1976_National_Audio_Handbook 1976 National Audio Handbook

User Manual: 1976_National_Audio_Handbook

Open the PDF directly: View PDF .
Page Count: 202

Download
Open PDF In Browser	View PDF

AUDIO
HANDBOOK
Technical Editor & Contributing Author:

Dennis Bohn
Consumer Application Engineer
Contributing Authors:

John Wright
Ron Page
Tim Regan
Thomas B. Mills
John Maxwell
Tim D. Isbell
Nello Sevastopou los
Jim Sherwin

National Semiconductor Corporation • 2900 Semiconductor Drive • Santa Clara, CA 95051
©

1976 National Semiconductor Corp.

DA-A70M46/Printed in U.S.A.

Section Edge Index

Introduction
Preamplifiers
AM, FM and FM Stereo
Power Amplifiers
noobydusl
Appendices
Index

Manufactured under one or more of the following U.S. patents: 3083262,3189758,3231797,3303356,3317671, 3323071, 3381071,
3408542, 3421025, 3426423, 3440498, 3518750, 3519897, 3557431, 3560765,3566218,3571630,3575609,3579059,3593069,
3597640,3607469,3617859,3631312,3633052, 3638131, 3648071, 3651565, 3693248.

National does not assume any responsibility for use of any circuitry described; 00 circuit patent licenses are implied;
and National reserves the right, at any time without notice, to change said circuitry.

Table of Contents
1.0 Introduction
1.1
1.2

Scope of Handbook .
IC Parameters Applied to Audio.

1-1
1-1

2.0 Preamplifiel'S
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
2.13
2.14
2.15
2.16
2.17
2.18
2.19
2.20

Feedback - To Invert or Non-Invert
Design Tips on Layout, Ground Loops and Supply Bypassing
Noise.
Audio Rectification -- or, "How Come My Phono Detects AM?"
Dual Preamplifier Selection
LM381
LM381A
LM387 and LM387 A
LM382. . .
LM1303. . .
Phono Preamps and R IAA Equalization
Tape Preamps and NAB Equalization
Mic Preamps.
.....
Tone Controls - Passive and Active
Scratch, Rumble and Speech Filters
Bandpass Active Filters
Octave Equalizers
Mixers
Driving Low Impedance Lines
Noiseless Audio Switching

2-1
2-1
2-3
2-10
2-11
2-12
2-15
2-19
2-20
2-24
2-25
2-31
2-37
2-40
2-49
2-52
2-53
2-59
2-61
2-62

3.0 AM, FM and FM Stereo
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9

AM Radio .
LM1820
FM-I F Amplifiers/Detectors
Simple Limiters . .
Gain Blocks
Complete I F Amplifier and Detectors
LM3089 - Today's Most Popular FM-IF System
FM Stereo Multiplex - LM1310/1800 .
Definition of Terms

3-1
3-4
3-8
3-8
3-11
3-13
3-18
3-23
3-27

4.0 Power Amplifiel'S
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4.11
4.12
4.13
4.14

Inside Power Integrated Circuits.
Design Tips on Layout, Ground Loops and Supply Bypassing
Power Amplifier Selection.
LM377 /378/379
LM380
LM384
LM386
LM389
LM388
LM390
Boosted Power Amps/LM391 .
Power Dissipation
Effect of Speaker Loads
Heatsinking

4-1
4-4
4-4
4-8
4-21
4-28
4-30
4-33
4-37
4-41
4-42
4-43
4-45
4-46

Table of Contents
(continued)

5.0 Fioobydust*
5.1
5.2
5.3
5.4
5.5
5.6
5.7

5-1
5-1
5-7

Biamplification
Active Crossover Networks
Reverb. .
Phase Shifter
Fuzz. . .
Tremolo. .
Acoustic Pickup Preamp

5-10
5-11
5-11
5-12

6.0 Appendices
A1
A2
A3
A4
A5
A6
A7
A8

Power Supply Design
Decibel Conversion .
Wye-Delta Transformation
Standard Building Block Circuits
Magnetic Phono Cartridge Noise Analysis.
General Purpose Op Amps Useful for Audio .
Feedback Resistors and Amplifier Noise
Reliability

6-'
6-11
6-11
6-12
6-13
6-16
6-17
6-18

7.0 Index

*"Floobydust" is a contemporary term derived from the archaic Latin miscellaneus, whose disputed history probably springs
from Greek origins (influenced, of course, by Egyptian linguists) - meaning here "a mixed bag."

Device Index
LM171
LM377
LM378
LM379
LM380
LM381
LM381A
LM382
LM384
LM386
LM387
LM387A
LM388
LM389 .

3-9
4-8
4-8
4-8
4-21
2-12
2-15
2-20
4-28
4-30
2-19
2-19
4-37
4-33

LM390
LM391
LM703
LM1303.
LM1310.
LM1351.
LM1800·
LM1820·
LM2111·
LM3011.
LM3065·
LM3075·
LM3089·
MM5837·

4-41
4-43
3-9
2-24
3-23
3-13
3-23
3-4
3-13
3-11
3-15
3-15
3-18
2-56

1.0 Introduction
it must be done noiselessly - in the sun, and in the snow forever.

In just a few years time, National Semiconductor Corporation has emerged as a leader - indeed, if not the leader in
all areas of integrated circuit products. National's wellknown linear and digital ICs have become industry standards
in all areas of design. This handbook exists to acquaint
those involved in audio systems design with National
Semiconductor's broad selection of integrated circuits
specifically designed to meet the stringent requirements
of accurate audio reproduction. Far from just a collection
of data sheets, this manual contains detailed discussions,
including complete design particulars, covering many areas
of audio. Thorough explanations, complete with real-world
design examples, make clear several audio areas never before
available to the general public.

Unfortunately, this IC doesn't exist; we're working on it,
but it's not ready for immediate release. Meanwhile, the
problem remains of how to choose from what is available.
For the most part, DC parameters such as offset voltages
and currents, input bias currents and drift rates may be
ignored. Capacitively coupling for bandwidth control and
single supply operation negates the need for concern about
DC characteristics. Among the various specifications applicable to AC operation, perhaps slew rate is the most
important.
1.2.1 Slew Rate
The slew rate limit is the maximum rate of change of the
amplifier's output voltage and is due to the fact that the
compensation capacitor inside the amplifier only has finite
currents l available for charging and discharging (see Section
4.1.2). A sinusoidal output signal will cease being small
signal when its maximum rate of change equals the slew
rate limit Sr of the amplifier. The maximum rate of
change for a sine wave occurs at the zero crossing and may
be derived as follows:

1.1 SCOPE OF HANDBOOK
Between the hobbyist and the engineer, the amateur and
the professional, the casual experimenter and the serious
product designer there exists a chaotic space filled with
Laplace transforms, Fourier analysis, complex calculus,
Maxwell's equations, solid-state physics, wave mechanics,
holes, electrons, about four miles of effete mysticism, and,
maybe, one inch of compassion. This audio handbook
attempts to disperse some of the mist. Its contents cover
many of the multidimensional fields of audio, with emphasis
placed on intuition rather than rigor, favoring the practical
over the theoretical. Each area is treated at the minimum
depth felt necessary for adequate comprehension. Mathematics is not avoided - only reserved for just those areas
demanding it. Some areas are more_ "cookbook" than
others, the choice being dictated by the material and
Mother Nature.

vo = V p sin 211 ft

(1.2.1 )

dvo
= 211fVpcos211ft
dt

(1.2.2)

I

General concepts receive the same thorough treatment as do
specific devices, based upon the belief that the more
informed integrated circuit user has fewer problems using
integrated circuits. Scanning the Table of Contents will
indicate the diversity and relevance of what is inside.
Within the broad scope of audio, only a few areas could be
covered in a book this size; those omitted tend to be ones
not requiring active devices for implementation (e.g., loudspeakers, microphones, transformers, styli, etc.).

dvo
= 211fVp
dt t=O

(1.2.3)

Sr = 211 f max Vp

(1.2.4)

where:

Vo = output voltage
V p = peak output voltage
dvo
Sr = maximumdt

The maximum sine wave frequency an amplifier with a given
slew rate will sustain without causing the output to take on
a triangular shape is therefore a function of the peak
ampl itude of the output and is expressed as:

Have fun.
1.2 Ie PARAMETERS APPLIED TO AUDIO
Audio circuits place unique requirements upon IC parameters which, if understood, make proper selection of a
specific device easier. Most linear integrated circuits fall
into the "operational amplifier" category where design
emphasis has traditionally been placed upon perfecting
those parameters most applicable to DC performance. But
what about AC performance? Specifically, what about audio
performance?

Sr
f max = - 211 Vp

(1.2.5)

Equation (1.2.5) demonstrates that the borderline between
small signal response and slew rate limited response is not
just a function of the peak output signal but that by trading
off either frequency or peak amplitude one can continue to
have a distortion free output. Figure (1.2.1) shows a quick
reference graphical presentation of Equation (1.2.5) with
the area above any VPEAK line representing an undistorted
small signal response and the area below a given VPEAK
line representing a distorted sine wave response due to slew
rate limiting.
As a matter of convenience, amplifier manufacturers often
give a "full-power bandwidth" or "large signal response" on
their specification sheets.

Audio is really a rather specialized area, and its requirements
upon an integrated circuit may be stated quite concisely:
The Ie must process complex AC signals comprised of
frequencies ranging from 20 hertz to 20k hertz, whose
amplitudes vary from a few hundred microvolts to several
volts, with a transient nature characterized by steep,
compound wavefronts separated by unknown periods of
absolute silence. This must be done without adding distortion of any sort, either harmonic, amplitude, or phase; and

1 -1

1.2.4 Noise
100

"'-"~

10

Jp;~~ "1'6~
VPEAK"8VM

~"

VPEAK

~

~

The importance of noise performance from an integrated
circuit used to process audio is obvious and needs little
discussion. Noise specifications normally appear as "Total
Equivalent Input Noise Voltage," stated for a certain source
impedance and bandwidth. This is the most useful number,
since it is what gets amplified by the closed loop gain of the
amplifier. For high source impedances, noise current
becomes important and must be considered, but most
driving impedances are less than 600n, so knowledge of
noise voltage is sufficient.

SMALL SIGNAL
RESPONSE AREA

'"

4V

~

VPEAK '" 2V:::T 1Ji!
VPEAK

0.1

=lV

/1/
SLEW RATE

.01

'J,fiYy
100

lk

LIMITING AREA

10k

lOOk

1M

1.2.5 Total Harmonic Distortion

SINE WAVE FREQUENCY (Hz)

Need for low total harmonic distortion (THO) is also
obvious and need not be belabored. THD performance for
preamplifier ICs will state the closed loop gain and frequency
at which it was measured, while audio power amplifiers will
also include the power output.

FIGURE 1.2.1 Sine Wave Response

This frequency can be derived by inserting the amplifier
slew rate and peak rated output voltage into Equation
(1.2.5). The bandwidth from DC to the resulting f max is
the full-power bandwidth or "large signal response" of the
amplifier. For example, the full-power bandwidth of the
LM741 with a 0.5V/j1s Sr is approximately 6kHz while the
full-power bandwidth of the LF356 with a Sr of 12V/j1s
is approximately 160 kHz.

1.2.6 Supply Voltage
Consideration of supply voltage limits may be more important than casual thought would indicate. For preamplifier
ICs and general purpose op amps, attention needs to be
directed to supply voltage from a dynamic range, or
"headroom," standpoint. Much of audio processing roquires
headroom on the order of 20-40dB if transient clipping is to
be avoided. For a design needing 26dB dynamic range with
a nominal input of 50mV and operating at a closed loop
gain of 20dB, a supply voltage of at least 30 V would be
required. It is important, therefore, to be sure the IC has a
supply voltage rating adequate to handle the worst case
conditions. These occur for high power line cases and low
current drain, requiring the IC user to check the "absolute
maximum" ratings for supply voltage to be sure there are
no conditions under which they will be exceeded. Remember, "absolute maximum" means just that - it is not the
largest supply you can apply; it is the value which, if
exceeded, causes all bets to be cancelled. This problem is
more acute for audio power devices since their supplies tend
to sag greatly, i.e., the difference between no power out
and full power out can cause variations in power supply
level of several volts.

1.2.2 Open Loop Gain
Since virtually all of an amplifier's closed loop performance
depends heavily upon the amount of loop-gain available,
open loop gain becomes very important. Input impedance,
output impedance, harmonic distortion and frequency
response all are determined by the difference between open
loop gain and closed loop gain, i.e., the loop gain (in dB).
Details of this relationship are covered in Section 2.1. What
is desired is high open loop gain - the higher the better.
1.2.3 Bandwidth and Gain-Bandwidth
Closely related to the slew rate capabilities of an amplifier
is its unity gain bandwidth, or just "bandwidth." The
"bandwidth" is defined as the frequency where the open
loop gain crosses unity. High slew rate devices will exhibit
wide bandwidths.

1.2.7 Ripple Rejection

Because the size of the capacitor required for internally
compensateq devices determines the slew rate - hence, the
bandwidth - one method used to design faster ampl ifiers is
to simply make the capacitor smaller. This creates a faster
IC but at the expense of unity-gain stability. Known as a
decompensated (as opposed to uncompensated - no
capacitor) amplifier, it is ideal for most audio applications
requiring gain.

An integrated circuit's ability to reject supply ripple is
important in audio applications. The reason has to do with
minimizing hum within the system - high ripple rejection
means low ripple bleedthrough to the output, where it adds
to the signal as hum. Relaxed power supply design (i.e.,
ability to tolerate large amounts of ripple) is allowed with
high ripple rejection parts.
Supply ripple rejection specifications cite the amount of
rejection to be expected at a particular frequency (normally
120Hz), or over a frequency band, and is usually stated in
dB. The figure may be "input referred" or "output
referred." If input referred, then it is analogous to input
referred noise and this amount of ripple will be multiplied
by the gain of the amplifier. If output referred, then it is
the amount of ripple expected at the output for the given
conditions.

The term gain-bandwidth is used frequently in place of
"unity gain bandwidth." The two terms are equal numerically but convey slightly different information. Gainbandwidth, or gain-bandwidth product, is a combined
measure of open loop gain and frequency response - being
the product of the available gain at any frequency times
that frequency. For example, an LM381 with gain of
around 2000V/V at 10kHz yields a GBW equal to 20MHz.
The GBW requirement for accurate audio reproduction may
be derived for general use by requiring a minimum loopgain of 40dB (for distortion reduction) at 20kHz for an
amplifier with a closed loop gain of 20dB. This means a
minimum open loop gain of 60dB (1000V/V) at 20kHz, or
a GBW equal to 20MHz. Requirements for lo-fi and mid-fi
designs, where reduced frequency response and higher
distortion are allowable, would, of course, be less.

REFERENCES

1. Solomon, J. E.. Davis, W. R., and Lee, P. L., "A SelfCompensated Monolithic Operational Amplifier with
Low Input Current and High Slew Rate," ISSCC Digest
Tech. Papers, February 1969, pp. 14-15.
1-2

-

----

2.0 Preampli6ers
2.1

FEEDBACK - TO INVERT OR NON·INVERT

2.2 DESIGN TIPS ON LAYOUT, GROUND LOOPS, AND
SUPPLY BYPASSING

The majority of audio applications of integrated circuits
falls into two general categories: inverting and non·inverting
amplifiers. Both configurations employ feedback of a frac·
tion of the output voltage (or current) back to the input.
A general discussion of feedback amplifier theory will not
be undertaken in this handbook; the interested reader is
referred to the references cited at the end of this section.
What follows is an abbreviated summary of the important
features of both types of amplifiers so the user may develop
an intuitive feel for which configuration best suits any given
application.

The success of any electronic circuit depends on good
mechanical construction as well as on sound electrical
design. Because of their high gain·bandwidth, high input
impedance characteristics, ICs tend to be less forgiving of
improper layout than their discrete counterparts. Many
excellent "paper" circuits wind up not worth the solder
they contain when improperly breadboarded, and are need·
lessly abandoned in frustration; this experience can be
avoided with proper breadboard techniques.

inverting amplifiers use shunt·shunt feedback, while non·
inverting amplifiers use series·shunt feedback. These names
derive from whether the feedback is in series or shunt with
the input and output. Thus, a series·shunt scheme has feed·
back that is in series with the input and is in shunt
(parallel) with the output.

Good layout involves logical placement of passive compon·
ents around the IC, properly dressed leads, avoidance of
ground loops, and adequate supply bypassing. Consult the
following list prior to breadboarding a circuit to familiarize
yourself with its contents:

2.2.1 Layout

•

Make overall layout compact.

•

Keep all component lead lengths as short as possible.

•

Route all inputs and input related components away
from any outputs.

•

Separate input and output leads by a ground or supply
trace where possible.

•

Low level high impedance signal carrying wires may
require shielded cable.

An important concept in understanding feedback amplifiers
is that of "loop gain." If the gain of an amplifier is
expressed in decibels then the loop gain equals the algebraic
difference between the open loop and closed loop gains
(e.g., an amplifier with 100dB open loop gain and 40dB
closed loop gain has 60dB of loop gain).
Table 2.1.1 is provided as a summary of the most important
amplifier parameters and the effect of feedback upon them.
AVCL = c1osed·loop gain

•

Make good solder connections, removing all excess flux.

GBW = gain bandwidth product = unity·gain frequency

•

Avoid using the popular plug·in socket strips. (These
units are excellent for digital ICs but troublesome for
linear breadboarding.)

Rf
Rin

= feedback resistor
= open·loop differential input impedance

2.2.2 Ground Loops

Ro = open·loop output impedance
T

"Ground Loop" is the term used to describe situations
occurring in ground systems where a difference in potential
exists between two ground points. Ideally a ground is a
ground is a ground. Unfortunately, in order for this to be
true, ground conductors with zero resistance are necessary.

= loop gain

THO = open·loop total harmonic distortion (%)

Observe (Table 2.1.1) that feedback affects output imped·
ance and harmonic distortion equally for both amplifier
types. Input impedance is high for non·inverting and low
for inverting configurations. The noise gains differ only
by unity and become significant for low gain applications,
e.g., in the unity gain case an inverting amplifier has twice
the noise gain of a non·inverting counterpart. (See Section
2.3 for detailed discussion of noise performance.) Band·
widths are similarly related, i.e., for the unity gain case a
non·inverting amplifier will have twice the bandwidth of
the inverting case.

2.1 REFERENCES

1. Graeme, J. G., Tobey, G. E., and Huelsman, L. P.,
Operational Amplifiers: Design and Applications,
McGraw·Hill, New York, 1971.
2. Jung, W. G., IC Op·Amp Cookbook, H. W. Sams & Co.,
Inc., Indiana, 1974.

3. Millman, J., and Halkias, C. C., Integrated Circuits:
Analog and Digital Circuits and Systems, McGraw·Hill,
New York, 1972.

TABLE 2.1.1 Summary of Feedback Amplifier Parameters.

Amplifier
Type

Input
Impedance

Output
Impedance

Non·inverting

(1 +T)Rin

--

--

1+T

1+T

Inverting

Rf

-

T

Harmonic
Distortion

Ro

Noise
Gain

THO

Ro

THO

--

--

1+T

1+T

2·1

Bandwidth
(closed·loop)
GBW

AVCL

--AVCL

AVCL + 1

GBW
AVCL+1

The single-point ground concept should be applied rigorously to all components and all circuits. Violations of singlepoint grounding are most common among printed circuit
board designs_ Since the circuit is surrounded by large
ground areas the temptation to run a device to the closest
ground spot is high. This temptation must be avoided if
stable circuits are to result.

Real-world ground leads possess finite resistance, and the
currents running through them will cause finite voltage
drops. If two ground return lines tie into the same path at
different points there will be a voltage drop between them_
Figure 2.2.1 a shows a common-ground example where the
positive input ground and the load ground are returned to
the supply ground point via the same wire. The addition of
the finite wire resistance (Figure 2.2.1 b) results in a voltage
difference between the two points as shown.

A final rule is to make all ground returns low resistance and
low inductance by using large wire and wide traces.

2_2_3 Supply Bypassing
Many IC circuits appearing in print (including many in this
handbook) do not show the power supply connections or
the associated bypass capacitors for reasons of circuit
clarity. Shown or not, bypass capacitors are always required.
Ceramic disc capacitors (0_1 J,tF) or solid tantalum (1 J,tF)
with short leads, and located close (within one inch) to the
integrated circuit are usually necessary to prevent interstage
coupling through the power supply internal impedance.
Inadequate bypassing will manifest itself by a low frequency
oscillation called "motorboating" or by high frequency
instabilities. Occasionally multiple bypassing is required
where a 10J,tF (or larger) capacitor is used to absorb low
frequency variations and a smaller 0.1 J,tF disc is paralleled
across it to prevent any high frequency feedback through
the power supply lines_

SUPPLY
GROUND

la)

In general, audio ICs are wide bandwidth (- 10MHz)
devices and decoupling of each device is required. Some
applications and layouts will allow one set of supply bypassing capacitors to be used common to ·several ICs. This
condition cannot be assumed, but must be checked out
prior to acceptance of the layout. Motorboating will be
audible, while high frequency oscillations must be observed
with an oscilloscope.

SUPPLY
GROUND

Ib)
FIGURE 2_2.1 Ground Loop Example

r - -if- -,

Load current I L will be much larger than input bias current
11, thus Vl will follow the output voltage directly, Le., in
phase. Therefore the voltage appearing at the non-inverting
input is effectively positive feedback and the circuit may
oscillate_ If there were only one device to worry about then
the values of R 1 and R2 would probablY be small enough
to be ignored; however, several devices normally comprise a
total system. Any ground return of a separate device, whose
output is in phase, can feedback in a similar manner and
cause instabilities. Out of phase ground loops also are
troublesome, causing unexpected gain and phase errors.

I

-=
la) Unity-Gain Stable Device

The solution to this and other ground loop problems is to
always use a single-point ground system. Figure 2.2.2 shows
a single-point ground system applied to the example of
Figure 2.2.1. The load current now returns directly to the
supply ground without inducing a feedback voltage as
before.

(b) ·Decompensated Device

FIGURE 2_2_3 Addition of Feedback Capacitor

2_2,4 Additional Stabilizing Tips
If all of the previous rules are followed closely, no instabilities should occur within the circuit; however, Murphy being
the way he is, some circuits defy these rules and oscillate
anyway. Several additional techniques may be required
when persistant oscillations plague a circuit:

SUPPl Y
GROUND

FIGURE 2.2_2 Single-Point Ground System

2-2

•

Reduce high impedance positive inputs to the minimum
allowable value (e.g., replace 1 Meg biasing resistors with
47k ohm, etc.).

•

Add small « 100pF) capacitors across feedback resistors
to reduce amplifier gain at high frequencies (Figure 2.2.3).
Caution: this assumes the amplifier is unity-gain stable.
If not, addition of this capacitor will guarantee oscillations. (For amplifiers that are not unity-gain stable, place
a resistor in series with the capacitor such that the gain
does not drop below where it is stable.)

•

is known as excess noise. Excess noise has a 1/f spectral
response, and is proportional to the voltage drop across the
resistor. It is convenient to define a noise index when
referring to excess noise in resistors. The noise index is the
RMS value in fl V of noise in the resistor per volt of DC drop
across the resistor in a decade of frequency. Noise index
expressed in dB is:
Eex
NI = 20 log ( - x 10 6~ dB
VOC

Add a small capacitor (size is a function of source
resistance) at the positive input to reduce the impedance
to high frequencies and effectively shunt them to ground.

where:

Eex = resistor excess noise in flV per frequency
decade.
VOC = DC voltage drop across the resistor.

Excess noise in carbon composition resistors corresponds to
a large noise index of +10dB to -20dB. Carbon film
resistors have a noise index of -10dB to -25dB. Metal film
and wire wound resistors show the least amount of excess
noise, with a noise index figure of -15dB to -40dB. For a
complete discussion of excess noise see Reference 2.

2.3 NOISE
2.3.1 Introduction
The noise performance of IC amplifiers is determined by
four primary noise sources: thermal noise, shot noise, 1/f,
and popcorn noise. These four sources of noise are briefly
discussed. Their contribution to overall noise performance
is represented by equivalent input generators. In addition to
these equivalent input generators, the effects of feedback
and frequency compensation on noise are also examined.
The noise behavior of the differential amplifier is noted
since most op amps today use a differential pair. Finally
noise measurement techniques are presented.

2.3.3 Noise Bandwidth
Noise bandwidth is not the same as the common amplifier
or transfer function -3dB bandwidth. Instead, noise bandwidth has a "brick-wall" filter response. The maximum
power gain of a transfer function T(jw) multiplied by the
noise bandwidth must equal the total noise which passes
through the transfer function. Since the transfer function
power gain is related to the square of its voltage gain we
have:

2.3.2 Thermal Noise
Thermal noise is generated by any passive resistive element.
This noise is "white," meaning it has a constant spectral
density. Thermal noise caJ:LjJe represented by a meansquare voltage generator eR 2 in series with a noiseless
resistor, where eR 2 is given by Equation (2.3.1).

(2.3.2)
where:

eR 2 = 4k TRB (volts)2
where:

T = temperature in oK

B = noise bandwidth in Hz

R = resistor value in ohms
B

TMAX = maximum value of T(jw)
T(jw) = transfer function voltage gain

For a single RC roll-off, the noise bandwidth B .is
71/2 L3dB, and for higher order maximally flat filters, see
Table 2.3.1.

noise bandwidth in Hz

k = Boltzmann's constant (1.38 x 1Q-23W-sec/o K)
The RMS value of Equation (2.3.1) is plotted in Figure
2.3.1 for a one Hz bandwidth. If the bandwidth is
increased, the plot is still valid so long as eR is multiplied by

TABLE 2.3.1 Noise Bandwidth Filter Order

Filter Order

VB.

Noise Bandwidth B

1
2
3
4
"Brick-wall"

1000 _ _

1.57L3dB
1. 11L3dB
1.05L3dB
1.025L3dB
1.00L3dB

2.3.4 Shot Noise
Shot noise is generated by charge crossing a potential
barrier. It is the dominant noise mechanism in transistors
and op amps at medium and high frequencies. The mean
square value of shot noise is given by:
IS2 = 2q IOC B (amps)2
where:

(2.3.3)

q = charge of an electron in coulombs
IOC = direct current in amps

FIGURE 2.3.1 Thermal Noise of Resistor

B = noise bandwidth in Hz
Like thermal noise, shot noise has a constant spectral
density.

Actual resistor noise measurements may have more noise
than shown in Figure 2.3.1. This additional noise component
2-3

2.3.5 l/f Noise
1/f or flicker noise is similar to shot noise and thermal
noise since its amplitude is random. Unlike thermal and
shot noise, llf noise has a llf spectral density. This means
that the noise increases at low frequencies. llf noise is
caused by material and manufacturing imperfections, and is
usually associated with a direct current:
(lDC)a

K - - B (amps)2
f

where:

100

1000

I~

100

~'"

10

~

."

(2.3.4)

IDC = direct current in amps

,'C

~
10

c- '"

r--..

5::
1.0

~I

1.0 L.-'-J..=WL..-'-.u...l.LWl--l....J..J-U.WJ 0.1
100
loOk
10k
10

K and a = constants

FREUUENCY (Hz)

f = frequency in Hz
B

=

noise bandwidth in Hz

FIGURE 2.3.3 Noise Voltage and Current for an Op Amp

2.3.6 Popcorn Noise (peN)
Noise Current, in, or more properly, equivalent open-circuit
RMS noise current, is that noise which occurs apparently at
the input of the noiseless amplifier due only to noise
currents. It is expressed in "picoamps per root Hertz"
(pA/y'HZ) at a specified frequency or in nanoamps in a
given frequency band. It is measured by shunting a capacitor or resistor across the input terminals such that the noise
current will give rise to an additional noise voltage which is
in x Rin (or XCin). The output is measured, divided by
amplifier gain, and that contribution known to be due to
en and resistor noise is appropriately subtracted from the
total measured noise. If a capacitor is used at the input,
there is only en and in XCin. The in is measured with a
bandpass filter and converted to pA/YHZ if appropriate.
Again, note the 1If and shot noise regions of Figure 2.3.3.

Popcorn noise derives its name from the popcorn·like
sound made when connected to a loudspeaker. It is characterized by a sudden change in output DC level, lasting
from microseconds to seconds, recurring randomly.
Although there is no clear explanation of PCN to date, it is
usually reduced by cleaner processing (see Reference 5).
In addition, extensive testing techniques are used to screen
for PCN units.
2.3.7 Modelling
Every element in an amplifier is a potential source of noise.
Each transistor, for instance, shows all three of the above
mentioned noise sources. The net effect is that noise sources
are distributed throughout the amplifier, making analysis
of amplifier noise extremely difficult. Consequently, amplifie( noise is completely specified by a noise voltage and a
noise current generator at the input of a noiseless amplifier.
Such a model is ~hown in Figure 2.3.2. Correlation between
generators is neglected unless otherwise noted.

Now we can examine the relationship between en and in at
the amplifier input. When the signal source is connected,
the en appears in series with the esig and eR. The in flows
through Rs, thus producing another noise voltage of value
in x Rs. This noise voltage is clearly dependent upon the
value of Rs. All of these noise voltages add at the input of
Figure 2.3.2 in RMS fashion, that is, as the square root of
the sum ofthe squares. Thus, neglecting possible correlation
between en and in, the total input noise is:

r------------,
,
INPUT'

1
L.. -

-

-

T
,

'"

I

~,."¢.---I"-""'>---O*

'OUTPUT

V ~~~l~~i:~
GAIN 'A,

I.!i.0!!:!': £!!A.!:!!!El, ___ ~ G!,. __

(2.3.5)

T
,

2.3.8 Effects of Ideal Feedback on Noise

--1

Extensive use of voltage and current feedback are common
in op amps today. Figures 2.3.4a and 2.3.4b can be used
to show the effect of voltage feedback on the noise performance of an op amp.

FIGURE 2.3.2 Noise Characterization of Amplifier

Figure 2.3.4a shows application of negative feedback to an
op amp with generators;;-;;2 and 1;;2. Figure 2.3.4b shows
that the noise generators can be moved outside the feedback
loop. This operation is possible since shorting both amplifiers' inputs results in the same noise voltage at the
outputs. Likewise, opening both inputs gives the same
noise currents at the outputs. For current feedback, the
same result can be found. This is seen in Figure 2.3.5a and
Figure 2.3.5b.

Noise voltage en, or more properly, equivalent short-circuit
input RMS noise voltage, is simply that noise voltage which
would appear to originate at the input of the noiseless
amplifier if the input terminals were shorted. It is expressed
in "nanovolts per root Hertz" (n V1y'HZ) at a specified
frequency, or in microvolts for a given frequency band.
It is measured by shorting the input terminals, measuring
the output RMS noise, dividing by amplifier gain, and
referencing to the input - hence the term "equivalent input
noise voltage." An output bandpass filter of known characteristic is used in measurements, and the measured value is
divided by the square root of the bandwidth if data are to
be expressed per unit bandwidth.

The significance of the above result is that the equivalent
input noise generators completely specify ci rcuit noise.
The application of ideal negative feedback does not alter
the noise performance of the circuit. Feedback reduces the
output noise, but it also reduces the output signal. In other
words, with ideal feedback, the equivalent input noise is
independent of gain.

Figure 2.3.3 shows en of a typical op amp. For this amplifier, the region above 1 kHz is the shot noise region, and
below 1 kHz is the amplifier's llf region.

2-4

(a) Feedback Applied to Op Amp with Noise Generators

(b) Noise Generators Outside Feedback Loop

FIGURE 2.3.4

la) Current Feedback Applied to Op Amp

Ib) Noise Generators Moved Outside Feedback Loop
FIGURE 2.3.5

la) Practical Voltage Feedback Amplifier

(b) Voltage Feedback with Noise Generators Moved
Outside Feedback Loop
FIGURE 2.3.6

2.3.9 Effects of Practical Feedback on Noise
1. Thermal noise from Rs + R111R2 "" 2k is 5.65nV/y'HZ.

Voltage feedback is implemented by series·shunt feedback
as shown in Figure 2.3.6a.

2. Read en from Figure 2.3.3 at 1 kHz; this value is
9.5nV/y'HZ.

The noise generators can be moved outside the feedback
loop as shown in Figure 2.3.6b if the thermal noise of
R111R2 is included in e2 2 . In addition, the noise generated
by in x (R 111 R2) must be added even though the (-) input
is a virtual ground (see Appendix 6). The above effects can
be easily included if R111R2 is considered to be in series
with Rs.

