the traces. The farther apart the signal traces are, the less the capacitance, and the smaller the crosstalk. Space constraints on the board may put limits on how far apart the signal lines can be placed. An alternate approach is to put a ground trace between adjacent-signal lines, as shown in Figure 32. The signal is now coupled to ground, not to the adjacent-signal line. Note that the ground trace must be a solid ground. If it is only connected to the ground plane at the trace ends, the trace has a relatively high impedance. For good grounding, the ground trace should be connected to the ground plane with taps separated a quarter wavelength (A/4) of the highest frequency component of the signal. 16356A-032A Figure 32. Isolating Traces with a Ground Trace The wavelength is the distance the signal travels in a single period or: It = vel * Perio'd =_1 * __ 1_ tpD freq With digital signals, the highest significant frequency harmonic of interest is usually assumed to be 1httR. Consider an example where tR = 1.25 ns (possible for PAL 16R8-4 devices). The upper-frequency component is: f MAX = _ _~1,---_ _ l.25 ns * n 255 MHz The distributed load delay for our example in section 2' was 4.14 ns/ft. A. is equal to the period divided by tPD. It = 1 255 MHz * _--=1"---_ * 12 in 4.14...Qg ft ft 1l. 4 in 1t/4 =~ 4 = 2.8 in 3.2 Inductive Crosstalk Inductive crosstalk can be thought of as the coupling of signals between the primary and secondary coils of an unwanted transformer (Figure 33). The transformer windings are the current loops on the board (or system). These can be either artificial loops inadvertently created by inefficient layout (Figure 34a) or natural loops resulting from the combination of the signal path and the signal return path (Figure 34b). Artificial loops are sometimes hard to locate, but can be eliminated as shown in Figure 34c. The amount of unwanted signal coupled to the load depends on the proximity and size of the loops, as well as the impedance of the affected load. The amount of energy transferred increases as the loops become larger and get closer together. The size of the signal seen at the load, on the secondary loop, increases with the load impedance. 3.2.1 Loop Size and Proximity The inductance of a loop (L) increases with loop size. When two loops interact, one will have a primary inductance (LP) and the other will have a secondary inductance (LS), as shown in Figure 33b. Because the signal lines are not purposely designed to be transformers, the coupling is loose; however, it can create interference on the secondary loop. For maximum isolation, the ground trace must have taps to the ground plane no more than 2.8 inches apart. High-Speed-Board Design Techniques 5-129 ~AMD 16356A-033A b) a) Figure 33. a) Inductive Crosstalk; b) Transformer Equivalent C?( ......... , It;> ~, , ~ ,, ' '' ,, C?: : , I I ~ (,-,' ~ a) ,. I ~ 11:;> ',_/) ~ c) b) 16356A-034A Figure 34. a) Artificial Loops; b) Schematic Equivalent; c) Solution 16356A-035A a) b) c) Figure 35. a) Common Return Path; b) Loop; c) Autotransformer Equivalent Circuit If portions of the return paths of two signal lines coincide, the resulting loops might form an auto-transformer (Figure 35 a and c). An example of this is the VME-backplane example discussed above. Ensuring that each signal has its own return path can eliminate this source of crosstalk. 3.2.2 Load Impedance If inductive crosstalk comes about due to artificial loops, the solution is to open the loops. Unfortunately, locating the loops can often be a challenge. If the crosstalk is generated by natural signal/return-signal loops, then 5-130 clearly the loop cannot broken. But by keeping the load impedance low, the effect of the crosstalk can be minimized. Figure 36 shows a simplified schematic representation of a secondary "natural" loop with a load. Here Zs is the intrinsic impedance of the secondary loop. Note the series current (is). Because the impedances are in series, is is the same everywhere in the loop. With a constant is, the voltage drop is largest across the largest impedance. On an unterminated line, this usually is the load at the end of the line; i.e., at the input of the receiving device. Hlgh-Speed-Board Design Techniques AMD~ Zs 4.1 Loops Signal Line fg--r--i-S "M..------='-----1 § Ground Return Path f Current loops are an unavoidable part of every design. RIN = 10 ill to 100 kn They act as antennae. Minimizingthe effects of loops on or greater 16356A-036A Figure 36. Series Inductive Loop Large noise signals are most unwanted at the inputs, where noise signals should be minimaL If the maximum signal is developed across the largest impedance, the signal developed at the input can be reduced by terminating the signal line at the receiver end, which reduces RIN to RT. RT is usually in the 30 .n to 150 .n range. This reduction in RIN is at least two orders of magnitude. The voltage drop across RIN is reduced accordingly. The exact drop is difficult to predict because it depends on the value of Zs , which is difficult to determine. But reducing RIN by orders of magnitude should have a significant effect. 3.3 Crosstalk Solutions Summary The following steps summarize the ways in which the effects of crosstalk can be minimized. 1. The effect of both capacitive and inductive crosstalk increases with load impedance. Thus all lines susceptible to interference due to crosstalk should be terminated at the line impedance. 2.. Keeping the signal lines separated reduces the energy that can be capacitively coupled between signal lines. 3. Capacitive coupling can be reduced by separating the signal lines by a ground line. To be effective, the ground trace should be connected to the ground plane every Al4 inches. 4. For inductive crosstalk, the loop size should be reduced as much as possible. Where possible, loops should be eliminated. 5. For inductive crosstalk, avoid situations where signal return lines share a common path. 4. ELECTRO-MAGNETIC INTERFERENCE (EMI) . EMI is becoming more critical with speed. High-speed devices are naturally more susceptible to interference. They accept fast glitches, which slower devices ignore. Even if the board or system is not susceptible, the FCC in the United States, along with VDE and CCITT in Europe, places severe limitations on the high-frequency noise (both radiated and line noise) that the board can generate. The designer can reduce EMI through shielding, filtering, eliminating current loops, and reducing device speed where possible. Although shielding is outside the scope of this article, all the other issues are discussed as follows. EMI means minimizing the number of loops and the antenna efficiency of the loops. Do not create artificial loops; and keep the natural loops as small as possible. 1. Avoid artificial loops by ensuring that each signal line has only one path between any two points. 2. Use power planes whenever possible. Ground planes automatically result in the smallest natural current loop. When using ground planes, ensure that the signal-return line path is not blocked. If power buses are necessary, have the fast-signal lines run either over or next to a power bus. 4.2 Filtering Filtering is standard for power lines. It can also be used on signal lines, but is recommended only as a last resort, when the source of the signal noise cannot be eliminated. Three options are available for filtering: bypass capacitors, EMI filters, and ferrite beads. Bypass capacitors are discussed in section 1. EMI filters are commercially available filters; they are available over a wide frequency range. Ferrite beads are ferrite ceramics that add inductance to any wire within their proximity. They are used as high-frequency suppressors. 4.2.1 EMf Filters EMI filters are commercially manufactured devices designed to attenuate high-frequency noise. They are used primarily to filter out noise in power lines. They act to isolate the power outside the system (referred to as the line) from the power inside the system (referred to as the load). Their effect is bi-directional: they filter out noise going into, and coming out of, the device or board. EMI filters consist of combinations of inductors and capacitors. In general, the configuration to use depends on the impedance of the nodes to be connected. A capacitor should be connected to a high-impedance node; an inductor should be connected to a low-impedance node. EM I filters are available in variations of the following configurations: feedthrough capacitor, L-Circuit, PI-Circuit, and T-Circuit. • • The feedthrough capacitor's only component is a capacitor (Figure 37a). It is a good choice when the impedances connected to the filter are high. Note that it provides no high-frequency current isolation between nodes. The L-Circuit has an inductoron one side of a capacitance (Figure 37b). It works best when the line and load have a large difference in impedance. The inductive element is connected to the lowest impedance. High-Speed-Board Design Techniques 5-131 ~AMD • The PI-Circuit has an inductor surrounded by two capacitors (Figure 37c). PI filters are best when the line and load impedances are high and when high levels of attenuation are needed. • The T-filter has inductors on either side of the capacitor in a T fashion (Figure 37d). It is a good choice when both line and load impedances are low. LC filters are rated according to insertion loss, which is the amount of signal lost due to the insertion of the filter. Insertion loss is usually stated in decibels. Filter manu"-~r--",,, 0 I facturers provide graphs of their filters over prescribed frequency ranges. 4.2.2 Ferrite Noise Suppressors Ferrite noise suppressors are ferrite ceramics placed in proximity to the conducting material. They are available as beads for single wires and clamps for cables. When using beads, the wire is placed through a hole in the bead (Figure 38a). When using clamps, the ferrite material is clamped around the cable (Figure 38b). Clamps are popular with ribbon cable. 0 b) a) 0 I ml5' I ~ 0 I d) c) 16356A-037A Figure 37. LIne-Noise Filters. a) Capacitor; b) LC Filter; c) PI Filter; d) T Filter a) b) Figure 38. a) Ferrite Bead; b) Ferrite Clamp 5-132 High-Speed-Board Design Techniques AMD~ Ferrite suppressors work by adding inductance in series with the line (Figure 39). Ferrite manufactures supply graphs similar to those in Figure 40, which shows the added impedance as a function of frequency. The system designer must determine the insertion loss. The formula is: where Zs = Source Impedance ZL = Load Impedance ZF = Ferrite Impedance Figure 39. Ferrite Filter Equivalent Circuit 1000 9: CD 100 u c: CIl "0 CD a. .§ 10 1 1M 10M 100M Frequency (Hz) 1G 16356A·040A Figure 40. Frequency Response of Ferrite Filter Ferrite suppressors add inductance to the line without adding DC resistance. This makes an ideal choice for line-noise suppressors on the Vcc pins of devices. Because ferrite beads are small and easy to handle, they can sometimes be used in signal lines to suppress high-frequency noise signals. This is not recommended for two reasons: first, it masks the cause of most problems; second because it might affect the edge rates of the signal. However, when the board is already laid out, ferrite beads can be used on noisy signal lines as a last resort.- 4.3 Device Speed Faster devices, by definition, have shorter transition times. Because shorter transition times have more energy in the high-frequency range, faster devices can generate more high-frequency noise. Figure 41 a is an outline of the Fou rier transform of a square wave (Figure 41 b). There are two corners of interest: 1IntL(this frequency is determined by the period of the signal) and 1/ntf (determined by the transition time of the signal; this is also the frequency we used in determining the wavelength in the discussion on capacitive coupling). After 1/ntf, the curve drops off very rapidly. For practical purposes, 1/7ttf is the highest significant frequency component of the signal. The less energy a device generates in a given frequency range, the less noise can be radiated in that range. Hlgh-Speed-Board Design Techniques 5-133 ~ AMD a) 1 1ttL 1 1 : 1ttR or 1ttF ,, b) Figure 41. a) Single Pulse; b) Fourier Transform of Pulse For example, the PAL 16R8-4 series has a typicaltransition time of 2 ns. It can be as short as 1.25 ns. The frequency component of the edge is: f = _ _~1L-_ _ 1t * 1.25 ns 254 MHz The output signal has a high-frequency component of 254 MHz, regardless of the clock frequency. Because of the high-frequency component, the board might require extra filtering and possibly shielding in order to comply with EMI emissions restrictions required by regulatory agencies. If the system speed requirements are high enough (for example, clock rates over 80 MHz), devices this fast must be used, and the extra effort required to meet compliance is justified. However, if a slower device can meet system requirements, it should be used. By virtue of the longer transition time, the slower device generates less energy at the higher frequencies. In general, try to use devices that are fast enough to meet the system requirements, but no faster. SUMMARY While faster technologies provide the theoretical possibility of faster systems, extra care must be taken to turn 5-134 16356A-041 A this possibility into reality. The largest noise components can be eliminated by addressing the fOllowing: • • integrity and stability of power and ground; termination and careful layout of transmission lines to eliminate reflections; • termination and careful layout to reduce the effects of capacitive and inductive crosstalk; • noise suppression for compliance with radiation regulations. There are many other second-order issues that could be addressed, but that are beyond the scope of the application note. Some references for additional information aie listed below. 1. Sherman lee, Mark McClain, Dave Stoenner. "Am29000 32-Bit Streamlined Instruction Processor Memory Design Handbook," Advanced Micro Devices Inc., Sunnyvale, CA, Appendix A, Memory Array loading Calculations. 2. William R. Blood Jr. "ECl Systems Design Handbook," Motorola Semiconductor Products Inc., Mesa, AZ, May, 1983 (Fourth Edition) Chapters 3 and 7. 3. Ramo, Whinnery, and Van Duzer, "Fields and Waves in Communications Electronics," John Whilley & Sons, 1965, Chapter 1. Hlgh-Speed-Board Design Techniques ~ Minimizing Power Consumption with Zero-Power PLDs Advanced Micro Devices Application Note by Shawn D. Worsell, Senior Applications Engineer Zero-Power Programmable Logic Devices (PLDs) are advanced PAL devices designed with ultra low-power, high-speed, electrically-erasable CMOS technology. ZPAL1M devices provide zero-standby power and high speed for a variety of applications. At 15 JlA maximum standby current, Zero-Power devices allow battery powered operation for an extended period of time. ZeroPower CMOS devices can significantly reduce system power consumption by replacing equivalent CMOS and TTL devices. ZPAL devices are available in the following configurations: the industry-standard 20-pin PALC E16V8Z family and the AMD-patented 24-pin PALCE22V1 OZ family. The PALCE16V8Z family is functionally compatible with all CMOS half- and quarter-power PALCE16V8 devices and will directly replace the bipolar PAL16R8 and PAL 10H8 series devices with the exception of the PAL16C1. The PALCE22V10Z family is functionally compatible with all 24-pin 22V1 0 devices, and it provides user-programmable logic for replacing conventional low-power CMOS SSI/MSI gates and flip-flops ata reduced chip count. POTENTIAL APPLICATIONS ZPAL devices may be used in any application where a standard 22V1 0 or 16V8 device would be used. DeSigns that are currently in a 20V8 will also fit in the 24-pin 22V10 device. In addition, they are ideal for low-frequency or low-duty-cycle environments such as line card and peripheral applications, where low power consumption is a priority. Laptop computers and other battery-operated or backed-up equipment, such as hand-held meters and portable communication units, would benefit from the Zero-Power devices. The PALCE16V8Z and PALCE22V10Z feature a zerostandby power mode. When none of the inputs switch for an extended period, the device will go into standby mode, shutting down most of its internal circuitry, causing the current consumption to drop to almost zero (15 JlA). The outputs will maintain the states held before the device went into the standby mode. When any input switches, the internal circuitry is fully enabled, and power consumption returns to normal. Since all of the features which cause the device to be a Zero-Power PLD are internal, the 16V8Z and the 22V10Z PAL devices are pin-for-pin and JEDEC-file compatible with existing devices of the same families. Publication# 16948 Rev. A Issue Date: June 1992 Amendment/O HINTS ON MINIMIZING POWER CONSUMPTION The quintessential feature of the ZPAL device's current reducing operation is the "sleep mode." When the device is inactive for a period of time, certain portions of the PLD can be disabled or "put to sleep" by the presence of input transition detection circuitry. Any switching delays of about 50 nanoseconds or more, forthe entire device, will place the PLD in sleep mode. This means that none of the inputs, including the clock, can be switching in order to utilize the Significant power saving features. Therefore, during the deSign phase, special attention must be given to every signal that is being transferred to the device, so that the sleep mode feature may be engaged as much as possible. Refer to Figure 1 for ZPAL device features. Inputs Transition Detectors Input Buffer Programmable AND Array OR Plane Array Latches Macrocells, Output Buffers Transition Detectors liD Pins 16948A-001A Figure 1. ZPAL Device Features 5-135 ~ AMD As sleep mode is asserted, all outputs will latch to the state they were in when the last input transition occurred. This state will be held as long as the device is asleep. When an input does experience a transition, the device will ''wake up." The wake-up delay associated with the initial transition is included in the determined propagation delay of the device. Therefore, there are no extra system delays to consider when the device utilizes the standby feature, since it quickly returns to full operation. However, if a designer is more interested in high speed, such as a burst mode application, a typical propagation delay would be closer to 5 nanoseconds faster while the device is awake. The inputs and I/O pins are monitored by an input transition detection circuit. Any transition on any pin, including noise spikes, will disengage the sleep mode. Thus, during the design phase, care should be taken to ensure that al/ circuitry associated with the Zero-Power device is as quiet as possible. Effects of Frequency The sleep mode benefits are best realized in combinatorial applications since sequential functions will be powered·up with every edge of the clock. However, significant power savings can still be realized when the clock is stopped or operating at a modest rate. For instance, when used in an application with a 5-MHz clock (200-ns period), the power consumption of the ZPAL device will be reduced by 50% from the consumption at the maximum frequency. The designer must also be careful when considering the true operating frequency. If the fastest input is 1 MHz, but there are two inputs 90 degrees out of phase, even though a transition occurs on each input every half-cycle (500 ns); there is a transition on some input every quarter cycle (250 ns). Thus, the effective frequency is doubled, and the 2-MHz point should be used to calculate the power consumption. This is not as important when the signal in question is greater than 10 MHz, since with a 50% duty cycle the part would never enter sleep mode. Another way of realizing power savings is by decreasing the duty cycle of the clock and input signals. Dynamic Icc is a function of duty cycle and frequency. The two are mutually exclusive. As previously mentioned, at a frequencyof 10 MHz with a 50% duty cycle, the part would always be powered up. However,if the duty cycle was decreased to 20%, a 10-MHz Signal would cause the part to shut down for 30 nanoseconds (80-50 ns), thereby reducing the Dynamic Icc of the device. Referring to the typical Icc versus Frequency Graphs of the corresponding data sheets, will clearly indicate the power savings that may be realized by optimizing the ZPAL device's operating frequency. 5-136 Effects of Product Terms To further reduce power consumption, unused product terms are permanently disabled during programming. Each product term has a Sense Amp Off (SAOFF) bit, which will be programmed when the corresponding product term is unused, thereby shutting off the sense amp to save power. Note that the SAOFF bit is automatically configured and has no effect on the JEDEC file, so the designer does not even have to think about it. A typical power savings of approximately 300 J.lA per unused product term will be achieved. Thus, a logic design that utilizes a minimum number of product terms will result in the maximum amount of power savings. Inverting simplified logic by using DeMorgan's theorem and changing the output polarity is one way that a designer may easily reduce product term reqUirements. For example, in the equation z=X+y the "OR" function denoted by the "+" sign, requires two product terms, one for each variable. However, if both sides of the equation are inverted to become fZ=/(X+Y) using DeMorgan's theorem yields /(X+Y) = /X*/Y Here the "AND" function denoted by the "*" sign does not use the second product term, because one product term is shared by both variables. The resultant equation is now fZ=/X*/Y and by switching the output polarity, the logic behaves the same way as it was originally intended as well as reducing a product term requirement. The choice of output polarity itself does not save power, but if chosen wisely, it may help to reduce product term usage. It should also be noted, however, that switching output polarity will also invert the Synchronous Preset (SP) and Asynchronous Reset (AR) functions at the output of the register of a 22V10. 1/0 Characteristics The output stage of the Zero-Power PAL devices consist of a P-channel pull-up transistor and an N-channel pull-down transistor, a true CMOS output with rail-torail switching. A P-channel pull-up is better for lowpower applications than an N-channel transistor, since P-channel outputs can be driven up to the Vcc level. This ensures that the following input buffer draws no current. The same amount of current is available for both high and low outputs with a P-channel pull-up, unlike the unequal, although slightly faster N-channel transistor used in other PLD products. Minimizing Power Consumption with Zero-Power PLDs AMol1 These devices are capable of driving and being driven by either TTL or CMOS devices. They are compatible with both HC and HCT standard specifications as illustrated in Table 1. Table 1 Parameter HC HCT ZPAL VIH 3.15 V 2.0V 2.0V VIL 0.9V 0.8 V 0.9V Va..@6mA 0.33 V 0.33 V 0.33 V Va..@20~ 0.1 V 0.1 V 0.1 V VOH@6mA 3.84 V 3.84 V 3.84 V VOH@ 20~ ously, it is important to ensure that the ground path on the circuit board has low inductance, and to reduce the loading on the outputs. The lower-lead-inductance PLCC package will also reduce the possibility of ground bounce since the bonding wires are 1/4 the length of a DIP's bond wires. Effects of Loading Power dissipation of the outputs is greatly affected by the load. To minimize power dissipation, ZPAL device loads should have no DC components. If termination is required, an AC terminator like the one in Figure 2 should be used to eliminate DC power drain. VOPinT: Vcc-0.1 V Vcc -0.1 V Vcc -0.1 V One minor disadvantage of true CMOS outputs is the intrinsic SCR circuit that is developed in the CMOS structure and cannot be eliminated. The SCR has been made as difficult as possible to turn on by using guard rings and carefully laying out all input and output circuits. An excess current of 100 rnA would be required on an individual pin to induce "latch-up". However, there is a potential to latch-up a ZPAL device through hot-socket insertion. If there is a possibility of the device or board being instantly powered-up, design care must be taken. Effects of Reducing 1/0 Switching Since each input will draw up to a maximum of 5 rnA each time it switches, additional power savings are possible by reducing the number of inputs used. A CMOS transistor pair only draws current when it is switching or floating in an intermediate region, so unused inputs should be externally tied either HIGH or LOW. The number of outputs switching will also affect the amount of current consumption. Therefore, minimizing output switching will also help to reduce the amount of current required. The potential for ground bounce problems will also be reduced by limiting the number of outputs switching. Built-in slew-rate-limiting circuits will help to slow down the fast CMOS falling edges that contribute to ground bounce. A fall rate of 1.25 V/ns is typical of the faster N-channel pull-down transistor, while a slower rise rate of 1 V/ns can be expected from the P-channel pull-up. If most of the outputs are required to switch simultane- JC 16948A-002A Figure 2. Typical AC Terminator The capacitance should generally be kept as low as possible since the output stage will go through a process of constantly charging and discharging the capacitor. The formula for current consumption due to loading is i = CLVsjO where i is the current, CL is the capacitive load, Vs is the voltage swing, and to is the frequency at which the output is switching. Therefore, current is consumed every high transition since the capacitor has to recharge. SUMMARY ZPAL devices provide zero-standby power and high speed for a variety of applications. Zero-Power CMOS devices can significantly reduce system power consumption by replacing equivalent CMOS and TTL devices. Since all of the featu res which cause the device to be a Zero-Power PLD are internal, the 16V8Z and the 22V10Z PAL devices may be used in any application where a standard 22V1 0 or 16V8 device would be used. With a little extra attention given to the particular design involving a Zero-Power PAL device, a designer can realize significant system power savings. Minimizing Power Consumption with Zero-Power PLOs 5-137 ~ Designing with the PALCE16V8HD Advanced Micro Devices Application Note The PALCE16V8HD is a High Drive PLD that combines the popular PALCE16V8 architecture with 64-mA current drivers. It has application anywhere a reasonably complicated function must control a high-current signal line. Examples of this are control signals for buses such as the VME bus, combination decoder and LED driver, 3 V translator, and control lines for memory. device. The extra device increases part count and increases total propagation delay. The PALCE16V8HD is an ideal replacement forthe pair of devices. First, it combines the full functionality of a PAL device and 64-mA current sink capability into one device. Second, it provides input latches to store information. Third, it provides the choice of either totem-pole oropen-drain output drivers. Finally, it has a minimum of 200-mV hysteresis on all input and 1/0 pins for increased noise immunity. The macrocell is shown in Figure 1. Previously, this problem was solved by using a programmable device, such as a PAL device, followed by a highcurrent driver. The drawback to this solution is that it requires two devices to perform the functionality of one ~==~------------------------------------~11 Vcc 1 01-----, To Adjacent 00 Macrocell 01 SLOx SG1 I/OX From Adjacent Pin ·SG1 a) 1ssnA-1 ·'n macrocells MCo and MC7 SGt is replaced by SGO on ,the feedback multiplexer. b) Figure 1. a) PALCE16V8HD 1/0 Macrocell; b) PALCE16V8HD Input Macrocell 5-138 AmendmentlO AMD~ The purpose of this application note is to guide the engineer in using the PALCE16V8HD high drive-current PAL device. What follows is a discussion of its features and hints on how to maximize their benefits. Input Latches ently in the state opposite the desired initialized state. An example is shown below. OUT!. T = A*B*OUTl * /OUT2 iapplication iequations + /A*/B*/OUT1* All inputs to the device (including 1/0 pins) can be programmed as a transparent latch. The latch enable (LE) signal is pin 4 (pin 5 for PLCC). The latches are transparent when LE is HIGH and latched when LE is LOW. Input latches provide a convenient way to speed up data transfer while providing storage at the same time. With latches, the new data is available instantaneously, while with registers, the data is not available until the next clock edge. The input-latch setup time is also short because the product-term array is bypassed. The register-setup time is 10 ns, while the latch-setup time is only 4 ns. This last feature makes it possible to capture signals which tend to occur late in the clock cycle. The PALCE16V8HD is designed so that the transparent latch does not add extra delay to the signal. This makes the PALCE16V8HD faster than the usual PAL device and input latch combination. /OUT2 + INIT LOW*OUTl + INIT_HIGH*/OUTl ; " itoggles to iLOW if HIGH itoggles to iHIGH if LOW Output Driver Configurations The output macrocells can be independently configured as either totem-pole outputs or open-drain outputs. In the totem-pole configuration the driver has an n-channel pull-down transistor and an n-channel pull-up transistor. The output-LOW voltage is typically 0.3 V and the output-HIGH voltage is typically 3.5 V. In the open-drain configuration, the n-channel pull-up transistor off. The output-LOW voltage is the same as in the totem-pole configuration. The output-HIGH voltage is determined by the termination voltage and the load on the signalline. Totem-Pole Output Configuration Register Configurations The output registers can be configured as either D-type or T-type flip-flops. The D-type register's Q output is equal to the D input at the rising edge of the clock. The T-type registers output toggles when the corresponding T input is HIGH and maintains the current state when the T input is LOW. The T-type register's action is also determined at the rising edge of the clock signal. The D-type register has the advantage of simplicity. It is straight forward and the easiest to determine the desired results. In many applications, such as counters, the T-type flipflop requires fewer product terms. If the application is large enough, this can mean the difference between fitting or not fitting on the part. T-type flip-flops tend to reduce the number of product terms required in sequential applications such as counters and state-machines. They are not optimal for all applications. For example, to load data takes two product terms with a T-type flip-flop and only one product term with a D-type flip-flop. Because the PALCE16V8HD can be configured as either type, it has a wider range of application than either one alone. Note that because there is no global initialization function, deSigners should be sure that T-type flip-flops have a direct way of being initialized. This can most easily be done by ensuring that there is at least one LOAD product term. A LOAD product term is one that can be activated with no other product term active, and which will have the flip-flop toggle only if the flip-flop state is pres- The totem-pole output driver conforms to TTL specifications. Because the pull-up transistor is n-channel, the output-HIGH voltage is 1 to 1.5 volts below Vcc, even under zero-load conditions. If rail-to-rail swings are required, VOH can be raised by terminating the output with a resistor to Vee. Open-Drain Outputs The open-drain