3. Read in from Figure 2.3.3 at 1 kHz; this value is
0.68pA/y'HZ. Multiply this noise current by Rs + R111R2
to obtain 1.36nV /y'HZ.
4. Square each term and enter into Equation (2.3.5).

.je2 2 + i22 (Rs + R111R2)2 nV/y'HZ

e2 2 = en 2 + 4k T (Rs + R111R2)

.jen 2 + 4 k T (R s + R1!1R2) + in 2 (Rs + R111R2)2

i22 = in 2

.j(9.5)2 + (5.65)2 + (1.36)2
Example 2.3.1
eN

Determine the total equivalent input noise per unit band·
width for the amplifier of Figure 2.3.6a operating at 1 kHz
from a source resistance of 1 kn. R1 and R2 are 100 kn and
1 kn respectively.

11.1 nV/y'HZ

This is total RMS noise at the input in one Hertz band·
width at 1 kHz. If total noise in a given bandwidth is
desired, one must integrate the noise over a bandwidth as
specified. This is most easily done in a noise measurement
set-up, but may be approximated as follows:

Solution:
Use data from Figure 2.3.1 and Figure 2.3.3.
2-5

1. If the frequency range of interest is in the flat band, i.e.,
between 1 kHz and 10kHz in Figure 2.3.3, it is simply a
matter of multiplying eN by the square root of the
noise bandwidth. Then, in the 1 kHz·10kHz band, total
noise is:

First, move the noise generators outside feedback R1. To
do this, represent the thermal noise generated by R 1 as a
noise current source (Figure 2.3.7b):

so:
1.05J,lV

2. If the frequency band of interest is not in the flat band
of Figure 2.3.3, one must break the band into sections,
calculating average noise in each section, squaring,
multiplying by section bandwidth, summing all sections,
and finally taking square root of the sum as follows:
eN =

JiiR 2 B + ~ (ej\j2 + 1;;2 Rs2)i Bi

where:

1

Now move these noise generators outside Rs + R2 as shown
in Figure 2.3.7c to obtain ii2 2 and 122:
~2 = ;;;;2 +4k T (R s + R2) B

(2.3.6)

i2 2

= 1;;2 + 4k T

(2.3.7)

-.!. B

(2.3.8)

R1

i is the total number of sub·blocks

e22 and f22 are the equivalent input generators with feed·
back applied. The total equivalent input noise, eN, is the

For details and examples of this type of calculation, see
application note AN·104, "Noise Specs Confusing?"

sum of the noise produced with the input shorted, and the
noise produced with the input opened. With the input of
Figure 2.3.7c shorted, the input referred noise is e2 2 . With
the input opened, the input referred noise is:

Current feedback is accomplished by shunt·shunt feedback
as shown in Figure 2.3.7a.

RS

The total equivalent input noise is:
esig

(a) Practical Current Feedback Amplifier

Example 2.3.2
Determine the total equivalent input noise per unit band·
width for the amp of Figure 2.3.7a operating at 1 kHz from
a 1 kQ source. Assume R1 is 100kQ and R2 is 9kQ.
Solution
Use

d~ta

from Figures 2.3.1 and 2.3.3.

1. Thermal noise from Rs + R2 is 12.7 nV IVHz.
2. Read en from figure 2.3.3 at 1 kHz; this value is
9.5nV/y'RZ. Enter these values into Equation (2.3.7).
(b) Intermediate Move of Noise Generators

3. Determine the thermal noise current contributed by R1:
1.61 x 10- 20
lOOk

= 0.401 pA/y'HZ

4. Read in from Figure 2.3.3 at 1 kHz; this value is
O.68pA/y'RZ. Enter these values into Equation (2.3.7).

(c) Current Feedback with Noise Generators Moved Outside

Feedback Loop
FIGURE 2.3.7

eN = 17.7nV/VHz
For the noise in the bandwidth from 1 kHz to 10kHz,
eN = 17.7nV.J9000 = 1.68J,lV. If the noise is not constant
with frequency, the method shown in Equation (2.3.6)
should be used.

ii;;2 and 1;;2 can be moved outside the feedback loop if the
noise generated by R1 and R2 are taken into account.
2·6

TABLE 2.3.2 Equivalent Input Noise Comparison

NON-INVERTING AMPLIFIER
R2

INVERTING AMPLIFIER

eN (nV-yfHz)

AV

Rs

R1

eN (nVyHz)

AV

Rs

R1

R2

101

1k

100k

1k

11.1

100

1k

100k

0

11

lk

lOOk

10k

17.3

10

lk

100k

9k

17.7

2

lk

lOOk

lOOk

46.0

2

lk

lOOk

49k

49.5

1

lk

100k

00

80.2

1

lk

100k

99k

89.1

10.3

Example 2.3.3
Compare the noise performance of the non·inverting
amplifier of Figure 2.3.6a to the inverting amplifier of
Figure 2.3.7a.

The undesirable consequence of a single-pole roll-off, wideband design is the excess gain beyond audio frequencies,
which includes the AM band; hence, noise of this frequency
is amplified and delivered to the load where it can radiate
back to the AM (magnetic) antenna and sensitive RF
circuits. A simple and economical remedy is shown in
Figure 2.3.8c, where a ferrite bead, or small R F choke is
added in series with the output lead. Experiments have
demonstrated that this is an effective method in suppressing
the unwanted RF signals.

Solution:
The best way to proceed here is to make a table and
compare the noise performance with various gains.
Table 2.3.2 shows only a small difference in equivalent
input noise for the two amplifiers. There is, however, a
large difference in the flexibility of the two amplifiers.
The gain of the inverting amplifier is a function of its input
resistance, R2. Thus, for a given gain and input resistance,
R 1 is fixed. This is not the case for the non·inverting
amplifier. The designer is free to pick Rl and R2 independent of the amplifier's input impedance. Thus in the
case of unity gain, where R2 = 00, R 1 can be zero ohms.
The equivalent input noise is:

10.3nV/-yfHz
There is now a large difference in the noise performance of
the two amplifiers. Table 2.3.2 also shows that the equivalent input noise for practical feedback can change as a
function of closed loop gain A V. This result is somewhat
different from the case of ideal feedback.

(a) Typical Compensation

Example 2.3.4
Determine the signal-to-noise ratio for the amplifier of
Example 2.3.2 if eSIG has a nominal value of 100mV.
60dB

Solution:
Signal to noise ratio is defined as:

SIN = 20 log eSIG

(2.3.9)

10kHz

eN

AM

10M

BAND

20 log 100mV
1.68/LV

95.5dB

(b) Source of RF Interference

2.3.10 R F Precautions
A source of potential RF interference that needs to be
considered in AM radio applications lies in the radiated
wideband noise voltage developed at the speaker terminals.
The method of amplifier compensation (Figure 2.3.8a)
fixes the point of unity gain cross at approximately 10MHz
(Figure 2.3.8b). A wideband design is essential in achieving
low distortion performance at high audio frequencies, since
it allows adequate loop-gain to reduce THO. (Figure 2.3.8b
shows that for a closed-loop gain of 34dB there still exists
26dB of loop-gain at 10kHz.)

(c) Reduction of RF Interference

FIGURE 2.3.8

2-7

In order to find the input noise current generator, in, open
the input and equate the output noise from Figure 2.3.9a
and Figure 2.3.9b. The result of this operation is in : inl.
Thus, from a high impedance source, the differential pair
gives similar noise current as a single transistor.

2.3.11 Noise in the Differential Pair
Figure 2.3.9a shows a differential amplifier with noise
generators en 1, in 1, en2, and in2.

2.3.12 Noise Measurement Techniques
This section presents techniques for measuring en, in, and
eN. The method can be used to determine the spectral
density of noise, or the noise in a given bandwidth. The
circuit for measuring the noise of an LM387 is shown in
Figure 2.3.10.

INPUT

The system gain, VOUT/en, of the circuit in Figure 2.3.10
is large - 80dB. This large gain is required since we are
trying to measure input referred noise generators on the
order of 5nV/y'RZ, which corresponds to 50/N/y'Hz at
the output. Rl and R2 form a 100: 1 attenuator to provide
a low input signal for measuring the system gain. The gain
should be measured in both the en and in positions, since
LM387 has a 250k bias resistor which is between input and
ground. The LM387 of Figure 2.3.10 has a closed loop gain
of 40dB which is set by feedback elements R5 and R6.
40dB provides adequate gain for the input referred
generators of the LM387. The output noise of the LM387 is
large compared to the input referred generators of the
LM381; consequently, noise at the output of the LM381
will be due to the LM387. To measure the noise voltage
en, and noise current in x R3, a wave analyzer or noise filter
set is connected. In addition the noise in a given bandwidth
can be measured by using a bandpass filter and an RMS
voltmeter. If a true RMS voltmeter is not available, an
average responding meter works well. When using an
average responding meter, the measured noise must be
multiplied by 1.13 since the meter is calibrated to measure
RMS sine waves. The meter used for measuring noise should
have a crest factor (ratio of peak to RMS value) from 3 to
5, as the peak to RMS ratio of noise is on that order. Thus,
if an average responding meter measures 1 mV of noise, the
RMS value would be 1.13mVRMS, and the peak-to·peak
value observed on an oscilloscope could be as high as
11.3mV (1.13mV x 2 x 5).

(a) Differential Pair with Noise Generators

INPUT

(bl Differential Pair with Generators Input Referred

FIGURE 2.3.9

To see the intrinsic noise of the pair, short the base of T2
to ground, and refer the four generators to an input noise
voltage and noise current as shown in Figure 2.3.9b. To
determine en, short the input of 9(a) and 9(b) to ground.
en is then the series combination of enl and en2' These add
in an RMS fashion, so:

Some construction tips for the circuit of Figure 2.3.10 are
as follows:
1. R4 and R6 should be metal film resistors, as they
exhibit lower excess noise than carbon film resistors.

Both generators contribute the same noise, since the
transistors are similar and operate at the same current;
thus, en : ~, i.e., 3dB more noise than a single ended
amplifier. This can be significant in critical noise applica·
tions (see Section 2.71.

2. Cl should be large, to provide low capacitive reactance
at low frequency, in order to accurately observe the 1/f
noise in en.

lOOn

lOOn
r-~./\I'.;y..----o+l0V

50pF

+ ~VOUT
200n ~

20k

lOOk
51k

lk

+

Il00.uF

-=FIGURE 2.3.10 Noise Test Setup for Measuring en and in of an lM387

2-8

WAVE ANALYZER
OR FILTER SET

3. C2 should be large to maintain the gain of 80dB down
to low frequencies for accurate 1/f measurements.

The equivalent input noise is:

4. The circuit should be built in a small grounded metal
box to eliminate hum and noise pick·up, especially in in.

VOUT

= 0.18mV =

AV

100

5. The LM387 and LM381 should be separated by a metal
divider within the metal box. This is to prevent output
to input oscillations.

1.81.N in a 20kHz bandwidth.

If this preamp had an NAB or RIAA playback equalization,
the output noise, VOUT, would have been divided by the
gain at 1 kHz.

Typical LM387 noise voltage and noise current are plotted
in Figure 2.3.11.

Typical values of noise, measured by the technique of
Figure 2.3.12, are shown in Table 2.3.3. For this data,
B = 10kHz and Rs = 600ri.

10.0

100

TABLE 2.3.3 Typical Flat Band Equivalent Input Noise

~

I£'

in

~ 10.0

:sIi

1.0

=

'"

I:::

Type

eN (IlV)

-a

LM381
LM381A
LM382
LM387
LM387A

0.70
0.50
0.80
0.80
0.65

~

.-

1.0 L-L..J...l..llJJlL....LJ..l.l.illJl-l'-'..WWJ 0.1
10

100

1k

10k

FREOUENCY (Hz)

REFERENCES
1. Meyer, R. G., "Notes on Noise," EECS Department,
University of California, Berkeley, 1973.

FIGURE 2.3.11 LM387 Noise Voltage and Noise Current

2. Fitchen, F. C., Low Noise Electronic Design, John Wiley
& Sons, New York, 1973.

Many times we do not care about the actual spectral dis·
tribution of noise, rather we want to know the noise
voltage in a given bandwidth for comparison purposes. For
audio frequencies, we are interested only in a 20 kHz band·
width. The noise voltage is often the dominant noise source
since many systems use a low impedance voltage drive as
the signal. For this common case we use a test set·up as
shown in Figure 2.3.12.

INPUrT

3. Cherry, E. M. and Hooper, D. E., Amplifying Devices and
Low Pass Amplifier Design, John Wiley & Sons, New
York, 1968.
4. Sherwin, J., Noise Specs Confusing?, Application Note
AN·104, National Semiconductor, 1975.
5. Roedel, R., "Reduction of Popcorn Noise in Integrated
Circuits," IEEE Trans. Electron Devices (Corresp.), vol.
ED·22, October 1975, pp. 962·964.

~
~

'------'

FIGURE 2.3.12 Test Setup for Measuring Equivalent Input Noise

for a 20 kHz Bandwidth

Example 2.3.5
Determine the equivalent input noise voltage for the preamp
of Figure 2.3.12. The gain, AV, of the preamp is 40dB and
the voltmeter reads 0.2mV. Assume the voltmeter is
average responding and the 20 kHz low-pass filter has a
single R·C roll·off.
Solution:
Since the voltmeter is average responding, the RMS voltage
is VRMS = 0.2mV x 1.13 = 0.226mV. Using an average reo
sponding meter causes only a 13% error. The filter has a
single R·C roll·off, so the noise bandwidth is 71/2 x 20kHz
= 31.4 kHz, i.e., the true noise bandwidth is 31.4 kHz and
not 20kHz. Since RMS noise is related to the square root
of the noise bandwidth, we can correct for this difference:
VOUT

=

jO.226
-71/2

= 0.18mV
2·9

2.4 AUDIO RECTIFICATION
Or, "How Come My Phono Detects AM?"
Audio rectification refers to the phenomenon of R F signals
being picked up, rectified, and amplified by audio circuits
- notably by high-gain preamplifiers_ Of all types of interference possible to plague a hi-fi system, audio rectification
remains the most slippery and troublesome_ A common
occurrence of audio rectification is to turn on a phonograph
and discover you are listening to your local AM radio
station instead. There exist four main sources of interference, each with a unique character: If it is clearly audible
through the speaker then AM radio stations are probably
the source; if the interference is audible but garbled then
suspect SSB and amateur radio equipment; a decrease in
volume can be produced by FM pickup; and if buzzing
occurs, then RADAR or TV is being received. Whatever the
source, the approaches to eliminating it are similar.

/

RF CHOKE OR FERRITE BEAD l-l0pH)

0-..fYYY"l- .....- -......~

,...

..L

..L

-,-

OUTPUT

CERAMiY ",--""",,,"""-""
l-l0·300,F)

FIGURE 2.4.1 Audio Rectification Elimination Tips

Commonly, the rectification occurs at the first non-linear,
high gain, wide bandwidth transistor encountered by the
incoming signal. The signal may travel in unshielded or
improperly grounded input cables; it may be picked up
through the air by long, poorly routed wires; or it may
enter on the AC power lines. It is rectified by the first
stage transistor acting as a detector diode, subsequently
amplified by the remaining circuitry, and finally delivered
to the speaker. Bad solder joints can defect the R F just as
well as transistors and must be avoided (or suspected).

A particularly successful technique is uniquely possible with
the LM381 since both base and emitter points of the input
transistor are available. A ceramic capacitor is mounted
very close to the IC from pin 1 to pin 3, shorting base to
emitter ait RF frequencies (see Figure 2.4.2).

The following list should be consulted when seeking to
eliminate audio rectification from existing equipment. For
new designs, keep input leads short and shielded, with the
shield grounded only at one point; make good clean solder
connections; avoid loops created by multiple ground points;
and make ground connections close to the IC or transistor
that they associate with.

RF CHOKE OR FERRITE BEAD l-l0pH)
/IONLY IF NECESSARY)

OUTPUT

Audio Rectification Elimination Tips (Figure 2.4.1).
•

Reduce input impedance.

•

Place capacitor to ground close to input pin or base
(~ 10-300pF).

•

Use ceramic capacitors.

•

Put ferrite bead on input lead close to the device input.

•

Use RF choke in series with input

•

Use R F choke (or ferrite bead) and capacitor to ground.

•

Pray.

I
(~

10MH)_
FIGURE 2.4.2 LM381 Audio Rectification Correction

2-10

2.5 DUAL PREAMPLIFIER SELECTION
National Semiconductor's line of integrated circuits designed
specifically to be used as audio preamplifiers consists of the
LM381, LM382, LM387, and the LM1303. All are dual
amplifiers in recognition of their major use in two channel
applications. In addition there exists the LM389 which has
three discrete NPN transistors that can be configured into

a low noise monaural preamplifier for minimum parts count
mono systems (Section 4.11). Table 2.5.1 shows the major
electrical characteristics of each of the dual preamps offered.
A detailed description of each amplifier follows, where the
individual traits and operating requirements are presented.

TABLE 2.5.1 Dual Preamplifier Characteristics

LM381N
(14 Pin DIP)

PARAMETER
MIN
Supply Voltage

TYP

9

Quiescent Supply Current

LM382N
(14 Pin DIP)
MAX

40

MIN

TYP

LM387N
(8 Pin DIP)
MAX

9

40

MIN

TYP

9

16

LM1303N
(14 Pin DIP)
MAX
30'

MIN

TYP

±4.5

10

UNITS
MAX

±15

V

15

rnA

10

10

100k
200k

100k
200k

Open Loop Gain

104

100

104

76

80

dBV

Output Voltage Swing
RL" 10k~

V, - 2

V, - 2

V, - 2

11.3

15.6

V p.p

S'
2

S'
2

8'

0.6
0.6

O.S
0.8

150

150

150

4k
5.0'

V Ill'

100

kHz
kHz

Input Resistance (open loop)

Positive Input
Negative Input

50k

~

25k
25k

lOOk
200k

~

Output Current

Source
Sink

2

rnA
rnA

Output Resistance
(open loop)
Slew Rate
(Av = 40dB)

4.7

4.7

4.7

Power Bandwidth
20V p .p (V, = 24V)
11.3V p. p (V,=±13V)

75

75

75

Unity Gain Bandwidth

15

Input Voltage
Positive Input

15

20

15

300

300

±5

Supply Rejection Ratio
(Input Referred. 1 kHz)

120

Channel Separation
(f = 1 kHz)

60

Total Harmonic Distortion
(f" 1 kHz)'

0.1

Total Equivalent Input Noise
600~,

MHz

300

Either Input

(R, =

~

10·10k Hz)

Total NAB 8 Output Noise
(R, = 600~, 10·10k Hz)

0.54
0.5 4 .,

120
40

1.04
0.7 4 • 5

dBV

110
40

60

60

60

0.1

0.3

0.1

0.5

O.S

1.2

0.8
0.65 6

1.2
0.9'

230
180'

190
140'

1. Specifications apply for T A'" 25°C with Vs '" +14V for LM381/382/387 and Vs '" ± 13V for LM1303, unless otherwise noted.
2. DC current; symmetrical AC current = 2mA p .p .
3. l-M381 & LM387: Gain .. 60dS; LM382: Gain == 60dS; LM1303: Gain'" 40dB.
4. Single ended input biasing.
5. LM381AN.
6. 40V for LM387AN.
7. Frequency Compensation: C '" 0.0047.uF, Pins 3 to 4.
B. NAB reference level: 37dBV Gain at 1 kHz. Tape Playback Circuit.

2·11

rnVRMS
V

70
0.1

dBV

%
IlVRMS
IlVRMS
IlVRMS
IlVRMS

2.6 LM381 LOW NOISE DUAL PREAMPLIFIER
R,
200K

2.6.1 Introduction
The LM381 is a dual preamplifier expressly designed to
meet the requirements of amplifying low level signals in
low noise applications. Total equivalent input noise is
typically O.5,uVRMS (R s = 600rl, 10·10,OOOHzl.

+

Each of the two amplifiers is completely independent, with
an internal power supply decoupler·regulator, providing
120dB supply rejection and 60dB channel separation.
Other outstanding features include high gain (112dBI, large
output voltage swing (VCC - 2VI p.p, and wide power
bandwidth (75kHz, 20V p. p l. The LM381 operates from a
single supply across the wide range of 9 to 40 V. The
amplifier is internally compensated and short·circuit
protected.

•

~

Q3, Q4 provides level shifting and current gain to the
common-emitter stage (Q51 and the output current sink
(Q71. The voltage gain of the second stage is approximately
2,000, making the total gain of the amplifier typically
160,000 in the differential input configuration.
The preamplifier is internally compensated with the polesplitting capacitor, Cl. This compensates to unity gain at
15 MHz. The compensation is adequate to preserve stability
to a closed loop gain of 10. Compensation for unity gain
closure may be provided with the addition of an external
capacitor in parallel with Cl between pins 5 and 6, 10 and 11.

With the low output level of magnetic tape heads and
phonograph cartridges, amplifier noise becomes critical in
achieving an acceptable signal·to·noise ratio. This is a major
deficiency of the op amp in this application. Other inade·
quacies of the op amp are insufficient power supply
rejection, limited small'signal and power bandwidths, and
excessive external components.

Three basic compensation schemes are possible for this
amplifier: first stage pole, second stage pole and polesplitting. First stage compensation will cause an increase in
high frequency noise because the first stage gain is reduced,
allowing the second stage to contribute noise. Second stage
compensation causes poor slew rate (power bandwidth I
because the capacitor must swing the full output voltage.
Pole-splitting overcomes both these deficiencies and has the
advantage that a small monolithic compensation capacitor
can be used.

2.6.2 Circuit Description
To achieve low noise performance, special consideration
must be taken in the design of the input stage. First, the
input should be capable of being operated single ended,
since both transistors contribute noise in a differential stage
degrading input noise by the factor
(See Section 2.3.1
Secondly, both the load and biasing elements must be
resistive, since active components would each contribute as
much noise as the input device.

V2'.

The output stage is a Darlington emitter-follower (08, Q91
with an active current sink (Q71. Transistor 010 provides
short-circuit protection by limiting the output to 12mA.

The basic input stage, Figure 2.6.1, can operate as a differen·
tial or single ended amplifier. For optimum noise perfor·
mance 02 is turned OFF and feedback is brought to the
emitter of Ql.

The biasing reference is a zener diode (Z21 driven from a
constant current source (Ql11. Supply decoupling is the
ratio of the current source impedance to the zener impedance. To achieve the high current source impedance
necessary for 120dB supply rejection, a cascade configuration is used (Ql1 and 0121. The reference voltage is used
to power the first stages of the amplifier through emitterfollowers Q14 and Q15. Resistor Rl and zener Zl provide
the starting mechanism for the regulator. After starting,
zero volts appears across 01, taking it out of conduction.

In applications where noise is less critical, Ql and Q2 can be
used in the differential configuration. This has the advantage
of higher impedance at the feedback summing point,
allowing the use of larger resistors and smaller capacitors in
the tone control and equalization networks.
The voltage gain of the single ended input stage is given by:

where:

=

RL
re

re =

=

~

200k _ 160
1.25k

(2.6.11

2.6_3 Biasing
Figure 2.6.3 shows an AC equivalent circuit of the LM381.
The non-inverting input, Ql, is referenced to a voltage
source two VBE above ground. The output quiescent point
is established by negative DC feedback through the external
divider R4/R5 (Figure 2.6.41,

"" 1.25 x 10 3 at 25°C, IE"" 20,uA

qlE

The voltage gain of the differential input stage is:
1 RL q IE
1 RL
AV = - - = - - - - "" 80
2 re
2
KT

RT
10K

FIGURE 2_6.1 Input Stage

Attempts have been made to fill this function with selected
operational amplifiers. However, due to the many special
requirements of this application, these recharacterizations
have not adequately met the need.

AV(ACI

U2

Ul

For bias stability, the current through R5 is made ten times
the input current of 02 ("" 0.5,uAI. Then, for the differential
input, resistors R5 and R4 are:

(2.6.21

The schematic diagram of the LM381, Figure 2.6.2, is
divided into separate groups by function - first and second
voltage gain stages, third current gain stage, and the bias
regulator.

2VBE
1.3
= 260krl maximum
R5 = - - =
10lQ2
5 x 10- 6

The second stage is a common-emitter amplifier (Q51 with a
current source load (Q61. The Darlington emitter-follower

VCC
R4 = ( - -1
2.6

~

2-12

R5

(2.6.31

(2;6.41

Vee

----I
I
I

1-----I
II Rl

I
I
I
I
I
I
I
I
I

I

01

I
I

I
I

' - - - - -.....-+-0(7,81

ZI

I

I

R,

I
_ _ _ _ _ .L _ _ _ _ -.J

10k

I

L___ ~ __ _
(41

FIGURE 2.6.2 Schematic Diagram
Vee

R,

R2

200K

R3
Z2

+o--r-+--i

02

R,
10K

FIGURE 2.6.3 AC Equivalent Circuit

R4

Z2

+0--+---411---1

1--o---+--I--.,.3V

R5

FIGURE 2.6.4 Differential Input Biasing

2-13

gain now approaches open loop. The low frequency 3dB
corner, fo, is given by:

Vee

(2.6.7)
where:

Ao = open loop gain

2.6.4 Split Supply Operation
Although designed for single supply operation, the LM381
may be operated from split supplies just as well. (A tradeoff exists when unregulated negative supplies are used since
the inputs are biased to the negative rail without supply
rejection techniques and hum may be introduced.) All that
is necessary is to apply the negative supply (VEE) to the
ground pin and return the biasing resistor R5 to VEE
instead of ground. Equations (2.6.3) and (2.6.5) still hold,
while the only change in Equations (2.6.4) and (2.6.6) is to
recognize that VCC represents the total potential across the
LM381 and equals the absolute sum of the split supplies
used, e.g., VCC = 30 volts for ±15 volt supplies. Figure
2.6.7 shows a typical split supply application; both differential and single ended input biasing are shown. (Note that
while the output DC voltage will be approximately zero
volts the positive input DC potential is about 1.3 volts
above the negative supply, necessitating capacitive coupling
into the input.)

R4

Z2

+0--1-....-1

RS

FIGURE 2.6.5 Single Ended Input Biasing

When using the single ended input, 02 is turned OFF and
DC feedback is brought to the emitter of 01 (Figure 2.6.5).
The impedance of the feedback summing point is now two
orders of magnitude lower than the base of 02 ("" 10kn).
Therefore, to preserve bias stability, the impedance of the
feedback network must be decreased. I n keeping with
~easonable resistance values, the impedance of the feedback
voltage source can be 1/5 the summing point impedance.

(1.81

The feedback current is < 100ilA worst case. Therefore,
for single ended input, resistors R5 and R4 are:
VBE

0.65

51FB

5 x 10-4

= 1300n maximum

~ R5

RS

(2.6.5)
VEE

R4 = ( VCC
--1
1.3

(2.6.6)
Differential Input Biasing

17.81

R4

l'

RS

C2

-=
Single Ended Input Biasing

FIGURE 2.6.6 AC Open Loop
vooe~OVOlTS

VINoe ~ VEE + 1.2 VOLTS

The circuits of Figures 2.6.4 and 2.6.5 have an AC and DC
gain equal to the ratio R4/R5. To open the AC gain,
capacitor C2 is used to shunt R5 (Figure 2.6.6). The AC

FIGURE 2.6.7 Split Supply Operation

2-14

2.6.5 Non·lnverting AC Amplifier

Since the LM381 is a high gain amplifier, proper power
supply decoupling is required. For most applications a
0.1 jJ.F ceramic capacitor (Cs ) with short leads and located
close (within one inch) to the integrated circuit is sufficient.
When used non·inverting, the maximum input voltage of
300mVRMS (850mV p.p ) must be observed to maintain
linear operation and avoid excessive distortion. Such is not
the case when used inverting.

Perhaps the most common application of the LM381 is as a
flat gain, non·inverting AC amplifier operating from a single
supply. Such a configuration is shown in Figure 2.6.8.
Resistors R4 and R5 provide the necessary biasing and
establish the DC gain, AVDC, per Equation (2.6.8).
(2.6.8)

2.6.6 Inverting AC Amplifier
The inverting configuration (2.6.9) is very useful since it
retains the excellent low noise characteristics without the
limit on input voltage and has the additional advantage of
being inherently unity gain stable. This is achieved by the
voltage divider action of R6 and R5 on the input voltage.
For normal values of R4 and R5 (with typical supply
voltages) the gain of the amplifier itself, i.e., the voltage
gain relative to pins 2 or 13 rather than the input, is always
around ten - which is stable. (See Section 2.8.7 for details.)
The real importance is that while the addition of C3 will
guarantee unity gain stability (and roll·off high frequencies),
it does so at the expense of slew rate.

AC gain is set by resistor R6 with low frequency roll·off at
fa being determined by capacitor C2.
(2.6.9)

(2.6.10)

vs
Vs

I

-.L

FIGURE 2.6.8 Non·inverting AC Amplifier
FIGURE 2.6.9 Inverting AC Amplifier

The small'signal bandwidth of the LM381 is nominally
20MHz, making the preamp suitable for wide·band instru·
mentation applications. However, in narrow·band applica·
tions it is desirable to limit the amplifier bandwidth and
thus eliminate high frequency noise. Capacitor C3 accom·
plishes this by shunting the internal pole·splitting capacitor
(Cll. limiting the bandwidth of the amplifier. Thus, the
high frequency -3dB corner is set by C3 according to
Equation (2.6.11).
C3 =
where:

1
_ 4 x 10-12
2rrf3reAVAC

Using Figure 2.6.9 without C3 at any gain retains the full
slew rate of 4.7V/jJ.s. The new gain equations follow:

(2.6.11 )

AVDC

=

AVAC

=

R4

(2.6.13)

R5
R4

(2.6.14)

R6

Capacitor C2 is still found from Equation (2.6.10), and Cc
and Cs are as before. Capacitor CB is added to provide AC
decoupling of the positive input and can be made equal to
0.1 jJ.F. Observe that pins 3 and 12 are not used, since the
inverting configuration is not normally used with single
ended input biasing techniques.

f3 = high frequency -3dB corner
re = first stage small·signal emitter resistance
'" 1.3kQ
AVAC = mid·band gain in V/V

Capacitor Co acts as an input AC coupling capacitor to
block DC potentials in both directions and can equal 0.1 jJ.F
(or larger). Output coupling capacitor Cc is determined by
the load resistance and low frequency corner f 0 per
Equation (2.6.12).

2.7 LM381A DUAL PREAMPLIFIER FOR ULTRA·LOW
NOISE APPLICATIONS
2.7.1 Introduction
The LM381 A is a dual preamplifier expressly designed to
meet the requirements of amplifying low level signals in
noise critical applications. Such applications include hydro·

(2.6.12)

2·15

phones, scientific and instrumentation recorders, low level
wideband gain blocks, tape recorders, studio sound equipment, etc_

(2.7.1)
where:

The LM381 A can be externally biased for optimum noise
performance in ultra-low noise applications. When this is
done the LM381 A provides a wideband, high gain amplifier
with noise performance that exceeds that of today's best
transistors.

en; amplifier noise voltage/YHZ
in ; amplifier noise current/YHZ
Rs ; source resistance in n
k ; Boltzmann's constant ; 1.38 x 10- 23 Jt K
T ; source resistance temperature in 0 K

The amplifier can be operated in either the differential or
single ended input configuration. However, for optimum
noise performance, the input must be operated single ended,
since both transistors contribute noise in a differential stage,
degrading input noise by the factor
(See Section 2.3.)
A second consideration is the design of the input bias
circuitry_ Both the load and biasing elements must be
resistive, since active components would each contribute
additional noise equal to that of the input device_ Thirdly,
the current density of the input device should be optimized
for the source resistance of the input transducer.

B.W. ; noise bandwidth
Figure 2.7.3 shows a plot of input transistor (01) collector
current versus source resistance for optimum noise performance of the LM381 A. For source impedances less than
3 kn the noise voltage term (en) dominates and the input
is biased at 170pA, which is optimum for noise voltage. In
the region between 3kn and 15kn, both the en and in Rs
terms contribute and the input should be biased as indicated
by Figure 2.7.3. Above 15kn, the in Rs term is dominant
and the amplifier is operated without additional external
biasing.

V2:

2_7_2 Optimizing Input Current Density
Figures 2.7.1 and 2.7.2 show the wide-band (10Hz-10kHz)
input noise voltage and input noise current versus collector
current for the single ended input configuration of the
LM381A. Total input noise of the amplifier is found by:

200
180

~.::.S~~NANT

160

t

I_I

1\

I.

~

I

"'- .......

120
100
80

10 Hz -10 kHz_
As'" 0

1'.

140

60
40
20

.....

NORMAL SING LE

I\t.n Rs DOMINANT
I IIIII

ENDED BIAS LEVEL
3k

5k

10k

20k

40k

lOOk

Rs(n)-.-

o
20

60

100

140

180

FIGURE 2.7.3 Collector Current vs Source Resistance for
Optimum Noise Performance

220

IctuA)

Figure 2.7.4 shows the input stage of the LM381A with the
external components added to increase the current density
of transistor 01. Resistors Rl and R2 supply the additional
current (12) to the existing collector current (Ill. which
is approximately 18pA.

FIGURE 2_7_1 Wideband Equivalent Input Noise Voltage
vs Collector Current

The sum of resistors Rl and R2 is given by:
(Rl + R2) ;
1.2
(10 Hz -10 kHz)

Vs - 2.1

(2.7_2)

Ic -18xlO- 6

1.0

~

!

l/

0.8

......

0.6
0.4

L,..f..-

For DC considerations, only the sum (Rl + R2) is
important. When considering the AC effects, however, the
values of Rl and R2 become significant.

f..-f..-

f..- ......

Since resistors R 1 and R2 are biased from the power supply,
the decoupling capacitor, Cl, is required to preserve supply
rejection. The value of Cl is given by:

0.2

20

60

100

140

180

lOP.S. R./20
Cl; - - - 21ffs Rl Al

220

Ie (PAl

where:

P.S.R_ ; supply rejection in dB referred to input
fs ; frequency of supply ripple

FIGURE 2_7_2 Wideband Equivalent Input Noise Current

Al ; voltage gain of first stage

vs Collector Current

2-16

(2.7_3)

For DC stability let: (For production use, R3 is made equal
to a 2.5 kQ trimpot, allowing process variations while
preservi ng output DC level. I
R1

R3 = 1 kQ nominal

(2.7.81

Rf can then be found from:

t---...

I "
I

I

T Cl

R1

"

1[
Rf =

(5111

"2

(2.7.91

V s = supply voltage

where:

+5.6V

J

Vs x 10 7
VE (1.1 x 1041- lc x 107

I

Ic = 01 collector current

I', ~

The AC closed loop gain is set by the ratio:

I

(7,81

(2.7.101

1 1.,V

(1,141

I
I

/'

I
I

-/
1---//
(3,

/

/

/'

/'

/

/

/'

Capacitor C2 sets the low frequency 3dB corner where:
f

R,

_

o - 2

IT

1
C2 R4

(2.7.111

_-=-_ ______"_'_'__-.

11, ...

R3

R4

Rl

R1

FIGURE 2.7.4 LM381 A with Biasing Components for Increasing
01 Current Density
V1N

As R1 becomes smaller capacitor C1 increases for a given
power supply rejection ratio. Conversely, as R2 becomes
smaller the gain of the input stage decreases, adversely
affecting noise performance. For the range of collector
currents over which the LM381 A is operating, a reasonable
compromise is obtained with:

f

'l'Cl
4

>-(7_,8_,

'\....

BIAS
ADJ

_-0

Vo

(3111

R,

R3

R4

(2.7.41

'l'C1

The gain of the input stage is:

FIGURE 2.7.5 Single Ended Input Configuration with External

Biasing Components

(2.7.51

Figure 2.7.5 shows the LM381A in the single ended input
configuration with the additional biasing components.
Capacitor C3 may be added to limit the amplifier band·
width to the frequency range of interest, thus eliminating
excess noise outside the pertinent bandwidth.

Adding current to 01 increases the base current flowing
through the 250k bias resistor. This voltage drop affects 01
emitter voltage VE as follows:
VE = 0.8 -

(~
x 2500
130
)

(2.7.61
(2.7.121

Resistor divider Rf/R3 provides negative DC feedback
around the amplifier establishing the quiescent operating
point. Rf is found by:
where:
(2.7.71

f1

high frequency 3dB corner

Ic

01 collector current

A = mid·band gain in dB
2·17

Input capacitor C4 plays an important role in reducing the
effect of llf noise. Noise due to llf is predominantly a
current phenomenon, so making C4 large presents a small
impedance to the 1If current, creating a smaller equivalent
noise voltage. A value of C4 ; 10MF has been found
adequate.

C2 ; _ _1_ _ ; _ _ _ __
2 IT fo R4
6.28 x 20 x 36
; 2.21 x 10-4

Example 2.7.1

9. From Equation (2.7.5) the gain of the input stage is:

Design an ultra-low noise preamplifier with a gain of 1,000
operating from a 24 V supply and a 600n source impedance.
Bandwidth of interest is 20 Hz to 10kHz.

(2 x 10 5 ) R2

0.026

Solution:

1

--+----1
1
1
Ic
-+-+-

1. From Figure 2.7.3 the optimum collector current for
600n source resistance is 170MA.

104

2. From Equation (2.7.2),

R4

2 x 10 5 x 10 5

Vs - 2.1
Rl + R2 ; - - - - Ic - 18 x 10-6

10 5 +2x 10 5
0.026

1

1.7x 10-4

~+_1_+~
104 103 36

----+-----

24 - 2.1
(170 - 18) x 10-6
Rl

R3

A1 ; 355.

+ R2 ; 1.44 x 105

10. For 100dB supply rejection at 120Hz, Equation (2.7.3):
3. From Equation (2.7.4),
lOP.S.R./20
C1; - - - 2 IT f R 1 A 1
R2 "" 100kn

C1

10 5

10100/20

2

IT

x 120 x 39 x 103 x 355

9.6 x 10-6

1.04 x 10 10

R 1 ; 36 x 10 3 "" 39 kn
4. From Equation (2.7.6),
VE ; 0.8 _ (170 x 10-6 x
130

11. For a high frequency corner, f1, of 10kHz, Equation
(2.7.12):

250~
'}

1

C3 ;

VE ; 0.47

2 IT f1

5. From Equation (2.7.8) let R3 ; 1 kn. (Use 2.5kn trimpot and adjust for Vo ; V s/2.)

C3 ;

6. From Equation (2.7.9),

(0.~:6) 10A/20

- 4 x 10- 12

1
- 4 x 10-12
6.28 x 104 x 1.53 x 102 x 103

C3 ; 1.0 x 10- 10 "" 100pF

J

Vs x 10 7
Rf ; -1 [
2 VE (1.1 x 104 ) - Ic x 10 7

r.

The noise performance of the circuit of Figure 2.7.6 can be
found with the aid of Figures 2.7.1 and 2.7.2 and Equation
(2.7.1). From Figures 2.7.1 and 2.7.2 the noise voltage ~
and noise current (in) at 170MA are: en ; 3.0nV/y'Hz,
in; O.72pA/VHZ. From Equation (2.7.1):

1

R
1
24 x 10 7
f ; "210.47 (1.1 x 104 ) - 1.7 x 103J
Rf

=

3.46 x 104 "" 36kn

7. For a gain of 1,000, Equation (2.7.10):

; J·U3.0 x 10-9 )2 + (7.2 x 10-13 x 600)2 + 9.94 x 10-18] 104

(Rf + R4)
Amplifier Gain; - - - ; 1,000
R4

Total Wide band ; 4.37 x 10-7V
NOise Voltage
Wide band
Noise Figure

R4 ; 36x 10 3 ; 36n
103

4 KT Rs + en 2 + (in Rs)2
1010g--------4 KT Rs
10-18+90x 10-18+186x 10-19
10 log 994x
.
.
.
9.94 x 10- 18
10 log 1.92 ; 2.83dB

8. For a low corner frequency, fo, of 20 Hz, Equation
(2.7.11):
2-18

Vs
1.V

O.'"Fl'
C5

V1N

lOOK

+ c'
Tl0uF

f'
'\...,

R'
39K

R1

C'
lO!JF

,

V,

(3'1)

-L

R,
36K

-=

-=

BIAS
AOJ

R3
2.5K

-=

R4

l-

AVAC '" 1 +

36

T200

=(~-1)R5
1.6

RS '" 240kn MAXIMUM

R,
As

C2=-'211 foR6

C1

pF

Cc

=-'211fO Rl

fa '" LOW FREOUENCY -3dB CORNER

FIGURE 2.7.6 Typical Application with Increased Current
Density of Input Stage

FIGURE 2.8.1 LM387 Non-inverting AC Amplifier

2.8

LM387/387A LOW NOISE MINI DIP DUAL PRE·
AMPLIFIER

Vs

2.8.1 Introduction
The LM387 is a low cost, dual preamplifier supplied in the
popular 8 lead minidip package. The internal circuitry is
identical to the LM381 and has comparable performance.
By omitting the external compensation and single ended
biasing pins it has been possible to package this dual
amplifier into the 8 pin minidip, making for very little
board space requirement. Like the LM381, this preamplifier
is 100% noise tested and guaranteed, when purchased
through authorized distributors. Total equivalent input
noise is typically 0.65INRMS (R s = 600[7" 100Hz·10kHz)
and supply rejection ratio is typically 110dB (f = 1 kHz).
All other parameters are identical to the LM381. Biasing,
compensation and split·supply operation are as previously
explained.

R4 =(~-1\
Rs
1.6 j
R5 '" 240kn MAXIMUM

AVAC '"

R,

-Aij

2.8.2 Non-Inverting AC Amplifier
C1'-'2nfoR6

For low level signal applications requiring· optimum noise
performance the non-inverting configuration remains the
most popular. The LM387 used as a non-inverting AC
amplifier is configured similar to the LM381 and has the
same design equations. Figure 2.8.1 shows the circuit with
the equations duplicated for convenience.

Cc

=-'2:rr oAl
f

fa '" LOW FREOUENCY -3dB CORNER

FIGURE 2.8.2 LM387 Inverting AC Amplifier

2.8.3 Inverting AC Amplifier

2.8.4 Unity Gain Inverting Amplifier

For high level signals (greater than 300 mVI. the inverting
configuration may be used to overcome the positive input
overload limit. Voltage gains of less than 20dB are possible
with the inverting configuration since the DC biasing
resistor R5 acts to voltage divide the incoming signal as
'previously described for the LM381. Design equations are
the same as for the LM381 and are duplicated along with
the inverting circuit in Figure 2.8.2.

The requirement for unity gain stability is that the gain of
the amplifier from pin 2 (or 7) to pin 4 (or 5) must be at
least ten at all frequencies. This gain is the ratio of the
feedback resistor R4 divided by the total net impedance
seen by the inverting input with respect to ground. The
assumption is made that the driving, or source, impedance
is small and may be neglected. In Figure 2.8.2 the net
impedance looking back from the inverting input is R511 R6,
2-19

at high frequencies. (At low frequencies where loop gain is
large the impedance at the inverting input is very small and
R5 is effectively not present; at higher frequencies loop
gain decreases, causing the inverting impedance to rise to
the limit set by R5. At these frequencies R5 acts as a
voltage divider for the input voltage guaranteeing amplifier
gain of 10 when properly selected.) If the ratio of R4
divided by R511R6 is at least ten, then stability is assured.
Since R4 is typically ten times R5 (for large supply voltages)
and R6 equals R4 (for unity gain), then the circuit is stable
without additional components. For low voltage applications where the ratio of R4 to R5 is less than ten, it becomes
necessary to parallel R5 with a series R-C network so the
ratio at high frequencies satisfies the gain requirement.
Figure 2.8.3 shows such an arrangement with the constraints
on R7 being given by Equations (2.8.1 H2.8.3).

4. From Equation (2.8.2):
RY = R IIR = 5.6k x 20k = 4,375
5 6
5.6k + 20k
5. From Equation (2.8.3):
4375 x 20 x 10 3

= 3684

10 x 4375 - (20 x 103 )
Use R7 = 3.6k
6. For fo = 20 Hz,
= _ _ _1_ _ = 3.98

x 10-7

211' x 20 x 20k
Use C2 = 0.5pF.

R4

AV (pin 2 to 4) - -

;;, 10

(2.8.1)

Cc = __1__ =
1
= 7.95 x 10-8
211'foRL
211'x20x100k

R511 R611 R7
RY

= R511R6

(2.8.2)
Use Cc

RY R4

R7 .;;;

(2.8.3)

=

0.1pF.

7. The selection of C3 is somewhat arbitrary, as its effect is
only necessary at high frequencies. A convenient frequency for calculation purposes is 20 kHz.

10RY - R4

Vs

C3 =

1
211' (20kHz) R7

= ___1_ _ _ =

2.21 x 10-9

211' x 20k x 3.6k

Use C3 = 0.0022pF.

2.8.5 Application to Feedback Tone Controls
One of the most common audio circuits requiring unity
gain stability is active tone controls. Complete design
details are given in Section 2.14. An example of modified
Baxandall tone controls using an LM387 appears as Figure
2.14.10 and should be consulted as an application.of the
stabilizing methods discussed in Section 2.8.4.

FIGURE 2.8.3 Unity Gain Amplifier for Low Supply Voltage

Example 2.8.1
2.9 LM382 LOW NOISE DUAL PREAMPLIFIER WITH
RESISTOR MATRIX

Design a low noise unity gain inverting amplifier to operate
from Vs = 12V, with low frequency capabilities to 20Hz,
input impedance equal to 20 kn, and a load impedance of
100kn.

2.9.1 Introduction
The LM382 is a dual preamplifier patterned after the LM381
low noise circuitry ·but with the addition of an internal
resistor matrix. The resistor matrix allows the user to
select a variety of closed loop gain options and frequency
response characteristics such as flat-band, NAB (tape), or
RIAA (phonograph) equalization. The LM382 possesses all
of the features of the LM381 with two exceptions: no
single ended input biasing option and no external pins for
adding additional compensation capacitance. The internal
resistors provide for biasing of the negative input automatically, so no external resistors are necessary and use of
the LM382 creates the lowest parts count possible for
standard designs. Originally developed for the automotive
tape player market with a nominal supply voltage of +12V,
the output is self queuing to about +6V (regardless of
applied voltage - but this can be defeated, as will be
discussed later). A diagram of the LM382 showing the
resistor matrix appears as Figure 2.9.1.

Solution:
1. Rin = R6 = 20kn.
2. For unity gain R4 = R6, R4 = 20k.
3. From Figure 2.8.2:
R4

=( 2.6
Vs _ 1) R5 =(E -1) R5
2.6

R4

= 3.62

R5

Therefore:
R4
R5 = 3.62

20k
3.62

= 5,525n

Use R5 = 5.6k.
2-20

Since bias currents are small and may be ignored in gain
calculations, the 50k input resistor does not affect gain.
Therefore, the gain is given by:

J

1',

vs

I y, lM3B;' ,
I

Av1 = 1 +~~ = 101"" 40dB
500

'111)

110,""14-)t-:---i+

"

'-

12,13)

,,

,
".-

15k

".-'"
-/L

".-

17,81

-/

/
SDk 15,10) SDk
15,10)
16,9)

500

15k

13,12)
13,12)

FIGURE 2.9.1 LM382 Resistor Matrix

FIGURE 2.9.4 Equivalent Circuit for 55dB Gain (C2 Only)

2.9.2 Non-Inverting AC Amplifier
The fixed·gain flat·response configuration of the LM382
(Figure 2,9.2) shows that with just two or three capacitors a
complete high gain, low noise preamplifier is created,

With C2 only, the redrawn equivalent circuit looks like
Figure 2,9.4, Since the feedback network is wye·connected,
it is easiest to perform a wye·delta transformation (see
Appendix A3) in order to find an effective feedback
resistor so the gain may be calculated. A complete trans·
formation produces three equivalent resistors, two of which
may be ignored. These are the ones that connect from the
ends of each 50k[1 resistor to ground; one acts as a load on
the amplifier and doesn't enter into the gain calculations,
and the other parallels 500[1 and is large enough to have no
effect, The remaining transformed resistor connects directly
from the output to the input and is the equivalent feedback
resistor, Rf. Its value is found from:

Vs

...L

-=GAIN

40dB
55d8
BOd8

-=-

*REnUIRED

CAPACITORS
Cl ONLY
C2 ONl Y
Cl & C2

,
(50k)2
Rf (eqUivalent) = 50k + 50k + - 15k

267k

The gain is now simply

FIGURE 2.9.2 LM382 as Fixed Gain·Flat Response
Non-inverting Amplifier

AV2 = 1 + 267k = 535 "" 55dB
500
To understand how the gains of Figure 2,9.2 are calculated
it is necessary to redraw each case with the capacitors
short·circuited and include only the relevant portion of the
resistor network per Figure 2,9,1. The redrawn 40dB gain
configuration (C1 only) appears as Figure 2.9.3.

Adding both C1 and C2 gives the equivalent circuit of
Figure 2,9.5,

SDk
15,10)
SDk
SOD
16,9)

~

FIGURE 2.9.3 Equivalent Circuit for 40dB Gain (C, Only)

13,12)

-= 16,9)

FIGURE 2.9.5 Equivalent Circuit for SOdB Gain IC, and C2)

2·21

Treating Figure 2.9.5 similarly to Figure 2.9.4, an equivalent
feedback resistor is calculated:
.
(50k)2
Af (equivalent) = 50k + 50k + - 500

done by adding a resistor at pin 5 (or 10) which parallels
the internal 15 k£2 resistor and defeats its effect (Figure
2.9.7).

5.1 Meg

/',
I LM382

Therefore, the gain is:
Av12 = 1 +

5.1 Meg

50()

I

'-....

11.141

= 10201 "" 80dB.

v

-....,

S
-"",1111

I
2.9.3 Adjustable Gain for Non-Inverting Case

........

12.131

'-....

L-'-./\i'v-.....-¥ltv-....------~r;nOvsl2

As can be learned from the preceding paragraphs, there are
many combinations of ways to configure the resistor
matrix. By adding a resistor in series with the capacitors it
is possible to vary the gain. Care must be taken in attempting low gains « 20dB), as the LM382 is not unity gain
stable and should not be operated below gains of 20dB.
(Under certain specialized applications unity gain is possible,
as will be demonstrated later.) A general circuit allowing
adjustable gain and requiring only one capacitor appears as
Figure 2.9.6.

50k

/

5.10
15k

,/

-=-

I

1---/

//

/

//141--

11.8

/

/

FIGURE 2.9.7 Internal Bias Override Resistor

vs

PINS 3. 5. 6. 9, 10, 12
ALL NO CONNECTION

Since the positive input is biased internally to a potential
of +1.3V (see circuit description for LM381), it is necessary
that the DC potential at the negative input equal +1.3 V
also. Because bias current is small (0.5!1A), the voltage drop
across the 50k resistor may be ignored, which says there is
+1.3V across RQ. The current developed by this potential
across RQ is drawn from the output stage, through the 50k
resistor, through RQ and to ground. The subsequent voltage
drop across the 50k resistor is additive to the +1.3 V and
determines the output DC level. Stated mathematically,

I
I
...L

~2

GAIN'" 1 + 267k
Rl

= (50k)1.3V + 1.3V

(2.9.3)

RX

where:

RX = RQI115k

Cl=_l_

2rrfo Rl

From Equation (2.9.3) the relationships of RX and RQ may
be expressed.

fo = LOW FREQUENCY -Jd8 CORNER

50k

RX =

FIGURE 2.9.6 Adjustable Gain Non-inventing Amplifier

(2.9.4)

Vs _ 1
2.6
Referring to Figure 2.9.1, it is seen that the R1-C1 combination is used instead of the internal 500£2 resistor and that
the remaining pins are left unconnected. The equivalent
resistance of the 50k-50k-15k wye feedback network was
found previously to equal 267k£2, so the gain is now given
by Equation (2.9.1).
Gain = 1 + 267k
R1

RQ = RX (15k)
15k - RX

(2.9.5)

Example 2.9.1
Select RQ such that the output of a LM382 will center at
12VDC when operated from a supply of Vs = 24VDC.

(2.9.1)

Solution
1. Calculate RX from Equation (2.9.4).

And C1 is found from Equation (2.9.2):
RX =
(2.9.2)

50 x 103

~-1

= 6075£2

2.6
where:

fo = low frequency -3dB corner.

2. Calculate RQ from Equation (2.9.5).

2.9.4 Internal Bias Override

RQ = (6075)(15 x 103 ) = 10210£2
(15 x 103 ) - 6075

As mentioned in the introduction, it is possible to override
the internal bias resistor which causes the output quiescent
point to sit at +6 V regardless of applied voltage. This is

Use RQ = 10k£2.
2-22

Since RQ parallels the 15k resistor, then the AC gains due
to the addition of capacitor Cl or C2 (or both) (as given in
Figure 2.9.2) are changed. The new gain equations become
a function of RQ and are given as Equations (2.9.6).(2.9.8)
and refer to Figure 2.9.8.
Cl 0nl y:

Gain ""

1+~

(2.9.6)

C2 Only:

Gain

201 + 5 x 10 6
RX

(2.9.7)

C1 & C2:

6
Gain "" 201 + 5 x 10
RQI1500

(2.9.8)

where:

With C1:

Gain

=

(_~)(,05+2.5Xl09)
R1 \

(2.9.10)

RQI1500

and the circuit is shown in Figure 2.9.11.

Vs

RQI1500

RX and RQ are given by Equations (2.9.4) and
(2.9.5).

Vs
GAIN" - 267' I> 20dB FOR STABILITY)
Rl
Co =_1_

2rrfo Rl

to " lOW FREQUENCY -3dB CORNER
INPUT IMPEDANCE" Rl
PINS 3. 5. 6. 9. 10. 12 NOT USED

FIGURE 2.9.9 LM382 as Inverting AC Amplifier

-=

- -

-=
Vs

* - IF REQUIRED
PINS 2 & 13 NO CONNECTION
FIGURE 2.9.8 Fixed Gain Amplifier with Internal Bias Override

Continuing the previous example to find the effect of RQ
on the gain yields:
3. Cl Only:

Gain

1+~=

53.6dB

10kl1500
4. C2 Only:

Gain

201 + 5 x 10 6
6075

GAIN" _5,1,10 6
Rl

60.2dB

Co =_1_

2nfoRl

Gain = 201 + 5 x 106 = 80.6dB
lOkl1500

to " lOW FREQUENCY -3dB CORNER
INPUT IMPEDANCE" Al

2.9.5 Inverting AC Amplifier

PINS 3. 5. 10. 12 NOT USED

Examination of the resistor matrix (Figure 2.9.1) reveals
that an inverting AC amplifier can be created with just one
resistor (Figure 2.9.9).

FIGURE 2.9.10 High Gain Inverting AC Amplifier

Vs

The gain is found by calculating the equivalent feedback
resistance as before, and appears in Figure 2.9.9. Higher
gains are possible (while retaining large input resistance = R 1)
by adding capacitor Cl as shown in Figure 2.9.10.
The internal bias override technique discussed for the non·
inverting configuration may be applied to the inverting case
as well. The required value of RQ is calculated from
Equations (2.9.4) and (2.9.5) and affects the gain relation
shown in Figures 2.9'.9 and 2.9.10. The new gain equations
are:
Without C1 :

Gain

t

* -IF REQUIRED

~)~05 + 2.5 x 109 )
Rl

PINS 3 & 12 NOT USED

(2.9.9)

FIGURE 2.9.11 Inverting Amplifier with Internal Bias Override

RQI115k

2·23

depending upon supply voltage. If done DC (tied from pin
2 (or 13) directly to ground), then it becomes RQ (from
Figure 2.9.7) and affects the output DC level. Placing a
capacitor in series with this resistor makes it effective only
for AC voltages and does not change the output level. The
required resistor equals 9.1 kn, which is close enough to the
required RQ for Vs ~ 24 V. Two examples of unity gain
amplifiers appear as Figure 2.9.13 and should satisfy the
majority of applications.

Example 2.9.2
Design an inverting amplifier to operate from a supply of
Vs ~ 24 VDC, with output quiescent point equal to 12VDC,
gain equal to 40dB, input impedance greater than 10kn,
low frequency performance flat to 20 Hz, and a load
impedance equal to 100 kn.
1. From the previous example RQ ~ 10kn.
2. Add Cl for high gain and input impedance.
3. Calculate Rl from Equation (2.9.10).

+24V

Rl ... ( 1 )(,0 5 2.5 x 10 9 )
... Gain \
+ RQI1500

h05 + 2.5 x 109 )
(_!_\
102/ \
10kl1500

(Note: 40dB

=

o-It--"v
..........r-1\;7,;\
Co
R1

102VIV)

0.15

SOk

131 ~ 5.35 x 104
I

Use Rl

~

...L

56kn.

4. Calculate Co from equation shown in Figure 2.9.9.
Co ~ __1_ _
211" fo Rl
Use Co

~

(al Supply Voltage = 24 Volts

_ _ _1_ _ ~ 1.42 x 10-7
(211") (20) (56k)

t12V

0.15gF.

5. Calculate Cc from Equation (2.6.12).
Cc ~ __1_ _ ~
211" fo Rl
(211") (20) (105)
Use Cc

~

Co

Rt

0.15

SDk

o-I~..........~~
'2

7.96 x 10-8

9.1k

I

O.lgF.

~

The complete amplifier is shown in Figure 2.9.12.

(bl Supply Voltage = 12 Volts
+24V

FIGURE 2.9.13 Unity Gain Inverting Amplifier

2.9.7 Remarks

o-!I-'VIN-'--I
Co

The above application hints are not meant to be all-inclusive,
but rather are offered as an aid to LM382 users to familiarize
them with its many possibilities. Once understood, the
internal resistor matrix allows for many possible configurations, only a few of which have been described in this
section.

0.15

FIGURE 2.9.12 Inverting Amplifier with Gain = 40dB and
Vs = +24V

2.9.6 Unity Gain Inverting Amplifier

2.10 LM1303 STEREO PREAMPLIFIER

Referring back to Figure 2.9.1, it can be seen that by
shorting pin 2 (or 13) to 5 (or 10) the feedback network
reduces to a single 50kn resistor connected from the
output to the inverting input, plus the 15 kn biasing
resistor from the inverting input to ground. To create unity
gain then, a resistor equal to 50kn is connected to the
minus input. Simple enough; however, the amplifier is not
stable. Since the 15k resistor acts as a voltage divider to
the input, the gain of the amplifier (pin 7 to pin 2) is only
50k divided by 15k, or 3.33VIV. Minimum required gain
for stability is 10V/V, so it becomes necessary to shunt the
15k resistor with a new resistor such that the parallel
combination equals 5kn. This may be done AC or DC,

The LM1303 is a dual preamplifier designed to be operated
from split supplies ranging from ±4.5V up to ±15V. It has
"op amp" type inputs allowing large input signals with low
distortion performance. The wide band noise performance
is superior to traditional operational amplifiers, being
typically 0.9VRMS (10kHz bandwidth). Compensation is
done externally and offers the user a variety of choices,
since three compensation points are brought out for each
amplifier. The LM1303 is pin-for-pin compatible with
"739" type dual preamplifiers and in most applications
serves as a direct replacement.

2.10.1 Introduction

2-24

2.10.2 Non·lnverting AC Amplifier

2.10.3 Inverting AC Amplifier

The LM1303 used as a non·inverting amplifier (Figure
2.10.1) with split supplies allows for economical direct·
coupled designs if the DC levels between stages are main·
tained at zero volts. Gain and C1 equations are shown in the
figure. Resistor R3 is made equal to R1 and provides DC
bias currents to the positive input. Compensation capacitor
C2 is equal to O.022/1F and guarantees unity gain stability
with a slew rate of approximately 1 V //1s. Higher slew rates
are possible when higher gains are used by reducing C2
proportionally to the increase in gain, e.g., with a gain of
ten, C2 can equal 0.0022/1F, increasing the slew rate to
around 10 V//1s. Some layouts may dictate the addition of
C3 for added stability. It should be picked according to
equation (2.10.1) where fH is the high frequency -3dB

For applications requiring inverting operation, Figure 2.10.2
should be used. Capacitors C2 and C3 have the same
considerations as the non-inverting case. Resistor R3 is made
equal to R 1 again, minimizing offsets and providing bias
current. The same slew rate-gain stability trade-ofts are
possible as before.
2.11 PHONO PREAMPLIFIERS AND RIAA
EQUALIZATION
2.11.1 Introduction
Phono preamplifiers differ from other preamplifiers only
in their frequency response, which is tailored in a special
manner to compensate, or equalize, for the recorded
characteristic. If a fixed amplitude input signal is used to
record a phonograph disc, while the frequency of the signal
is varied from 20Hz to 20kHz, the playback response curve
of Figure 2.11.1 will result. Figure 2.11.1 shows a plot of
phono cartridge output amplitude versus frequency, indicating a severe alteration to the applied fixed amplitude signal.
Plavback equalization corrects for this alteration and reo
creates the applied flat amplitude frequency response. To
understand why Figure 2.11.1 appears as it does, an
explanation of the recording process is necessary.

corner.

(2.10.1)
Vee

2.11.2 Recording Process and R IAA

AVAC =

The grooves in a stereo phonograph disc are cut by a chisel
shaped cutting stylus driven by two vibrating systems
arranged at right angles to each other (Figure 2.11.2). The
cutting stylus vibrates mechanically from side to side in
accordance with the signal impressed on the cutter. This is
termed a "lateral cut" as opposed to the older method of
"vertical cut." The resultant movement of the groove back
and forth about its center is known as groove modulation.
The amplitude of this modulation cannot exceed a fixed
amount or "cutover" occurs. (Cutover, or overmodulation,
describes the breaking through the wall of one groove into
the wall of the previous groove.) The ratio of the maximum
groove amplitude possible before cutover, to the minimum
amplitude allowed for acceptable signal·to·noise performance (typically 58dB), determines the dynamic range of a
record (typically 32·40dB). The latter requirement results
from the grainy characteristic of the disc surface acting as a
noise generator. (The cutting stylus is heated in recording
to impart a smooth side wall to minimize the noise.) Of
interest in phono preamp design is that the record noise
performance tends to be ten times worse than that of the
preamp, with typical wide band levels equal to 10/1V.

,+~
R2

c,=-'21ffoRZ
I. = LOW FREQUENCY -JdB CORNER

• - MAY BE OMITTED FOR
DIRECT·CDUPLED DESIGNS.

FIGURE 2.10.1 LM1303 Non·inverting AC Amplifier
VCC

+

Cs

.... -it--,

'OpFI

I

CJ

I

Amplitude and frequency characterize an audio signal. Both
must be recorded and recovered accurately for high quality
music reproduction. Audio amplitude information trans·
lates to groove modulation amplitude, while the frequency
of the audio signal appears as the rate of change of the
groove modulations. Sounds simple enough, but Figure
2.11.1 should, therefore, be a horizontal straight line
centered on OdB, since it represents a fixed amplitude input
signal. The trouble results from the characteristics of the
cutting head. Without the negative feedback coils (Figure
2.11.2) the velocity frequency response has a resonant peak
at 700Hz due to its construction. Adding the feedback coils
produces a velocity output independent of frequency;
therefore, the cutting head is known as a constant velocity
device (Figure 2.11.2a) .

...L

AVAC =

_~

VEE

R2

Co : _ ' 21ffo RZ
I. = LOW FREQUENCY -JdB CORNER
• - MAY BE OMITTED FOR
DIRECT·COUPLED DESIGNS.

Figure 2.11.1 appears as it does because the cutting
amplifier is pre·equalized to provide the recording character·

FIGURE 2.10.2 LM13031nverting AC Amplifier

2·25

+20

H+H:tHII--++++-HlIf-+l-Httfll-+-I-+t11t11

+to

H+H:tHII--++++-HlIf-+1+I>4IlI-+l-+t11t11

-t 0

H+HHlIf-M+lfjlll-++i-HlllH-++I-IlHI

-20

1-I''FHl#ll--++ftttIlHtttHlll--+t+ttHll
to

tOO

tk

10k

The not-so-simple answer begins with the drilling coils of
the cutting head_ Being primarily inductive, their impedance
characteristic is frequency dependent_ If a fixed amplitude
input signal translates to a fixed voltage used to drive the
coils (called "constant amplitude") then the resulting
current, Le_, magnetic field, hence rate of change of
vibration, becomes frequency dependent (Figure 2_11.2b);
if a fixed amplitude input signal translates to a fixed current,
Le_, fixed rate of vibration, used to drive the coils (called
"constant velocity") then the resulting voltage, Le_, cutting
amplitude, becomes frequency dependent (Figure 2_11_2al.
With respect to frequency, for a given input amplitude the
cutting head has only one degree of freedom: vibrating rate
(constant velocity = current drive) or vibrating distance
(constant amplitude = voltage drive).

tOOk

FREQUENCY (Hz)

The terms constant velocity and constant amplitude create
confusion until it is understood that they have meaning only
for a fixed amplitude input signal, and are used strictly to
describe the resultant behavior of the cutting head as a
function of frequency. It is to be understood that changing
the input level results in an amplitude change for constant
velocity recording independent of frequency. For example,

FIGURE 2.11.1 Typical Phono Playback Characteristic for a Fixed
Amplitude Recorded Signal

istic shown. Two reasons account for the shape: first, low
frequency attenuation prevents cutover; second, high frequency boosting improves signal-to-noise ratio. The unanswered question is why is all this necessary?