configuration has two types of applications. The first is as a switch to apply power to devices such as a small lamp or even a small motor. The second is as an active-LOW signal source on a signal line with multiple asynchronous drivers. Because the pull-up transistor is always off, the risk of device damage due to contention is eliminated. If the output structure were totem-pole only, the state of the output would have to be set internally and then the output would be enabled. The open-drain structure eliminates this extra complexity. An example of asynchronous signals on one line is the bus requestforVME buses. For open-drain outputs, the request line simply goes LOW. Simultaneous requests must be handled by an arbiter, but there is no risk of circuit damage due to bus contention. Output Terminations The anticipated applications for the PALCE16V8HD include relatively long signal-line lengths. At the rise and fall times of this device (2 to 3 ns), the signal lines resemble transmission lines. The transmission line's propensity for reflections requires the use of terminating techniques. It is beyond the scope of this article to pre- Designing with the PALCE16V8HD 5-139 ~ AMD sent a detailed discussion of transmission lines, but we will discuss some of the more popular termination techniques that lend themselves well to high-drive devices. Those interested in studying transmission lines in more detail are encouraged to check out the references at the end of this application note. Terminating Totem-Pole Outputs The most effective termination is a resistor at the load end of the signal line that is equal to the line impedance (Figure 2a). Because the input impedance of all PLDs is greater than 10 kn, this is effectively an exact match. Because parallel termination increases the DC load, the output will be degraded for both VOH and VOL. The IN curves for the device (Figure 3) can be used with a load line to determine the degraded output levels. The dashed lines indicate simple termination to Vcc or ground. RT RTh = R1 ~VBias II R2 VTh=Vee~ R1+ R2 c) b) a} Vec Vee d) e) 16677A-2 Figure 2. Termination Alternatives: a) Parallel; b) Active; c) Th~venln; d) AC; e) Multiple Drivers 0 2 VOH (Volts) 3 4 5 6 0 100n -20 75n 50n --o_ _ _zo-..JRT a) b) VSias c) 16677A-6 Figure 6. Termination for Open-Collector Outputs: a) Single Driver; b) Multiple Drivers; c) Alternative If Load Exceeds If multiple drivers are used, terminating resistors should be placed at both ends of the signal line as in Figure 6b. The DC load is the parallel combination of the loads at both ends of the signal line, effectively doubling the DC load. The DC current load can be reduced by connecting the terminating resistors to a bias voltage that is less than Vcc as in Figure 6c. Because the PAL device provides no pullup in the open-collector configuration, VSIAS is VOH; therefore VSIAS must be high enough to ensure a valid logic HIGH to all the device inputs on the signal line. signal line. In addition, series-terminated signals can hover at a mid-range until reflections cause them to settle to their final value. The mid-range is usually dangerously close to the input thresholds of most devices. One technique that increases the devices tolerance to noise is hysteresis. Hysteresis moves the input threshold in the direction that requires a larger signal. For ex-' ample, when the input signal crosses the threshold on a positive transition, the threshold moves lower (Figure 7). The input signal would have to move lower to cross the new threshold. Therefore, even a noisy Signal will cause a single croSSing per transition, not the multiple crossings that might occur without hysteresis. Hysteresis Buses tend to operate in a noisy environment. Voltage spikes from crosstalk can be expected on almost every 5-142 10L Designing with the PALCE16V8HD AMD~ v Input Signal --- ----------- --- ---- '~~~t~~~~ --Th~~~h~ld- Threshold Crossings a) v Input Signal Single Threshold Crossing ~ I -- --- --- --- --- --- --- --- --- --- --- --- --- --- --- -_ .. _Threshold b) 166nA-7 Figure 7. Response to Noisy Inputs: a) Device with no Input Hysteresis; b) Device with Input Hysteresis The PALCE16V8HD comes with a minimum of 200 mV hysteresis. Therefore, if the noise occurs when the input signal level is at the exact middle of the threshold range, the device can tolerate 200 mV peak-to-peak of noise before detecting multiple transitions. ing current. Because there is less current in anyone ground pin, there is less energy available to generate ground bounce. Slew rate limiting reduces the current surge in the ground pin. Together these features keep ground bounce down to a tolerable level. Ground Bounce The following table shows ground-bounce data measured underworst-case conditions: a" eight macroce"s in the registered configuration and seven outputs switching simultaneously. The measured pin is in the LOW state during the test. The test was performed with the fo"owingACtest load:80nto Vec, 160ntoground, and 50 pF to ground. An issue that is associated with high-current CMOS is ground bounce. This phenomenon is usually first noticed as a pulse or ringing on a LOW output. If these pulses are large enough they can generate false clocking or false data. The design of the PALCE16V8HD takes ground bounce into consideration so that it is minimized. The PALCE16V8HD has two features which make it resistant to ground bounce: multiple ground pins and slew rate limiting. The multiple ground pins share the switch- Note that the maximum transient VOL is 1.3 V, which shows that ground bounce on the PALCE16V8HD is equal to, or better than, ground bounce on devices with only 24-mA drive current capability. Designing with the PALCE16V8HD 5-143 ~ AMD transistors, which will not conduct or cause latchup when the output signal is higher than Vec. Ground Bounce Peak Voltage with Seven Outputs Switching LOW Pint Peak Voltage 23 22 20 19 16 1.2V 1.1 V 1.0 V 1.1 V 1.2 V 1.3 V 1.3 V 1.2 V 15 14 13 . Hot socketing has two manifestations that concern us here: signal and ground connected before Vee, or signal . and Vee connected before ground. If ground and signal lines are connected to the device before Vce is, the effect on the device is similar to that presented by the power-down situation; the device inputs and outputs do not conduct current. Power Supply Considerations There are two special situations which should be considered for devices used in bus applications: power down with live signals and "hot-socketing." Disabling Vce of certain sections of a system is a common technique for power conservation. The devices on the disabled board should be able to tolerate live signals on the input and I/O pins while power is down. "Hot-socketing," the technique of inserting or removing boards while power is applied, is never recommended; however, almost every system is subjected to it at one time or another. The PALCE16VSHD can withstand the stresses brought about by power shutdown and hot socketing. During power shutdown, Vee is either open or shorted to ground. The PALCE16VSHD has n-channel pull-up 0 2 VI/O Volts 4 3 5 When the ground pin is open with Vee and the inputs or outputs connected, the pull-up transistors conduct. This is because a solid ground is necessary to establish the proper bias voltages to allow the output transistors to turn off. If ground is floating the output transistors are biased on. Figure Sa shows the IN curves on an I/O pin with Vec on and ground open. The driver starts to conduct at 4 Vand crosses 0 V at -110 rnA. It is interesting to note that there is also conduction at the input pins (Figure Sb). The ESD structure is a totem-pole configuration resembling an output driver; therefore, the ESD-protection transistor also conducts when ground floats. If the ground is disconnected for only a second or two, the device will not be damaged. 6 0 0 0 20 20 40 40 « 2 VI Volts 3 4 5 6 « E 60 E 60 ~ .::- 80 80 100Y 100 t 120 120 a) b) 16677A-8 Figure S. IN Curves when Ground is Disconnected: a) I/O Pin; b) Input 5-144 Designing with the PALCE16VSHD AMD~ SUMMARY • The PALCE16V8HD incorporates a number of features which make it ideal for applications involving high-current signal lines such as bus control lines. The output drivers generate fast signals (2-ns to 3-ns rise and fall times). At these edge rates transmission line terminating procedures should be used. • The drivers can be individually programmed as either totem-pole or open drain. For open-drain outputs, the terminating voltage determines VOH. • All input pins have a minimum of 200 mV hysteresis. This allows the PALCE16V8HD to operate more robustly in noisy environments such as buses. • The PALCE16V8HD incorporates slew-rate limiting and multiple grounds. This reduces ground bounce to levels more common in low-power devices. • The PALCE16V8HD has n-channel pullup transistors. Therefore, it is resistant to latchup caused by signals on input and I/O pins while Vce is off, or by hot -socketing. • • There are input latches on every input and I/O pin. The latches allow the earlier capture of data, which effectively shortens the cycle time in many applications. The use of the latch does not add any extra delay. Thus the PALCE16V8HD is faster than the PLD device and external latch combination. The PALCE16V8HD macrocell can be programmed as either D-type or T-type flip-flops. The T-type flipflops can reduce product-term usage in many applications. Larger counters and state-machines can often be built with T-type flip-flops than with D-type flip-flops. Designing with the PALCE16V8HD 5-145 ~ AMD 5-146 6 APPENDICES Logic Reference Guide ........................................ . . . . . . . . . . . .. 6-3 Signal Polarity Glossary ............................................................... 6-25 Worldwide Application S~pp~rt' :::::::::::::................................. ................................. Appendices 6-30 6-37 6-1 ~AMD 6·2 Appendices ~ Logic Reference Guide Advanced Micro Devices INTRODUCTION Throughout this data book and design guide we have assumed that you have a good working knowledge of logic. Unfortunately, there always comes a time when you are called on to remember something which can only be found in that logic textbook which you threw away years ago. This section is intended to provide a quick review and reference of the basic principles of digital logic. We will cover three general areas: • Basic logic elements ~ ==D==D- NOT AND OR 101730- Figure 1. Schematics Symbols for the Three Fundamental Gates • Basic storage elements • The standard schematic symbols used to represent these gates are shown in Figure 1. Binary numbers Throughout the text, we will use the notation that was used throughout this book. If you are unfamiliar with the syntax, you will probably find it easy to understand as you read; if you wish for a more detailed explanation of the symbols, please refer to the Basic Design with PLDs section where they are defined. As this is a logic reference only, we cannot take on lengthy discussions, nor can we train you in the basic principles of digital logic if you have not previously been trained. In such a case, we must refer you to your favorite logic textbook. The AND and NOT functions can be combined into the NAND function. This is equivalent to an AND gate followed by an inverter, as shown in Figure 2a. Likewise, the OR and NOT gates can be combined into the NOR function, as shown in Figure 2b. Each of these gates is functionally complete; any logic function can be expressed solely as a function of NAND or NOR gates. a. The NAND Function BASIC LOGIC ELEMENTS In this section, we will discuss the concepts surrounding combinatorial logic functions. b. The NOR Function The Three Basic Gates There are three basic logic gates from which all other combinatorial logic functions can be generated. These functions are NOT, AND, and OR. A truth table indicating these functions is shown in Table 1. Since they can be used to generate any function, they are said to be functionally complete. Table 1. Truth Table for the NOT, AND, and OR Functions A B fA A*B a a a 1 1 1 0 0 0 1 1 1 0 0 1 a A+B 0 101730- Figure 2. The NAND and NOR Functions Precedence of Operators Logic functions may be created with any combination of the three basic functions. How those functions are expressed affects the evaluation· of the function. The normal order of evaluation is: NOT, AND, OR Evaluation proceeds in order from left to right. 1 1 1 6-3 ~ AMD This order may be altered by inserting parentheses in the function. The contents of the parentheses will always be evaluated before the rest of the expression, from left to right. Some example functions are evaluated in Table 2. Table 2. Using Parentheses to Change the Order of Evaluation 8 0 1 0 1 A 0 0 1 1 C 0 1 0 1 A*B+/A* C+D 0 0 0 0 1 1 A*B+/A* (C+D) 0 1 0 1 1 1 1 Commutative,· Associative, and Distributive Laws A*(B+/A)* C+D 0 0 1 1 A*(B+/A)* (C+D) 0 0 0 1 Duality The AND and OR functions are commutative and assoCiative. This means that the operands can appear in any order without affecting the evaluation of the function. This is illustrated in Tables 3 and 4. The two distributive laws give an example of the concept of duality. This principle states that: Any identity will also be true if the following substitutions are made: * for + + for * 1 for 0 o for 1 Table 3. Commutativity A 0 0 1 1 8 0 1 0 A*B 0 0 1 B*A 0 A+B 0 B+A 0 0 0 0 1 1 1 1 1 1 1 1 Table 4. Associativity C 0 A 0 0 8 0 1 1 1 a 1 1 1 1 (A*B)*C A*(B*C) 0 0 0 0 a 0 1 1 (A+B)+C A+(B+C) 0 0 1 1 1 1 1 1 Thus, it is only necessary to prove the first of the distributive laws; the second one will then be true by duality. Note that duality is not required to prove the second law; it can also be proven by truth table or by logic manipulation. Manipulating Logic Logic functions may be manipulated by the use of Boolean algebra. The logic functions may be expressed in one of the two canonical forms, or by using a simplified expression. There are actually two distributive laws; one of them resembles standard algebra more than the other. These two laws state that: A*(B+C) (A*B) + (A*C) A+(B*C) = (A+B) * (A+C) 6-4 Logic Reference Guide AMD~ Canonical Forms There are two fundamental canonical forms: sum-otminterms and product-ot-maxterms. The former is by far the most widespread. These are special cases of what are more generally referred to as sum-ot-products and product-ot-sums forms. Minterms and maxterms are products and sums of the variables involved in a function. Each particular combination of noninverted and inverted variables in a product or su m is given a minterm or maxterm number, as shown in Table 5. Within each minterm or maxterm, the individual variables are referred to as literals. ' For the case of sum-of-minterms form, the expression for a function may be found by ~Ring the minterms which correspond to the 1's in the function's truth table. Likewise, the product-of-maxterms expression may be found by ANDing the maxterms which correspond to the O's in the truth table. This is illustrated in Figure 3. Table 5. Mlnterms and Maxterms Table of Minterms for Three Variables Minterm /x*Iy*/z /x*/y·z /x*y*/z /x*y·z x·/y·/z x·/y·z x·y*/z x·y·z Name Maxterm Name mO m1 m2 m3 m4 m5 m6 m7 x+y+z x + y +/z x +/y + z x +/y + /z /x+y+z /x+y+/z /x+ly+z /x+/y+/z MO M1 M2 M3 M4 M5 M6 M7 Conversi.on Between Canonical Forms It is a simple matter to convert between canonical forms. Given a truth table for a function F, there are four different representations that can be used: • Table of Maxterms for Three Variables One can convert back and forth between these representations by using the rules shown in Table 6. Sum-of-minterms form of F • Product-of-maxterms form of F • Sum-of-minterms form of IF • Product-of-maxterms form of IF Logic Reference Guide 6-5 ~AMD Minterml Maxterm Number B 0 0 0 0 1 1 1 1 0 0 0 C y 0 0 1 1 0 0 1 1 0 0 1 0 0 