ELECTROMECHANICAL TRANSDUCERS

"'"~'·"'~A~""'·"'
FEEOBACK_

COI~ ~"---ORIVING

COIL

:~~
FIGURE 2.11.2 Stereo Cutting Head

1 - - - - '.....--; (GROOVE VElOCITY)
v (GROOVE AMPLITUDE)
FREQUENCY

FIGURE 2_11.2A Constant Velocity Recording

; (GROOVE VElOCITY)

I-...,.~----v (GROOVE AMPLITUDE)

FREQUENCY

s· MAXIMUM SLOPE

FIGURE 2.11.2B Constant Amplitude Recording

2-26

reference points and are sometimes referred to as time
constants. This is a carryover from the practice of specifying
corner frequencies by the equivalent RC circuit (t = RC)
that realized the response. Conversion is done simply with
the expression t = 1/21ff and results in time constants of
3180l.1s for fl, 3181.1s for f2, and 751.1s for f3. Frequency f2
is referred to as the turnover frequency since this is the
point where the system changes from constant amplitude to
constant velocity. (Likewise, f3 is another turnover frequency.) Table 2.11.1 is included as a convenience in
checking phono preamp R IAA response.

if an input level of 10mV results in 0.1 mil amplitude
change for constant amplitude recording and a velocity of
5cm/s for constant velocity recording, then a change of
input level to 20mV would result in 0.2 mil and 10cm/sec
respectively - independent of frequency.
Each of these techniques when used to drive the vibrating
mechanism suffers from dynamic range problems. Figures
2.11.2a and 2.11.2b diagram each case for two frequencies
an octave apart. The discussion that follows assumes a
fixed amplitUde input signal and considers only the effect
of frequency change on the cutting mechanism.
Constant velocity recording (Figure 2.11.2a) displays two
readily observable characteristics. The amplitude varies
inversely with frequency and the maximum slope is constant
with frequency. The second characteristic is ideal since
magnetic pickups (the most common type) are constant
velocity devices. They consist of an active generator such
as a magnetic element moving in a coil (or vice versa) with
the output being proportional to the speed of movement
through the magnetic field, i.e., proportional to groove
velocity. However, the variable amplitude creates serious
problems at both frequency extremes. For the ten octaves
existing between 20 Hz and 20 kHz, the variation in
amplitude is 1024 to 1! If 1 kHz is taken as a reference
point to establish nominal cutter amplitude modulation,
then at low frequencies the amplitudes are so great that
cutover occurs. At high frequencies the amplitude becomes
so small that acceptable signal-to-noise ratios are not
possible - indeed, if any displacement exists at all. So much
for constant velocity.

TABLE 2.11.1 RIAA Standard Response

30
pi,1

50Hz
f2'" 500Hz

li3=2120H,

l~
i2
-10

-30
10Hz

CONSTANT

I II
I /I
100Hz

1 kHz

+0.7
0.0'
-1.4
-2.6
-4.8
-6.6
-8.2
-9.6
-11.9
-13.7
-17.2
-19.6

Magnetic cartridges have very low output levels and require
low noise devices to amplify their signals without appreciably
degrading the system noise performance. With low noise
integrated circuits like the LM387 or LM381, the dominant
noise source becomes the cartridge and loading resistor and
not the active device (see Appendix A5).

'.

r- AMPlITUOE
-20

dB

2.11.4 LM387 or LM381 Phono Preamp

VElOCITY i3

~

800
lk
1.5k
2k
3k
4k
5k
6k
8k
10k
15k
20k

Before getting into the details of designing RIAA feedback
networks for magnetic phono cartridges, a few words about
crystal and ceramic cartridges are appropriate. In contradistinction to the constant velocity magnetic pickups,
ceramic pickups are constant amplitude devices and therefore do not require equalization, since their output is
inherently flat. Referring to Figure 2.11.3 indicates that
the last sentence is not entirely true. Since the region
between f2 and f3 is constant velocity, the output of a
ceramic device will drop 12dB between 500 Hz and 2000 Hz.
While this appears to be a serious problem, in reality it is
not. This is true due to the inherently poor frequency
response of ceramic and restriction of its use to lo-fi and
mid-fi market places. Since the output levels are so large
(100mV-2V), a preamp is not necessary for ceramic pickups;
the output is fed directly to the power amplifier via
passive tone (if used) and volume controls.

CONSTAN~'+---

10

Hz

+19.3
+18.6
+17.8
+17.0
+16.1
+14.5
+13.1
+10.3
+8.2
+5.5
+3.8
+2.6

2.11.3 Ceramic and Crystal Cartridges

-ttt+i!' '"
III I

dB

20
30
40
50
60
80
100
150
200
300
400
500

* Reference frequency.

Looking at Figure 2.11.2b, two new observations are seen
with regard to constant amplitude. Amplitude is constant
with frequency (which corrects most of the ills of constant
velocity), but the maximum slope varies directly with
frequency, i.e., groove velocity is directly proportional to
frequency. So now velocity varies 1024 to lover the audio
band - swell! Recall that magnetic cartridges are constant
velocity devices, not constant amplitude, so the output will
rise at the rate of +6dB/octave. (6dB increase equals twice
the amplitude.) To equalize such a system would require
60dB of headroom in the preamp - not too practical. The
solution is to try to get the best of both systems, which
results in a modified constant amplitude curve where the
midband region is allowed to operate constant velocity.

20

Hz

10kHz

100kHz

Typical cartridge output levels are given in Table 2.11.2.
Output voltage is specified for a given modulation velocity.
The magnetic pickUp is a velocity device, therefore output
is proportional to velocity. For example, a cartridge producing 5mV at 5cm/s will produce 1 mV at 1 cm/s and is
specified as having a sensitivity of 1 mV /cm/s.

FIGURE 2.11.3 RIAA Playback Equalization

The required RIAA (Record Industry Association of
America) playback equalization curve (Figure 2.11.3) shows
the idealized case dotted and the actual realization drawn
solid. Three frequencies are noted as standard design

In order to transform cartridge sensitivity into useful
preamp design information, we need to know typical and
maximum modulation velocity limits of stereo records.
2-27

TABLE 2.11.2

Manufacturer

Model

Example 2.11.1
Design a phonograph preamp operating from a 30V supply,
with a cartridge of 0.5mV/cm/s sensitivity, to drive a power
amplifier of 5 V RMS input overload limit.

Output at 5 em/sec

Empire Scientific

999
888

5mV
8mV

Shure

V·15
M91

3.5mV
5mV

Solution

Pickering

V-15AT3

5mV

2. From Equation (2.6.4):

1. From Equation (2.6.3) let R5 =

100k~L

R4 = (VCC _ 1) R5
2.6

The R IAA recording characteristic establishes a maximum
recording velocity of 25cm/s in the range of 800 to 2500 Hz.
Typically, good quality records are recorded at a velocity of
3 to 5cm/s.

= (30 _ 1) 105

2.6

Figure 2.11.3 shows the R IAA playback equalization. This
response is obtained with the circuit of Figure 2.11.4.

R4 = 10.5 x 10 5 "" 1.0Mn
3. Equation (2.11.2):

6.28 x 50x 1.0 x 106
3.18x 10-9
C7 "" 0.003/lF
4. Equation (2.11.3):

R6
R5

'1'C2

C7 - - - 2 7r f2 RlO
RlO =

FIGURE 2.11.4 RIAA Phono Preamp

1
6.28 x 500 x 3 x 10-9
1.06 x 105

Resistors R4 and R5 set the DC bias (Section 2.6). The 0 dB
reference gain is set by the ratio:
Rl0 + R6
OdB Ref Gain = - - - R6

RlO "" 100kn.
5. The maximum cartridge output at 25cm/s is
(0.5mV/cm/s) x (25cm/s) = 12.5mV. The required midband gain is therefore:

(2.11.1)

5VRMS

The corner frequency, f1 (Figure 2.11.3), is established
where XC7 = R4 or:

6. Equation (2.11.1):

(2.11.2)

RlO + R6
OdB Ref Gain = - - - ' R6

Likewise, frequency f2 occurs where XC7 = RlO or:

R6 = ]OOk
399

(2.11.3)

1
2 7r f3 RlO

400

251 "" 240n

RZ = 10R6 = 2400n

The third corner frequency, f3, is determined where
XC8 = RlO:
C8 =

= 400

12.5mVRMS

7. Equation (2.6.10):

(2.11.4)
2 7r fo R6

Resistor RZ is used to insert a zero in the feedback loop
since the LM381 is not compensated for unity gain. Either
RZ is required to provide a zero at or above a gain of 20dB
(R Z = 10 R61. or external compensation is provided for
unity gain stability.
2-28

6.28 x 20 x 240

3.3 x 10-5

2.11.6 LM1303 Phono Preamp

8. Equation (2.11.4):
C8 =

The LM1303 allows a convenient low noise phono preamp
design when operating from split supplies. The circuit
appears as Figure 2.11.7. For trimm ing purposes and/or
gain changes the relevant formulas follow:

--~-

6.28 x 2120 x 10 x 104

2rr f3 RlO
7.51 x 10-10

R2
OdB Ref Gain = 1 + R3

C8 '" 750pF
The completed design is shown in Figure 2.11.5 where a
47 kD- input resistor has been included to provide the R IAA
standard cartridge load.

f1

----

f2

------

30V

U

f3 =

"::"

(2.11.6)

2rr Rl Cl
(2.11.7)

2rr R2 Cl
(2.11.8)
21T R2 C2

As shown in Figure 2.11.7, the OdB reference gain (1 kHz)
equals about 34dB and the feedback values have been
altered slightly to minimize pole·zero interactions.

PC

41k

_

(2.11.5)

12.7)

lOOk

Vee
tOOk

III

FIGURE 2.11.5 LM387 Phone Preamp. (RIAA)

A,

"2
51k

The LM381 integrated circuit may be substituted for the
LM387 in Figure 2.11.5 by making the appropriate pin
number changes.

R3

0.0015

-=

2.11.5 LM382 Phono Preamp
By making use of the internal resistor matrix, a minimum
parts count low noise phono preamp is possible using the
LM382 (Figure 2.11.6). The circuit has been optimized for a
supply voltage equal to 12·14 V. The midband OdB reference
gain equals 46dB (200VN) and cannot easily be altered.
For designs requiring either gain or supply voltage changes,
the required extra parts make selection of a LM381 or
LM387 more appropriate.

I

+eJ
25 !J F
VEE

-=

FIGURE 2.11.7 LM1303 Phono Preamp. (RIAA)

2.11.7 LM381A Ultra-Low Noise Mini Preamp
By increasing the current density of the first stage of the
LM381A (see Section 2.7), it is possible to obtain optimum
noise performance for magnetic cartridge pickups. A
complete phono preamp using this technique is given in
Figure 2.11.8 with provisions for tuner and tape inputs,
selector switch and ganged volume control. Tone controls
are omitted but may be easily added (see Section 2.14). The
RIAA frequency response is within ±O.6dB of the standard
values shown in Table 2.11.1. The OdB reference gain at
1kHz is 41.6dB (120VNl, producing 1.5VRMS output
from a nominal 12.5mVRMS input. With the given supply
voltage of 33VDC, this gives better than +25dB headroom
(dynamic range) for a typical 5mV input at 1 kHz. Input
overload limit equals 91 mV at midband frequencies. Signalto-noise ratio is better than -85dB referenced to a 10mV
input level, with unweighted total output noise less than
100llV (input shorted). Metal film resistors and close
tolerance capacitors should be used to minimize excess
noise (see Section 2.3.2) and maintain R IAA frequency

+12V

[f

I

'1

1k

"F
0.33

47k

lk

accuracy.

FIGURE 2.11.6 LM382 Phono Preamp. (RIAA)

2·29

2.11.8 Inverse RIAA Response Generator

Break frequencies of the filter are determined by Equations
(2.11.9)·(2.11.11).

A useful test box to have handy while designing and building
phono preamps is one which will yield the opposite of the
playback characteristic, i.e., an inverse RIAA (or record)
characteristic. The circuit (Figure 2.11.9) is achieved by
adding a passive filter to the output of an LM387, used as a
flat·response adjustable gain block. Gain is adjustable over
a range of 24dB to 60dB and is set in accordance with the
OdB reference gain (1 kHz) of tMe phono preamp under
test. For example, assume the preamp being tested has
+34dB gain at 1 kHz. Connect a 1 kHz generator to the
input of Figure 2.11.9. The passive filter has a loss of -40dB
at 1 kHz, which is corrected by the LM387 gain, so if a
1 kHz test output level of 1 V is desired from a generator
input level of 10mV, then the gain of the LM387 is set at
+46dB (+46dB - 40dB + 34dB = Xl 00; 10mV x 100 = 1 V).

fl = 50Hz =

1
2rr R9 C4

(2.11.9)

500Hz

(2.11.10)

2120Hz

(2.11.11)

The R7,C3 network is necessary to reduce the amount of
feedback for AC and is effective for all frequencies beyond
20 Hz. With the values shown the. inverse R IAA curve falls
within 0.75dB of Table 2.11.1.

Vs = 33V

TAPE

L TUNER R

Rl
51k

+

R2

150k

R

Cl

Il0~

PHONO

R~IGHT:~:
__

TUNER

R7

Cs
0.1

--

-

-

TAPE

SlA

47k

-=-

PHONO

-=R4*
39.2

R3

BIAS
Uk

-=-

-=-

* - METAL FILM. 1% TOLERANCE

Vs = 33V
RS
51k

R9

150k

I

+ C9

10~

TAPE
TUNER

S18

PHONOO

C15
0.1

OUT
LEFT

lOOk

~

(LOGI

_

\~

1

- ~ -

FIGURE 2.11.8 LM381A Ultra-Low Noise Mini Preamp. (RIAAI

2-30

tape; thus each one grows shorter. As their effective length
decreases, more and more magnetic cancellation occurs due
to the close proximity of north and south poles - hence,
self·demagnetization. In lay language, the higher the
frequency, the weaker the signal (field).

40

V

I

'"=>
'"=>

0

w

:>

\

30
20

V

;::

~

v

l/

10

V
10

1k

100

10k

lOOk

FREQUENCY - Hz

FIGURE 2.11.9 Inverse RIAA Response Generator

FIGURE 2.12.1 Typical Tape Playback Head Response

2.12 TAPE PREAMPLIFIERS AND NAB EQUALIZATION

l~,toJ Glp

2.12.1 Introduction
Tape recorder playback preamplifiers require special fre·
quency shaping networks in their feedback paths in order
to equalize, or correct, the signal coming off the tape head.
Magnetic tape is recorded "constant current" (i.e., constant·
current for all frequencies) and the recording head is
primarily inductive. The impedance of the head, therefore,
rises at a 6dB/octave rate with respect to increasing fre·
quency, resulting in a corresponding rise in output voltage
amplitude, i.e., the output voltage varies in direct proportion to frequency. So the signal fed to the playback preamp
does not have a flat frequency response, but instead shows
a steadily increasing level with increasing frequency (Figure
2.12.1). At high frequencies Figure 2.12.1 shows an abrupt
change in response resulting in severe decrease in amplitude
with continuing increase in frequency. There are several
reasons for this phenomenon - all different and unrelated,
but each contributing to the loss of high frequency response.
The first area of degradation is due to the effects of the
decreasing wavelengths of the higher frequencies. Two
factors are important in minimizing wavelength problems:
recording speed (Figure 2.12.2) and head gap (width)
(Figure 2.12.3). The first of these is accounted for by the
fact that the faster the tape is moved past the recording
head, the more magnetic material (normally iron oxide
deposited on a plastic tape backing) is available for use in
capturing the rapidly changing magnetic field. With slowly
moving tape, a point is reached where there just is not
enough iron available to be magnetized. The second factor
is true because when the width of the gap in a playback
head equals the recorded wavelength, no output signal is
possible since the edges of the gap are at equal magnetic
potential.

151PS I :

I I

11111

,,/'

I

f\

7·1/2tPS

I I

3-3/4

V
100

"

~

IPS

~,

I II

\

'tBI"IS

1\ \
10k

lk

lOOk

FREUUENCY (Hz)

FIGURE 2.12.2 Effect of Tape Speed on Response

1 MICRON GAP

11111

J!JlIIPS

,/

"
,-'"
\,\

/

2 MlfR9NI GI~~

.I. .LU.IJ ~.\

4n7111m I
100

lk

10k

lOOk

FREQUENCY (Hz)

FIGURE 2.12.3 Effect of Head Gap on Response

Still another deleterious effect is due to the use of bias
current. High frequency bias current (typically - 70kHz) is
used in recording the audio signal to help correct for the
inherent nonlinearities of the magnetic material, improving
both distortion and signal-to-noise ratio. It is also used in
higher quality machines (at about 20dB higher levels) to
drive the erase head. The problem arises that a side effect
of the distortion minimizing record bias current is high
frequency erasure! The technical term is bias erasure. It
is more noticeable at high frequencies because they are put
onto the tape weaker and are more susceptible to being
erased.

Another area of serious high frequency loss is related more
to the formulation of the tape itself than to the dynamics
of recording. This is the fundamental problem of magnetic
saturation, i.e., as magnetic variations increase in intensity,
a point is reached where the tape begins to be saturated and
a subsequent drop-off in level occurs. The trade term used
to describe this effect is self-demagnetization and refers to
the fact that the recorded material effectively consists of
bar magnets in line with each other. The higher the
frequency, the more bar magnets are recorded per inch of
2-31

~+M~A+~-W~l~/~

15

~ 11 B - PREAMP
1I' 1/ ~mr1;

105

o
10 Hz

100 Hz

1 kHz

10 kHz

100 kHz

10 Hz

FIGURE 2.12.4 NAB Equalization Characteristic

100 Hz

1 kHz

Il.t r.
11w
3

10 kHz

f,

100 kHz

FIGURE 2.12.6 Recording Head & Preamp Response for NAB
Equalization

Of the many factors contributing to high frequency roll·
off, those due to self·demagnetization and bias erase are
the most troublesome. This makes universal equalization
difficult, since the qual ity of the tape used and proper
adjustment of bias current ultimately determine flat
response. Nevertheless, a standard does exist and is known
as NAB (National Association of Broadcasters) equalization
and appears as Figure 2.12.4. The four most used tape
speeds are given along with the necessary design frequencies.

MICRO~~)~

~

P
1 ....

INPUT

14 5)
M387A'

12.7)

R9

r

:,.'.

~ECOROING

jCE-.-J HEAD

R4

2.12.2 LM381 or LM387 Tape Record Preamp
When recording, the frequency response is the complement
of the NAB playback equalization, making the composite
record and playback response flat. Figure 2.12.5 shows the
record characteristic superimposed on the NAB playback
response.

FIGURE 2.12.7 Tape Recording Preamp

Section 2.6.) Resistor R6 and capacitor C2 set the mid·band
gain as before (Section 2.6). Capacitor C5 sets the high
frequency 3dB point, f3 (Figure 2.12.6). as:
30

~

(2.12.1)

25

1\

20
15

The preamp gain increases at 6dB/octave above f3 until
RS = XC5'

f-1-+l+HllI--+OO++lll-NAB
PLAYBACK

10

5

o

f-1-+l+HllI---lo<++lM-A+>-4+HI
f
J
II1!I{ f,
f-I-+~r.+JlIlF++H+lfIH'Illllt.-H4+AJJ.l+HI