1 0 1 0 1 0 1 0 1 0 x 0 0 0 0 0 0 0 0 1 1 1 1 0 1 1 0 1 0 1 1 1 0 1 1 1 1 1 0 0 1 1 1 0 10 1 1 1 1 0 0 15 A 0 1 2 3 4 5 6 7 8 9 a. Truth Table x = = = /A */B * /C * + + + + + + y x = M1*M4*M6*M10*M11*M12*M13*M14*M15 mO+m2+m3+m5+m7+m8+m9 LIn (0,2,3,5,7,8,9) /A /A /A /A A A */B */B * B * B */B */B * C * * C * * /C * * C * * /C * * /C * nM = (A+B+C+/O) ;mO ;m2 /0 /0 0 0 0 /0 0 * (A+/B+C+D) * (A+/B+/C+O) * (/ A+B+/C+O) * (/A+B+/C+/O) * (/ A+/B+C+O) * (/A+/B+C+/O) * (/A+/B+/C+O) * (/A+/B+/C+/O) ;~3 ;m5 ;m7 ;m8 ;m9 mO+m1+m2+m3+m4+m7+m8+m9 2.m (0,1,2,3,4,7,8,9) = /A */B * /C * + + + + + + + /A /A /A /A /A A A */B */B */B * B * B */B */B * * * * * /C C C /C C * /C * /C y ;mO ;m1 ;m2 ;m3 ;m4 ;m7 ;m8 ;m9 /0 0 /0 0 /0 0 * * * * * * /0 * 0 b. The Sum-of-Minterms Expression (1,4,6,10,11,12,13,14,15) = M5*M6*M10*M11*M12*M13*M14*M15 = nM (5,6,10,11,12,13,14,15) = (A+/B+C+/O) * (A+/B+/C+O) * (/ A+B+/C+O) * (/A+B+/C+/O) * (/A+/B+C+O) * (/ A+/B+C+/O) * (/A+/B+/C+D) * (/A+/B+/C+/O) ;M5 ;M6 ;M10 ;Mll ;M12 ;M13 ;M14 ;M15 c. The Product-of-Maxterms Expression Figure 3. Finding the Canonical Form from the Truth Table 6-6 ;M1 ;M4 ;M6 ;M10 ;Mll ;M12 ;M13 ;M14 ;M15 Logic Reference Guide AMO;t1 Table 6. Conversion of Forms Table Desired Form Given Form Minterm Expansion of F Minterm expansion of F - Maxterm expansion of F Minterm numbers are those numbers not on the Maxterm list of F Maxterm Expansion of F Maxterm numbers are those numbers not in the Minterm list of F - Inverted Minterm Expansion of F List Minterms not present in F Minterm numbers are the same as Maxterm numbers of F Simplifying Logic Inverted Maxterm Expansion of F Maxterm numbers are the same as Minterm numbers ofF List Maxterms not present in F There are four basic postulates, two of which are the commutative and distributive laws which were discussed above. From these postulates, it is possible to derive nine basic theorems. The postulates and theorems are listed in Table 7. Canonical forms are convenient in that it is easy to derive and convert them. However, the representation is bulky, since all variables must appear in each sum or product. These expressions can be simplified by applying the basic laws and theorems of Boolean algebra. Table 7. Postulates and Theorems of Boolean Algebra Postulate 1 (A) (B) Postulate 2 (A) (B) Postulate 3 (A) (B) Postulate 4 (A) (B) Theorem 1 (A) (B) x + FALSE X*TRUE = x = X x + !X = TRUE = FALSE X * !X x + Y = Y x*y = y*x x x * (y + Z) + x = (x*y) X + X X * X = = X + TRUE = FALSE X*FALSE = FALSE (A) (B) X + (y + Z) X * (y*Z) = Theorem 5 (A) (B) ! ! Theorem 6 (A) (B) X + X * Theorem 7 (A) (B) (x*y) + (x*!y) = X (X + y) * (X + !y) Theorem 8 (A) (B) X + (lX*Y) = X + Y X * (IX + Y) x*y Theorem 9 (A) (B) Theorem 3 Theorem 4 (A) ! (IX) (x*Z) (X + X) X X (B) Theorem 2 + + (y*Z) = (X + y) * = (X + y) (X * y) X = = (X * y) (X + y) = (X + y) + Z (x*y) * Z !X * !y !X + !y = = (x*y) + (lX*Z) (X + Y) * (IX X X = X + (y*Z) = (x*y) + (lX*Z) + Z) * (Y + Z) = (X + Y) * Notice that each theorem and postulate (with the exception of theorem 3) has two forms. This is a result of the duality principle; once one form of a theorem is established, the dual representation follows immediately. Theorem 3 has no dual because it does not involve any of the elements that have duals (+, *, 1, or 0). (IX + Z) As the logic expression is simplified, it no longer contains minterms (or maxterms), since some of the minterms and literals are being eliminated. What was a sum-of-minterms (product of maxterms) representation is now simplified to a sum-of-products (product of sums). Logic Reference Guide 6-7 ~ AMD dIf dr ---...----..---".-- Values of A DeMorgan's Theorem Once an expression has been simplified, it is no longer possible to invert the function by using Table 6. Inverting simplified logic requires DeMorgan's theorem: f A C B 00 01 11 10 /(x*y) = /x + /y /(x + y) = /x*/y ValuesolB Moving to an Adjacent Cell Changes the Value of one Variable only, This is theorem 5 in Table 7. There is one shortcut which can be used. The effect of inversion can be accomplished by inverting all literals and then using the dual representation. For example, given the expression [l Values of C 10173D- /(A*/B + A*C + /A*B*D) we can invert to obtain: /A*B + /A*/C + A*/B*/D (/A + B)*(/A + /C)* (A + /B + /D) Groups Can WrapAround Figure 4. A Karnaugh Map for Three Variables ;step one, invert literals ;step two, take dual This expression must still be simplified to obtain a sumof-products representation, but this shortcut eliminates , some of the early steps. Karnaugh Maps: Minimizing Logic Simplifying by hand by using algebraic manipulation can be a tedious and error-prone procedure. When only a few variables are' used (generally less than 5 or 6), Karnaugh maps (also called K-maps) provide a simpler graphical means of simplifying logic. K-maps not only allow for logic simplification, but for logic minimization, where an expression has a minimal number of product terms (or sum terms) and literals. A Karnaugh map consists of a box which has one cell for each minterm. These cells are arranged so that only one literal is inverted when moving from one cell to an adjacent cell. The headings placed by each row and column indicate the polarities of the literals for that row or column. The literals themselves are indicated in the top left corner of the map. An example of a Karnaugh map for three variables is shown in Figure 4. The truth table for a function is then transferred to the K-map by plaCing the 1's and D's in the appropriate cells. Since each cell differs from its neighbor only in the polarity of one of the literals, 1's in adjacent cells can be combined by theorem 7a, which says that x*y + x*/y = x In this manner, two product terms are combined into one. This procedure can conceptually be repeated to allow groupings of two, four, eight, or any group of adjacent cells whose size is a power of two. A cell may appear in more than one group. Just enough groups are found to include all of the 1'So The groups should be as large as possible. This process provides a minimal sum of products. The product-of-sums form can be obtained by grouping D's instead of 1's and inverting the header for each cell. The two functions from Figure 3 have been placed into K-maps in Figure 5. The groups are then used as individual product terms. When reading the product terms from the map, the only literals which will appear in the product term are the ones whose values are constant for each cell in the group. If that value is 1, then the noninverted form of the literal is used. If the value is 0, then the inverted form of the literal is used. For active-LOW functions, the same procedure is used, except that the D's are grouped instead of the 1'so The active-LOW version of the functions from Figure 3 are derived in Figure 6. Hand simplification and minimization is not needed as frequently today as in the past, since software is now available for handling these logic manipulations. Most software can perform logic simplification and minimization automatically. 6-8 Logic Reference Guide AMD~ A*S*/C S*/C*D y x X Y IA*/S*ID IA*ID + IC*ID + IA*/S*/C + A*S*/C + A*/C*ID + S*/C*D + IA*C*ID 10173D- Figure 5. Using a K-map to Minimize the Functions in Figure 3 A S C 00 D IS*D 1 00 o o 10 x 1 I V1 ~, 01 1 11 reo 10 1 C*D 01 0 1 0.) 0 1 0 A*C IX = C*D 10 1 ~ o~ V A*/S*D V A*C 0 y IY = + IS*D + A*C + IA*S*/C*ID IA*S*D COD + A*D + IA*S*D + A*/S*D 10173D- Figure 6. Finding Inverse Functions Logic Reference Guide 6-9 ~ AMD Comparison and Equivalence: the XOR and XNOR Gates ~D-XOR The Exclusive-OR (XOR) and Exclusive-NOR (XNOR) gates are two special gates which are relatively common. These gates have schematic symbols as shown in Figure 7a. They are actually compound gates, and can be generated by AND, OR, and NOT gates using the functions: ==:}D- x :+: y = x*/y + Ix*y x :*:y = x*y + Ix*/y ;XOR gate ;XNOR gate The XOR and XNOR functions are actually inverses of each other; that is, XNOR 101730- a. Schematic Symbols A B A:+:B A:*:B 0 0 1 1 0 0 1 1 1 0 1 1 0 0 0 1 b. XOR and XNOR Truth Table x :+: y = I(x :*: y) The truth tables for these gates are shown in Figure 7b. Note that the XOR function is true if and only if the operands are different. For this reason, it is useful as a comparator. The XNOR function is true if and only if its operands are the same; therefore it is used as an equivalence indicator. Figure 7. The Exclusive-OR and Exclusive-NOR Functions Some basic properties of the XOR and XNOR functions are listed in Table 8. Table 8. Properties of the XOR and XNOR Functions XNOR XOR x :+: 0 =x x:*:O=/x x :*: 1 = x x :*: x = 0 x :*: Ix = 1 x .*. y = y :*: x x:*:y:*:z = (x :*: y) :*: z) = x :*: (y :*: z) x :*: y =Ix :*: Iy I (x :*: y) = Ix :*: y = x :*: Iy =x :+: y x :*: y = x* y + Ix*ly x :*: x* y = Ix + y x :*: Ix*y =Ix * Iy x + (y :*: z) ::: (x + y) :*: (x + z) Ix + (y :*: z) = (x*y) :*: (x*z) :+ 1 =Ix :+: x = 0 :+: Ix = 1 :+: y =y :+: x :+: y = :+: z = (x :+: y) :+: z = x :+: (y :+: z) x :+: y =Ix :+: Iy I (x :+: y) = Ix :+: y = x :+: Iy = x :+: y x :+: y = x* Iy + Ix*y x :+: x* y = x*ly x :+: Ix*y = x + y x* (y :+: z) = (x*y) :+: (x*z) Ix*(y :+: z) = (x + y) :+: (x + z) x x x x x When deriving equations from a Karnaugh map, XOR and XNOR functions can usually be identified by their characteristic pattern. Exactly what the operands are mayor may not be obvious for more complicated functions. Some examples are shown in Figure 8. The XOR gate can be used as an "UNLESS" operator. In other words, the function, A = X :+: Y can be interpreted as: "A will have the same value as X UNLESS Y is true." This can be helpful when trying to derive a logic equation for a function which can be described in words. 6-10 Logic Reference Guide AMD~ P P 0 R 00 J r-.... 00 01 0 R 1 0 10 0 11 10 0 ~ 0 0 1 '-.,...I 11 01 1 1 0 1 0 00 01 0 '-.....I r-.... S r-.... 1 11 1 10 00 01 0 0 11 0 10 1 "-.,...I ,........... 1'"1' 0 1 0 0 0 '-.....I 1 0 1 0 '-" ~ '-.,...I '-.,...I 0 K J J + /P·/O·/R P·O·/R /P·O·R P·/O·R ((/P·/O) + (P·O))*/R ((/p·O)+(p·/O))·R (P:·:O)·/R (P:+:O)·R /(P:+:O)*/R (P:+:O)*R = (P:+:O):·:R + + + + + I 1 K /P·/O·S + p·O·S + p·/O·/S ((/p·/O) + (p·O))·S + p·/O·,S I 101730- Figure 8. Finding XOR and XNOR Functions In Karnaugh Maps Basic Storage Elements Storage elements provide circuits with the capability of remembering past conditions or events. The prototypical storage element is just a pairof cross-coupled NAND gates, as shown in Figure 9. These elements are normally called flip-flops. Flip-flops can also be characterized by their con'trol scheme. There are fourtypesof flip-flops, each of which can be unclocked or clocked: • S-R • J-K • D • T The discussion below will be divided between unclocked and clocked flip-flops. Each of the four flip-flop types will be treated for each section. Unclocked Flip-Flops-Latches 10173D- An S-R latch can be built out of NOR gates as shown in Figure 10, and behaves according to the truth table in Table 9. "S' stands for 'set' and 'R' stands for 'reset,' as suggested by the truth table. Figure 9. Basic Storage Element In general, there are two primary classes of flip-flops: • Unclocked flip-flops, or latches • ClockedfJip-flops S-R Latches Clocked flip-flops are sometimes referred to as registers, although technically speaking, a register is a bank of several flip-flops with a common clock signal. Note that the latch actually has two outputs, which are complementary. These are referred to as Q and O. If both Sand R are raised at the same time, then both Q and 0 will be HIGH; although this is physically possible, it does not make sense if Q and Q are to be complementary signals. Thus, this condition is not allowed. Logic Reference Guide 6-11 ;t1 R AMD -I----~ D-.._--+- a s D-..--+--o 1 R I a I 1 I S ---11- - - 1 L_....;. _ _ _ _ _ _ _ .J 101730- Figure 10. An S-R Latch There are some applications where it is desirable forthe input data to be effective only when another signalusually called a control signal-is active. The circuit of Figure 10 can be modified to give an S-R latch with a control input,'as shown in Figure 13. The operation of this circuit is summarized in Table 10 and Figure 14. The S-R latch is somewhat restrictive, since both inputs cannot be HIGH at the same time. The other latch types are based on the S-R latch, but have additional logic which removes the input restrictions. Table 9. S-R Latch Truth Table 5 0 0 1 1 s R 0 1 0 Q+ 0 Not allowed a 0 1 R The transfer function for this latch can be derived with a Karnaugh map, as shown in Figure 11. By choosing either 1 's or D's, we can obtain two representations: a 101730- a. Q+ = S+/R*Q b. /Q+ = R+/S*/Q s a R 00 01 a a a 1 Figure 12. S-R Latch Behavior s a a R 1 00 a 01 r---. 1 0 1 a 0 s-r---r-'" c R-t----L-..J ~ 11 X 10 1 X 11 X X 10 1 1 r-1 101730- a. 0+ =S + iR·Q b. iO+ = R + /S·/O Figure 11. Karnaugh Map for an S-R Latch Waveforms illustrating the operation of the S-R latch are shown in Figure 12. 6-12 101730- Figure 13. Adding a Control Input to an S-R Latch Table 10. Truth Table for an S-R Latch with a Control Input S X 0 0 1 1 Logic Reference Guide R X C Q+ a a 1 1 1 1 a a 1 0 1 0 1 Not allowed AMD~ s R c a 101730- Figure 14. Behavior of an S-R Latch with a Control Input 0-Type Latches (Transparent Latches) A single-input latch can be formed by adding some logic to the controlled S-R latch in Figure 13; this gives rise to the D-type latch in Figure 15. This latch is often called a transparent latch, since data on the input passes right through to the output as long as the control input is HIGH. If the control input is set LOW, then the latch holds whatever data was present when the control went LOW. With this type of latch, the control is usually called a gate. The behavior of the D-type latch is shown in Table 11 and Figure 16. The basic transfer function for a D-type latch can be derived from the Karnaugh map in Figure 17. D G a 101730- Figure 16. 0-Type (Transparent) Latch Behavior 0+ = D*G + O*/G 10+ = ID*G + 10*IG 10 10 r----------, I I D S G a a C I I IL _ _ _ _ _ _ _ _ _ _ .JI Q U a G 101730- Figure 15. A 0-Type (Transparent) Latch 0 0+ Table 11. Truth Table for a D-Type Latch ~ I ~ a. 0+ I =O*G + O*/G a o 1 0 10+ b. /0+ =/O*G + /O*/G 101730- Figure 17. Karnaugh Maps for a D-Type Latch Logic Reference Guide 6-13 ~ AMD If realized exactly as the transfer function indicates, the result is actually a glitchy circuit. J-K Latches Another two-input latch can be derived from the S-R latch as shown in Figure 18. This is called a J-K latch, and operates in the same manner as an S-R latch, except that the condition where both inputs are HIGH is now allowed. The truth table is shown in Table 12; the waveforms are shown in Figure 19. ~ ---u101730- There are still some potential problems here forthe case where J and K are both HIGH. If J and K are left HIGH for too long, the output may change more than one time; if left HIGH forever, the output will oscillate. Thus, J and K should not be asserted for a time longer than the propagation delay of the latch. There are also potential race conditions if J and K are not asserted and removed at exactly the same time. If one of the inputs is raised slightly ahead of the other, it may give the output time to react, giving the wrong output once the second input is raised. The same problem can occur if one input is lowered slightly before the other. This is illustrated in Figure 20. There are several ways to derive transfer functions for J-K latches. Two can be derived directly from Kamaugh maps, as shown in Figure 21 ; the others are not as obvious, and make use of the XOR gate described before. The basic transfer functions are listed in Table 13. Table 13. Transfer Functions for a J-K Latch Figure 18. A J-K Latch 0+ = + Table 12. Truth Table for a J-K Latch J K 0 1 0 1 0 0 1 1 = 0 /0+ 0+ :+: (J* /0 + K*O) = /0 :+: (/J*/O + /K*O) /0+ Q a /0+ 0+ Q+ 0 1 J * /0 /K*O = /J* /0 + K*Q = /0 :+: (J*/O + K*O) = 0 :+: (/J*/O + /K*O) J K Q 101730- Figure 19. Behavior of a J-K Latch 6-14 Logic Reference Guide AMD~ J K Q Expected Levels a. Falling Edge Race Conditions J K Q -----r---- ... ... Expected Levels b. Rising Edge Race Conditions J K Q I- -I tPD OF Latch ••• ~ ~ xxxx c. Possible Oscillation 10173D- Figure 20. Hazards Inherent in a J-K Latch Logic Reference Guide 6~15 ~ J AMD a J 0 K 00 r-------, a 0 I o K T 00 01 11 10 10 a a -EJ= T I a L. _ _ _ _ _ _ _ .J 01 11 I I II _ Q 101730- Figure 22. A T-Type Latch Table 14. The Truth Table for a T-Type Latch T /0+ a. Q+ = J*/O + /K*O b. /0+ =/J*/O + K*O 101730- Figure 21. Karnaugh Maps for a J-K Latch T-Type Latches T-type latches are formed by connecting the J and K inputs of a J-K latch together to form a single input. as shown in Figure 22. This latch has two possible functions: hold the present state or invert the output. as summarized in Table 14. 