RS = _ _1__
21T f4 C5

~~~~~~~~~~~

10Hz

100 Hz

1 kHz

(2.12.2)

10 kHz

where:
FIGURE 2.12.5 NAB Record & Playback Equalization

f4 = desired high frequency cutoff

Resistor Rg is chosen to provide the proper recording head
current.
Vo

Rg = iRECORD HEAD

The NAB record characteristic is the sum of the record head
response plus the record amplifier equalization response.
Design of record amplifiers therefore requires accurate
knowledge of the record head frequency response. The
difference between the head response and the NAB record
curve, then, determines the shape of the equalization
required of the amplifier.

(2.12.3)

L1 and C6 form a parallel resonant bias trap to present a
high impedance to the recording bias frequency and prevent
intermodulation distortion.
Example 2.12.1
A recorder having a 24 V power supply uses recording heads
requiring 30pA AC drive current. A microphone of 10mV
peak output is used. Single ended input is desired for
optimum noise performance.

Curve A of Figure 2.12.6 shows the response characteristics
of a typical laminated core, quarter·track head.
Curve B shows the required preamplifier response to make
the composite, A + B, provide the NAB recording charac·
teristic. This response is obtained with the circuit of Figure
2.12.7. Resistors R4 and R5 set the DC bias as before. (See

Solution
1. From Equation (2.6.5) let R5 = 1200n.
2·32

2. Equation (2.6.6):

24V

VCC
)
( - - -1 R5

R4

~

R4

~ (~4
_ 1) 1200
1.3

1.3

R4 ~ 2.09 x 104 '" 22kQ
1200

3. The maximum output of the LM381 is (VCC - 2V)p_p.
For a 24 V power supply, the maximum output is 22V p _p
or 7.8VRMS. Therefore, an output swing of 6VRMS is
reasonable.

~20"F~0.27"F

FIGURE 2.12.8 Typical Tape Recording Amplifier

From Equation (2.12.3).
2.12.3 LM387A or LM381 Tape Playback Preamp
R9

R9

The NAB response is achieved with the circuit of Figure
2.12.9. Resistors R4 and R5 set the DC bias and are chosen
according to section 2.6.

iRECORD HEAD
~

6V
-30llA

~

200kQ

4. Let the high frequency cutoff f4 ~ 16 kHz (Figure 2.12.6).
The recording head frequency response begins falling
off at approximately 4 kHz. Therefore, the preamp gain
must increase at this frequency to obtain the proper
composite characteristic. The slope is 6dB/octave for the
two octaves between f3 (4 kHz) and the cutoff frequency
f4 (16kHz). Therefore, the mid·band gain lies 12dB
below the peak gain.

II

We are allowing 6VRMS output voltage swing. Therefore,
the peak gain ~ 6V/l0mV ~ 600 or 55.6dB.
The mid-band gain

~

43.6dB or 150.

5. From Equation (2.6.9) the mid-band gain is:

FIGURE 2.12.9 NAB Tape Preamp

150
R4
R6 ~ 149

22 x 103
149

~

The reference gain of the preamp, above corner frequency
f2 (Figure 2.12.4). is set by the ratio:
147.7
OdB reference gain

R6"" 150Q

~

R7 + R6
--R6

(2.12.4)

The corner frequency f2 (Figure 2.12.4) is determined
where XC4 ~ R 7 and is given by:

6. Equation (2.6.10):
2.12 x 10-5
6.28 x 50 x 150

(2.12.5)
7. Equation (2.12.1):

Corner frequency f1 is determined where XC4

~

R4:
(2.12.6)

6.28 x 4 x 103 x 150
2.66 x 10- 7

The low frequency 3dB roll-off point, fo, is set where
XC2 ~ R6:

C5 '" 0.27 IlF
(2.12.7)

8. Equation (2.12.2):
Example 2.12.2

6.28 x 16 x 10 3 x 2.7 x 10-7

Design a NAB equalized preamp for a tape player requiring
O.5VRMS output from a head sensitivity of 800llV at
1 kHz, 3% IPS. The power supply voltage is 24 V and the
differential input configuration is used.

36.8
R8 '" 33Q
2-33

Solution
~

1. From Equation (2.6.3) let R5

An example of a LM387 A tape playback preamp designed
for 12 volt operation is shown in Figure 2.12.11 along with
its frequency response.

240kQ.

2. Equation (2.6.4):
R4 = (VCC -1)R5
2.6
R4

l"F
~+12V
I-r~+ (6)

~ (24
_ 1)2.4 x 105
2.6

LM387A
(2.7)

R4 ~ 1.98 x 16 6 "" 2.2MQ

l~o

C4 _ _ _1 _ = _ _ _-'-1_ __
1T

fl R4

.".

6.28 x 50 x 2.2 x 106

4. From Figure 2.12.4, the corner frequency f2 ~ 1770 Hz
at 3-3/4 IPS. Resistor R7 is found from Equation
(2.12.5).

220k

2O "F
.".

65
60

"

55
50

~

~04~

(a) NAB Tape Circuit

= 1.44 x 10-9 "" 1500pF.

R7

i

(4,5t-o

(3)

680k

3.3k

3. For a corner frequency, fl, equal to 50 Hz, Equation
(2.12.6) is used.

2

-

_ _ _----'1_ _ _ _ ~ 6 x 104
6.28 x 1770 x 1.5 x 10-9

=

45

"~

40

N~B}LAJAJ'\

I'\.

35

-

.....

30
25
20
15
20

5. The required voltage gain at 1 kHz is:

50 100200 500 lk 2k

5k 10k 20k

FREQUENCY (Hz)

AV =

0.5VRMS ~ 6.25 x 102V/V ~ 56dB
800,uVRMS

(b) Frequency Response of NAB Circuit

6. From Figure 2.12.4 we see the reference frequency gain,
above f2, is 5dB down from the 1 kHz value or 51 dB
(355V/V).

FIGURE 2.12.11 LM387 Tape Playback Preamp

From Equation (2.12.4):
OdB Ref Gain

2.12.4 Fast Turn-On NAB Tape Playback Preamp

R7 + R6

~

The circuit shown in Figure 2.12.10 requires approximately
5 seconds to turn on for the gain and supply voltage chosen
in the example. Turn-on time can closely be approximated
by:

- - - = 355

R6
R7
62k
= R6 = 355 _ 1
354

175

~

2.4
tON "" -R4 C2 In ( 1 - -)
VCC
7. For low frequency corner fa

~

40Hz, Equation (2.12.7):

As seen by Equation (2.12.8), increasing the supply
voltage decreases turn-on time. Decreasing the amplifier
gain also decreases turn-on time by reducing the R4C2
product.

_ _ _1____ = 2.21 x 10-5
6.28 x 40 x 180

J
II

Where the turn-on time of the circuit of Figure 2.12.9 is too
long, the time may be shortened by using the circuit of
Figure 2.12.12. The addition of resistor R D forms a voltage
divider with R6'. This divider is chosen so that zero DC
voltage appears across C2. The parallel resistance of R6' and
RD is made equal to the value of R6 found by Equation
(2.12.4). In most cases the shunting effect of RD is
negligible and R6' ~ R6.

1(CI1,8),:,2t:)

800"V @
1kHz

I+LM~>,(4::c.5::L)_>-C

(2.7)1-7

~3) I
-

L-1::

1

(2.12.8)

0.5VRMS

2.2M
1500 F
62k

It

For differential input, RD is given by:
(2.12.9)

FIGURE 2.12.10 Typical Tape Playback Amplifier

2-34

Equations (2.12.4), (2.12.5), and (2.12.7) describe the
high frequency gain and corner frequencies f2 and fo as
before.

For single ended input:
RD =

(VCC - 0.6) R6'

(2.12.10)

0.6

Frequency fl now occurs where XC4 equals the composite
impedance of the R4, R6, C2 network as given by Equation
(2.12.11).

In cases where power supply ripple is excessive, the circuit
of Figure 2.12.12 cannot be used since the ripple is coupled
into the input of the preamplifier through the divider.
The circuit of Figure 2.12.13 provides fast turn-on while
preserving the 120dB power supply rejection.

(2.12.11)

The DC operating point is still established by R41R5.
However, Equations (2.6.3) and (2.6.5) are modified by a
factor of 10 to preserve DC bias stability.
The turn-on time becomes:

,.------4.-------- Vee

. ~-.
2.4)
tON"" -2y
R4C2 In ( 1 - VCC

(2.12.12)

Example 2.12.3
Design an NAB equalized preamp with the fast turn-on
circuit of Figure 2.12.13 for the same requirements as given
in Example 2.12.2.

II

Solution
1. From Equation (2.6.3a) let R5 = 24 k£1.
2. Equation (2.6.4):
R4 =(VCC_ 1)R5
2.6

FIGURE 2.12.12 Fast Turn·On NAB Tape Preamp

=

(2.4.
- 1) 24 x 103
2.6

1.98 x 10 5

3. From Example 2.12.2, the reference frequency gain,
above f2, is 51 dB or 355VIV.

(7 8)

Equation (2.12.4):
C4

II

R7 + R6 = 355
R6

R4

4. ThS corner frequency f2 is 1770 Hz for 3-3/4 IPS.
R8

R6

Equation (2.12.5) :

R5

"::'

1: C2

*,C2
"::'

5. The corner frequency f1 is 50 Hz and is given by
Equation (2.12.11).

FIGURE 2.12.13 Two-Pole Fast Turn-On NAB Tape Preamp

C4

For differential input, Equation (2.6.3) is modified as:

=
21T f1 R6[(R4:6 R6y -

2VBE

1.2

100lQ2

50 x 10-6

(2.6.3a)

~

6. Solving Equations (2.12.4), (2.12.5), and (2.12.11) simultaneously gives:

24k£1 maximum
(2.12.13)

For single ended input:
VBE

0.6

50lFB

50 x 10-4

f2 (Ref Gain)
2.2 x 10 5 (50 + V2500 + 50 x 1770 x 355)

(2.6.5a)

1770 x 355

120£1 maximum

1.98 x 103
2-35

7. From Equation (2.12.4):
Vs '" +33V

R7 = 354R6 = 708 x 103
R7 "" 680kn
8. Equation (2.12.5):

6.2!! x 1770 x 680 x 10 3
C4 = 1.32 x 10- 10 "" 120pF
9. Equation (2.12.7):
R3
BIAS
2.5k

6.28 x 40 x 2 x 103

"METAL FILM, 1% TOLERANCE

C2 = 1.99 x 10-6 '" 2f.lF
This circuit is shown in Figure 2.12.14 and requires only
0.1 seconds to turn on.

FIGURE 2.12.15 LM381A Ultra-Low Noise Tape Preamp
(NAB, 1-718 & 3-3/4 IPSI

2.12.6 lM382 Tape Playback Preamp
With just one capacitor in addition to the gain setting
capacitors, it is possible to design a complete low noise,
NAB equalized tape playback preamp (Figure 2.12.16). The
circuit is optimized for automotive use, i.e., Vs = 10-15V.
The wideband OdB reference gain is equal to 46dB (200V IV)
and is not easily altered. For designs requiring either gain or
supply voltage changes the required extra parts make
selection of a LM387 a more appropriate choi ceo

24V

17,81

120 pF

II

220k

+12V

2k
24k

FIGURE 2.12.14

FIGURE 2.12.16 LM382 Tape Preamp (NAB, 1-7/8 & 3-3/4 IPS)

2.12.5 lM381A Ultra low Noise Tape Playback Preamp

2.12.7 lM1303 Tape Playback Preamp

Optimum noise performance will be obtained by using a
LM381 A biased single-ended, with the current density
increased per instructions given in Section 2.7. A typical
circuit (Figure 2.12.15) is shown for the popular tape speeds
of 1-7/8 and 3-3/4 IPS. Metal film resistors should be used
where indicated to reduce excess noise. The OdS reference
gain is 41 dB and produces an output level equal to 200mV
from a head output of 1 mV at 1 kHz. Notice that the twopole fast turn-on configuration has not been used. While it
could be used, its advantages are not as evident in single
ended biasing schemes since turn-on is inherently faster due
to the lower voltage required at pin 3 (- 0.5 V compared to
- 1.2V for differential scheme). The high supply voltage
also results in faster turn-on as discussed earlier. Figure
2.12.15 requires approximately 0.6 seconds to turn on.

For split supply applications, the LM1303 may be used as a
tape preamp as shown in Figure 2.12.17. Design equations
are given below for trimming or alteration purposes.
(Frequency points refer to Figure 2.12.4.)
OdB Ref Gain

(2.12.14)

(2.12.15)

(2.12.16)

2-36

As shown, the OdB reference gain equals 34dB. Due to the
limited open loop gain of the LM1303, this should be
treated as a maximum value allowed.

+24V

vcc
+

::r:

'Op

O.OO47pF

VOUT

-=

f'
-=

R,'
lBOk

R,'

-=

R3
'DO

+C2

1

* FOR 7·1(2 & 151PS

100 1-1

Av " 52dB

SUBSTITUTE

-=

-=

lI- _

Rl '" R4 = 330k
Cl =- O.D1J.lF

METAL FILM

NOISE:

~69dB

BELOW

2mV (-121dBm)
THO >(, 0.1%

VEE

(al LM381 AS. E. Bias

FIGURE 2.12.17 LM1303 Tape Preamp (NAB, '-7/8 & 3-3/41PSI

+24V

2.13 MIC PREAMPS
2.13.1 Introduction
Microphones classify into two groups: high impedance
(~ 20kQl, high output (~ 200mVI; and low impedance
(~ 200QI, low output (~ 2mVI. The first category places
no special requirements upon the preamp; amplification is
done simply and effectively with the standard non-inverting
or inverting amplifier configurations. The frequency respon5e
is reasonably flat and no equalization is necessary. Hum and
noise requirements of the amplifier are minimal due to the
large input levels. If everything is so easy, where is the
hook? It surfaces with regard to hum and noise pickup of
the microphone itself. Being a high ·impedance source, these
mics are very susceptible to stray magnetic field pickup
(e.g., 60 Hz), and their use must be restricted to short
distances (typically less than ten feet of cable length).
Because of this problem, high impedance mics are rarely
used.

VOUT

R,
l20k

+

C2

I'D"
Av = 52dB
* - METAL FILM
NOISE: -67dB BELOW
2mVH19dBml
THO";; 0.1%

Low impedance microphones also have a flat frequency
response, requiring no special equalization in the preamp
section. Their low output levels do, however, impose rather
stringent noise requirements upon the preamp. For a signal·
to-noise ratio of 65dB with a 2mV input signal, the total
equivalent input noise (EIN) of the preamp must be 1.12pV
(10·lOk Hz). National's line of low noise dual preamps with
their guaranteed EIN of';; 0.7pV (LM381AI and';; 0.9pV
(LM387A) make excellent mic preamps, giving at least
67dB SIN (LM387AI performance (re: 2mV input levell,
or -119d8m.

(bl LM3B7A

FIGURE 2.13.1 Transformerless Mic Preamps for Unbalanced Inputs

2.13.2 Transformerless Unbalanced Designs
Low impedance unbalanced (or single-ended 1 mics may be
amplified with the circuits appearing in Figure 2.13.1. The
LM381 A (Figure 2.13.1 al biased single-ended makes a
simple, quiet preamp with noise performance -69 dB below
a 2mV input reference point. Resistors R4 and R5 provide
negative input bias current and establish the DC output
level at one-half supply. Gain is set by the ratio of R4 to
R2, while C2 establishes the low frequency -3dB corner.
High frequency roll-off is done with C3. Capacitor C1 is
made large to reduce the effects of lit noise currents at low

Low impedance mics take two forms: unbalanced two wire
output, one of which is ground, and balanced three wire
output, two signal and one ground. Balanced mics predominate usage since the three wire system facilitates
minimizing hum and noise pickup by using differential
input schemes. This takes the form of a transformer with a
center·tapped primary (grounded), or use of a differential
op amp. More about balanced mics in a moment, but first
the simpler unbalanced preamps will be discussed.
2-37

2.13.3 Transformer-Input Balanced Designs

frequencies. (See Section 2.6 for details on biasing and gain
adjust.)

Balanced microphones are used where hum and noise must
be kept at a minimum. This is achieved by using a three
wire system - two for signal and a separate wire for ground.
The two signal wires are twisted tightly together with an
overall shield wrapped around the pair, acting as the ground.
Proper grounding of microphones and their interconnecting
cables is crucial since all noise and hum frequencies picked
up along the way to the preamplifier will be amplified as
signal. The rationale behind the twisted-pair concept is that
all interference will be induced equally into each signal wire
and will thus be applied to the preamp common-mode,
while the actual transmitted signal appears differential.
Balanced-input transformers with center-tapped primaries
and single-ended secondaries (Figure 2.13.2) dominate
balanced mic preamp designs. By grounding the center-tap
all common-mode signals are shunted to ground, leaving the
differential signal to be transformed across to the secondary
winding, where it is converted into a single-ended output.
Amplification of the secondary signal is done either with
the LM381A (Figure 2.13.2a) or with the LM387A
(Figure 2.13.2b). Looking back to Figure 2.13.1 shows
the two circuits being the same with the exception of a
change in gain to compensate for the added gain of the
transformer. The net gain equals 52dB and produces
- OdBm output for a nominal 2 mV input. Selection of the
input transformer is fixed by two factors: mic impedance
and amplifier optimum source impedance. For the cases
shown the required impedance ratio is 200:10k, yielding a
voltage gain (and turns ratio) of about seven h/10k/200) .

The LM387A (Figure 2.13.1b) offers the advantage of
fewer parts and a very compact layout, since it comes in the
popular 8'pin minidip package. The noise degradation
referenced to the LM381 A is only +2dB, making it a desirable alternative for designs where space or cost are dominant
factors. Biasing and gain resistors are similar to LM381 A.
(See Section 2.8 for details.)
+24V

Cs

~0.1

VOUT

MIC
INPUT

Av '" 52dB
.. - METAL FILM

NOISE, -B6dS BElOW

Assuming an ideal noiseless transformer gives noise performance -86dB below a 2mV input level. Using a carefully
designed transformer with electrostatic shielding, rejection
of common-mode signals to 60dB can be expected (which
is better than the cable manufacturer can match the
twisting of the wires).

2mV (-138dBm)
THO":;; 0.1%

(a) LM381A S. E. Bias

+24V

+15V

;3

200n'10[pkn;~+

MIC
INPUT

-=

Your

R1*
10k

Mit

X7.0J

INPUT

R2*

AS"
27k

lk

0.1%

VOUT

A2*

lk
0.1%

R.
4.7k

>-:::L
-=

220k

R5
lOOk

-::Av '" 52dB
.. - METAL FILM
NOISE: -84dB BElOW
2mV {-136dBml
THO..;; 0.1%

-15V

Av" 52dB
* -METAL FILM
NOISE: -64dB BElOW

(b) LM387A

2mV{-115dBml
THO";; 0.1%

FIGURE 2.13.3 Transformerless Mic Preamp for Balanced Inputs

FIGURE 2.13.2 Transformer-Input Mic Preamps for Balanced Inputs

2-38

ances is not conducive to low noise design and should be
avoided.! The common·mode rejection ratio (CMRR) ofthe
LF357 is 100dB and Gan be viewed as the "best case"
condition, i.e., with a perfect match in resistors, the CMRR
will be 100dB. The effect of resistor mismatch on CMRR
cannot be overemphasized. The amplifier's ability to reject
common·mode assumes that exactlv the same signal is
simultaneously present at both the inverting and non·
inverting inputs (pins 2 and 3). Any mismatch between
resistors will show up as a differential signal present at the
input terminals and will be amplified accordingly. By using
0.1 % tolerance resistors, and adjusting R5 for minimum
output with a common·mode signal applied, a CMRR near
100dB is possible. Using 1% resistors will degrade CMRR to
about 80dB. The LF356 may be substituted for the LF357
if desired with only a degradation in slew rate (12V/lls vs.
50V/lls) and gain bandwidth (5MHz vs. 20MHz).

2.13.4 Transformerless Balanced Designs
Transformer input designs offer the advantage of nearly
noise·free gain and do indeed yield the best noise perfor·
mance for microphone applications; however, when the
total performance of the preamplifier is examined, many
deficiencies arise. Even the best transformers will introduce
certain amounts of harmonic distortion; they are very
susceptible to hum pickup; common·mode rejection is not
optimum; and not a small problem is the expense of quality
input transformers. For these reasons, transformerless
designs are desirable. By utilizing the inherent ability of an
operational amplifier to amplify differential signalS while
rejecting common·mode ones, it becomes possible to
eliminate the input transformer.
Figure 2.13.3 shows the FET input op amp, LF357 (selected
for its high slew rate and CMRR) configured as a difference
amplifier. As shown, with Rl; R2 and R3 ; R4 + R5 the
gain is set by the ratio of R3 to R 1 (see Appendix A4) and
equals 52dB. The LF357 is selected over the quieter
LM387A due to its high common·mode rejection capability.
The LM387A (or LM381A) requires special circuitry when
used with balanced inputs since it was not designed to reject
common·mode signals. (A design trade·off was made for
lower noise.) See Section 2.13.4.

Due to the thermal noise of the relatively large input
resistors the noise performance of the Figure 2.13.3 circuit
is poorer than the other circuits, but it offers superior hum
rejection relative to Figure 2.13.1 and eliminates the costly
transformer of Figure 2.13.2.

Input resistors R1 and R2 are made large compared to the
source impedance, yet kept as small as possible, to achieve
an optimum balance between input loading effects and
low noise. Making R1 + R2 equal to ten times the source
impedance is a good compromise value. Matching imped·

An improvement in noise performance over Figure 2.13.3 is
possible by using a LM387 A in front of the LF356 (or
LF357) as shown in Figure 2.13.4. This configuration is
known as an instrumentation amplifier after its main usage
in balanced bridge instrumentation applications. In this

2.13.5 Low Noise Transformerless Balanced Designs

+15V

I

C4
0.1

+15V

-=C5
O.lI

AI'

lk

5%

AS'
10k
S%

AU'
lk

A3

A12

50k
0.1%

lOOk
0.1%
A'0

0.1%

10k

MIC
INPUT

0.1%

A6'
10k
5%
A2'
lk

VOUT

A11

2.5k
AS'
lk
0.1%

I

+ C3
470pf

-=-A4
5Dk
0.1%

-lSV

5%

Av '" 54dB

* - METAL FILM
ADJ. R7 FOR VOUT = OVoc
ADJ. R14 FOR MAX CMRR
NOISE, -67dB BELOW
2mV INPUT H 19dBm)
THO" 0.1%

FIGURE 2.13.4 Low Noise Transformerless Balanced Mic Preamp

2·39

I
-=-

C6
0.1

design each half of the LM387 A is wired as a non-inverting
amplifier with bias and gain setting resistors as before.
Resistors R1 and R2 set the input impedance at 2 kO
(balanced). Potentiometer R7 is used to set the output DC
level at zero volts by matching the DC levels of pins 4 and 5
of the LM387 A.

relative magnitude of the signal has been reduced (or
increased) by 3dB.
Passive tone controls require "audio taper" (logarithmic)
potentiometers, i.e., at the 50% rotation point the slider
splits the resistive element into two portions equal to 90%
and 10% of the total value. This is represented in the
figures by "0.9" and "0.1" about the wiper arm.

This allows direct coupling between the stages, thus
eliminating the coupling capacitors and the associated
matching problem for optimum CMRR. AC gain resistors
R8 and R9 are grounded by the common capacitor, C3,
eliminating another capacitor and assuring AC gain match.
Close resistor tolerance is necessary around the LM387 A in
order to preserve common-mode signals appearing at the
input. The function of the LM387 A is to amplify the low
level signal adding as little noise as possible, and leave
common-mode rejection to the LF356.

111111111

By substituting a LM381 A and increasing its current density
(see Section 2.7) a professional quality transformerless
balanced mic preamp can be designed. With the exception
of the additional components necessary to increase the
curre(1t density, the circuit is the same as Figure 2.13.4. The
improvement in noise performance is 7dB, yielding noise
-74dB below a 2mV input level.

•

f2

FREQUENCY (H,)

eio--

REFERENCES

~~C1

BOOST

1. Smith, D. A. and Wittman, P. H., "Design Considerations
of Low-Noise Audio Input Circuitry for a Professional
Microphone Mixer," Jour. Aud. Eng. Soc., vol. 18, no. 2,
April 1970, pp. 140-156.

t

0,9 = a

R2

eo

(LOG) ~O.1 =1-0

l

CUT

C2

2.14 TONE CONTROLS

12=_1_=_1_
27rR3C2
211R1 C,

2.14.1 Introduction

ASSUME R2

There are many reasons why a user of audio equipment may
wish to alter the frequency response of the material being
played. The purist will argue that he wants his amplifier
"flat," i.e., no alteration of the source material's frequency
response; hence, amplifiers with tone controls often have a
FLAT position or a switch which bypasses the circuitry.
The realist will argue that he wants the music to reach his
ears "flat." This position recognizes that such parameters as
room acoustics, speaker response, etc., affect the output of
the amplifier and it becomes necessary to compensate for
these effects if the Iistener is to "hear" the music "flat,"
i.e., as recorded. And there is simply the matter of personal
taste (which is not simple): one person prefers "bassy"
music; another prefers it "trebley."

»

R1

;t>

R3

FIGURE 2.14.1 Bass Tone Control - General Circuit

I

1~

~

z

~

!\

Al/A1---

II

"3/"2

~

2.14.2 Passive Design

'3

Passive tone controls offer the advantages of lowest cost
and minimum parts count while suffering from severe
insertion loss which often creates the need for a tone recovery amplifier. The insertion loss is approximately equal
to the amount of available boost, e.g., if the controls have
+20dB of boost, then they will have about -20dB insertion
loss. This is because passive tone controls work as AC
voltage dividers and really only cut the signal.

eiO--

"1

2.14.3 Bass Control
The most popular bass control appears as Figure 2.14.1
along with its associated frequency response curve. The
curve shown is the ideal case and can only be approximated.
The corner frequencies fl and f2 denote the half-power
points and therefore represent the frequencies at which the

11 =

2rr~1 C1

'2 =

2rr~3C1

'3 =

2rr~2C1

ASSUME R2" "1 .. "3

FIGURE 2.14.2

2-40

Minimum~Parts

Bass Tone Control

changed, and gives analogous performance. The amount of
boost or cut is set by the following ratios:

For designs satisfying R2 ?> R1 ?> R3, the amount of
available boost or cut of the signal given by Figure 2.14.1 is
set by the following component ratios:

(2.14.7)

treble boost or cut amount
bass boost or cut amount

(2.14.1 )

The turnover frequency f2 occurs when the reactance of C1
equals R 1 and the reactance of C2 equals R3 (assuming
R2?> R 1 ?> R3):

Treble turnover frequency f1 occurs when the reactance of
C1 equals R 1 and the reactance of C2 equals R3:

(2.14.2)
__ 1

(2.14.3)

J

The frequency response will be accentuated or attenuated
at the rate of ±20dB/decade = ±6dB/octave (single pole
response) until f1 is reached. This occurs when the limiting
impedance is dominant, i.e., when the reactance of C1
equals R2 and the reactance of C2 equals R 1 :

~

_C,/Cz

z

~

\
(2.14.4)
FREOUENCY IHd

Note that Equations (2.14.1)-(2.14.4) are not independent
but all relate to each other and that selection of boost/cut
amount and corner frequency f2 fixes the reamining para·
meters. Also of passing interest is the fact that f2 is
dependent upon the wiper position of R2. The solid·line
response of Figure 2.14.1 is only valid at the extreme ends
of potentiometer R2; at other positions the response
changes as depicted by the dotted line response. The
relevant time constants involved are (1 - ~) R2C1 and
~R2C2, where ~ equals the fractional rotation of the wiper
as shown in Figure 2.14.1. While this effect might appear to
be undesirable, in practice it is quite acceptable and this
design continues to dominate all others.

R2
R1

R1

=-

R3

=

bass boost or cut amount

fZ

BOOST

t
~

'1 '"

0.9

IL~lJS·D·.l--""''''''''O.,

21f~3C2 '" Zll~lCl

f2,,_I_
Zrr Rle,

RZ y Rl

)0

R3

CUT

Figure 2.14.2 shows an alternate approach to bass tone
control which offers the cost advantage of one less capacitor
and the disadvantage of asymmetric boost and cut response.
The degree of boost or cut is set by the same resistor ratios
as in Figure 2.14.1.

-

• •

fl

FIGURE 2.14.3 Treble Tone Control- General Circuit

Cl

(2.14.5)

(2.14.8)

(2.14.9)
The boost turnover frequency f2 occurs when the reactance
of C1 equals R3:

The amount of available boost is reached at frequency f2
and is determined when the reactance of C1 equals R3.

(2.14.6)
(2.14.10)
Maximum boost occurs at fl, which also equals the cut
turnover frequency. This occurs when the reactance of Cl
equals Rl, and maximum cut is achieved where XC1 = R2.
Again, all relevant frequencies and the degree of boost or
cut are related and interact. Since in practice most tone
controls are used in their boost mode, Figure 2.14.2 is not
as troublesome as it may first appear.

In order for Equations (2.14.8) and (2.14.9) to remain
valid, it is necessary for R2 to be designed such that it is
much larger than either R1 or R3. For designs that will not
permit this condition, Equations (2.14.8) and (2.14.9) must
be modified by replacing the Rl and R3 terms with R111R2
and R311R2 respectively. Unlike the bass control, f1 is not
dependent upon the wiper position of R2, as indicated by
the dotted lines shown in Figure 2.14.3. Note that in the
full cut position attenuation tends toward zero without the
shelf effect of the boost characteristic.

2.14.4 Treble Control
The treble control of Figure 2.14.3 represents the electrical
analogue of Figure 2.14.1, i.e., resistors and capacitors inter·
2·41

~,

1-+~IHII-++HIftI"'/++tIttIt~++-tIttIIIC1/CZ

f

i

• •

+!

I,
IZ
FREQUENCY (Hz)

'i~

'i~

'9
C'iO
RZ

CZ

I

~eo

D.'

~

+

I, 13
IZ
FREQUENCY (Hz)

f1 =

IZ •

::1"1
t·,
09
•

2n-~2C2

cZ

Z.~ZC,

I

RL

12

~ Z':LC,

13

~

ZI,

ASSUMES RZ

--

FIGURE 2.14.4 Minimum-Parts Treble Tone Control

'0

I'~Z.R'LCZ

~

'0 RL

--

FIGURE 2.14.5 Effect of Loading Treble Tone Control

It is possible to omit R1 and R3 for low cost systems.
Figure 2.14.4 shows this design with the modified equations
and frequency response curve. The obvious drawback
appears to be that the turnover frequency for treble cut
occurs a decade later (for ±20dB designs) than the boost
point. As noted previously, most controls are used in their
boost mode, which lessens this drawback, but probably
more important is the effect of finite loads on the wiper of
R2·
Figure 2.14.5 shows the loading effect of RL upon the
frequency response of Figure 2.14.4. Examination of these
two figures shows that the presence of low impedance
(relative to R2) on the slider changes the break points
significantly. If RL is 1/10 of R2 then the break points
shift a full decade higher. The equations given in Figure
2.14.5 hold for values of R2;;;' 10 RL. A distinct advantage
of Figure 2.14.5 over Figure 2.14.4 is seen in the cut
performance. R L tends to pull the cut turnover frequency
back toward the boost corner - a nice feature, and with two
fewer resistors. Design becomes straightforward once R L is
known. C1 and C2 are calculated from Equations (2.14.11)
and (2.14.121.

Solution
1. For symmetrical controls, combine Figures 2.14.1 and
2.14.3.
BASS (Figure 2.14.1):
2. From Equation (2.14.1):

~

(-20dB)

10

f1

=

50 Hz and f2

=

500 Hz

3. Let R2 = 100k (audio taper).
4. From Step 2:
R1 =

R2

10

R1
R3 = 10

100k
10

10k

10k = 1 k
10

5. From Equation (2.14.2) and Step 2:
C1 = - - - 2 1T f2 RL

(2.14.11 )
C1 = _ _
1_
21T f2 R1
(2.14.12)
Use C1 = 0.0331*

Here again, gain and turnover frequencies are related and
fixed by each other.
Example 2.14.1
Design a passive, symmetrical bass and treble tone control
circuit having 20dB boost and cut at 50Hz and 10kHz,
relative to midband gain.
2-42

C2

lOC1

C2

0.33/J F

(21T)(500)(10k)

3.18x 10-8

TREBLE (Figure 2.14.3):

2.14.5 Use of Passive Tone Controls with LM387 Preamp

6. From Equation (2.14.7):

A typical application of passive tone controls (Figure 2.14.7)
involves a discrete transistor used following the circuit to
further amplify the signal as compensation for the loss
through the passive circuitry. While this is an acceptable
practice, a more judicious placement of the same transistor
results in a superior design without increasing parts count
or cost.

1

(-20dB)

10
fl = 1 kHz, f2 = 10kHz

Placi ng the transistor ahead of the LM387 phono or tape
preamplifier (Figure 2.14.8) improves the SIN ratio by
boosting the signal before equalizing. An improvement of at
least 3dB can be expected (analogous to operating a
LM381 A with single·ended biasing). The transistor selected
must be low'noise, but in quantity the difference in price
becomes negligible. The only precaution necessary is to
allow sufficient headroom in each stage to minimize
transient clipping. However, due to the excellent open·loop
gain and large output swing capability of the LM387, this is
not difficult to achieve.

7. Let R2 = 100k (audio taper).
8. Select R 1 = 10k (satisfying R2}> Rl and minimizing com·

ponent spread).
Then:
R
Rl _ 10k _ lk
3 =
10

10

9.

~rom

Equation (2.14.8) and Step 6:

(2rr)(lk)(10k)
Use Cl

=

= 1.59

An alternative to the transistor is to use an LM381 A
selected low-noise preamp. Superior noise performance
is possible. (See Section 2.7.) The large gain and output
swing are adequate enough to allow sufficient single-stage
gain to overcome the loss of the tone controls. Figure
2.14.9 shows an application of this concept where the
LM381 A is used differentially. Single-ended biasing and
increased current density may be used for even quieter
noise voltage performance.

x 10- 8

0.015!1F

C2 = lOCl

the completed design appears as Figure 2.14.6, where RI
has been included to isolate the two control circuits, and
Co is provided to block all DC voltages from the circuit insuring the controls are not "scratchy," which results from
DC charge currents in the capacitors and on the sliders. Co
is selected to agree with system low frequency response:

2.14.6 Loudness Control
A loudness control circuit compensates for the logarithmic
nature of the human ear. Fletcher and Munson! published
curves (Figure 2.14.10) demonstrating this effect. Without
loudness correction, the listening experience is characterized
by a pronounced loss of bass response accompanied by a
slight loss of treble response as the volume level is
decreased. Compensation consists of boosting the high and

C 1
= 7.17 x 10-8
0- (2rr)(20Hz)(10k+l00k+lk)
Use Co

=

O.l!1F

Co

';0--1

0

0.1

0.015:!::

10k

'0

~~
R,

lOOk
(lOG)

lOOk
(lOG)

10k

Mr
lk

~

-10

10k

-

Ull~~OJT

r-,

i'-

-20

-30

K

V

IIII I

1/2800ST

r-,\

i-":?

V

i-"

I-'

>1"-

1/2 CUT

II I I

lk

0.15

FMlfuJ

1

~

-40
10 Hz

100 Hz

1 kHz

10 kHz

100 kHz

Bass & Treble Tone Control Response

FIGURE 2.14.6 Complete Passive Bass & Treble Tone Control

2-43

SPEAKER

FIGURE 2.14.7 Typical Passive Tone Control Application

PASSIVE
TONE
CONTROLS

SPEAKER

FIGURE 2.14.8 Improved Circuit Using Passive Tone Controls

30V

U

+

I"F

47k

"::"

"::"

T

0.1
0.015

(7,8)

(2,13)

10k

2400"

1.2M

0.033

BASS

10k
10k

lOOk
lk

24012

VOLUME
50k

lOOk
0,33

tk

BALANCE

lOOk

1

T

+-0 TO POWER
AMP

O.15

TO CH 2

FIGURE 2.14.9 Single Channel of Complete Phono Preamp

low ends of the audio frequency band as an inverse function
of volume control setting. One commonly used circuit
appears as Figure 2.14.11 and uses a tapped volume pot
(tap @ 10% resistance), The switchable R·C network paral·
leling the pot produces the frequency response shown in
Figure 2.14,12 when the wiper is positioned at the tap point
(i.e" mid'position for audio taper pot). As the wiper is
moved further away from the tap point (louder) the
paralleling circuit has less and less effect, resulting in a
volume sensitive compensation scheme.

2.14.7 Active Design
Active tone control circuits offer many attractive advantages:
they are inherently symmetrical about the axis in boost and
cut operation; they have very low THD due to being incor·
porated into the negative feedback loop of the gain block,
as opposed to the relatively high THD exhibited by a tone
recovery transistor; and the component spread, i.e., range
of values, is low.

2-44

R1 + R2

AVB

140

=

;::

LEVE

~

lZ0

~

100

./ zo

BD

z.o

/.

w

g;

60

~

O.oz

Q

Z

./

ZO

~

0.002

~z

-~

100

ZO

---

(max bass cut)

(2.14.14)

R1 + R2

~

w

At very high frequencies the impedance of the capacitors is
small enough that they may be considered short circuits,
and the gain is controlled by the treble pot, being equal to
Equations (2.14.15) and (2.14.16) at the extreme ends of
travel.

'"

~
~

'"
Q.

O.oooz

0

~

AVB

Q

~

(2.14.13)

R1
R1

1

1;

O.z

40

- - - (max bass boost)

zoo

;

w

~

1000 Z 3 4510000
FREQUENCY IN Hz

AVT

~

R3+ R1+ 2R 5

(max treble boost)

(2.14.15)

(max treble cut)

(2.14.16)

R3

FIGURE 2.14.10 Fletcher·Munson Curves (USA). (Courtesy,
Acoustical Society of America)

R3

1
-~

~

R3+ R1+ 2R 5

AVT

Equations (2.14.15) and (2.14.16) are best understood by
recognizing that the bass circuit at high frequencies forms
a wye·connected load across the treble circuit. By doing a
wye·delta transformation (see Appendix A3), the effective
loading resistor is found to be (R1 + 2R5) which is in
parallel with (R3 + R4) and dominates the expression. (See
Figure 2.14.13b.) This defines a constraint upon R4 which
is expressed as Equation (2.14.17).
AVB --H~-Iffi-+-Htl.+I+tttHfill
...~++I-++IIH-- AVT

RZ
3.3k

FIGURE 2.14.11 Loudness Control

'i

lIAVIl-

It

fL

=~~ H++14!11!:+tttlttt-+ttHt1lf-:.HftlHH

=~: H-l+I-Iffi-+-!otH1tt-+l++lI!II--H'fIltlII
-Z8

Cl*

*Cl

rnll=m~ttI=+I+1
10

100

"

10k

-

CUT

BASS
Rz

Rl

~

ej'"

H-l+I-Iffi-+-t1-H1tt-+l+H!!II-++-fIltlII

~

-

BOOST
Rl

!

fH

FREQUENCY (Hz)

111111
11111
1111111
111111
1111
111111
VOLUME CONTROL MIO·POSITION
(10% RESISTANCE)
-18

fLB fHB

_l/AVT

~

R5

lOOk

FREQUENCY (Hd

T
C3

FIGURE 2.14.12 Loudness Control Frequency Response

R4
TREBLE

R3

-::-

~

BOOST
BASS

The most common active tone control circuit is the so·
called "Americanized" version of the Baxandall (1952)2
negative feedback tone controls. A complete bass and treble
active tone control circuit is given in Figure 2.14.13a. At
very low frequencies the impedance of the capacitors is
large enough that they may be considered open circuits,
and the gain is controlled by the bass pot, being equal to
Equations (2.14.13) and (2.14.14) at the extreme ends of
travel.

CUT
TREBLE

fL=_'_
ZrrRz C,

1
fH = 2 rrR 3 C3

fLB=Zrr~'Cl

fHB;:

AVB =

1+~
R,

ASSUMES RZ" Rl

21T(Rl+R~+2R5)C3
R1 +2 RS

AVT=I+~

ASSUMES R4 ,. Rl + R3 + 2 RS

FIGURE 2.14.13a Bass and Treble Active Tone Control

2-45

that the flat (or midband) gain is not unity but approxi·
mately ±2dB. This is due to the close proximity of the
poles and zeros of the transfer function. Another effect of
this close proximity is that the slopes of the curves are not
the expected ±6dB/octave, but actually are closer to ±4dBI
octave. Knowing that fL and fLB are 14 dB apart in
magnitude, and the slope of the response is 4dB/octave, it
is possible to relate the two. This relationship is given as
Equation (2.14.22).
(2.14.22)
Example 2.14.2

(a) High Frequency Max Treble Boost Equivalent Circuit

Design a bass and treble active tone control ci rcuit having
±20dB gain with low frequency upper 3dB corner at 30 Hz
and high frequency upper 3dB corner at 10kHz.

NO EFFECT ON
GAIN IF SOURCE
ZR, + R1Z/R5 _'MPEDANCE IS LOW.

r-----...A./II'v------,

11111111

..

+ZO
+17

,

+10
iii

+3
0
-3

;s
z

i:

111111111 1111111

'*~~~~~~PEIs~~WJ~~V~

...

..-

..

~

..

it.

.....

..

-10
-17

-zo

(Rl + 2 Rslll(RJ + R4)
R3 + Rl + 2 RS
Av '"
(Rl + 2 R51llRJ
'" - - R 3 - IFR4

~

!

!

IL

Rl+R3+2RS

!

IH
ILB
IHB
FREQUENCY IHzl

(b) High Frequency Circuit After Wye-Delta Transformation
ILB
IH
~. ~~10
IL
IHB

FIGURE 2.14.13b Development of Max Treble Gain

FIGURE 2.14.14 Relationship Between Frequency Breakpoints of

(2.14.17)

Active Tone Control Circuit

At low·to-middle frequencies the impedance of Cl decreases
at the rate of -6dB/octave, and is in parallel with R2, so
the effective resistance reduces correspondingly, thereby
reducing the gain. This process continues until the resistance
of Rl becomes dominant and the gain levels off at unity.

Solution
BASS DESIGN:

The action of the treble circuit is similar and stops when
the resistance of R3 becomes dominant. The design equa·
tions follow directly from the above.
Cl = - - 21TfLB Rl

assumes R2 ~ R 1

1. Select R2 = lOOk (linear). This is an arbitrary choice.
2. From Equation (2.14.13):
R2
AVB = 1 + - = 10 (+20dB)
Rl

(2.14.18)

(2.14.19)

R2

100k

10 - 1

9

1.11 x 104

Rl = 11k
(2.14.20)
3. Given fL = 30Hz and from Equations (2.14.22) and
(2.14.18):
(2.14.21)

fLB

The relationship between fL and fLB and between fH and
fHB is not as clear as it may first appear. As used here these
frequencies represent the ±3dB points relative to gain at
midband and the extremes. To understand their relationship
in the most common tone control design of ±20 dB at
extremes, reference is made to Figure 2.14.14. Here it is seen
what shape the frequency response will actually have. Note

= 10fL

Cl

300l;lz

(21T) (300)( 11 k)

4.82 x 10-8

Cl = 0.05pF
TREBLE DESIGN:
4. Let R5 = R 1 = 11 k. This also is an arbitrary choice.
2·46

impedance for the tone control circuit and creates a high
input impedance (100kQ) for the source. The LM349 was
chosen for its fast slew rate (2.5 VIllS), allowing undistorted,
full-swing performance out to > 25kHz. Measured THD
was typically 0.05% @ OdBm (O.77V) across the audio
band. Resistors R6 and R7 were added to insure stability at
unity gain since the LM349 is internally compensated for
positive gains of five or greater. R6 and R7 act as input
voltage dividers at high frequencies such that the actual