'T' stands for 'trigger' or 'toggle' depending on who you talk to. That is, when T is HIGH. a change at the output is triggered; or. put another way, raising T causes the output to toggle. T a :?j-------,r~ o a 1 10 This Latch also has the problem that if T is left HIGH for too long, the output will oscillate. However. since there is only one input, the race condition problems of the J-K latch have been eliminated. Unfortunately. this comes at the cost of initialization. There is now no way to get the output into a fixed state without knowing what the previous state was. Thus. this device is not very useful without some kind of initialization circuit. The general waveforms for a T-type latch are shown in Figure 23. ... ~ ~ ~ ... 'llIXX ~ tPD OF Latch F!gure 23. Behavior of a T-Type Latch 6-16 Logic Reference Guide 101730- AMD~ From the Karnaugh map in Figure 24, we can generate the following transfer functions: T Q+ = T*/Q + /T*Q /Q+ = T*Q + /T*/Q Q+ =Q:+:T Q+ = /Q:+: /T /Q+ /Q+ T Q o o '1 0 0 00 a. 0+ = T*/O + /T*O The clock provides two basic advantages. It removes the hazards inherent in the J-K and T flip-flops, since all inputs will have settled by the time the clock edge arrives, and only one transition is possible for each clock edge. The clock also allows the design of synchronous systems, where all signals are coordinated with other signals. The entire system is then regulated by the clock. /Q :+: T Q: +: /T Q o b. /0+ sit ion is. detected. A device is classified as positive edge-triggered or negative edge-triggered, depending on whether it responds to the rising or falling edge of the clock signal, respectively. The behavior to a clocked S-R flip-flop is illustrated in Figure 25. = T*O + /T*/O 10173D- Figure 24. Karnaugh Maps for a T-Type Latch Clocked Flip-Flops Latches can be modified by adding a clock input. The purpose of the clock is to delay any output changes until the clock signal changes. Whereas latch control inputs (such as the gate) are level-sensitive, clock inputs are generally edge-sensitive (or edge-triggered), meaning that output transitions can occur only when a clock tran- The basic behavior of the four flip-flops types does not change with the addition of a clock; the output changes are merely made to wait forthe clock edge. Thus, the basic transfer equations for most of the flip-flops are the same. We can indicate the clocked nature of the flipflops by using the "registered" assignment ':=' instead of'=.' 0-Type Flip-Flops This is the only flip-flop type whose basic transfer characteristic changes, because the clock input replaces the gate input. Thus the transfer equations become: Q+:= D/Q+ := /D That is, whatever data appears on the input will be transferred to the output after the next clock edge. The input is not changed in any way. Logic Reference Guide 6-17 ~ AMD The simplicity of this flip-flop makes it the most widely used flip-flop. However, functions are sometimes more conveniently expressed using J-K flip-flops, or using T-type flip-flops. If we replace the D signal with the transfer function for one of the other flip-flop types, we can then emulate that flip-flop type in the D-type flip-flop. This is equivalent to taking a latch and placing a clocked D-type flip-flop after the latch output for synchronization. Figure 26 illustrates how each flip-flop can be emulated in a D-type flip-flop. The standard schematic symbols for the flip-flop types are also shown. Table 15 summarizes the transfer functions for all of the flip-flop types. These functions can directly be used to emulate a particular flip-flop type in a D-type flip-flop. This can be particularly useful since D-type flip-flops are available in most registered PLDs. s R Clock Op On 10173D- Figure 25. Behavior of a Clocked S-R Flip-Flop for Positive (Qp) and Negative (Qn) Edge-triggered S-R Flip-Flops 6-18 Lagle Reference Guide AMD~ 0 a a Clock a. Clocked 0-Type Flip-Flop r------------------,I J I I K Clock J K a 0 a I I I I a a J a a K a a T a I L __________________ JI b. Clocked J-K Flip-Flop r------------------,I T I I I I a 0 a I I I T a Clock I a a I L __________________ JI c. Clocked T-Type Flip-Flop r------------------,I 5 R Clock I I 5 a T a 0 a I I L __________________ JI a 5 a a R a 101730- d. Clocked 5-R Flip-Flop Figure 26. Clocked Flip-Flops. All can be Emulated with a D-Type Flip-Flop Logic Reference Guide 6-19 ~ AMD Table 15. Clocked Flip-Flop Transfer Functions D-Type 0+ 0+ 0+ J-K-Type 0+ 0+ T-Type S-R-Type 0+ 0+ 0+ '- /0+ /0+ D ,- J*/O + /K*O ,- 0 :+: (J*/O + K*O) ,= /0 :+: (/J*/O + /K*O) := T*/O +IT*O .= O:+:T .= /0 :+: IT .- S + /R*O /D /J* /0 K*O := /0 :+: (J*/O + K*O) '- 0 :+: (/J*/O + /K*O) = T*/O + IT*'O .- /0 :+:T ,- + /0+ /0+ /0+: /0+ /0+ /0+ Binary Numbers The concept of a number is taken for granted by most people. And most people equate numbers in general with the decimal system, with which we are most familiar, However, there is nothing particularly special about the decimal system; the choice of system is actually rather arbitrary. History has chosen the decimal system for most humans. For electronic systems, the binary system is more appropriate. It makes possible arithmetic and logical calculations that would be much more difficult-likely impractical-if implemented directly in a decimal system. Closely related to the binary system are the octal and hexadecimal systems, which will also be discussed here. Arithmetic is normally performed using binary numbers in a computer. Octal and hexadecimal representations are generally used as a way to "abbreviate" what might otherwise be lengthy binary numbers. This will be seen when conversion is discussed below. There are several terms which must be defined before proceeding further. A number is an abstract entity which is used to describe quantity, There are· many ways of representing a number. Normally, the representation is designed around a base. The number is expressed as a sum of multiples of the powers of the base. Thedecima! system is a base-10 system, meaning that 10 is used as the base, The binary system is base-2; the octal system is base-8; and the hexadecimal system is base-16, The binary, octal, and hexadecimal systems are closely related because 8 and 16 are both powers of 2. When different bases are being used, a number will often be followed by its base in subscript, to indicate exactly what the base is. For example, the 6-20 ,- .- o :+:IT '- R + /S*/O decimal number25would be written 2510 if its base were in doubt. A number can thus be expressed in terms of some base x as follows: anXn+an_1X n- 1 + ... +alx1+aoxo+a-lx- 1 + ..• +a-rnX- rn (1) The numbers an ... a-m are called digits. The value of each digit can range from 0 to x-1. Each digit is represented by a symbol, called a numeral. X numerals are required to represent a number in base x. The most familiar numerals are the symbols '0,' '1,· .. .'9.' There are ten of them, since they are used for the decimal system. For binary numbers, only '0' and '1' are used; for octal numbers, the numerals '0' through '1' are used, Hexadecimal numbers are more difficult, since sixteen numerals are required. Therefore, the numerals '0' through '9' are used to represent the quantities 010 through 910; the letters A through F are used to represent the quantities 1010 through 1510. The number expressed by equation 1 is normally represented as a string of digits: anan-l .•• alaO . a-l ... a-rn The digits representing negative powers of the base are separated from those representing non-negative powers by a point. In the decimal system, this is referred to as a decimal point; in the binary system. it is referred to as a binary point. There are two basic classes of manipulation which will be discussed: conversions between bases and arithme- . tic within a base. Logic Reference Guide 'AMD~ Here we see that the fraction will repeat, since we have already multiplied 0.6 earlier. Thus Converting Between Bases Base-2 <-> Base-1 0 Converting a binary number to a decimal number is accomplished by using equation 1 directly. Example: Converting 110100.0112 to decimal: y 0.162510 = 0.00101001100110011 ... 2 For mixed numbers, it is necessary to calculate the whole and fractional portions separately. Thus, for example, we know that 61.162510 = 111101.0010100110011 ... 2 110100.011 1.25 + 1.24 + 0.23 + 1.22+ 0.2 1+ 0.20+ 0.2-1+ 1.2-2 + 1-2..,3 32 + 16 + 4 +.25 + .125 52.375 These are actually general procedures which can be used to convert a decimal number into any base, and vice versa. Examples: When converting whole numbers from decimal to binary, the decimal number is repeatedly divided by 2. Integer division is used, so the quotients are "rounded down" to the next integer. The remainders form the digits of the number. The least significant digit is the first one calculated. 321.548 = 209.687510 Example: 2. Converting 106.1037510 to octal: Converting 6110 to binary: 61/2 = 30 30/2 = 15 15/2 = 7 7/2 = 3 3/2 = 1 1/2 = 0 remainder = 1 remainder = 0 remainder = 1 remainder = 1 remainder = 1 remainder = 1 LSB 1. Converting 321.548 to decimal: Y = 3.8 2 + 2-8 1+ 1-80+ 5-8-1+ 4.8-2 = 192+16+1+.625+.0625 = 209.6875 106/8 = 13 13/8 = 1 118 = 0 LSB remainder = 2 remainder = 5 remainder = 1 MSB Thus, the whole portion is 1518. MSB 6110 =1111012 When converting a decimal fraction into a binary fraction, the decimal number is multiplied by 2. This results in a whole number and a fraction. The whole number is a digit; the procedure is repeated on the new fraction. This procedure is repeated until the fractional portion is zero. If the procedure does not terminate, then the result is a repeating fraction. The first digit calculated is the most significant digit. 0.10375.8 = 0.83 0.83.8 = 6.64 0.64.8 = 5.12 0.12.8 = 0.96 0.96.8 = 7.68 0.68-8 = 5.44 whole whole whole whole whole whole portion portion portion portion portion portion =0 =6 =5 =0 =7 =5 MSB At this point we have enough significant digits. We could continue either until the procedure terminated, or until the pattern started repeating. However, those last digits are not likely to be significant. Thus, we can approximate by saying that... 106.1037510=152.0650758 Example: 3. Converting 31 F.A216 to decimal: Converting .162510 to binary: 0.1625.2 = 0.3250 0.3250.2 = 0.65 0.65.2 = 1.3 0.3.2 = 0.6 0.6.2 = 1.2 0.2.2 = 0.4 0.4.2 = 0.8 0.8.2 = 1.6 0.6.2 = 1.2 whole whole whole whole whole whole whole whole whole portion portion portion portion portion portion portion portion portion =0 =0 =1 =0 =1 =0 =0 =1 =1 MSB Y = 31F.A216 = 3.162+ 1.161 + 15.160+ 10.16-1+2.16-2 = 768 + 16 + 15 + 0.625 + 0.0078125 = 799.6328125 31F.A216= 799.632812510 4. Converting 7689.10085410 to hexadecimal: 7689/16 = 480 480/16 =30 30/16 = 1 1/16 = 0 Logic Reference Guide remainder = 9 remainder = 0 remainder = E remainder = 1 LSB MSB 6·21 ~·AMD Examples: Thus, the whole portion is 1 E0916. 0.100854.16 = 1.613664 0.613664.16 = 9.818624 0.818624.16 = 13.097984 0.097984.16 = 1.567744 0.567744.16 = 9.083904 0.083904.16 = 1.342464 whole whole whole whole whole whole portion portion portion portion portion portion =1 =9 =D =1 =9 =1 MSB 1. Convert 7324.348 to binary: 7 3 2 3 4 4 011 100 111 011 010 100 Thus 7324.348 = 111011010100.01112 Again, we likely have enough digits atthis point. The exact fraction could be either very long or a long repeating pattern. For our purposes, we can approximate the overall result as: 2. Convert 1A2.3F516 to binary: 1 A 2 0001 1010 0010 3 F 5 . 0011 1111 0101 Thus 1A2.3F516= 110100010.0011111101012 7689.10085410 = 1E09.19D19116 Binary Arithmetic Binary <-> Octal, Hexadecimal Converting between the binary-related systems is very easy. The procedure consists of dividing the binary digits into groups, and replacing each group with an appropriate digit. For this reason, octal and hexadecimal numbers are often used to shorten long binary numbers. To convert from binary to octal, group the digits by three, starting on each side of the binary point, and then convert each group of three digits into its corresponding octal digit. Leading and trailing zeroes may have to be added to the left of the whole portion and the right of the fractional portion, respectively, to make complete groups of three binary digits. Example: Converting 11011010110101.0010011012 to octal: Divide into groups of three digits: 011 011 3 3 010 2 110 101 6 5 001 001 101 1 1 5 Thus 11011010110101.0010011012 = 33265.1158 To convert from binary to hexadecimal, the digits are divided into groups of four digits, and then given their corresponding hexadecimal digits. Again, leading and/ or trailing zeroes may be needed. Example: Converting 100101011101100.1101100012 to hexadecimal: Divide into groups of four digits: 0100 101 0 111 0 11 00 4 A E C . One's Complement Representation The one's complement of a binary number can be calculated by inverting all of the bits of the number. Fractions are handled exactly the same way, although this is convenient only for fixed-point arithmetic. Floating-point arithmetic requires other methods, which will not be discussed here. Example: Finding the one's complement of 110111.0101 : 110111.0101 001000.1010 (Inverting each bit) Thus, the one's complement of 110111.0101 001000.1010. D 8 0000 1111 8 To convert from octal or hexadecimal to binary, merely expand each digit into its corresponding binary representation. is The sign of a number is determined by the most significant bit. If the MSB is 0 the number is positive; if the MSB is 1 , then the number is negative. Zero is represented by all bits being zero. However, one normally thinks of zero as being its own complement. But if we take the one's complement of zero, 11011000 1000 Thus 100101011101100.1101100012 = 4AEC.D8816 6-22 Positive binary arithmetic is very simple, and completely analogous to decimal arithmetic. However, if we are restricted to positive numbers, then we are also restricted to addition. We need a means of representing negative numbers. Using a dash '-' is unacceptable for representation in a computer. There are two general schemes which can be used. In binary systems, they are referred to as 15 complement and 25 complement representation, although they can be generalized for any base system as diminished-radix complement and radix complement representation. we see that 1111 is another representation of zero. Thus, in an eight-bit representation, positive numbers range from 00000001 to 01111111; negative numbers range from 10000000 to 11111110. Note that there are just as many negative numbers as positive numbers. Logic Reference Guide AMD~ This eight-bit code allows us to represent the numbers . from -127 to +127. When performing addition with one's complement numbers, it is important to watch for overflow results. Whenever an overflow occurs, a correction must be made by adding 1 to the result. In some cases, the results of an operation will not be meaningful, since the intended result cannot be represented .. For instance, in the eight-bit system above, adding 127 to 127 will give a meaningless result, since 254 cannot be represented in this system. Thus, the operation must be evaluated to ensure that the result is meaningful. Add 3 + 2: + 3 + 2 ---5 result meaningful Add 7 + 7 (14 cannot be represented): 0111 ....:...+_---=-01..:...1:....:,1 1110 7 _+_ _7 -1 result meaningless Subtract 3 from 7: + Examples: All examples will use 4-bit systems. Thus, the range of representable numbers is from -7 to +7. 0011 0010 0101 0111 1100 10011 +1 0100 7 ~ 4 overflow - add 1, discard overflow bit Subtract 5 from 2: + 0010 101 0 1100 2 ....:.+_---'-5:.,. -3 result meaningful Subtract 6 from -5 (-11 cannot be represented): 1010 +,!....-_1.,:..::0=0...:...1 10011 +1 0100 -5 ±--=.2.. 4 overflow - add 1, discard overflow bit result meaningless Subtract 5.25 from 3.5 (fixed point; requires 6 bits): 0011.10 + 1010.10 1110.00 3.5 + -5.25 -1.75 result meaningful Subtract 7 from 7: + Logic Reference Guide 0111 1000 1111 7 . . :.+_ _ -..:...7 0 one of the representations of 0 6-23 ~ AMD The advantage of one's complement code is the fact that it is easy to compute the complement. However, the fact that there are two representations for zero is a problem. In addition, the results of subtraction frequently have to be adjusted for overflow by adding 1. Two's Complement Representation The two's complement of a binary number is more difficult to calculate. It is generated by taking the one's complement, and then adding 1. Any overflow is discarded. Fractions are again handled in the same way, although 1 is added to the least significant bit. Examples: Add 3 + 2: + 110111.0101 001000.1010 +1 001000.1011 + + is + 7 7 --=2 result meaningless 0111 1101 10100 7 -3 ---4 + overflow - discard overflow bit 0010 1011 1101 2 + -5 ---=3 result meaningful Subtract 6 from -5 (-11 cannot be represented): The sign of a number is again determined by the most significant bit. If the MSB is 0 the number is positive; if the MSB is 1, then the number is negative. Zero is represented by all bits being zero. In this case, if we take the two's complement of zero, we get: 1011 + 1010 ------:--:10::-"1""!"0~1 -5 + -6 ---5 overflow - discard overflow bit result meaningless Subtract 5.25 from 3.5 (fixed point; requires 6 bits): 0011.10 + 1010.11 1110.01 (overflow is discarded) giving only one representation for zero. 3.5 + -5.25 -1.75 result meaningful Subtract 7 from 7: Thus, in an eight-bit representation, positive numbers range from 00000001 to' 01111111; negative numbers range from 10000000 to 11111111. This means that there is one more negative number than there are positive numbers. So this eight-bit code allows us to represent the numbers from -128 to +127. Addition is handled in the same fashion as with one's complement code, except that when an overflow occurs, the overflow bit is disregarded. No correction must be made to the results. . + 0111 1001 10000 7 -7 --0 + overflow - disregard overflow bit The benefits of two's complement lie in the fact that there is only one representation for zero, and the fact that the results of operations never need adjusting due to overflow. The disadvantage is the fact that it is harder to generate the two's complement of a number. After any operation, one must still make sure that the results are meaningful. 6·24 0111 0111 1110 Subtract 5 from 2: 001000.1011. 0000 1111 +1 0000 result meaningful Subtract 3 from 7: + (take one's complement) Thus, the two's complement of 110111.0101 3 + 2 ---5 Add 7 + 7 (14 cannot be represented): Example: Finding the two's complement of 110111.0101 : 0011 0010 0101 Logic Reference Guide Signal Polarity The polarity of signals, simple as it seems, turns out to be a potentially confusing issue. With such phrases as positive and negative logic, active HIGH, and active LOW, and with one person saying "asserted," another saying "active," and another saying "enabled," all of which mayor may not be well defined, it is very difficult to explain the relationships between signals. This can also make the generation of the design file more difficult. outputs. The possibilities are shown in Table 2. As an example, if a signal X is to go LOW only when inputs A and B are HIGH, and this function is to be implemented in an active-HIGH device, then from the third row of Ta- ( ble 2, declare the output pin as X in the design file, and use the equation: In an attempt to sidestep the ambiguities in the language, this discussion contains tables instead of vague descriptions. The tables list the various possibilities. If you know what you want, you should be able to find how to specify your equations from the tables. The issues of input signal polarity, output signal polarity, and feedback signal polarity are treated separately. Feedback Polarity x = / (A*B) Input Pin Polarity Using feedback combines some of the polarity issues of inputs with some of the polarity issues of outputs. It is more difficult to use a simple example forthis type of circuit. In Table 3, an output is assumed to be fed back to itself. The basic principles can be extended to any output feeding back to any other output. The waveform shows the output level that is considered to be "TRUE," or "active." . . Table 1 shows the relationships between the input pin names and the use of the input in a Boolean equation. As an example of how this table can be used, if you have a signal called IA on your schematic, and you wish for the output to go HIGH when both IA and B are HIGH, then from the second row of Table 1, declare the pins as IA and B in the design file, and use the equation: As an example, if the equation for a pin IX has to contain the inverse of the output, the output signal is to be active when HIGH, and an active-LOW device is to be used, then from the sixth row of Table 3, declare the output pin as IX in the design file, and specify the Boolean expression as: Ix : x = IA*B The basic function A *B has been used throughout for the purpose of illustration. The same procedure holds regardless of the waveforms being used or generated. = f (A, X) meaning that the Boolean equation uses X as an input term. Output Pin Polarity The issue of output polarity is slightly more complicated because of the issue of active-HIGH and active-LOW Signal Polarity 6·25 ;r1 AMD Table 1. Input Pin Polarity Desired Waveform Schematic A :8:8:8:8- x X x x S I L ~ A ~ S I L X ~ A ~ L S I X n A ~ S Boolean Equation ~ X X 6-26 Input Pin Definition I L ~ Signal Polarity A,S X = A'S lA, S X = IA'S A,S X = IA'S lA, S X =A*S AMD~ Table 2. Output Pin Polarity Desired Waveform Schematic x :8:=E}-x :=E}-x x :8- A S Output Pin Definition Boolean Equation Device Restriction X X X = A*S Active-HIGH Devices ~ ~ X ~ A ~ S ~ X ~ A ~ S ~ X n A ~ S ~ X ~ Signal Polarity IX = I(A*S) IX IX IX = A*S X = I(A*S) X X X = I(A*S) IX = A*S IX IX IX = I(A*S) X = A*S Active-LOW Devices Active-HIGH Devices Active-LOW Devices Active-HIGH Devices Active-LOW Devices Active-HIGH Devices Active-LOW Devices 6·27 ~ AMD Table 3. Feedback Signal Polarity Output Pin Definition Boolean Equation xJL X X X=f(A,X) IX = If(A,X) xJL X X X = f(A,/X) IX = If(A,IX) XLJ X X X=If(A,X) IX = f(A,X) XLJ X X X=If(A,/X) IX = f(A,/X) Desired Waveform Schematic Device Restriction -po- Ar- X --~ po- Ar- X --~ po- A- - --- X ~ po- Ar- 6-28 X --- ~~ Signal Polarity Active-HIGH Devices Active-LOW Devices Active-HIGH Devices Active-LOW Devices Active-HIGH Devices Active-LOW Devices Active-HIGH Devices Active-LOW Devices AMD~ Table 3. Feedback Signal Polarity (continued) Schematic Ar-- ~ X ~ """-- Ar-- A- - A- - x ~ ~> - t - ~ "- Output Pin Definition xJL IX IX Boolean Equation IX = f(A,IX) X =1f(A,IX) xJL IX IX IX =f(A,X) X = If(A,IX) x---U- IX IX IX = If(A,IX) X = f(A,IX) X ~> r---- Desired Waveform It x x---U- Signal Polarity IX IX Device Restriction Active-HIGH Devices Active-LOW Devices Active-HIGH Devices Active-LOW Devices Active-HIGH Devices Active-LOW Devices Active-HIGH Devices IX = If(A,X) X = f(A,IX) Active-LOW Devices 6-29 Glossary Advanced Micro Devices 10KH (adj.) A family of ECl devices. Circuits are temperature compensated. See also: ECl, 100K, temperature compensation. 100K (adj.) A family of ECl devices. Circuits are both temperature and voltage compensated. They have lower power dissipation and higher speed than their 10KH counterparts. See also: ECl, temperature compensation, voltage compensation, power dissipation, 10KH. A active high (adj.) See polarity. active low (adj.) See polarity. B BCD (n.) Binary Coded Decimal. Decimal numbers in 4-bit binary. binary (adj.) Having only two possible states, which can be variously called on/off, I/O, true/false, high/low, etc. bipolar (adj.) One of the two basic types of transistor. In logiC design, used for TTL, ECl, and 12 l families. See also: TTL, ECl, Fl, MOS. bistable (adj.) Describes a system which has 2 stable states. Any other state is unstable, and will eventually change to one of the stable states. A flip-flop ,is the most common electronic bistable circuit. See also: flip-flop, astable, monostable. ALS (adj.) Advanced l~w-power Schottky TTL family. Characterized as a lower power version of the AS family, and actually faster and lower power than the lS family. See also: AS, lS, TTL, Schottky TTL. bit 1. (n.) Binary Digit. One unit of binary information 2. (n.) A measure of the storage capacity of a memory chip. See also: binary. AND 1. (adj.) One of the three elementary logic functions. Result of the AND operation is true if and only if all operands are true. 2. (v.t.) To perform the AND operation. blank (adj.) Describes the state of a program~able c~1I after manufacturing, and before any programming, or, In the case of an erasable device, after erasure. Opposite of "programmed". See also: programmable cell, programmed, program, erase. , AS (adj.) Advanced Schottky TTL family. High-speed versions of the standard Schottky TTL family. Generally use oxide isolated technology for very high speed. See also: Schottky TTL, TTL, oxide isolation. buffer (n.) A logic gate which performs the logic identify function; i.e., the input is passed through unchanged. Used to isolate various parts of a system, or to provide voyage or current amplification. assertive high (adj.) Same as "active high". See polarity. assertive low (adj.) Same as "active low". See polarity. astable (adj.) Describes a system which has no stable state. Such a system will oscillate. Astable circuits can be used to generate timing and synchronizing clock signals. See also: bistable, monostable. asynchronous 1. (adj.) Describes a sequential logic system wherein operations are not synchronized to a common clock. 2. (adj.) Describes signals whose behavior and timing are completely unrelated to a particular clock. Such signals can either be random or based on another clock which has a different frequency. 3. (adj.) Describes a communication protocol whereby the timing of various operations is not determined by a system clock, but rather by events whose relationships are known, but whose exact timing cannot be precisely predicted. See also: sequential, clock, synchronous. 6-30 C chip (n.) A single piece of semiconductor material which contains an integrated circuit. Sometimes called a die if not in a package. See also: integrated circuit, die, package. clock 1. (adj.) A signal used to synchronize the operation of a system. 2.(adj.) An input to a clocked flip-flop. The flip-flop will not change state until an appropriate pulse appears at the clock input. 3. (n.) A circuit which generates a clock signal. 4. (v.t.) To pulse the clock signal or the clock input of a clocked flip-flop. See also: flip-flop, clocked flip-flop. clocked flip-flop (n.) A flip-flop that does not change state until a clock signal is received. See also: flip-flop, unclocked flip-flop, clock. CMOS (n., adj.) Complementary MOS. A type of circuit which makes use of both N-channel and P-channel AMD~ MOS transistors. Many CMOS logic circuits consume no power when not actually switching. See also: MOS, NMOS, PMOS, standby power. combinational (adj.) See combinatorial. combinatorial (adj.) Refers to a logic circuit which implements logic functions of present input signals only. Also called combinational. See also: sequential. . complement 1. (adj). Refers to a signal which is identical to some reference Signal, except that it is of opposite polarity. Opposite of '1rue". 2. (v.t.) To invert. See also: true, polarity, invert. complementary (adj.) Refers to logic device outputs which implement identical logic functions, but with opposite polarities. Used on some PlDs and ECl devices. See also: polarity, PlD, ECL. D decimal (adj.) Based on the number 10. die (n.; plural: dice) Same as a chip, particularly before being placed in a package. See also: chip, package. digit (n.) Any number from 0 to 9. DIP (n.) Dual In-line Package. The most common integrated circuit package. It is rectangular in shape, with widths ranging from .300 inch to .900 inch, and has vertical leads along the length. See also: integrated circuit, package. disable 1. (v.t.) To turn off a three-state output. 2. (v.t.) To inhibit another function, such as "disabling the clock". See also: three-state, enable. download 1. (v.t.) To pass data from one machine to a less complex machine. 2. (n.) The act of downloading data. See also: upload. E Eel (n., adj.) Emitter Coupled Logic family. An extremely high-speed family of bipolar logiC and memory devices. See also: bipolar. EE cell (E2 cell) (n.) A floating gate cell which can be both programmed and erased with electrical signals. EEPROM (n.) Electrically Erasable Programmable Read-Only Memory. A nonvolatile read-only memory device which can be erased and reprogrammed, both with special electrical signals. See also: program, erase, EPROM, PROM, ROM, RAM, nonvolatile. enable 1. (v.t.) To turn on a three-state output. 2. (adj.) By itself, usually refers to a pin which is used to enable a three-state output. Also called "output enable". 3. (adj.) Used with other function names, indicates a qualifier or inhibitor of the function. For example, "clock enable" is a function which qualifies the clock function. 4. (v.t.) To allow a signal which has been disabled to function; for example, "enabling the clock" removes any restraint . which may disable the clock signal. See also: threestate, disable. EPROM (n.) Erasable Programmable Read-Only Memory. A non-volatile read-only memory device which can be erased and reprogrammed. Erasure is accomplished by exposing the die to ultraviolet light for a period of time. Die must be packaged in a windowed package to allow erasure. See also: program, erase, EEPROM, PROM, ROM, RAM, non-volatile, windowed package. erase 1. (v.t.) To return a programmed device to its blank state. Opposite of "program". 2. (v.t.) To return an individual programmable cell to its blank state. See also: blank, programmable cell, program. • ESD (n.) Electrostatic Discharge. The natural physical event of the transferring of electrical charges. If uncontrolled, ESD can destroy or degrade both CMOS and bipolar semiconductor devices with inadequate on-chip protection circuitry and/or insufficient packaging and handling protection. See also: ESDS Device, CMOS, bipolar. ESDS Device (n.) Electrostatic Discharge Sensitive Device. A device which is sensitive to damage at certain levels of ESD. Three classes exist at ESD levels of up to 1999 V, to 3999 V and above 4000 V. See also: ESD. F finite state machine (FSM) (n.) A machine which can be in one of a finite number of states. Often used for logic circuits which sequence through various states. Such a circuit is referred to as sequential. See also: sequential. flip-flop (n.) A bistable digital circuit. The simplest variety is called an S-R flip-flop. Other types are J-K,T, and D-type. May be unclocked or clocked. See also: bistable, unclocked flip-flop, clocked flip-flop. floating gate (n.) A gate on an MOS transistor which is not connected to anything. Used to store charge; forms the basis of UV cells and EE cells. See also: MOS, gate, UV cell, EE cell. FPGA 1. (n.) Field Programmable Gate Array. A highdensity PlD with multiple levels of logic and programmable interconnect. 