input-to-output gain is never less than five (four if used
inverting). Coupling capacitors C4 and C6 serve to block DC
and establish low-frequency roll-off of the system; they
may be omitted for direct-coupled designs.

5. From Equation (2.14.15):
AVT = 1 +

R1 + 2 R5

= 10 (+20dB)

R3
R) + 2R5

11k + 2(11k)

10 - 1

9

3.67 x 10 3

R3 = 3.6k
6. Given fH = 10kHz and from Equation (2.14.20):

(211)(10kHz)(3.6k)

4.42 x 10-9

2.14.8 Alternate Active Bass Control
Figure 2.14.16 shows an alternate design for bass control,
offering the advantage of one less capacitor while retaining
identical performance to that shown in Figure 2.14.13. The
development of Figure 2.14.16 follows immediately from
Figure 2.14.13 once it is recognized that at the extreme
wiper positions one of the C1 capacitors is shorted out and
the other bridges R2.

7. From Equation (2.14.17):

;;;, 10(3.6k+11k+22k)
;;;, 3.66 x 10 5
R4

The modifications necessary for application with the
LM387 are shown in Figure 2.14.17 for a supply voltage of
24 V. Resistors R4 and R5 are added to supply negative
input bias as discussed in Section 2.8. The feedback coupling
capacitor Co is necessary to block DC voltages from being
fed back into the tone control circuitry and upsetting the
DC bias, also to insure quiet pot operation since there are
no DC level changes occurring across the capacitors, which

500k

The completed design is shown in Figure 2.14.15, where the
quad op amp LM349 has been chosen for the active
element. The use of a quad makes for a single IC, stereo
tone control circuit that is very compact and economical.
The buffer amplifier is necessary to insure a low driving

~JFT o---j

BASS
lOOk

+

C4
0.1

11k

C5
1"

11k
+15V

0.05

-= -=

I

11k

DUPLICATE FOR RIGHT CHANNEL

T
3.6k

0.005

500k
TREBLE

LEFT
OUT

R7
750
J.6k

-=
-lSV

TH61~~~\%

+20

L BOO

V

OdBm LEVel
10Hz - 50kHz

+15
+10

'\

+5

FlA
~
r.!I

-5

/

-10

-IS
-20

III~Ull ~ilI
10

100

1k

10k

lOOk

FREQUENCY IHzl

FIGURE 2.14.15 Typical Active Bass & Treble Tone Control with Buffer

2-47

O.l

-=

would cause "scratchiness." The R7-C3 network creates
the input attenuation at high frequencies for stability.

While the additional circuitry appears simple enough, the
resultant mathematics and design equations are not. In the
bass and treble deSign of Figure 2.14.13 it is possible to
include the loading effects of the bass control upon the
treble circuit, make some convenient design rules, and
obtain useful equations. (The treble control offers negligible
load to the bass circuit.) This is possible, primarily because
the frequencies of interest are far enough apart so as not
to interfere with one another. Such is not the case with the
midrange included. Any two of the controls appreciably
loads the third. The equations that result from a detailed
analysis of Figure 2.14.18 become so complex that they are
useless for design. So, as is true with much of real-world
engineering, design is accomplished by empirical (Le., trialand-error) methods. The circuit of Figure 2.14.18 gives the
performance shown by the frequency plot, and should be
optimum for most applications. For those who feel a
change is necessary, the following guidelines should make it
easier.

For other supply voltages R4 is recalculated as before,
leaving R5 equal to 240kD. It is not necessary to change
R7 since its value is dictated by the high frequency equivalent impedance seen by the inverting input (equals 33kD).
2.14.9 Midrange Control
The addition of a midrange control which acts to boost or
cut the midrange frequencies in a manner similar to the bass
and treble controls offers greater flexibility in tone control.
The midrange control circuitry appears in Figure 2.14.18.
It is seen that the control is a merging together of the bass
and treble controls, incorporating the bass bridging capacitor and the treble slider capacitor to form a combined
network. If the bass control is, in fact, a low pass filter, and
the treble control a high pass filter, then the midrange is a
combination of both, i.e., a bandpass filter.

1. To increase (or decrease) midrange gain, decrease
(increase) R6. This will also shift the midrange center
frequency higher (lower). (This change has minimal effect
upon bass and treble controls.)

R4

2. To move the midrange center frequency (while preserving gain, and with negligible change in bass and treble
performance), change both C4 and C5. Maintain the
re1ationship that C5 "" 5C4. Increasing (decreasing) C5
will decrease (increase) the center frequency. The
amount of shift is approximately equal to the inverse
ratio of the new capacitor to the old one. For example,
if the original capacitor is C5 and the original center
frequency is fo, and the new capacitor is C5' with the
new frequency being fo', then

R3

BASS

fL=-'211' R2 Cl

ILB

C5'

h~' C,

=

AVB = ,

C5

The remainder of Figure 2.14.18 is as previously described
in Figure 2.14.15.

IH=-'-

hR3 C3

fHB

AVT

=

fo'

+~

R,

TREBLE

fo

""

The temptation now arises to add a fourth section to the
growing tone control circuitry. It should be avoided. Three
paralleled sections appears to be the realistic limit to what
can be expected with one gain block. Beyond three, it is
best to separate the controls and use a separate op amp with
each control and then sum the results. (See Section 2.17 on
equalizers for details.)

211(Rl+R~+2R5)C3

= ,+R,+ 2R 5
R3

ASSUMES R4 .. Rt< R3 + 2 R5

FIGURE 2.14.16 Alternate Bass Design Active Tone Control

+24V

A,
11k
11k

T
C3

R7
3.3k

o005
.

R3
3.6k

500k

TREBLE

R3
3.6k

C3

1:0.002
-=-

-=-

-=-

-=-

FIGURE 2.14.17 LM387 Feedback Tone Controls

2-48

...L

C,

LEFT o-ll-'W\-..---i

IN

C7

>_"'I~--">IY'_"""W""""''''''---'l>IY'v----,BASS

0.1

C6

'"

Rl

Rt

11k

11k
C4

11k

0.005

..."";.........-'IIIt'v-_-+-.I\II/V-----..... MIDRANGE
R6
3.6k

3.6k

R6

R3
1.Bk

1.8k

DUPLICATE FOR RIGHT CHANNEl
TREBLE

R3
+15V

1C3
0.005
lEFT
OUT

RB
270

1
-::-

~lO"

ClJ"

+5

II"
It-..

:s
z

~

-::-

m\

+10

C]) All CONTROLS FLAT
ClJ BASS & TREBLE BOOST, MID flAT
C!l BASS & TREBLE CUT, MID flAT
® M'O BOOST, BASS & TREBLE FLAT
@ MID CUT, BASS & TREBLE FLAT

-::-

III

+20
,,5

iii

-15V

Cs
0,001

-5

C!l.(

-'0

i5>

-15
-20

V
10

100

10k

lk

lOOk

FREQUENCY (Hz)

FIGURE 2.14.18 Three Band Active Tone Control (Bass, Midrange & Treble)

REFERENCES

1. Fletcher, H., and Munson, W, A., "Loudness, Its Defini·
tion, Measurement and Calculation," J. Acoust. Soc, Am.,
vol. 5, p, 82, October 1933,
2. Baxandall, p, J., "Negative Feedback Tone Control Independent Variation of Bass ·and Treble Without
Switches," Wireless World, vol. 58, no, 10, October 1952,
p.402,
2.15 SCRATCH, RUMBLE AND SPEECH FILTERS

2.15.2 Definition of Wc and Wo for 2-Pole Active Filters

2.15.1 Introduction

When working with active filter equations, much confusion
exists about the difference between the terms Wo and wc.
The center frequency, fo, equals wo/2IT and has meaning
only for bandpass filters. The term Wc and its associated
frequency, fc, is the cutoff frequency of a high or low pass
filter defined as the point at which the magnitude of the
response is -3dB from that of the passband (i.e., 0.707
times the passband value). Figure 2.15.1 illustrates the two
cases for two-pole filters.

Infinite·gain, multiple-feedback active filters using LM387
(or LM381) as the active element make simple low-cost
audio filters. Two of the most popular filters found in
audio equipment are SCRATCH (low pass), used to roll
off excess high frequency noise appearing as hiss, ticks and
pops from worn records, and RUMBLE (high pass), used to
roll off low frequency noise associated with worn turntable
and tape transport mechanisms. By combining low and high
pass filter sections, a broadband bandpass filter is created
such as that required to limit the audio bandwidth to
include only speech frequencies (300Hz·3kHz)

Equally confusing is the concept of "Q" in relation to high
and low pass two-pole active filters. The design equations
contain Q; therefore it must be determined before a filter
2-49

can be realized - but what does it mean? For bandpass
filters the meaning of a is clear; it is the ratio of the center
frequency, fo, to the -3dB bandwidth. For low and high

Always use Equations (2.15.1 )-(2.15.3) (or Table 2.15.1)
when a equals anything other than 0.707.
2.15.3 High Pass Design

pass filters, Q only has meaning with regard to the amount
of peaking occurring at fo and the relationship between the
-3d8 frequency, fe, and f o.

An LM387 configured as a high·pass filter is shown in
Figure 2.15.2. Design procedure is to select R2 and R3 per
Section 2.8 to provide proper bias; then, knowing desired
passband gain, Ao , the a and the corner frequency fc, the
remaining components are calculated from the following:

The relationship that exists between Wo and Wc follows:
High Pass
Low Pass

(2.15.1)

Calculate Wo from Wc = 21Tfc and a using Equations
(2.15.1) and (2.15.3) (or Table 2.15.1).

(2.15.2)

Wc = [3 Wo

Let Cl = C3
Then:

(2.15.3)

a

C1 = - - (2Ao+l)
Wo R2

A table showing various values of [3 for several different
values of a is provided for convenience (Table 2.15.1).
Notice that Wc = Wo only for the Butterworth case
(0 = 0.707). Since Butterworth filters are characterized by
a maximally flat response (no peaking like that diagrammed
in Figure 2.15.1 L they are used most often in audio systems.

~

C2

Rl

(2.15.4)

Cl

(2.15.5)

Ao
(2.15.6)

OWo Cl (2Ao + 1)

AO
RZ
FREOUENCY

C1
VIN

la) High Pass

C3

o-jH-l!-....-~
VOUT

z

~ AOr-----,

FIGURE 2.15.2 LM387 High Pass Active Filter

FREQUENCY
0.0033

Ib) Low Pass

2M
r-~P-o() +24V

FIGURE 2.15.1 Definition of we for Low and High Pass Filters

>:---.....-0 VOUT

TABLE 2.15.1 Wc vs. Q

Q

Wc
Low-Pass

Wo
High-Pass

0.707*
1
2
3
4
5
10
100

1.000wo
1. 272wo
1.498wo
1. 523wo
1. 537wo
1.543wo
1. 551wo
1.554wo

1.000wo
0.786wo
0. 668wo
0.657wo
0.651wo
0. 648wo
0. 645wo
0.644wo

fe'" 50Hz
SLOPE = -1ZdB/OCTAVE

Ao

=

-1

THO" 0.1%

FIGURE 2.15.3 Rumble Filter Using LM387

* Butterworth

Example 2.15.1
Design a two-pole active high pass filter for use as a rumble
filter. Passband gain, Ao = 1, a = 0.707 (Butterworth) and
corner frequency, fc = 50Hz. Supply Vs = +24V.

Substitution of fc for fo in Butterworth filter design
equations is therefore permissible and experimental results
will agree with calculations - but only for Butterworth.
2-50

Solution

1. Select R3 = 240k.

(2.15.131

2. From Section 2.8,
R2 =( Vs _ 1) R3 = (24 _ 1) 240k
2.6'
2.6

1.98 x 106
Example 2.15.2

Use R2 = 2M

Design a two-pole active low-pass filter for use as a scratch
filter. Passband gain, Ao = 1, Q = 0.707 (Butterworthl and
corner frequency fc = 10kHz. Supply Vs = +24V.

3. Since 0= 0.707, Wo = wc= 21Tfc (see Table 2.15.1).
4. Let Cl = C3.

Solution

5. From Equation (2.15.4):

1. From Equation (2.15.8):
(0.707)(2 + 1)
(21T)(50)(2 x 10 6 )

Cl =

3.38 x 10-9

K=

0.25
(4)(0.7071 2 (1 + 1)

Use Cl = C3 = 0.0033pF

2. Select Cl = 560pF (arbitrary choice).
6. From Equation (2.15.5):

3. From Equation (2.15.9):

Cl
C2 = - = Cl = 0.0033pF

C2 = KCl = (0.25)(560pF)

140pF

(1 )

Use C2 = 150pF

7. From Equation (2.15.6):

4. Since 0 = 0.707, Wo = Wc = 21Tfc (see Table 2.15.1).
Rl =

1

5. From Equation (2.15.101:

(0.707) (21T)(50) (0.0033 x 10-6 )(2 + 1)
= 45.5 x 104

R2 = (2)(0.707)(21T)(101kHZ)(560PF)(0.251

Use Rl = 470kD..

Use R2 = 82k

The final design appears as Figure 2.15.3. For checking and
trimming purposes Equation (2.15.7) is useful:
fc = . - - - - 21T Cl

6. From Equation (2.15.111:
82k
R3 = = 41k
2

(2.15.7)

v'R1R2

Use R3 = 39k

Capacitor C4 = 0.01 is included to guarantee high frequency
stability for unity gain designs (required for Ao .;; 10).

7. From Equation (2.15.121:
R2
Rl = = R2 = 82k

2.15.4 Low Pass Design

1

The low pass configuration for a LM387 is shown in Figure
2.15.4. Design procedure is almost the reverse of the high
pass case since biasing resistor R4 will be selected last.
Knowing Ao , 0 and fc, proceed by calculating a constant K
per Equation (2.15.8).
K =

1
402 (Ao + 1)

8. From Equation (2.15.13):

R - 82k + 39k = 14.7k
4 - (24 _ 1\
2.6
/

(2.15.81

Arbitrarily select Cl to be a convenient value.
Then:

C2 = KCl

The complete design (Figure 2.15.5) includes C3 for
stability and input blocking capacitor C4. Checking and
trimming can be done with the aid of Equation (2.15.14).

(2.15.9)

Calculate Wo from Wc = 21Tfc and 0 using Equations
(2.15.11 and (2.15.31 (or Table 2.15.11.

(2.15.141

Then:
R2 =

R3

Rl

80.4k

(2.15.10)

2.15.5 Speech Filter

2 OWo Cl K
R2
Ao+ 1
R2

A speech filter consisting of a highpass filter based on
Section 2.15.2, in cascade with a low pass based on Section
2.15.3, is shown in Figure 2.15.6 with its frequency response
as Figure 2.15.7. The corner frequencies are 300 Hz and
3kHz with roll-off of -40dB/decade beyond the corners.
Measured THD was 0.07% with a OdBm signal of 1 kHz.
Total output noise with input shorted was 150pV and is

(2.15.11 )

(2.15.121

Ao
2-51

82k

R2

+24 V
C4
V,N

o-ft-'I/III~~M""'-t---,;i
0.1

VOUT

82k

39k

>;;---4....a VOUT

R4
15k

-=-

Ao

=

-1

THO" 0.1%

FIGURE 2.15.4 LM387 Low Pass Active Filter

FIGU'!E 2.15.5 Scratch Filter Using lM387

R20
560pF
C,

V,N

r--'\IRV2........._ _ _ _ _...

-_--t
2M

C3

C20

270k
C,

15DpF

R,O

R30

270k

130k

o-!HHf....
560pF

56DpF

RI
430k

240k

VOUT

R40

R3
C10

FI

47k

560 P

-=-

+24V

-=-

FIGURE 2.15.6 Speech Filter (300Hz·3kHz Bandpass)

2.16 BANDPASS ACTIVE FI LTERS

;;;

'""
~

-10

Narrow bandwidth bandpass active filters do not require
cascading of low and high pass sections as described in
Section 2.15.4. A single amplifier bandpass filter using the
LM387 (Figure 2.16.1) is capable of Q .;;; 10 for audio
frequency low distortion applications. The wide gain band·
width (20MHz) and large open loop gain (104dB) allow
high frequency, low distortion performance unobtainable
with conventional op amps.

I

-20
-30
-40

~IIIIIII

10

100

THO = 0.07 %~
11~~,f @OdBm

111111

111111

Beginning with the desired fo, Ao and Q, design is straight·
forward. Start by selecting R3 and R4 per Section 2.8,
except use 24 kn as an upper limit of R4 (instead of
240kn). This minimizes loading effects of the LM387 for
high Q designs.

IIIIII IIIIII
lk

10k

lOOk

FREQUENCY (Hz)

Let C1 = C2. Then:
FIGURE 2.15.7 Speech Filter Frequency Response

R1 =
due mostly to thermal noise of the resistors, yielding SIN
of 74dBm. The whole filter is very compact since the
LM387 dual preamp is packaged in the 8'pin minidip,
making tight layout possible.

R3

(2.16.1)

2Ao
Q

C1

2·52

Ao Wo R1

(2.16.2)

o

R2 =

(2.16.3)

4. From Equation (2.16.1):

(20 2 - Ao) Wo Cl
200k

For checking and trimming, use the following:
R3

Ao

fo

2

(2.16.4)

Rl = lOOk

2 Rl

j

1

lOOk

5. Let Cl = C2; then, from Equation (2.16.2):
Rl + R2

o

(2.16.5)

Cl = - - Aowo R,

R1R2 R3

2rrCl

(2.16.6)

0= 2.woR3Cl
2

10

796pF

(1)(2rr)(20k)(1 x 105 )

Use C, = 820pF
6. From Equation (2.16.3):

Ct

R2 =

o
(202 - Ao) Wo Cl

Cz

Rt

VtN

10

o--'Vvv-.-; I-+---:i

488n

[(2)(10)2 - 1) (2rr)(20k)(820pF)

VOUT

Use R2 = 470n

RZ

The final design appears as Figure 2.16.2. Capacitor C3 is
used to AC ground the positive input and can be made equal
to O.lIlF for all designs. Input shunting capacitor C4 is
included for stability since the design gain is less than 10.

FIGURE 2.16.1 LM387 Bandpass Active Filter

8Z0pF r--'VRy3 V-_ _ _ _ _ _.....

2.17 OCTAVE EQUALIZER

+24V

V,N

Rt

Cz

tOOk

8Z0pF

o--'VVV....;t-....-

An octave equalizer offers the user several bands of tone
control, separated an octave apart in frequency with independent adjustment of each. It is designed to compensate
for any unwanted amplitude-frequency or phase-frequency
characteristics of an audio system.

...--:;i

>.:---....OVOUT

Example 2.16.1

The midrange tone control circuit described in Section 2.14
can be used separately to make a convenient ten band
octave equalizer. Design equations result from a detailed
analysis of Figure 2.17.1, where a typical section is shown.
Resistors R3 have been added to supply negative input DC
bias currents, and to guarantee unity gain at low frequencies.
This circuit is particularly suited for equalizer applications
since it offers a unique combination of results depending
upon the slider position of R2. With R2 in the flat position
(i.e., centered) the circuit becomes an all-pass with unity
gain; moving R2 to full boost results in a bandpass characteristic, while positioning R2 in full cut creates a bandreject (notch) filter.

Design a two-pole active bandpass filter with a center
frequency fo = 20kHz, midband gain Ao = 1, and a bandwidth of 2000Hz. A single supply, Vs = 24V, is to be used.

Writing the transfer function for Figure 2.17.1 in its general
form for max boost (assuming only R3 ~ Rl) results in
Equation (2.17.1).

RZ
470

R4
Z4k

Ao "" -1
fo '" 20kHz
0 to
0

THO" O.t%

FIGURE 2.16.2 20kHz Bandpass Active Filter

Solution
/', fo
1.0= BW

20kHz
2000Hz

10,

Wo

2rrfo

2. Let R4 = 24 kn.
3. R3 = (Vs _ 1\ R4 = (24 - 1\ 24k
2.6
2.6

'J

'J

1.98 x 10 5
(2.17.1)

Use R3 = 200k
2-53

Rewriting (2.17.7) and (2.17.8) yields:
R2 = 3(Ao -l)Rl

(2.17.9)

R2 = (9.6102 - 2) Rl

(2.17.10)

Combining (2.17.9) and (2.17.10) gives:
Ao

=(9.61~L2)+1

(2.17.11)

'0

From Equation (2.17.11) it is seen that gain and 0 are
intimately related and that large gains mean large Os and
vice versa. Equations (2.17.9) and (2.17.10) show that Rl
and R2 are not independent, which means one may be
arbitrarily selected and from it (knowing Ao and/or 0) the
other is found.

FIGURE 2.17.1 Typical Octave Equalizer Section

Equation (2.17.1) has the form of Equation (2.17.2):

Design
1. Select R2 = lOOk.

S2 + K2p woS + w02

(2.17.2)
2. R3

where:

0 =

~

2p

R3

Ao = gain @fo = K, Wo = 2nfo

10R2 = 10(100k)
1 Meg

3. Let Ao = 12dB = 4V!V and from Equation (2.17.9):

Equating coefficients yields Equations (2.17.3H2.17.5):
R2
Rl = - - - = ~ = 1.11xl04
3(Ao -1)
3 (4 - 1)

2R1 + R2
Wo

(2.17.3)

2 Rl R2Cl + R3 (Rl + R2) C2
Ao

0=

Use Rl = 10k.

Rl R2 R3Cl C2

4. Check 0 from Equation (2.17.8):
(2.17.4)

2 Rl R2Cl + Rl (R2 + R3)C2
(

2R1+R2

JR1R2R3C1C2

(

R2R3C1C2

o=
)

o

(Rl+R2)C2+2R2Cl+R3C2

= 1.12, which is satisfactory.

5. Calculate C2 from Equation (2.17.6) and Cl

(2.17.5)
In order to reduce these equations down to something
useful, it is necessary to examine what is required of the
finished equalizer in terms of performance. For normal
home use, ±12dB of boost and cut is adequate, which
means only a moderate amount of passband gain is
necessary; and since the filters will be centered one octave
apart in frequency a large 0 is not necessary (0 = 1·2 works
fine). What is desirable is for the passband ripple (when all
filters are at maximum) to be less than 3dB.

5.513 x 10- 7
C2 = - - - fo
A table of standard values for Cl and C2 vs. f 0 is given
below:
TABLE 2.17.1

to (Hz)

Cl

C2

32

0.18/lF
O.l/lF
0.047 /lF
0.022/lF
0.012/lF
0.0056/lF
0.0027/lF
0.OO15/lF
680pF
360pF

0.018/lF
O.Ol/lF
0.0047/lF
0.0022/lF
0.0012/lF
560pF
270pF
150pF
68pF
36pF

64

2nfo = - - M
2+1
10R2 C2
Rl

(2.17.6)

Ao

R2
1+-3 Rl

(2.17.7)

0=

2Rl + R2
9.61 Rl

2 + lOOk
10k

2n fo (10) (lOOk)

Examination of Equation (2.17.5) in terms of optimizing
the ratio of C1 and C2 in order to maximize 0 shows a good
choice is to let Cl = 10C2. A further design rule that is
reasonable is to make R3 = 10 R2, since R3 is unnecessary
for the filter section. Applying these rules to Equations
(2.17.3H2.17.5) produces some useful results:

Wo

2 (10k) + lOOk
(9.61) (10k)

125
250
500
lk
2k
4k
8k
16k

(2.17.8)

2·54

IS In order to maintain a unity gain system. Without it
the output would equal ten times the input, e.g., an input
of 1 V, with all pots flat, would produce 1 V at each
equalizer output - the sum of which is 10V. By scaling
R20 such that the input signal is multiplied by 9 before
the subtraction, the output now becomes 10V - 9V = 1V
output, i.e., unity gain. The addition of R4 to each section
is for stability. Capacitor C3 minimizes possibly large DC
offset voltages from appearing at the output. If the driving
source has a DC level then an input capacitor is necessary
(O.lIlFl, and similarly, if the load has a DC level, then an
output capacitor is required.

The complete design appears as Figure 2.17.2. While it
appears complicated, it is really just repetitious. By using
quad amplifier ICs, the whole thing consists of only three
integrated circuits. Figure 2.17.2 is for one channel and
would be duplicated for a stereo system. The input buffer
amplifier guarantees a low source impedance to drive the
equalizer and presents a large input impedance for the
preamplifier. Resistor Ra is necessary to stabilize the
LM349 while retaining its fast slew rate (2 V /Ils). The
output amplifier is a unity gain, inverting summer used to
add each equalized octave of frequencies back together
again. One aspect of the summing circuit that may appear
odd is that the original signal is subtracted from the
sum via R20. (It is subtracted rather than added because
each equalizer section inverts the signal relative to the
output of the buffer and R20 delivers the original signal
without inverting.) The reason this subtraction is necessary

It is possible to generate just about any frequency response
imaginable with this ten band octave equalizer. A few
possibilities are given in Figure 2.17.3.

C,
o.IB
Rl

R,o
32Hz

lOOk
641-1z

-=

R21

(DUPLICATE ABOVE FOR

'i OO....-VYI\r-...---::!

R6
lOOk

t

125Hz

A TOTAL OF 10 CIRCUITS.
SUBSTITUTING APPROPRIATE
CAP VALUES FROM TABLE

R8
24k

2.17.1.)
C,
360pF

!
2kHz

Rl

Rl
4kHz

8kHz

1. All RESISTORS %W. 5%.
2. POTS ARE LINEAR TAPER.
3. PIN 4 CONNECTEO TO VCC· +15V;
PIN 11 CONNECTEO TO VEE· -15V;
OECOUPLEO WITH o.lpF CAPS AT
EACH QUAO OP AMP.
4. CAPTOLERANCE ±10%.

>=___R'9
"",...16kHz
lOOk

11k

FIGURE 2.17.2 Ten Band Octave Equalizer

0-

+12

:

+9
+6
(j) ALL CONTROLS flAT
50oH, BOOST/CUT, ALL OTHERS FLAT
1 kH, BOOST/CUT, All OTHERS FLAT
l

FIGURE 2.17.3 Typical Frequency Response of Equalizer

2·55

-=

2.17.1 Pink Noise Generator
+15V

Once an equalizer is incorporated into a music system the
question quickly arises as to how best to use it. The most
obvious way is as a "super tone control" unit, where control
is now extended from the familiar two or three controls to
ten controls (or even 30 if 1/3 octave equalizers are used).
While this approach is most useful and the results are
dramatic in their ability to "liven" up a room, there still
remains, with many, the desire to have some controlled
manner in which to equalize the listening area without
resorting to the use of expensive (and complicated) spectrum
or real·time analyzers.

3k

300

VOUT

0.033

The first step in generating a self·contained room equalizing
instrument is to design a pink noise generator to be used as
a controlled source of noise across the audio spectrum.
With the advent of medium scale integration and MOS
digital technology, it is quite easy to create a pink noise
generator using only one IC and a few passive components.

0.27

0.047

0.047

FIGURE 2.17.6 Pink Noise Generator

The MM5837 digital noise source is an MOS/MSI pseudo·
random sequence generator, designed to produce a broad·
band white noise signal for audio applications. Unlike
traditional semiconductor junction noise sources, the
MM5837 provides very uniform noise quality and output
amplitude. Originally designed for electronic organ and
synthesizer applications, it can be directly applied to room
equalization. Figure 2.17.4 shows a block diagram of the
internal circuitry of the MM5837.

What is required to produce pink noise from a white noise
source is simply a -3dB/octave filter. If capacitive reactance
varies at a rate of -6dB/octave then how can a slope of less
than -6dB/octave be achieved? The answer is by cascading
several stages of lag compensation such that the zeros of
one stage partially cancel the poles of the next stage, etc.
Such a network is shown as Figure 2.17.5 and exhibits a
-3dB/octave characteristic (±1/4dB) from 10Hz to 40kHz.
The complete pink noise generator is given by Figure 2.17.6
and gives a flat spectral distribution over the audio band of
20Hz to 20kHz. The output at pin 3 is a 11.5V p. p random
pulse train which is attenuated by the filter. Actual output
is about 1 Vp. p AC pink noise riding on a 8.!;iV DC level.

The output of the MM5837 is broadband white noise. In
order to generate pink noise it is necessary to understand
the difference between the two. White noise is characterized
by a +3dB rise in amplitude per octave of frequency change
(equal energy per constant bandwidth). Pink noise has flat
amplitude response per octave change of frequency (equal
energy per octave). Pink noise allows correlation between
successive octave equalizer stages by insuring the same
voltage amplitude is used each time as a reference standard.

2.17.2 Room Equalizing Instrument
For a room equalizing instrument, a different type of
equalizer section is required than that previously described
under the Ten Band Octave Equalizer section. The difference
lies in the necessary condition that each section must pass
only its bandwidth of frequencies, i.e., the all·pass charac·
teristic of Figure 2.17.1 is unacceptable. The reason for this
is that to use this instrument all but one band will be
switch·ed out and under this condition the pink noise will
be passed through the remaining filter and it must pass only
its octave of noise. The filtered noise is passed on to the
power amplifier and reproduced into the room by the
speaker. A microphone with flat audio band frequency
response (but uncalibrated) is used to pick up the noise at
some central listening point. The microphone input is
amplified and used to drive a VU meter where some
(arbitrary) level is established via the potentiometer of the
filter section. This filter section is then switched out and
the next one is switched in. Its potentiometer is adjusted
such that the VU meter reads the same as before. Each
filter section in turn is switched in, adjusted, and switched
out, until all ten octaves have been set. The whole process
takes about two minutes. When finished the room response
will be equalized flat for each octave of frequencies. From
here it becomes personal preference whether the high end is
rolled off (a common practice) or the low end is boosted.
It allows for greater experimentation since it is very easy
to go back to a known (flat) position. It is also easy
to correct for new alterations within the listening room
(drape changes, new rugs, more furniture, different speaker
placement, etc.). Since all adjustments are made relative to
each other, the requirement for expensive, calibrated
microphones is obviated. Almost any microphone with flat
output over frequency will work.

OUTPUT

FIGURE 2.17.4 MM5837 Noise Source

6.8k

0.033

1k

VOUT

FIGURE 2.17.5 Passive -3dB/Octave Filter

2·56

ROOM EQUALIZING INSTRUMENT

MIC

(a) Stereo Application

PHONO

(c) Adding EQ to Receiver System

(b) Adding EQ to Component System

FIGURE 2.17.7 Typical Equalizing Instrument Application

For stereo applications, a two channel instrument is
required as diagrammed in Figure 2.17.7a. Figures 2.17.7b
and -c show typical placement of the equalizer unit within
existing systems.
While any bandpass filter may be used for the filter
sections, the multiple-feedback, infinite-gain configuration
of Figure 2.17.8 is chosen for its low sensitivity factors. The
design equations appear as follows:
R1

0

=

'.

(2.17.12)

2rrfoAoC1
0

R2

FIGURE 2.17.8 Bandpass Filter Section

(2.17.13)

Design

(2.17.14)

2. Select R 1 for desired input resistance. (Note that net
input impedance is (R1 + R2)/10, since there are 10
sections in parallel.)

(20 2 - Ao)2rrf o C1
R3

0
rr fo C1
R3

Ao

1. Select Ao = 4(12dB) and 0 = 2.

Let R1 = 120k.

(2.17.15)

2R1

3. Calculate R2 from Equations (2.17.13) and (2.17.12):

0

rrf o C1 R3

fo

--

1

2rrC1

1-R1 + R2

----

Q

o

(20 2 -Ao) 2rr fo Cl

[2(2)2-4]2rrfo C1

R2 =

(2.17.16)

o
(2.17.17)

R1 R2 R3

2·57

A table of standard values for Cl

4. Calculate R3 from Equation (2.17.15).
R3 = 2AoRl = 8Rl = 8(120k) = 960k

32
64
125
250
500
1k
2k
4k
8k
16k

5. Calculate Cl from Equation (2.17.12):
Q

2

2rrfoAoR1

(2rrfo) (4) (120k)

6.63 x 10-7

C1

C,

fo (Hz)

Use R3 = 1 Meg.

C1

fo is given below.

VS.

TABLE 2.17.2

fo

0.022tl F
0.011 tlF
0.0056tlF
0.0027 tlF
0.0015tlF
680pF
330pF
160pF
82pF
43pF

C,

A,

C,

'20k

0.022

S2
OUT

ei

1

R2'

FLAT

R2

R7

RS
20k

120k

-=

32Hz

4.7k

-=

-=

125Hz

0----0---

R17

250Hz

R20

NORM

64Hz

500Hz

+ C13

'OOk
OU'LiCATE ABOVE FOR A TOTAL
OF '0 CIRCUITS. SUBSTITUTING

0

20J.iF
1kHz

APPROPRIATE CAP VALUES FROM

TABLE 2.'7.2.

EQUALIZE

2kHz

S'A

4kHz

R22
5.Sk

8kHz

C,

-=-

-=

R,

C,

'20k

43,

y

S2
OUT
E

+24 V

'~~.:
A33
47k

+

N

R3

43,

-=

t··
'0

R2
'20k

R,S

RS
20k

-=

"::-

-=-

N

R24
R26
'k

Cs
0.21

6.ak
R25
3k

+

C6
lJ1

R3' MIC

~

SENSITIVITY

2k
1. ALL RESISTORS '!4W, ±5%.

R30
22

2. POTS ARE LINEAR TAPER
3. LM349: VCC ~ +ISV (PIN 4).
VeE ~ -ISV (PIN 11) OECOU'lEO
WITH O.lpF CAPS.
4. CAP TOLERANCE ±10%.

R29

1,2k

FIGURE 2.17.9 Room Equalizing Instrument

2·58

SID

I '00" 0
-=

GENERATOR

C'2
100jJ

E

+ C7

PINK NOISE.

R27
300

16kHz

4.7k

~+15V

0

F

For detailed discussions about room equalization, the
interested reader is directed to the references that follow
this section.

The complete room equalizing instrument appears as Figure
2.17.9. The input buffer and output summer are similar to
those that appear in Figure 2.17.2, with some important
differences. The input buffer acts as an active attenuator
with a gain of 0.25 and the output summer has variable gain
as a function of slider position. The purpose of these
features is to preserve unity gain through a system that is
really "cut-only" (since the gain of each filter section is
fixed and the output is dropped across the potentiometers).
The result is to create a boost and cut effect about the midpoint of the pot which equals unity gain. To see this,
consider just one filter section, and let the input to the
system equal 1 V. The output of the buffer will be 0.25V
and the filter output at the top of potentiometer R6 will
again be 1 V (since Ao ~ 4). The gain of the summer is given
by R17/R7 ~ 4 when the slider of R6 is at maximum, so
the output will be equal to 4V, or +12dB relative to the
input. With the slider at midposition the 4.7k summer input
resistor R7 effectively parallels 1/2 of R6 for a net resistance
from slider to ground of 4.7k111 Ok ~ 3.2k. The voltage at
the top of the pot is attenuated by the voltage divider
action of the 10krl (top of pot to slider) and the 3.2krl
(slider to ground). This voltage is approximately equal to
0.25 V and is multiplied by 4 by the summer for a final
output voltage of 1 V, or OdB relative to the input. With
the slider at minimum there is no output from this section,
but the action of the "skirts" of the adjacent filters tends
to create -12dB cut relative to the input. So the net result
is a ±12dB boost and cut effect from a cut only system.

REFERENCES
1. Davis, D., "Facts & Fallacies on Detailed Sound System
Equalization," AUDIO reprint available from AL TEC,
Anaheim, California.
2. Eargle, J., "Equalization in the Home," AUDIO, vol. 57,
no. 11, November 1973, pp. 54-62.
3. Eargle, J., "Equalizing the Monitoring Environment,"
Jour. Aud. Eng. Soc., vol. 21, no. 2, March 1973, pp.
103-107.
4. Engebretson, M. E., "One-Third Octave Equalization
Techniques and Recommended Practices," Technical
Letter No. 232, ALTEC, Anaheim, California.
5. Heinz, H. K., "Equalization Simplified:' Jour. Aud. Eng.
Soc., vol. 22, no. 9, November 1974, pp. 700-703.
6. Queen, D., "Equalization of Sound Reinforcement
Systems," AUDIO, vol. 56, no. 11, November 1972,
pp. 18·26.
7. Thurmond, G. R., "A Self·Contained Instrument for
Sound·System Equalization," Jour. Aud. Eng. Soc.,
vol. 22, no. 9, November 1974, pp. 695-699.

The pink noise generator from Figure 2.17.6 is included as
the noise s"urce to each filter section only when switch
Sl (3 position, 4 section wafer) is in the "Equalize"
position. Power is removed from the pink noise generator
during normal operation so that noise is not pumped back
onto the supply lines. Switch S2 located on each filter
section is used to ground the input during the equalizing
process. The LM381 dual low noise preamplifier is used as
the microphone amplifier to drive the VU meter. The
second channel is added by duplicating all of Figure 2.17.9
with the exception of the pink noise generator which can be
shared. Typical frequency response is given by Figure
2.17.10. While the system appears complex, a complete
two-channel instrument is made with just 8 ICs (6-LM349,
l-LM381, and l-MM5837).

2.18 MIXERS
2.18.1 Introduction
A microphone mixing console or "mixer" is an accessory
item used to combine the outputs of several microphones
into one or more common outputs for recording or public
address purposes. They range from simple four inputone output, volume-adjust-only units to ultra-sophisticated
sixteen channel, multiple output control centers that
include elaborate equalization, selective channel reverb,
taping facilities, test oscillators, multi-channel panning,
automatic mix·down with memory and recall, individual
VU meters, digital clocks, and even a built-in captain's
chair. While appearing complex and mysterious, mixing
consoles are more repetitious than difficult, being con·
structed from standard building-block modules that are
repeated many times.

+12
+9
+6

~

"~

+3

2.18.2 Six Input-One Output Mixer

-3

A detailed analysis of all aspects of mixer design lies beyond
the scope of this book; however, as a means of introduction
to the type of design encountered Figure 2.18.1 is included
to show the block diagram of a typical six input-one output
mixer. Below each block, the section number giving design
details is included in parentheses for easy cross reference.

-6
-9

-12
10

100

lk

10k

lOOk

FREQUENCY (Hz)

Individual level and tone controls are provided for each
input microphone, along with a choice of reverb. All six
channels are summed together with the reverb output by
the master summing amplifier and passed through the
master level control to the octave equalizer. The output of
the equalizer section drives the line amplifier, where monitoring is done via a VU meter.

(!) ALL CONTROLS flAT
(%i 1kHz BOOST. ALL OTHERS FLAT

!l> 500Hz, 1kHz, 2kHz, 4kHz BOOST,
All OTHERS flAT

FIGURE 2.17.10 Typical Frequency Response of Room Equalizer

2-59

Mle
PREAMP

CHANNEl
LEVEL

TONE
CONTROL

INPUT 1
MAIN

MIXING
BUS
REVERB
SEND
REVERB

MIXING
BUS

INPUT'>-INPUT'>--

MASTER

MASTER

SUMMER

LEVEL

OCTAVE
EQUALIZER

LINE

AMPLIFIER

}
.......-~~OUTPUT

INPUT4>-IOPUT5>-INPUT8>--

REVERB
RETURN

lUI

FIGURE 2.18.1 Six Input-One Output Microphone Mixing Console (Design details given in sections shown in parentheses.)

15k

INPUT

CHANNEll OUTPUT

O.l~AR~ >-...- -..- - -.....- - - 10k

CHANNEL 2 OUTPUT
15k

3.41R l
51k

FIGURE 2.18.2 Two Channel Panning Circuit

Expansion of the system to any number of inputs requires
only additional input modules, with the limiting constraint
being the current driving capability of the summing ampli·
fiers. (The summing amp must be capable of sourcing and
sinking the sum of all of the input amplifiers driving the
summing bus. For example, consider ten amplifiers, each
driving a 10k,Q summing input resistor to a maximum level
of 5VRMS. The summing amplifier is therefore required to
handle 5mA.) Expanding the number of output channels
involves adding additional parallel summing busses and
ampl ifiers, each with separate level, equalizer, and VU
capabilities. Other features (test oscillator, pink noise
generator, panning, etc.) may be added per channel or per
console as required.

2.18.3 Two Channel Panning Circuit
Having the ability to move the apparent position of one
microphone's input between two output channels often is
required in recording studio mixing consoles. Such a circuit
is called a panning circuit (short for panoramic control
circuit) or a pan-pot. Panning is how recording engineers
manage to pick up your favorite pianist and "float"
the sound over to the other side of the stage and back
again. The output of a pan circuit is required to have
unity gain at each extreme of pot travel (i.e., all input
signal delivered to one output channel with the other
output channel zero) and -3dB output from each channel
with the pan-pot centered. Normally panning requires two
2-60

oppositely wound controls ganged together; however, the
circuit shown in Figure 2.18.2 provides smooth and accurate
panning with only one linear pot. With the pot at either
extreme the effective input resistance equals 3041 R 1 (see
Appendix A3.1) and the gain is unity. Centering the pot
yields an effective input resistance on each side equal to
4.83Rl and both gains are -3dB. Using standard 5% resistor
values as shown in Figure 2.18.2, gain accuracies within

OAdB are possible; replacing Rl with 1% values (e.g., input
resistors equal 14.3 kn and feedback resistors equal 48.7 kn)
allows gain accuracies of better than 0.1 dB. Biasing resistor
R2 is selected per section 2.8 as a function of supply
voltage. Capacitor Cl is used to decouple the positive input,
while C2 is included to prevent shifts in output DC level
due to the changing source impedance.

VOUT

FIGURE 2.19.1 Preamp Current Booster

FIGURE 2.19.2 Discrete Current Booster Design

additional phase shift at 15MHz, thereby not appreciably
affecting the stability of the LM387 (Av;;' 10).

2.t9 DRIVING LOW IMPEDANCE LINES
The output current and drive capability of a preamp may
be increased for driving low impedance Iines by incorporat·
ing a LH0002CN current amplifier within the feedback loop
(Figure 2.9.1). Biasing and gain equations remain unchanged
and are selected per section 2.8. Output current is increased
to a maximum of ±100mA, allowing a LM387 to drive a
600n line to a full 24dBm when operated from a +36V
supply. Insertion of the LH0002C adds less than 10 degrees

Comparable performance can be obtained with the discrete
design of Figure 2.19.2 for systems where parts count is not
critical. Typical measured characteristics show a bandwidth of 10-200kHz at +20dBm output, with THD @ 1 kHz
equal to 0.01% rising to only 0.1% @ 20kHz. A maximum
output level of +23dBm can be obtained before c1ipping_
2-61

2.20 NOISELESS AUDIO SWITCHING
SWITCH
SElECTOR

TR = Re (TYPICALl Y '·10ms)

OV - ON
lOV - OFF

SIGNAL
INPUT

INPUT 1

5>
.

o-o ......-o--'I;VV"_-;!

'~
I

1

I

ADDITIONAL
SWITCHES

ADDITIONAL
INPUTS
MECHANICAL EQUIVALENT

FIGURE 2.20.1 Ooglitched Current Mode Switch
SWITCH
SelECTOR

lOOk
INPUT 1
SIGNAL
INPUT

~,

SIGNAL
QUTPUT

_ _ lRIN

P1087
OR J175

INPUT 2

OUTPUT

~

":"

T

-=1

ADDITIONAL
SWITCHES

ADDITIONAL
SWITCHES
MECHANICAL EQUIVALENT

FIGURE 2.20.2 A Oeglitched Voltage Mode Switch

2.20.1 Active Switching

Discrete JFETs may be used in place of the quad current
mode switch; or, they can be used as voltage mode switches
at a savings to the amplifier but at the expense of additional
resistors and a diode.

As prices of mechanical switches continue to increase, solid
state switching element costs have decreased to the point
where they are now cost effective. By placing the switch
on the PC board instead of the front panel, hum pickup