2. (n.) Field Programmable Gate Array. An array of logic gates whose configuration can be programmed by the customer. The gates are often NAND gates, but can also be NOR gates. See also: gate, program, NAND, NOR. FPlA (n.) Field Programmable Logic Array. See PLA. FPlS (n.) Field Programmable Logic Sequencer. A programmable logic device which is intended for sequencing or state machine applications. See also: finite state machine. Glossary 6·31 ~AMD functionally complete (adj.) Refers to a logic operation or group of operations from which any complex logic function can be built. The NAND and NOR operators are functionally complete. See also: NAND, NOR. fuse (n.) As used in programmable logic, usually refers to a lateral metal link fuse. See also: lateral fuse. fuse map (n.) A graphic representation of the contents of a PLD. The state of each connection (fuse or other programmable cell) is represented, usually with "X" indicating an intact connection, and ,,_If indicating an open connection. See also: PLD, programmable cell. G gate 1. (n.) A fundamental logic element. The elementary gates provide NOT, AND, and OR logic functions. 2. (n.) The control terminal of a gated D-type latch. See also: latch, gated latch. gate array (n.) A logic device which consists of an array of logic gates (usually NAND) which can be interconnected during fabrication. A custom metallization pattern is used to configure the desired functions. See also: gate, NAND, metallization. gate equivalency (n.) A rough measure of the complexity of a digital logic integrated circuit. Indicates the approximate number of discrete logic gates that would be needed to implement the same function. See also: gate. gated latch (n.) Generally refers to an unclocked D-type flip-flop which has a control signal called a gate. When the gate is "open", the flip-flop output follows the data input. When the gate is "closed", the output holds its current state. Also called a transparent latch. See also: flipflop, unclocked flip-flop, gate, latch. H HAL ~ device (n.) Hard Array Logic device. A version of a PAL device which is configured during fabrication with a custom metallization pattern. HAL is a registered trademark of Advanced Micro Devices. See also: PAL device, metallization. Invert (v.t.) To perform the logical NOT function on a digital signal. To reverse the polarity of a digital signal. See also: polarity, NOT. Inverter (n.) A logic gate which performs logical inversion, or the NOT operation. See also: gate, NOT. I/O (Input/Output) 1. (n.) The methods and equipment used to pass information into and/or out of a system or device. 2. (adj.) On a programmable logic device, a pin which can function as an input and/or an output. J JEDEC 1. (n.) Joint Electronic Device Engineering Council. A council which creates, approves, arbitrates, and/or oversees industry standards for electronic devices. 2. (adj.) In programmable logic, refers to a computer file containing information about the programming of a device. The file format is a JEDECapproved standard. Used for downloading to programmers. See also: program, programmer, download. junction Isolation (n.) A bipolar integrated circuit fabrication technique which uses P~N functions to isolate transistors. This is the original integrated circuit technology, and is being supplanted by oxide isolation in places where speed is critical. See also: oxide isolation, bipolar. K Karnaugh map (K-map) (n.) A graphic tool for minimizing sum-of-products or product-of-sums logic functions. Useful for up to six logic variables. See also: sum-ofproducts, product-of-sums. L latch 1. (n.) A type of flip-flop. Means different things to different people. In general, an unclocked flip-flop. Sometimes used to refer specifically to a gated D-type flip-flop. 2. (v.t.) To capture a signal in a latch. See also: flip-flop, unclocked flip-flop, gate, gated latch. latch up (v.t.) To enter the latch-up condition. See also: latch-up. J2L (ilL) (n., adj.) Integrated Injection Logic. A less common bipolar logic design technique which, when used, is found primarily in portions of LSI and VLSI circuits. See also: bipolar, LSI, VLSI. latch-up (n.) A condition in which a circuit draws uncontrolled amounts of current, and certain voltages are forced, or "latched-up" to some level. Used especially in reference to CMOS devices, which can latch up if the operating conditions are violated. See also: CMOS, latch up. Integrated circuit (n.) An electronic device which has many transistors and other semiconductor components integrated onto one piece of silicon. Often abbreviated IC. lateral fuse (n.) A thin metal link which is disconnected when programmed. Connected in the blank state, disconnected in the programmed state. Usually just called a "fuse". See also: program, programmed, blank. 6-32 Glossary AMD~ LCC (n.) Leadless Chip Carrier. A ceramic integrated circuit package having no leads. Connection is made to metal contacts which are flush with the package. See also: integrated circuit, lead, package. lead (n.) [led] A metal conductor which provides a connection from the inside of an integrated circuit package to the outside world for soldering or other mounting techniques. See also: integrated circuit. logic array (n.) Generally an array of programmable cells which attach inputs to logic gates of a specified type. See also: program, gate, programmable cell. logic simulation (n.) A means whereby a logic design can be evaluated on a computer before actually being built. The computer simulates the behavior of the components to predict the behavior of the overall circuit. LS (adj.) Low-power Schottky TTL family. Lower power version of the standard Schottky TTL family. See also: TTL, Schottky TTL. LSI (adj.) Large-Scale Integration. A rough measure of the complexity of a digital circuit. Characterized as having 100-5000 gate equivalents for logic chips, or 1 K16 K bits for memory chips. See also: gate equivalent, bit, VLSI, SSI, MSI. M macrocell (n.) Typically the output cell of a PLD, containing a flip-flop and path multiplexers. maxterm (n.) A sum in the canonical product-of-sums form. Each maxterm contains every input variable, in either true or complemented form. See also: product-ofsums, true, complement~ metallization (n.) The process of connecting the various elements of an integrated circuit or printed circuit board by placing a layer of metal overthe entire waferor board, and then selectively etching away unwanted metal. A photolit~ographic mask defines the pattern of connections. See also: integrated circuit, wafer, printed circuit board. mlnterm (n.) A product in the canonical sum-of-products form. Each minterm contains every input variable, either in true or complemented form. See also: sum-ofproducts, true, complement. monolithic (adj.) In the electronics industry, refers to a circuit which has been integrated onto one semiconductor chip. Integrated circuits are monolithic by definition. See also: integrated circuit. monostable (adj.) Describes a system which has 1 stable state. Any other state is unstable, and will eventually change to the stable state. The most common monostable circuit is a "one-shot". See also: bistable, astable. MOS (n., adj.) Metal-Oxide-Semiconductor transistor. One of the two basic types of transistor. In logic design, used for NMOS, PMOS, and CMOS families. See also: NMOS, PMOS, CMOS, bipolar. MSI (adj.) Medium-Scale Integration. A rough measure of the complexity of a digital logic circuit. Characterized as having 10-100 gate equivalents. See also: gate equivalent, SSI, LSI, VLSI. N NAND (adj.) Not AND. A commonly used logic gate which is equivalent to an AND gate followed by an inverter. The NAND logic operation is functionally complete.See also: gate, inverter, functionally complete, AND. negative logic (n.) A physical implementation of logic wherein a low voltage level represents a logic 1, or "true", and a high voltage level represents a logic 0, or '1alse". See also: positive logic, polarity. NMOS (n., adj.) N-channel MOS. A type of circuit which makes exclusive use of N-channel MOS transistors. See also: MOS, PMOS, CMOS. non-volatile (adj.) Refers to memory devices which do not lose their contents when power is removed. See also: volatile. NOR (adj.) Not OR. A logic gate which is equivalent to an OR gate followed by an inverter. The NOR logic operation is functionally complete. See also: gate, inverter, functionally complete, OR. NOT (adj.) One of the three elementary logic functions. Unary operation whose result is true if and only if the operand is false. o OR 1. (adj.) One of the three elementary logiC functions. Result of the OR operation is false if and only if all operands are false. 2. (v.t.) To perform the OR operation. OTP (adj.) One-Time Programmable. Refers to programmable devices which are UV-erasable, but which are not packaged in windowed packages. As a result, there is no way to erase the device, making it programmable only once. See also: program, erase, UV-erasable, windowed package. oxide Isolation (n.) A bipolar integrated circuit fabrication technique which uses silicon oxide to isolate transistors. This results in higher speed and density. See also: junction isolation, bipolar. Glossary 6-33 ~AMD p package (n.) The encasement which protects a die and provides convenient electrical contact to the die. Materials used are generally ceramic or plastic compounds. There are a variety of shapes and sizes. See also: die. PAL~ device (n.) Programmable Array Logic device. A PLD which implements logic via a programmable AND logic array driving a fixed OR logic array. PAL is a registered trademark of Advanced Micro Devices. See also: program, logic array, sum-of-products, PLD, AND, OR. PLA (n.) Programmable Logic Array. A programmable logic device which implements sum-of-products logic via programmable AND logic array driving a programmable OR logic array. See also: program, logic array, sum-of-products, AND, OR. a PLCC (n.) Plastic Leaded Chip Carrier. A molded plastic integrated circuit package with leads shaped like a "J" (J-leads).lntended for surface mounting. See also: integrated circuit, lead, surface mounting, package. PLD (n.) Programmable Logic Device. Generic term for a logic device whose function can be configured by the customer after purchase. See also: program. PMOS (n., adj.) P-channel MOS. A type of circuit which makes exclusive use ot P-channel MOS transistors. See also: MOS, NMOS, CMOS. polarity (n.) Specifies the sense of "active" and "inactive", or '1rue" and "false" in a digital signal. "Active high" represents '1rue" as a high signal; "active low" represents ''true'' as a low signal. positive logic (n.) A physical implementation of logic wherein a high voltage level represents a logic 1, or ''true'', and a low voltage level represents a logic 0, or ''false''. See also: negative logic, polarity. power dissipation (n.) The amount of electrical power used by a device. Calculated as the product of the operating voltage and current. Measured in watts (W) or milliwatts (mW), as appropriate. Sometimes incorrectly used to refer to the operating current only. printed circuit board (PC board, PCB) (n.) A board for assembling electrical components. Component connections are made by metal traces which have been fabricated through a metallization process. See also: trace, metallization. product-of-sums (PaS) (adj.) A representation of a logic function where the input signals are individually inverted (if necessary), then ORed together to form sums which are ANDed together. Any combinatorial logic function can be represented in product-ot-sums form. See also: sum-of-products, combinatorial, AND, OR. 6-34 product term (pterm, p-term) (n.) An AND gate in a PLD which implements sum-of-products logic. See also: sum-of-products, PLD, AND, gate. product term sharing (n.) See product term steering. product term steering (n.) A means whereby product terms in a PAL device can be routed to one of two device outputs, instead of being dedicated only to one output. Sometimes called "product term sharing". See also: product term, PAL device. program 1. (v.t.) As used in programmable logic, to configure a blank device so that it can perform some desired function. Applies to memory and logic devices. Opposite of "erase". 2. (v.t.) To change an individual programmable cell from a blank state to a programmed state. See also: blank, programmable cell, programmed, erase. programmable cell (n.) Any of a variety of cells which can be altered by applying certain electrical signals. Various types are lateral and vertical fuses, UV cells, E2 cells, and even RAM .cells. All but RAM cells are nonvolatile. See also: lateral fuse, vertical fuse, UV cell, E2 cell, RAM cell, non-volatile, volatile. programmed (adj.) Describes the state of a programmable cell or device after programming. OPPOSite "blank". programmer (n.) A device or machine used for configuring, or "programming", PLDs or PROMs. See also: program, PLD, PROM. PROM (n.) Programmable Read-Only Memory. A nonvolatile memory device whose contents are programmed by the customer. Once programmed, it cannot be erased. Also functions as a PLD with a fixed AND logic array which drives a programmable OR logic array. See also: program, erase, EEPROM, EPROM, ROM, RAM, non-volatile, AND, OR, logic array. R RAM (n.) Random-Access Memory. Sometimes called read/write memory. A type of memory device which can be written to and read at any time. Such memory is volatile. Actually a misnomer, since most types of memories can be accessed randomly. The distinguishing feature is the fact that RAM is designed specifically to be written to in normal usage. See also: ROM, volatile. RAM cell (n.) A cell which is used make one bit of volatile memory in a RAM. Can also form the basis of a programmable logic connectivity array. See also: RAM, volatile. Glossary AMD ;t1 ROM (n.) Read-Only Memory. A nonvolatile memory device which has its contents defined when manufactured. No changes can be made to the memory contents. See also: PROM, EPROM, EEPROM, RAM, nonvolatile. common clock. 2. (adj.) Describes signals whose behavior and timing are synchronized to a clock. 