and crosstalk are minimized, while at the same time
replacing the complex panel switch assemblies.

Driver rise times shown in the figures, in the 1·10 ms range,
will result in coupled voltage spikes of only a few mV when
used with the typical impedances found in audio circuits.

The CMOS transmission gate is by far the cheapest solid
state switching element available today, but it is plagued
with spiking when switched, as are all analog switches. The
switching spikes are only a few hundred nanoseconds wide,
but a few volts in magnitude, which can overload following
audio stages, causing audible pops. The switch spiking is
caused by the switch's driver coupling through its capaci·
tance to the load. Increasing the switch driver's transition
time minimizes the spiking by reducing the transient current
through the switch capacitance. Unfortunately, CMOS
transmission gates do not have the drivers available, making
them less attractive for audio use.

2.20.'2 Mechanical Switching
A common mechanical switching arrangement for audio
circuits involves a simple switch located after a coupling
capacitor as diagrammed in Figure 2.20.3. For "pop" free
switching the addition of a pull·down resistor, R 1, is
essential. Without R1 the voltage across the capacitor tends
to float up and pops when contact is made again; R1 holds
the free end of the capacitor at ground potential, thus
eliminating the problem.

Discrete JFETs and monolithic JFET current mode analog
switches such as AM97C11 have the switch element's input
available. This allows the transition time of the drive to be
tailored to any value, making noiseless audio switching
possible. The current mode analog switches only need a
simple series resistor and shunt capacitor to ground
between the FETswitch and the driver. (See Figure 2.10.1.)

FIGURE 2.20.3 Capacitor Pull·Down Resistor

2·62

3.0 AM, FM and FM Stereo
LB-Er-rl-,F-A-M-P-lI-F'-E-R'f-lETECTOR

1

-illB-Oj

(455kHz)

:~~IO

LOCAL I
990·2060kHz I OS.CllLATOR
FIGURE 3.1.1 Superheterodyne Radio

3.1 AM RADIO

Necessary design equations appear below:

3.1.1 Introduction

(3.1.1)

XL
QL =

RpllRL
XL

=

RT

(3.1.2)

--

XL

RL = N0 2 RIN

In the tuned RF, the incoming signal is amplified to a
relatively high level by a tuned circuit amplifier, and then
demodulated.

(3.1.3)

L

)

Controlled positive feedback is used in the regenerative
receiver to increase circuit Q and gain with relatively few
components to obtain a satisfactory measure of performance
at low cost.

L
/

II C

No

Both the TRF and regenerative circuits have been used for
AM broadcast, but are generally restricted to low cost toy
appl ications.
3.1.2

Rp

Qu =

Almost exclusively, the superheterodyne circuit reigns
supreme in the design of AM broadcast radio. This circuit,
shown in Figure 3.1.1, converts the incoming signal 535kHz to 1605kHz - to an intermediate frequency,
usually 262.5 kHz or 455 kHz, which is further amplified
and detected to produce an audio signal which is further
amplified to drive a speaker. Other types of receiver circuits
include tuned RF (TRFI and regenerative.

"v

iL:J
Rp
'"v

lO

II C
!

RL

I

---

~ R'N VtN

I

1

N,' TOTAL TURNS

Conversion of Antenna Field Strength to Circuit
Input Voltage

VTr

FIGURE 3.1.2 Ferrite Rod Antenna Equivalent Circuit

Looking at Figure 3.1.1, the antenna converts incoming
radio signals to electrical energy. Most pocket and table
radios use ferrite loop anterinas, while automobile radios
are designed to work with capacitive whip antennas.

VT = QL VID

(3.1.4)

VID = Heff E

(3.1.5)

Ferrite Loop Antennas
VIN =

The equivalent circuit of a ferrite rod antenna appears as
Figure 3.1.2. Terms and definitions follovy:
L = antenna inductance
C '" tuning capacitor plus stray capacitance (20·150pF typ.)
No = antenna turns ratio - primary to secondary

VT

-

(3.1.6)

No

The effective height of the antenna is a complex function
of core and coil geometry, but can be approximated! by:

RIN = circuit input impedance
Rp = equivalent parallel loss resistance (primarily a function
of core material)
R L = equivalent loading resistance

(3.1.7)

where:

VIN = volts applied to circuit
VID = volts induced to antenna
VT = voltage transferred across tank
Q u = unloaded Q of antenna coil
QL = loaded Q of antenna circuit
Heff = effective height of antenna in meters
E = field strength in volts/meter

N 1 = total number of turns
Mr = relative permeability of antenna rod
(primarily function of length)
A = cross sectional area of rod

A = wavelength of received signal

= 3 x 10 8 m/sec
freq (HZ)
3·1

g

5. Rearranging Equation (3.1.9) and solving for required
Heff:

Noise voltage is calculated from the total Thevenin equivalent loading resistance, RT = RpIIRL, using Equation
(3.1.8):

SIN ) 4 K T ~f RT
Heff = - - - - - - QLEm

(3.1.8)
~f =

where:

3dB bandwidth of IF
10)(4) (1.38 x 10-23 ) (300) (10kHz) (157k)

T = temperature in 0 K

K

(100) (100pV/m) (0.3)

Boltzmann's constant
= 1.7cm

1.38 x 10-23 joulest K

6. Rearranging Equation (3.1.7) and solving for N 1 :

The signal·to·noise ratio in the antenna circuit can now be
expressed as Equation (3.1.9):
SIN

=

VTm

=

QLHeffEm

(3.1.9)

v'4 K T ~f R-r

en

HeffA
Nl - - - 2rr Pr A
(0.017m) (3 x 108 m/sec)

wher~:

m

= index of modulation

70.7

(2rr) (65) (1 x 106 Hz) (rr) (7.5 x 10-3 m)2

Example 3.1.1
Nl "" 71 turns

Specify the turns ratio No, total turns N 1, effective height
Heff, and inductance required for an antenna wound onto a
rod with the characteristics shown, designed to match an
input impedance of 1 kQ. Calculate the circuit input voltage
resulting from a field strength of 100pV/m with 20dB SIN
in the antenna circuit. Assume a 15·365pF tuning capacitor
set at 100pF for an input frequency of 1 MHz.
Given:

RIN = 1 kQ

fo = 1 MHz

E = 100pV/m

rod dia. = 1.5cm

SIN = 20dB

Pr

C = 100pF

m = 0.3

Qu = 200

~f

7. Form Equation (3.1.5):
VID = Heff E
0.017m x 100pV/m
VID

1.7pV

8. Find VT from Equation (3.1.4):

= 65 (rod length = 19cm)

VT

QL VID
100 x 1.7pV

= 10kHz

VT = 170pV

Calculate L, No, Heff, Nl, VIN
1. Since the circuit is "tuned," i.e., at resonance, then
XL = XC, or

9. Using Equation (3.1.6), find VIN:

L =

100pF (2rr x 1 x 10 6 )2
2.53 x 10-4 H

Capacitive Automotive Antennas

2. From Equation (3.1.1):

A capacitive automobile radio antenna can be analyzed in a
manner similar to the loop antenna. Figure 3.1.3 shows the
equivalent circuit of such an antenna. Cl is the capacitance
of the vertical rod with respect to the horizontal ground
plane, while C2 is the capacitance of the shielded cable
connecting the antenna to the radio. In order to obtain a
useful signal output, this capacitance is tuned out with an
inductor, L. Losses in the inductor and the input resistance
of the radio form R L. The signal appearing at the input
stage of the radio is related to field strength:

Rp = Qu XL = 200 x 2rr x 1 MHz x 250pH
Rp "" 314k
3. For matched conditions and using Equation (3.1.3):
Rp

=

No =

RL

!

=

N0 2RIN

Rp = j314k = 17.7
RIN
lk

(3.1.10)

No"" 18:1
4. From Equations (3.1.1) and (3.1.2):
RpliRL = ~
XL
QL

2XL

where:

Qu
2

VID is defined by Equation (3.1.5)
QL is defined by Equation (3.1.2)

since Rp

CT = Cl + C2

100
3-2

FIGURE 3.1.3 Capacitive Auto Antenna Equivalent Circuit

Similar to the ferrite rod antenna, the signal-to-noise ratio
is given by:

3. From Equations (3.1.10) and (3.1.5):

0.5m x 100.uV/m x 80x
The effective height of a capacitive vertical whip antenna
can be shown 1 to equal Equation (3.1.12):

10pF
90pF

(3.1.12)
where:

h

4. Since matching requires Rp = RL, and resonance gives
XCT = XL, then using Equation (3.1.2):

= antenna height in meters

Example 3.1.2

For comparison purposes, calculate the circuit input voltage,
VIN, for an automotive antenna operating in the same
field as the previous example; assume same circuit input
impedance of 1 kQ and calculate the resultant SIN. Use the
given data for a typical auto radio antenna extended two
sections (1 meter).
Given:

RIN = 1 kQ

Af

10kHz

E = 100.uV/m

Cl

10pF

QL

= 80

2 x 80 x
211 (1 MHz) (90pF)
5. Using Equation (3.1.3):
No

VT
No

0.5m

211

X

1 MHz x 90pF

16.8

lk

= 444.uV
17

7. From Equation (3.1.1):
_ Rp
_
= 283k
Qu XCT

2. Rearranging Equation (3.1 11) and solving for SIN:

X

211 x 1 MHz x 90pF

160

It is interesting to note that operating in the same field
strength, the capacitive antenna will transfer approximately
three times as much voltage to the input of the circuit, thus
allowing the greater signal-to-noise ratio of 29dB.

SIN

(0.5) (100.uV/m) (0.3)

SIN

= j283k

6. From Equation (3.1.6):

1. Calculate Heff from Equation (3.1.12) and solve for XcT

2

(RL
j R;N

No '" 17:1

Calculate SIN, No, V I N.

=~ =

=

CT = 90pF
m = 0.3

fo = 1 MHz

Heff

283k

~x

10pF
90pF

v'aO
REFERENCES

10- 23 ) (300) (10k) (1768)

1. Laurent, H. J. and Carvalho, C. A. B., "Ferrite Antennas
for AM Broadcast Receivers," Application Note available
from Bendix Radio Division of The Bendix Corporation,
Baltimore, Maryland.

SIN = 27.55
SIN'" 29dB
3-3

AV

AV = 14V!V

=

45V/V

AV

=

36V!V

FIGURE 3.1.4 AM Radio Gain Stages

3.1.3 Typical AM Radio Gain Stages

useful for frequencies in excess of 50MHz. Figure 3.2.2a
shows the transconductance as a function of frequency.

The typical levels of Figure 3.1.4 give some idea of the gain
needed in an AM radio. At the IF amplifier output, a diode
det~ctor recovers the modulation, and is generally designed
to produce approximately 50mVRMS of audio with
m ~ 0.3: The gain required is therefore:

Transistors 04 and 05 make up the local oscillator circuit.
Positive feedback from the collector of 05 to the base of
04 is provided by the resistor divider Rg and RS. The
oscillator frequency is set with a timed circuit connected
between pin 2 and Vee. Transistors 04 and 05 are biased
at 0.5mA each, so the transconductance of the differential
pair is 10mmhos. For oscillation, the impedance at pin 2
must be high enough to provide a voltage gain greater than
the loss associated with the resistor divider network Rg, RS
and the input impedance of 04. Values of load impedance
greater than 400n satisfy this condition, with values of
10kn or greater being commonly used.

Av ~ 50mV ~ 23kV!V or S7dB
2.211V
3.2 LM1820 AM RECEIVER SYSTEM
The LM1S20 is a 3 stage AM radio Ie consisting of the
following functional blocks:
RF Amplifier
OScillator
Mixer

IF Amplifier
AGe Detector
Regulator

The differential pair 06 and 07 serve as a mixer, being
driven with current from the oscillator. The ·input signal,
applied to pin 1, is multiplied by the local oscillator
frequency to produce a difference frequency at pin 14.
This signal, the I F, is filtered and stepped down to match
the input impedance of the I F amplifier.

The RF amplifier section (Figure 3.2.1) consists of a cascode
amplifier 02 and 03, whose geometries are specially
designed for low noise operation from low source imped·
ances. The cascode configuration has very low feedback
capacitance to minimize stability problems, and high output
impedance to maximize gain. In addition, bias components
(01, etc.) are included. Biased at 5.6mA, the input stage is

13

12

Transistors Og and 010 form the I F amplifier gain stage.
Again, a cascode arrangement is used for stability and high
gain for a gm of gOmmhos.

14

RI
950

R17
BOO

05

RI2
Uk

04
GIO

03
R2
270

RI3
6BO
R7
5k

RIO
5.6k

RIB
10k

R3
25k
R4
25k

D6

RI4
5.5k
GI

RI5
5.5k

G5
GB
R9
3.3k

II

RII
3.3k

RIB
Ik

R5
520
10

FIGURE 3.2.1 LM1820 Schematic Diagram

3-4

Basically, three possibilities exist for using the LM1820 in
AM radio applications; these are illustrated in Figures 3.2.33.2.6. The mixer-I F-I F configuration results in an economical approach at some performance sacrifice because the mixer
contributes excess noise at the antenna input, which reduces
sensitivity. Since all gain is taken at the IF frequency,
stability problems may be encountered if attention is not
paid to layout .

w

u

TA" 25'C

""
g

e12 '"

Z

t--.

~Z

...""'"
::;

-4

w

-8

>

~

100~.tVRMS

OdO'" 120 mmho (typ)

C>
Z

1"\
\

-12

TABLE 3.2.1 Summary of Circuit Parameters

'"
0.1

0.5 1

5

10

50 100

Parameter

FREnUENCY - MHz

(a) RF Transconductance as a Function of Frequency

RF Section

Input Resistance

1k

IF

Mixer
1.4k

1k
70pF

Input Capacitance

80pF

8pF

Transconductance

120mmhos

2.5mmhos 90mmho

Input Noise Voltage,
6kHz Bandwidth

0.23JlV

0.5JlV

~

I

w

TA" 25'C

Z

el" lmVRMS

u

OdO '" 90 mmho (typ)

""

~

~
:;l
z

...'"'"
w

-4

-8

1\

>

;::

~

The RF-mixer-IF approach takes advantage of the low noise
input stage to provide a high performance receiver for
either automobile or high quality portable or table radio
applications. Another approach which sacrifices little in
performance, yet reduces costs associated with the three
gang tuning capacitor, is to substitute a resistor for the
tuned circuit load of the RF amplifier. The LM1820 has
sufficient gain to allow for the mismatch and still provide
good performance.

\

-12

'"

0.1

0.5

1

5 10

50 100

FREnUENCY - MHz

(b) IF Transconductance as a Function of Frequency

~

-2

I

z

~

-4

By appropriate impedance matching between stages, gain
in excess of 120dB is possible. This can be seen from
Figure 3.2.3c, where the correct interstage matching values
for maximum power gain are shown. The gain of the RF
section is found from:

~-bl"," ~

TA"2n
I MHz
!"
!p260kHz
f12 = 1MHz

~

tV'\)~~

~

//~.

w

> -6
;::

~

-8

where:

-10

N ~ turns ratio ~ v'Rsec/Rpri
K1

6dB loss @ output of RF amplifier due to
matching 500k output impedance

-12

K2
SUPPLY VOLTAGE IV31 - V

te) Relative Gain as a Function of Supply Voltage (V3)

~

6dB loss @ input to mixer due to matching
1.4k input impedance

For the values shown:
AV1 ~

FIGURE 3.2.2 LM1820 Performance Characteristics

~

An AGC detector is included on the clip. The circuit
consists of diodes D1 and D2 which function as a peak to
peak detector driven with I F signal from the output of the
IF amplifier. As the output signal increases, a greater
negative voltage is developed on pin 10 wh ich diverts
current away from the input transistor Q2. This current
reduction in turn reduces the gain of the input stage,
effectively regulating the signal at the I F output.

-1 (120 x 10-3 ) (500k) ~.4k
-- 1
2
500k 2
793.5 "" 58dB

Similarly, for the mixer:
AV2 ~

-1 (2.5 x 10- 3 ) (500k)
2
= 14 "" 23dB
And for the IF:
AV3

A zener diode is included on the chip and is connected from
VCC to ground to provide regulation of the bias currents on
the chip. However, the 1820 functions well at voltages
below the zener regulating voltage as shown in Figure
3.2.2c. Table 3.2.1 summarizes circuit parameters.

~ .!.

(90 x 10-3) (10k)

2

~

~k
--

1
500k 2

(5k

V10k

159 "" 44dB

Total gain ~ 1.8 x 16 6 "" 125dB
3-5

1
2

II
TOTAL GAIN = 25.9k OR BBdB

(a) Mixer-IF-IF Application

Ay = 16

...----,

Ay = 14.7

...---.,

A,= 225

II~

II
(b)
RF
gm1 '" 120mmhos
RIN '" 1k
ROUT = 500k

RF~Mixer~IF

Applications

MIXER

IF

gm2 '" 2.5 mmhos
RIN = 1.4k
ROUT = 500k

gm3 '" 90 mmhos
RIN = lk
ROUT = 10k

M IT

~ ~k

2

10~ ~k

VOUT

~5k

(c) Power Matching for Maximum Gain
FIGURE 3_2.3 Circuit Configurations for AM Radios Using the LM1820

This much gain is undesirable from a performance standpoint, since it would result in 1.5 V of noise to the diode
detector due to the input noise, and it would probably be
impossible to stabilize the circuit and prevent oscillation.

RF stage and mixer for less gain.
One example is shown in Figure 3.2.3a, a mixer-I F-I F
configuration. Gain is deliberately kept low to minimize
stability problems. A complete circuit of this radio is shown
in Figure 3.2.4, along with performance curves.

From a design standpoint, it is desirable to mismatch the

+12Vo------------.---------------------.__- - - - - - - ,

T5j- ---,

141.

m

I
455 kHz I

~

w

;

I

455 kHZ:

S
~
0

L ____ ..J

L __ _

-10

:s

I
I

I
I

-20
-30

-40
-50
10

100

lK

10K

RF INPUT ("Vrm.)

RZ
240K
14

13

12

II

Dl
IN914

LMI8Z0

RJ
IK
Vo

AM
ANTENNA
Cg
.005

L

_ _ _ _ ...l

~e_~~~~~-------o+12V
T3
T4
T5
T6

AM OSC TOKO RWO 6A6255
IF
TOKO RRC 3A6426N
IF
TOKO RRC 3A6427A
IF
TOKO RZC lA6425A

FIGURE 3.2.4 AM Radio Using Mixer-IF-IF

3-6

CIO
.005

AM
SENSITIVITY, 20pV fOR 20dB S+N/N
HARMONIC DISTORTION, 2%
HUM & NOISE, -45dB

CA

pOI

II

II

I
(

100

L2

LI

"f"

TI

I

1-

L': _____-_
11

+6V

II

II

lstlF

2ndiF

1-

.J

T2

II
II

(

O.OII

150.

"~"

II
II

0:-

0:-

1£

14

LM1820

+

AM PVC

VC

AMANT

OIP CHOKE

AM OSC

L2

L1
525KHz-1650KHz

L3
980kHz-2105kHz

IlL:

FERRITE BEAD

Y
Ii
ii
jrC-+-jrc
3T[

ii ".
J"[

11
110T
lUOx 8 mml I
11
I lOT

.n

I I

CA" 140pF

AM 2nd IF

AM 1st IF

T1

l" 40DJ-IH
Ilu = 80

leo 640"H
Ou" 200

Ca" 6DpF

5.5mm

d@
I
1.2mm

I

1

)t

3.5mm

SWG eo #32
TURNS" 3

AM 3rd IF

T2

T3

455KHz

455KHz

455KHz

15DpF EXT

70T

•

•

2T

70T

III

C = 150pF
Ou = 140

142T

____

1T

I I

71T

II

C = 47pf
Ou = 140

I 1
I I
11
11

71T

C" 18Dpf

au" 120

FIGURE 3.2.5 AM Radio Using RF-Mixer-IF

A higher quality approach is shown in Figure 3.2.5. The RF
amplifier is used with a resistor load to drive the mixer. A
double tuned circuit at the output of the mixer provides
selectivity, while the remainder of the gain is provided by

the I F section, which is matched to the diode through a
unity turns ratio transformer. The total gain in this design
is 57 k or 95dB from the base ofthe input stage to the diode
detector.
3-7

Ir- - - - - , - - - -y.-- 0047-'

I

I

I

/

II

~

330

T,-

/

'50k

1200

/

I

I

/

6
1

~

"

'J'3M
-

L_

t-_-_-I~-'A"'~"-C~_~'-"---.!'--+------~---'
9

~

10

-:r

-----

4"------ 8----

-:r

3pF

0.'5 p F

6-l

56pF

S+N

TRANSFORMERS T1:

~

C'" 130pF PRIMARY & SECONDARY
PRIMARY TO SECONDARY TAP RATIO - 30:1

~ -20~~HII-++t#llll-+HttHII-++ttllHl

a ~ 60

~~

COUPLING - CRITICAL
T2:

-, 0 I-lifHtHII-++t#llll-+HttHII-++ttllHl

C'" 130pF PRIMARY & SECONDARY
PRIMARY TAP RATIO - 8.5,'
SECONDARY TAP RATIO - 8.5,'

-30

I-+tHliHII--Pt-ijfj\II-Nc+1,-tItI :~==~~~~z
m =0.3

0-

~ ~o~+H~~~~~~-=~~

n~60

COUPLING - CRITICAL

-50 ~ffi~~~~HttHtll--+ttttttlf

'0

'DO

'k

'Ok

INPUT LEVEL (pVRMS)

FIGURE 3.2.6 AM Auto Radio

The major requirement of an FM I F is good limiting
characteristics, i.e., the ability to produce a constant output
level to drive a detector regardless of the input signal level.
This quality removes noise and amplitude changes that
would otherwise be heard in the recovered signal.

An AM automobile radio design is shown in Figure 3.2.6.
Tuning of both the input and the output of the RF
amplifier and the mixer is accomplished with variable
inductors. Better selectivity is obtained through the use of
double tuned interstage transformers. Input circuits are
inductively tuned to prevent microphonics and provide a
linear tuning motion to facilitate push-button operation.

Many integrated circuits have been developed for the FM
IF function and all fall into roughly three categories:
3.4 Simple Limiters
3.5 Gain Blocks
3.6 Complete I F and Detectors

3.3 FM IF AMPLIFIERS AND DETECTORS
In the consumer field, two areas of application exist for
FM I F amplifiers and detectors; in addition, applications
exist in commercial two way and marine VHF FM radios:

3.4 SIMPLE LIMITERS
Two especially useful RF/IF amplifiers are the "emitter
coupled" differential amplifier, Figure 3.4.1, and the
modified "cascode," Figure 3.4.2. Emitter coupled operation
is advantageous because of its symmetrical, non·saturated
limiting action, and corresponding fast recovery from large
signal overdrive, making a nearly ideal FM I F stage. The
"cascode" combines the large available stable gain and low
noise figure, for which the configuration is well known, with
a highly effective remote gain control capability .via a second
common·base stage, which overcomes many of the inter·
stage detuning and bandwidth variation problems found in
conventional transistor AGC stages.

TABLE 3.3.1 Application for FM-IF Amplifiers

Service

Frequency Deviation

Input
Distortion
Limiting

FM Broadcast

10.7MHz

75kHz

20/J.V

TV Sound

4.5MHz

25kHz

200/J.V

1.5%

5kHz

5/J.V

5%

Two-Way Radio various

0.5%

3-8

The "emitter coupled" and "cascode" configurations contain essentially the same components; they are available as
either type 703 (Figure 3.4.3). which is permanently connected as an emitter coupled amplifier in an economical six
pin package, or as the more versatile type LM171 (Figure
3.4.4). in which a ten pin package allows the user to select
either emitter coupled or cascode configurations. Since the
171, when externally connected as an emitter coupled
amplifier, is essentially identical in performance to the 703,
references will be made only to "cascode" or "emitter
coupled" configurations.

Vcc=12V

10
L1

r----

I

OUT

I

son

I

1=
R2

100H

I
I
I
IL _ _

=

D2

D1
Cl
C2

=
~

C3 = 9 36 pF TRIMMER
C4· 2 8 pf TRIMMER

L1 = Ll ~ 71 #IB a.w.g
SPACED 1 TURN, 1/4" INSIDE OIAM

FIGURE 3.4.1 Emitter Coupled RF Amplifier

FIGURE 3.4.3 LM703 Configuration

DC Biasing

VAGC=O
FOR GAIN TEST

Both the 703 and 171 are biased by using the inherent
match between adjacent monolithic components. They are
designed for use with conventional tuned interstages, in
which DC bias currents flow through the input and output
tuning inductances.

Vcc=+12V

11

In either case, a resistor forces DC current from the
positive supply into a chain of diodes (two for the 703,
three for the 171) proportional to the difference between
supply and forward diode-chain Voltages, and inversely to

r--50n
OUT

R2
2.SK
10

03
D2

50~ ;

IN

01

C1 =CJ
C2 = C4

9·36pFTR1MMER
28 pF TRIMMER

11 = L2=7t.-16a.IJY.g.
SPACEIl1 TURN, 1/4" INSIDE IlIAM

FIGURE 3.4.2 Cascode RF Amplifier

FIGURE 3_4.4 LM171 Configuration

3-9

the value of the resistor. The forced current, Ibias, establishes a voltage drop across the bottom diode (in reality, an
NPN transistor with collector-base short), which is identical
to the base-emitter voltage required to force a collector
current of Ibias in a matched common-emitter stage_ Since
the transistor is monolithically matched to the bottom
diode, and is of fairly high DC "beta," and efficient and
reliably biased current source is created.

This quantity, the transition width of an emitter coupled
amplifier is independent of supply voltage and current, and
proportional to absolute temperature, varying from 84 mV
at -55°C to 153mV at +125°C, and is approximately
114mV at +25°C. Forward transconductance, however, is
directly proportional to total supply current, taking the
approximate form:

at +25°C, 10.7MHz, for either 703 or emitter coupled 171.
Tllus, emitter coupled amplifier gain may be controlled by
externally varying "bias chain" current, changing the current
source by the same amount, but without affecting transition
width.
Because an emitter coupled amplifier's input impedance is
a function of drive level (Figure 3.4.6), interstages designed
with small-signal y-parameters may exhibit center frequency
shifts and bandwidth decreases as signal level increases. This
is less of a problem in FM IF strips, where input signal
amplitude is essentially constant, dictated by the limiting
characteristics of the previous stage (Figure 3.4.7).

Both 703 and 171 function as ordinary differential amplifiers, splitting available current source drive equally, when
base voltages are equal, and being capable of either complete cutoff or full conduction of available current into one
of the pair, depending on differential input. In emitter
coupled service, the input signal is injected in series with
the differential pair's DC bias, while, in the cascode,.it is in
series with the current source's base bias.

.'"

R;;- ~

I,

l' C

p

<
I;;

15

~

10

~

I..-'~

~

o
100

200

300

400

500

10.7MHz FM IF Using Emitter Coupled
Amplifiers
Complete design of a high quality FM IF strip is a painstaking process, in which a number of parameters must be
weighed against each other. Since design techniques are well
covered in the literature, only a brief discussion of design
considerations will be included in this section.
'
Maximum available power gain may be calculated for either
171 or 703 as emitter coupled amplifier using

-~+ = ~2V f -

1Y21 12
MAG = - - 4 911 922

f-

At 10.7MHz, 25°C, and VCC = 12V, using 703 values,

.......

o
50

150

= 2.9k, CIN = 9pF)

Y11

0.35 + j 0.61 mmho (RIN

Y21

-33.4 + j 5.88 mmho (note negative real part)

I\..
-50

!:;

Example 3.4.1

~

-150

~

:!

Impedance

\

-250

U

I'l"- l.<

r-o

~
~-

z
<
I-

FIGURE 3.4.6 Effect of Drive Level on Emitter Coupled Input

= 0.384T (mV)

,

riz

20

.!'

rms INPUT VOLTAGE (mV)

(3.4.2)

TA = 25'C

'"

10

o

q

i\.

12

25

~

Calculating the difference in VIN required to change this
ratio from 10% to 90%, it may be seen that:

\

30

!!

The transfer characteristic of Figure 3.4.5 is represented by
the equation:
qVIN
I(current source)
1 +e kT
(3.4.1)
I(output)

....

Vee'" +12V, 10.7 MHz

S

~

To assure symmetrical limiting and maximum small-signal
linearity, it is necessary that the differential pair be closely
balanced, so that quiescent operation occurs in the center
of the amplifier's transfer characteristic (figure 3.4.5).
Typical Vbe matches better than ±0.3mV, for both 703 and
171 assure this, provided that DC resistance of the input
inductor is so low that input bias currents in the 50MA
region do not induce appreciable input offset voltages.

r-..

14

35

Emitter Coupled Operation

kT
VIN (10%) - VIN (90%) = 2 - (In 9)

(3.4.3)

1Y21 1 = 3.6 (lsupply, mAl mmhos

Total current through an NPN differential pair is determined by the current source, while current "split" depends
on the differential base voltage. Common-mode base
voltage is readily available by using the tap at the top of the
diode chain. In the 703, the differential emitters operate at
a forced voltage of one forward diode drop, Vbe, the
current source still being effective with zero volts, collector
to base. Because the 171, as a cascode, requires high
frequency performance of the current source, three biasing
diodes are used, fixing the differential emitters at 2 Vbe.

250

Y12 "" 0.002 + jOmmho

INPUT VOLTAGE - mV

Y22 "" 0.03 + j 0.18 mmho (ROUT
COUT

FIGURE 3.4.5 Emitter Coupled Transfer Characteristic

3-10

33k,
2.6pF)

MAG

I Y21

12

2. Sharp skirt selectivity without phase/frequency nonlinearity within the passband. This usually implies
double-tuned interstage transformers. Stover, et. al? show
that a transformer coupling factor between 0.6 and 0.8
gives minimum phase nonlinearity, the higher value being
preferred for higher gain per stage.

(34 x 10-3 )2
4 (0.35 x 10-3 x 0.03 x 10-3

4 911 g22
2.75 x 103
34.4dB

3. Overall power gain of at least 100dB, or 25dB per stage
in a four stage strip, to obtain adequate sensitivity and
AM rejection.

(Due to somewhat different typical y-parameters, MAG for
an emitter coupled 171 = 39dB.)

4. A maximum value of load resistance across the output of
each stage, given by:

Calculating the stability criterion:
1Y12Y21 1

C=--------

2 (VCC - N Vbe)
RL";; - - - - - lOUT (MAX)

2911 922 - Re (Y12Y21)

where:
2 x 10-6 x 3.4 x 10-2
C

6.8 x 10-8

0.775

5. The input admittance used in making interstage calculations should be the value resulting from a given value of
input swing (see Figure 3.4.6), rather than the smallsignal value. The input swing, however, depends upon
the transformer ratio, so that transformer optimization
is a mUlti-approximation procedure.

For the conditions given, 0 < C < 1, making the device
unconditionally stable for all sources and loads. In a
practical 10.7MHz IF strip, however, external coupling,
especially fpJm the strip's output to its input, can cause
instability without careful physical design.

6. The interstages should be designed to minimize the
effects of varying drive levels upon center frequency and
bandwidth, since very weak signals may operate the
first one or two stages linearly, rather than as limiters.

A modern FM tuner I F strip capable of low distortion
multiplex reception requires:
1. Bandwidth of at least 300kHz. In a four stage design,
with five interstage networks, bandwidth per stage may
be calculated from overall bandwidth by use of the
"shrinkage" formula:
BW(overall)

J21In=1

2 for the 703, or 3 for the 171

This relationship assures that maximum output current
limiting is reached before the output transistor can
saturate, guaranteeing non-saturated limiting action.

2.1 x 10- 8 + 6.7 x 10-8

=

number of bias chain diodes

N

lOUT (MAX) is approximately 5mA, for both
types

2 (3.5 x 10-4 x 3 x 10- 5 ) - [2 x 10-6 x (-3.34) x 10- 2 ]

BW (per stage)

N

(3.4.5)

(n

300

REFERENCES

1. Linvill, J., and Gibbons, J., Transistors and Active
Circuits, McGraw-Hili, New York, 1961, ch. 9-18.
2. Stover, W., et. aI., Circuit Design for Audio, AM/FM, and
TV, McGraw-Hili, New York, 1967, ch. 7-11.

= number of interstages)
300

3. Gartner, W., Transistors: Principles, Design and Applications, D. Van Nostrand, Inc., New York, 1960, ch. 14-15.

(3.4.4)

0.388
733kHz

3.5 GAIN BLOCKS
+10

(REFERRED TO 1 MILLIWATT
INTO 50 OHMS)

~

BIAS"b"

/

I(

ItV

..-

vrrv

I-

The LM3011 (Figure 3.5.1) is a complete gain block
designed for FM limiter applications. It consists of 3
differential stages and associated biasing, with a current
source (free collector) output suitable for driving a variety
of loads. The circuit will provide 60dB of power gain to a
matched load (Figure 3.5.2), or 60dB of voltage gain to a 1 k
load resistor. The input impedance of the LM3011 is 3 kfl
in parallel with 7 pF; however, unless special care is used in
circuit board layout and shielding, oscillation problems will
occur if source terminations greater than 600 fl are used.

-

i
I I

Vic" 6V

10.7 MHz MATCHED TO 50 OHM SOURCE
AND LOAD. BANDWIDTH APPX 470 kHz.

I I

-20
-50

-40

-30

-20

While designed for FM I F applications, the LM3011 will
operate well at any frequency below 20MHz, and is useful
for a variety of low and medium frequency limiter
applications. Figure 3.5.3 shows the gain and input limiting
voltage characteristics of the circuit, while Figure 3.5.4
shows the input and output characteristics of the circuit.

-10

INPUT POWER Id8m)

FIGURE 3.4.7 Emitter Coupled Limiting Characteristics

3-11

r-----~t_--~t_----~t_------. .~10

01
02

03

04
05
06
Rl

07

R8

R9

R6
Rll

FIGURE 3.5.1 LM3011 Schematic Diagram

INPUT

nm

lOOn

10

FIGURE 3.5.2 Limiter for Driving 300n Ceramic Filter

.

72

AMBIENT TEMPERATURE (T A) '" 25°C
DC SUPPl Y VOLTAGE (Vee) '" 7.5V
SOURCE RESISTANCE (RS) =

son

700

LOAD RESISTANCE (RLl = 1 kn

~

70

600

z 68
<1

500

I

5

'"
'"'"~
~

66

400

64

'300

62

200

Q

>

100
0.1

.2

,4.6.8 1

4 6 810

15

FREQUENCY (II-MHz

j'::

~....

'"

~

~

u

~

z

!iii
=<
;;;

..
..

'"

10

0-

~....
'"lJl

u

~

<:

5~ 8

Q

~

0-1

=>..2

~

,.~

~~
a:
'"
~

~
~
1

"<:

~

;;

~
1

10

15

5

FREQUENCY (Q - MHz

FREQUENCY - MHz

FIGURE 3.5.3 LM3011 Voltage Gain and Input Limiting

Characteristics

FIGURE 3.5.4 LM3011 Input·Output Impedance Characteristics

3·12

R21

"0

o~

-:t--07
R1
1K

100

....

*31

R19

o~

...;,,,

I.,

AS
500

R1

AS

1K

500

r--o

Rll

2.5K

R6

R4
1K

1

.~~~~

'03 ~tO

4K

R1
SOO

I..,i

100

01
1K

),,,

100

RlB

R9

:1
~06

~3l- K~

01

"'0

R1'
100

~

R17

1

J

~ID

R16
8.BI<

).11

"

~

-:("

-

R15

04

r

05

r

RtO

450

~

"
R11
5K

O~

Rll

15K

(

R11

50

06

10

FIGURE 3.6.1 LM2111 Schematic Diagram

Referring to Figure 3.6.1, 01 and 02 form the first amplifier limiter, C4 and 05 form the second, and 07 and 08
form the third. 013 through 016 form the upper switches
for the quadrature demodulator, while 017-018 form the
lower switches. The signal from the I F output pin lOis fed
via a phase shift network (Figure 3.6.3) to the upper quad
pairs. At center frequency, 90° of phase shift occurs between
pins 10 and 12, and the output of the quad detector is at its
quiescent point. As the frequency of the I F signal varies, so
does the phase shift to pin 12, causing an output signal at
the quad detector output pin 14.

3.6 COMPLETE IF AMPLIFIERS AND DETECTORS
3.6.1 LM2111-LM1351 FM IF Amplifiers

Two very similar FM IF amplifier-detector combinations
are available in the LM2111-LM1351 circuits. These circuits
are designed to operate on supply voltages between 8 and
15V. Both circuits feature three stages of limiter/gain
blocks and employ a double balanced phase detector which
operates as a phase shift demodulator (Figure 3.6.2). In
addition, the LM1351 features an audio preamp with an
open loop gain of 40dB.
v'

h -=

O.liJF

r-------

13

--------

11

I
I
I
I
I

---------,I
I
I
I

f

50

2.

FIGURE 3.6.2 LM2111 Block Diagram

3-13

"' OUTPUT

Biasing is accomplished with a string of diodes D1 through
D5 that set the reference voltages for both the I F amplifier
and the quad detector. Feedback (to pin 5 through R6)
completes the bias on the input of the I F amplifier.
The careful reader will notice there are two IF outputs; one
high level (pin 10) produces approximately 1.2V p. p , while
the low level (pin 9) produces 120mV p. p . Thus, the
designer has an option whether to use high level or low level
injection to the quad coil and upper pair.

A '"

~
10

OR

2Ko

O.B

~
Iq

=

~ arctan a

~ = ! ----!......
Ko Iq

1T

(SW. MODE)

(LIN. MODE)

1 +a2

-4

Vee
a =

AUDIO
OUTPUT

2a~

'0

D.B

FIGURE 3.6.4 Theoretical Performance Curves

The designer has the freedom of selecting 0 for recovered
audio and distortion. A typical design for FM broadcast
would be:
10.7MHz "" 31
0.35MHz
where:

fo
Ll.f

center frequency
p-p separation = 0.35MHz (typ) for
±75kHz deviation

A fairly large capacitor should be chosen for C1 (Figure
3.6.3) to swamp out input capacitance variations to the IC,
120pF being a good choice:

FIGURE 3.6.3 Phase Shift Network for Discriminator Circuit

XC1 = 124[1 (atf= 10.7MHz)
The phase sh ift network consists of a small (h igh reactance)
capacitor C2 feeding a parallel tuned circuit C1, R1, L1.

R1 = (0) (X c 1) = 31 x 124 "" 3.9k

Figure 3.6.4 shows the theoretical curves. In the linear
mode, the well-known S curve appears, very similar to the
case of Foster·Seeley and ratio detectors. The relation
between frequency and output is straightforward and the
response shows well defined peaks at the 3dB frequency of
the tuned circuit.

L

18.4pHy

The injection capacitor (C2) should be a large reactance at
center frequency; 4.7pF is a suitable value. In this case, low
level injection is used. Figure 3.6.5 shows the complete
circuit with performance information.

In the switching mode, when limiting occurs after the phase
shift network, the bandwidth increases and the amplitude
response contribution vanishes. This produces a smooth
broad curve without well defined peaks and the trans·
formation of the S curve into a pure phase response of the
type arctan.

Certain precautions should be noted. Depending on the
frequency of operation, the value of the bypass capacitor on
pin 5 can be critical to the prevention of the bias loop in
the I F from oscillating. Layout around pin 12 has been
troublesome in the past since there is a tendency for the
transistor connected to this pin to oscillate around 200MHz.
National Semiconductor has made certain modifications
that eliminate this problem.

The switching mode has the advantages of higher linear
range, insensitivity to the amplitude of the injection voltage,
and can be used in afc systems or when side-responses are to
be avoided. On the other hand, the linear mode is preferred
for communication equipment, due to the preservation of
the tuned circuit bandwidth, affording better rejection to
Gaussian noise. Note the capability of the circuit to operate
in any of the modes or to combine both as a function of
the signal strength.

Narrowband FM I F amplifiers for scanners and two way
VHF radios can also be built with the LM2111. IF
amplifiers for this service are generally double conversion.
This system retains the good image rejection of a high IF
with the stability and higher percentage deviation (larger
recovered audio) of a low IF.
3-14

Cl

120pF

Rl = 3.9k
Ll = 1.B411Hy

r

12V

I

0.01

O•1

Y.

50l1F

AUDIO OUTPUT

2k

INPUT LIMITING VOLTAGE 300"V
RECOVERED AUDIO
450mVRMS
AM REJECTION (30% AM)
40dB
DISTORTION
0.3%

FIGURE 3.6.5 FM IF Amplifier

Typical values for the discriminator circuit, Figure 3.6.3,
for this application are:
Ll

90llHy

Cl

1500pF

C2

47pF

characteristic to drive the other input of the differential
peak detector. Transistors 022, 023, 026, and 027 form
the peak detectors, while capacitors C3 and C4 act as the
detector storage capacitors.
The electronic volume control consists of transistors 05
through 010. A zener diode 02 provides bias for a resistor
bridge RlO, R11, R14 and the external control potentio'
meter. This potentiometer, connected to pin 6, biases 06,
07, and 010 ON and Oa and 09 OFF when it is at a
maximum value, typically 50k.Q. As the pot is decreased,
transistors Oa and 09 begin to conduct, producing audio at
the output, pin a.

Rl = 5.6k.Q

150-

An audio preamplifier with a voltage gain of 10 is included,
with provision for addition of a tone control circuit to pin
13.

114MHz
455kHz

1st lOCAL
OSCillATOR

e.
t

Typical performance of the circuit shown in Figure 3.6.7
follows:

BOElECTOR

2nd LOCAL
OSCILLATOR

fo

= 4.5MHz, tof =

25kHz

Input limiting voltage

FIGURE.3.6.6 Narrowband FM Receiver

20llV

Recovered audio
3.6.2 LM3065-LM3075 SERIES IF DETECTORS
The LM3065 was originally developed for TV sound
applications and has certain advantages in this application
over the LM2111. In particular, it features a differential
peak detector which produces less harmonic energy of
(the 4.5MHz) carrier and hence fewer problems in the
crowded electromagnetic environment of the television
receiver. It has better AM rejection and features a DC
volume control for manufacturing economy.

700mV

AM rejection (30% AM)

50dB

Distortion

0.7%

The LM3075, shown in Figure 3.6.a, is similar in design to
the LM3065, but operates at higher IF frequencies (10.7
MHz and above), and does not have a DC volume control.
In a typical consumer FM application, the circuit as shown
has the following performance:
VCC = 12V, fo = 10.7 MHz, tof
Input limiting voltage

The circuit is shown in Figure 3.6.7. Like the LM2111,
there are three limiting stages of gain: an active filter, tuned
to approximately 5MHz (Cl, C2, etc.) limits the bandwidth and noise, while improving the AM rejection ratio.
Pin 9 is the signal output of the I F and drives one side of a
differential peak detector, while a pole-zero network
between pins 9 and 10 provides an amplitude vs. frequency

250/LV

AM rejection

55dB

Recovered audio

1.5V

THO
Audio preamp voltage gain
Power supply current
3-15

75kHz

1%
21dB
lamA

----~1O

R41
150

12

Schematic Diagram

"

t140V

R,
DC

VOLUME
CONTROL

:I

~-T- -----,
NO CONNECTION

I

I
I

:1)1
T

12pF

Block Diagram

fiGURE 3.6.7 LM3065 I F Detector

----------------3-16

'"'"
""

R26
15n

R2H
75K

'"

2.,K

""

'"

'"200

6

o

"o

All resistance values are 111 ohms
All Gall3Cllallce values are in pF

Schematic Diagram

v'" 11.2V

10

DETECTOR

,,'T3
'NPUTJ)
12K

""'1'
6.8K

Block Diagram

FIGURE 3.6.8 LM3075 FM IF-Detector-Preamp

3-17

3.7 THE LM3089 - TODAY'S MOST POPULAR FM IF
SYSTEM

r----'
I

I

I~I

IL- _ _ _ _ -"I
V+
10.7 MHz
INPUT

1

0.01

t--1---0g~~PUT
AUDIO

OUTPUT

51
2.7k

o,olr
M'.,L