3. (adj.) Describes a communication protocol whereby the timing of various operations is determined by a system clock. See also: sequential, clock, asynchronous. S T Schottky TTL (adj.) Family of TIL devices which make use of Schottky diodes for higher speed. See also: TI~. security fuse (n.) A PLD feature which allows a user to "secure" the PLD after programming. This prevents subsequent copying of the contents of the PLD. See also: PLD, program. semlcustom (adj.) Refers to a circuit which has been partially designed by the device vendor, and partially designed, or configured, by the customer. Primary types are PLDs, gate arrays, and standard cell circuits. See also: PLD, gate array, standard cell. sequential (adj.) Refers to a logic circuit whose operation depends both on present input signals and previous operations, or states. Requires some kind of memory (usually flip-flops) for remembering past states. See also: flip-flop, combinatorial. SSI (adj.) Small Scale Integration. A rough measure of the complexity of a digital logic circuit. Characterized as having less than 10 gate equivalents. See also: gate equivalent, MSI, LSI, VLSI. standard cell (n.) A method of designing semicustom or full custom circuits whereby predefined cells are brought together to provide the specified function. Unlike gate arrays, all· fabrication steps are customized, instead of just the metallization step. See also: semicustom, gate array, metallization. standby power (n.) The power consumed by a device when none of the device inputs are switching. Usually used in reference to CMOS devices, many of which consume practically no standby power. See also: CMOS. sum-of-products (SOP) (adj.) A representation of a logic function where the input signals are individually inverted (if necessary), then ANDed together to form products which are ORed together. Any combinatorial logic function can be represented in sum-of-products form. See also: product-of-sums, combinatorial, AND, OR. surface mounting (n.) A printed circuit board assembly technique whereby the integrated circuit packages are placed on the board with no leads protruding through to the other side. Packages can thus be mounted on both' sides of the board. See also: printed circuit board, lead, through-hole mounting. synchronous 1. (adj.) Describes a sequential logic system wherein all operations are synchronized to a temperature compensation (n.) A circuit feature which allows some electrical characteristics to remain relatively constant with some variation in operating temperature. three-state (adj.) A type of logic device output which can be in one of three-states: HIGH, LOW, and OFF, or High-Z (high impedance). When enabled (on), performs as a normal binary output. When disabled (off), acts as an open pin. See also: enable, disable, binary. through-hole mounting (n.) A printed circuit board assembly technique whereby the leads of the various components extend through holes in the board. These leads are then soldered from the opposite side of the board. See also: printed circuit board, lead, surface mounting. trace 1. (n.) During logic simulation, the behavior of a signal or group of signals. The results can sometimes be stored in a "trace file" on disk for later analysis. 2. (n.) A thin layer of metal on a printed circuit board which provides connections between components. Performs the function of a wire. See also: logic simulation, printed circuit board. transparent latch (n.) See gated latch. TRI-STATE$ (adj.) See three-state. TAl-STATE is a registered trademark of National Semiconductor Corp. true (adj.) Refers to a signal which is identical to some reference signal, with the same polarity. Opposite of "complement". See also: complement, polarity. TTL (adj.) Transistor-Transistor Logic family. The most widely used family of bipolar logic devices. The name refers to the particular circuit design technique used. See also: bipolar. u unclocked flip-flop (n.) A flip-flop that changes state as soon as the appropriate controls are applied.See also: flip-flop, clocked flip-flop. upload 1. (v.t.) To pass data from one machine to a more complex machine. 2. (n.) The act of uploading data. See also: download. UV cell (n.) A floating gate cell which can be erased by exposure to ultraviolet (UV) light. See also: floating gate, erase. Glossary 6-35 ~AMD UV-erasable (adj.) Refers to devices or programmable cells which can be erased when exposed to ultraviolet (UV) light for a period of time. See also: programmable cell, erase. v vertical fuse (n.) A transistor arranged such that the emitter and base are shorted together when programmed. Disconnected in the blank state, connected in the programmed state. See also: program, programmed, blank. VLSI (adj.) Very Large Scale Integration. A rough measure of the complexity of a digital circuit. Characterized as having 5000 or more gate equivalents for logic chips, or 16K or more bits for memory chips. See also: gate equivalent, bit, SSI, MSI, LSI. 6-36 volatile (adj.) Refers to memory devices which lose their contents when power is removed. See also: nonvolatile. voltage compensation (n.) A circuit feature which allows some electrical characteristics to remain relatively constant with some variation in the supply voltage. W wafer (n.) A round slice of very pure silicon which is used in the fabrication of integrated circuits. Several circuits can be built on one wafer. See also: integrated circuit. windowed package (n.) A package which has a quartz window in the lid directly overthe die. This makes it possible to expose the die to ultraviolet light for erasing the device.See also: erase, die, package. Glossary Worldwide Application Support Advanced Micro Devices The Worldwide Application Support Group is dedicated to making sure that the customer's need for technical support is met. problems which might occur when the design reaches production. Call your local Sales Office for an FAE in your area. It is the job of Applications Engineering to ensure that the customer engineer is familiar with our product line, that they have the information and tools necessary to get the appropriate part designed in, and that they have access to technical support throughout the lifetime of their product. Factory Applications Support-This group, located within the factory, is responsible for centralizedtechnical support of all AMD's products. Their duties include developing seminars for training of FAEs and customers, and hosting the twice yearly Applications Conference. They are responsible for the coordination of customertraining and giving the factory an insight into the customer's point of view. Support is provided by two groups: Field Applications Engineers-With Applications Engineers located in our field sales offices throughout the world, our customers always have access to an engineer when they have problems or questions. Working with Sales, our engineers teach seminars showing customers how to use our new parts, assist them during their design when technical questions come up, and provide troubleshooting support to help eliminate Additionally, the Factory Applications Support Group provides customer support for technical questions via a toll-free number: (800) 222-9323 (listed on the back of all databooks). Assistance is available from 7 a.m. to 5:30 p.m. (Pacific time) Monday thru Friday. Currently 80% of the questions are answered on the first call, 20% within 24 hours. 6-37 ~ AMD AMD~ ~AMD AMD~ International ALABAMA ............................................................... (205) 882·9122 ARIZONA ...............................................................; (602)· 242·4400 CALIFORNIA, Culver City ......................................................... (31 0) 645·1524 Newport Beach ................................................. (714) 752·6262 Sacramento(Roseville) .................................... (916) 786·6700 San Diego .......................................................... (619) 560·7030 San Jose ............................................................ (408) 452·0500 Woodland Hills .................................................. (818) 878·9988 CANADA, Ontario, Kanata ........................... :.................................... (613) 592·0060 Willowdale ......................................................... (416) 222·7800 COLORADO ........................................................... (303) 741·2900 CONNECTICUT ..................................................... (203) 264·7800 FLORIDA, Clearwater ......................................................... (813) 530·9971 Boca Raton ........................................................ (407) 361·0050 Orlando (Longwood) ........................................ (407) 862·9292 GEORGIA ............................................................... (404) 449·7920 IDAHO ..................................................................... (208) 377·0393 ILLINOIS, Chicago (Itasca) ............................................... (708) 773·4422 Naperville ........................................................... (708) 505·9517 MARYLAND ............................................................ (301) 381·3790 MASSACHUSETTS ............................................... (617) 273·3970 MINNESOTA .......................................................... (612) 938·0001 NEW JERSEY, Cherry Hill .......................................................... (609) 662·2900 Parsippany ......................................................... (201) 299·0002 NEW YORK, Brewster ............................................................. (914) 279·8323 Rochester .......................................................... (716) 425·8050 NORTH CAROLINA Charlotte ............................................................. (704) 875· 3091 Raleigh ................................................................ (919) 878· 8111 OHIO, Columbus (Westerville) .................................... (614) 891·6455 Dayton ................................................................ (513) 439·0268 OREGON ................................................................ (503) 245·0080 PENNSYLVANIA .................................................... (215) 398·8006 TEXAS, Austin ................................................................. (512) 346·7830 Dallas ................................................................. (214) 934·9099 Houston .............................................................. (713) 376·8084 International ____________ BELGIUM, Antwerpen ..... TEL ............................... (03) 248 43 00 FAX .............................. (03) 248 46 42 FRANCE, Paris ................ TEL ............................. (1) 49·75·10·10 FAX ............................. (1) 49·75·10·13 GERMANY, Bad Homburg .............. TEL ............................... (06172)·24061 FAX .............................. (06172)·23195 Munchen ....................... TEL .. ;............................... (089) 45053·0 FAX ................................. (089) 406490 HONG KONG, .................... TEL ................................ (852) 865·4525 Wanchai FAX ............................. (852) 865c4335 ITALY, Milano ................... TEL ................................. (02) 3390541 FAX .............................. (02) 38103458 JAPAN, Atsugi .......................... TEL ............................. (0462) 29·8460 FAX ............................. (0462) 29·8458 Kanagawa .................... TEL ............................. (0462) 47·2911 FAX ............................. (0462) 47·1729 (Continued) _ _ _ _ _ _ _ _ __ Tokyo ............................ TEL ............................. (03) 3346·7550 FAX ............................. (03) 3342·5196 Osaka ........................... TEL ................................ (06) 243·3250 FAX ............................... (06) 243·3253 KOREA, Seoul :................. TEL ............................ (82) 2·784·0030 FAX ............................ (82) 2·784·8014 LATIN AMERICA, Ft. Lauderdale ............. TEL ............................. (305) 484·8600 FAX ............................. (305) 485·9736 SINGAPORE ..................... TEL ................................. (65) 3481188 FAX ................................. (65) 3480161 SWEDEN, Stockholm area ........... TEL ................................ (08) 9861 80 (Bromma) FAX ............................... (08) 98 09 06 TAIWAN, Taipei ............... TEL ........................... (886) 2·7153536 FAX ........................... (886) 2·7122183 UNITED KINGDOM, Manchester area ......... TEL ............................... (0925) 830380 (Warrington) FAX .............................. (0925) 830204 London area ................ TEL ............................... (0483) 740440 (Woking) FAX .......... ,................... (0483) 756196 North American Representatives - - CANADA Burnaby, B.C. - DAVETEK MARKETING ........... (604) Kanata, Ontario - VITEL ELECTRONICS .......... (613) Mississauga, Ontario - VITEL ELECTRONICS. (416) Lachine, Quebec - VITEL ELECTRONICS ........ (514) ILLINOIS Skokie - INDUSTRIAL REPRESENTATIVES,INC ................................ (708) IOWA LORENZ SALES ............................................... (319) KANSAS Merriam - LORENZ SALES ............................ (913) Wichita - LORENZ SALES .............................. (316) MICHIGAN I Holland - COM·TEK SALES, INC .................. (616) Brighton - COM·TEK SALES, INC ................. (313) MINNESOTA Mel Foster Tech. Sales, Inc ............................ (612) MISSOURI LORENZ SALES ............................................... (314) NEBRASKA LORENZ SALES ............................................... (402) NEW MEXICO THORSON DESERT STATES ........................ (505) NEW YORK East Syracuse - NYCOM, INC ....................... (315) Hauppauge - COMPONENT CONSULTANTS, INC ...................... : ................ (516) OHIO Centerville - DOLFUSS ROOT & CO ............ (513) Columbus - DOLFUSS ROOT & CO ............. (614) Westlake - DOLFUSS ROOT & CO ............... (216) PENNSYLVANIA RUSSELL F. CLARK CO.,INC ........................ (412) PUERTO RICO COMP REP ASSOC, INC ................................ (809) UTAH Front Range Marketing .................................... (801) WASHINGTON ELECTRA TECHNICAL SALES ...................... (206) WISCONSIN Brookfield - INDUSTRIAL REPRESENTATIVES, INC ................................. (414) Advanced Micro Devices reserves the right to make changes in its product without notice in order to improve design or performance characteristics. The performance characteristics listed in this document are guaranteed by specific tests, guard banding, design and other practices common to the industry. For specific testing details, contact your local AMD sales representative. The company assumes no responsibility for the use of any circuits described herein. ~ ~. Advanced Micro Devices, Inc. 901 Thompson Place. P.O. Box 3453, Sunnyvale, CA 94088, USA Tel: (408) 732-2400 • T'NX: 910-339-9280 • TELEX: 34-6306 • TOLL FREE: (800) 538-8450 APPLICA T/ONS HOTLINE & LITERATURE ORDERING • TOLL FREE: (800) 222-9323 • (408) 749-5703 430·3680 592·0060 564·9720 636·5951 967·8430 377·4666 469·1312 721·0500 335·8418 227·0007 941·9790 997·4558 475·4660 883·4343 437·8343 273·5050 433·6776 885·4844 899·9370 242·9500 746·6550 288·2500 821·7442 574·9393 ~ RECYCLED & RECYCLABLE 10 1993 Advanced Micro Devices, Inc. 101730 6/18/93 BAN·SOM-7/93 Printed in USA Family Part Number Standard Packages STANDARD PAL DEVICES 16RS Technology tPD ns Icc mA Page TTL 4.5 210 2-3 PAL16LS-4 PAL16RS-4 PAL16R6-4 PAL16R4-4 28J PAL16LS-5 PAL16RS-5 PAL16R6-5 PAL16R4-5 20P,J 5 210 PAL16LS-7 PAL16R8-7 PAL16R6-7 PAL16R4-7 20P, J, D 7.5 1S0 20P, J 10 180 20N, J, NL 15 180 PAL16LSB-2 PAL16R8B-2 PAL16R6B-2 PAL16R4B-2 PAL16LSA PAL16R8A PAL16R6A PAL16R4A 25 90 25 180 PAL16L8B-4 PAL16R8B-4 PAL16R6B-4 PAL16R4B-4 35 55 PAL16L8D/2 PAL16RSDI2 PAL16R6D/2 PAL16R4D/2 PAL16L8B PAL16R8B PAL16R6B PAL16R4B 20RS I PAL20LS-5 PAL20R8-5 PAL20R6-5 PAL20R4-5 24P,2SJ 5 210 PAL20L8-7 PAL20R8-7 PAL20R6-7 PAL20R4-7 24P, 28J, 240 7.5 210 PAL20L8-10/2 PAL20R8-10/2 PAL20R6-10/2 PAL20R4-10/2 24P,2SJ 10 210 24NS, 28NL, 24JS 15 210 PAL20L8B-2 PAL20RSB-2 PAL20R6B-2 PAL20R4B-2 25 105 PAL20L8A PAL20R8A PAL20R6A PAL20R4A 25 210 PAL20L8B PAL20R8B PAL20R6B PAL20R4B 2-121 1SP8 AmPAL18P8B AmPAL18PSAL AmPAL1SPSA AmPAL1SP8L 20P,J 15 25 25 35 180 90 1S0 90 2-155 22P10 AmPAL22P10B Am PAL22P 1OAL AmPAL22P10A 24P,2SJ 15 25 25 180 90 180 2-249 ADVANCED MICRO DEVICES, INC. 901 Thompson Place P.D. Box 3453 Sunnyvale, California 94088-3453 (408) 732-2400 TWX: 910-339~9280 TELEX: 34-6306 APPLICA TlONS HOTLINE & LITERATURE ORDERING USA (800) 222-9323 USA (408) 749-5703 JAPAN 011-81-3-3346-7561 UK & EUROPE 44-(0)256-811101 TOLL FREE USA (800) 538-8450 FRANCE 0590-8621 GERMANY 0130-813875 ITALY 1678-77224 RECYCLED & RECYCLABLE Printed in USA BAN-80M-6/93-0 10173D
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 : 2011:08:30 15:17:12-08:00 Modify Date : 2011:08:30 19:39:24-07:00 Metadata Date : 2011:08:30 19:39:24-07:00 Producer : Adobe Acrobat 9.45 Paper Capture Plug-in Format : application/pdf Document ID : uuid:dd50edbb-67ec-4f0a-b732-429d2212a6b6 Instance ID : uuid:0cc11e27-b8e9-415d-8287-9cea7d1e6710 Page Layout : SinglePage Page Mode : UseNone Page Count : 764EXIF Metadata provided by EXIF.tools