~~~A,W

...~_--"154-o--I

RF AMPLIFIER

10k

14

13

12

470

33k

500
k
IF LEVEL
150 pA METER

MUTING

SENSITIVITY

FIGURE 3.7.1 LM30B9 Block and Connection Diagram

3.7.1 Introduction

IF Amplifier

LM3089 has become the most widely used FM IF amplifier
IC on the market today. The major reason for this wide
acceptance is the additional auxiliary functions not normally found in IC form. Along with the I F limiting amplifier
and detector th,e following functions are provided:

The I F amplifier consists of three direct coupled amplifierlimiter stages 01-022. The input stage is formed by a
common emitter/common base (cascode) amplifier with
differential outputs. The second and third I F amplifier
stages are driven by Darlington connected emitter followers
which provide DC level shifting and isolation. DC feedback
via R1 and R2 to the input stage maintains DC operating
point stability. The regulated supply voltage for each stage
is approximately 5 V. The IF ground (pin 4) is used only for
currents associated with the IF amplifiers. This aids in
overall stability. Note that the current through R9 and Zl
is the only current on the chip directly affected by power
supply variations.

1. A mute logic circuit that can mute or squelch the audio
output circuit when tuning between stations.
2. An I F level or signal strength meter circuit which
provides a DC logarithmic output as a function of IF
input levels from 10MV to 100mV (four decades).
3. A separate AFC output which can also be used to drive
a center-tune meter for precise visual tuning of each
station.

Ouadrature Detector and I F Output
FM demodulation in the LM3089 is performed accurately
with a fully balanced multiplier circuit. The differential IF
output switches the lower pairs 034. 026 and 039. 038.
The IF output at pin 8 is taken across 390n (R31) and
equals 300mV peak to peak. The upper pair-switching
(035. 023) leading by 90 degrees is through the externally
connected quad coil at pin 9. The 5.6V reference at pin 10
provides the DC bias for the quad detector upper pair
switching.

4. A delayed AGC output to control front end gain.
The block diagram of Figure 3.7.1 shows how all the major
functions combine to form one of the most complex FM
I F amplifier/limiter and detector ICs in use today.

3.7.2 Circuit Description (Figure 3.7.2)
The following circuit description divides the LM3089 into
four major subsections:

AFC. Audio and Mute Control Amplifiers

IF Amplifier
Ouadrature Detector and I F Output
A'FC. Audio and Mute Control Amplifiers
I F Peak Detectors and Drivers

The differential audio current from the quad detector circuit
is converted to a single ended output source for AFC by
"turning around" the 047 collector current to the collector
of 057. Conversion to a voltage source is done externally
3-18

IF Amplifier
V'
11

\'3.

G,

Zk

J

G.

°rz,
2k

R,O
2k

0,

R11
Zk

>--

~

SUBSTRATE

~
IF INPUT

G,

,

,

;,''

4::

G,

0,

~~2A

::

INPUT
BYPASSING

R,
IF

~

ro,

R,

".

R,

>--

Q1JA

~

M

r1i

Gzz

~

~

M

'"

'--

RI1
2.7k

R13
2.7k

R"

,

R"
48.

'80

R"

R"
2.7k

'50

R15
2.7k

R17

RZZ

'50

1.5k

r

040 039

114,

R"

1""

1.5k
~

2.2k

0"

~

-=

G"

'80

~

5.6V

Rl6

R"

'00
~

'00
~

REF,

10

OJJ 032

~

0"

48.

GNO

z,

-=
SIAS

GUA JRATURE
INPU T

9

;JJ

1""'1

Q25 024

R34
10k

R"

~

Rll
SOD

-=-

~

0,

R"

75'

r-

R,

]6'

>-K

II4,

-

R"
2k

~
011

0,

G"

.'~"

114,

Rll
500

R27
'50

..,

500

t1

0"

R"
2k

Q'4A

RIGA
2k

~

'i
30k

O'A

~

R18
2k

R1SA
2k

G'A

GNDAND

R19
2k

R"
50.

W"M""j

G15

;l

~

"'

Zk

''',

~

Quadrature Detector/IF Output

,

IF

RZ6
10k

~34

Q2~

-=-

-r

"":t

OUTPUT

038

"'I.

.,
~

't'
<0

c--

,.--5k
R51

,_~59

"li:?i6

'50

061

'f}

J'-~G"
15

059

AGe
OUTPUT

R52

U7

'00
~

r

J~60'00

Q3>

R61
Ok

4::

~

R57
10k

R54
5k

'00

""~
0"

i' ~

.,,"'-J
0"

R53

600
~

0"

"'r

R"

~

G67

114.

114,

'00
L...-

r-t:

G50

052
OUTPUT
~ AFe

6 AUDIO

R49
5k

R46
Uk

OUTPUT

0"

rK°

f!
0"

t-K

0"

78

"'IS

R64

'00

50'

013

R58
50

fJ'I
-

MUTE
CONTROL
OUTPUT

11

13
IF
lEVEl
METER

5
MUTE
INPUT

~81
~

VO"

~

0.,
R6J

'00

J

Figure 3.7.2 LM3089 Schematic Diagram

;

H"

055

Q79

-

L 0.57

..,

R4)

50.
-

OUTPUT

IF Peak Detectors and Drivers

ROJ

SOD

~

0"

".
-

500

114,

G53

R"

R55
13k

,"

114'

500

500

1146

G"

m''U~

':1; ~~

500

G))

AFC, Audio and Mute Control Amplifiers

-

50.

-

COMPONENT SIDE

150

JJA

METER

AUDIO

TO STEREO
OECODER

(a) PC Layout (Full Scale)

Vee
5.6k

IF INPUT

o--j ~~----'-I
0.01
5.6k

>--_...

r

O01
.

(b) Test Circuit
FIGURE 3.7.3 LM3089 Typical Layout & Test Circuit

3-20

~~~PUT

by adding a resistor from pin 7 to pin 10. The audio amplifier stage operates in a similar manner as the AFC amplifier
except that two "turn around" stages are used. This
configuration allows the inclusion of muting transistor 080.
A current into the base of 080 will cause transistors 079
and 081 to saturate, which turns off the audio amplifier;
the gain of the audio stage is set by internal resistor R49.
This 5 kQ resistor value is also the output impedance of the
audio amplifier. When the LM3089 is used in mono
receivers the 75f.1s de·emphasis (RC time constantl is
calculated for a 0.01 f.1F by including R49.
(RC ~ [R49 + Rll [Cll, Rl ~ 75f.1s/0.01f.1F - 5kQ"'2.7k,
Figure 3.7.1.1

I F Peak Detectors

~nd

require quad coil bandwidth equal to 800 kHz

Given:

fo

~

10.7MHz
~

0u (unloaded I
Find:

75

LCH and REXT

Find loaded 0 of quad coil for required BW (OLi
0L

~ ~ ~ ~0.7MHz ~
BW

13.38

0.8MHz

Find total resistance across quad coil for required BW (RTI

Find reactance of coupling choke (X LCHI

Drivers

Four I F peak or level detectors provide the delayed AGC,
I F level and mute control functions. An output from the
firs! IF amplifier drives the delayed AGC peak detector.
Since the first IF amplifier is the last IF stage to go into
limiting, 060 and 061 convert the first IF output voltage
swing to a DC current (for I F input voltages between 10mV
and 100mVI. This changing current (0.1 to 1 mAl is
converted to a voltage across R51. Emitter follower 058
buffers this output voltage for pin 15. The top of resistor
R51 is connected to a common base amplifier 074 along
with the output currents from the 2nd and 3rd stage IF
peak detectors (which operate for IF input voltages between
l0f.1V and 10mVI. The output current from 075 is turned
around or mirrored by 075, 076, and 077, cut in half, then
converted to a voltage across R61. Emitter follower 084
buffers this voltage for pin 13.

XLCH

RT V8
--V9

~

~

.1981xO.l10 ~ 1453Q
0.15

Find inductance of coupling choke (LCHI
X LCH
LCH ~ ~ ~

1453Q ~ 22f.1H
6.72 x 10 7

Find parallel resistance of the unloaded quad coil (Rpl
Rp

~

XUOUL

~

~

148Qx75

11.1kQ

Convert R31, LCH series to parallel resistance (RL311
(XLCHI2
RL31 - - - R31

The fourth peak detector "looks" at the IF voltage
developed across the quad coil. For levels above about
120 mV at pin 9, 073 will saturate and provide no output
voltage at pin 12. Because the I F level at pin 9 is constant,
as long as the last I F amplifier is in limiting, pin 12 will
remain low. Sudden interruptions or loss of the pin 9 IF
signal due to noise or detuning of the quad coil will allow
the collector of 073 to rise quite rapidly. The voltage at the
collector of 073 is buffered by 078 for pin 12.

~

1453 2

~

5413Q

390

Find REXTfor RT~ RpllRL3111REXT
~

REXT ~

Use REXT

~

4348

4.3k.

3.7.3 Stability Considerations
Because the LM3089 has wide bandwidth and high gain
(> 80dB at 10.7 MHzl. external component placement and
PC layout are critical. The major consideration is the effect
of output to input coupling. The highest IF output signal
will be at pins 8 and 9; therefore, the quad coil components
should not be placed near the I F input pin 1. By keeping
the input impedance low « 500QI the chances of output
to input coupling are reduced. Another and perhaps the
most insidious form of feedback is via the ground pin
connections. As stated earlier the LM3089 has two ground
pins; the pin 4 ground should be used only for the IF input
decoupling. The pin 4 ground is usually connected to the
pin 14 ground by a trace under the IC. Decoupling of
VCC (pin 111. AGC driver (pin 151. meter driver (pin 131,
mute control (pin 12) and in some cases the 5.6V REF
(pin 101 should be done on the ground pin 14 side of the
IC. The PC layout of Figure 3.7.3 has been used successfully
for input impedances of 500Q (1 kQ source/l kQ load).

lMJ089

R31

,

v,

v,

~ ,." ' t

110rnV
RMS

~
r

t ·11gl

IC1~t

-,

1100pF

1..._

"" 150mVRMS (REQUIRED FOR
MUTE CIRCUIT)

-1

"HI

REXT

.J

-l
FIGURE 3.7.4 Quad Coil Equivalent Circuit

3.7.5 Typical Application of the LM3089
The circuit in Figure 3.7.5 illustrates the simplicity in
designing an FM I F. The ceramic filters used in this
application have become very popular in the last few years
because of their small physical size and low cost. The filters
eliminate all but one IF alignment step. The filters are
terminated at the LM3089 input with 330Q. Disc ceramic
type capacitors with typical values of 0.01 to 0.02f.1F
should be used for IF decoupling at pins 2 and 3.

,3.7.4 Selecting Quad Coil Components
The reader can best understand the selection process by
example (see Figure 3.7.41:
3-21

+12V
VCC~------~-----------------------------------------, RIGHT

LEFT
OUTPUT

OUTPUT

CERAMIC FILTERS
IF INPUT
FROM TUNER

DELAYED
AGC
TO TUNER

_---+---,

0.33

10

19kHz
MONITOR

3.9
k

SIGNAL
STRENGTH

3.9k

150pA
FS

+ -

2-=-

3,uF
00pF

'-SEE SECTION 3.8.4
CENTER
TUNE
50·0·50
pA

MUTE
THRESHOLD

7.5k

r

O,
•

-=-

13k

-=- -=-

AFC
TO TUNER
(5.6 V ± 7 mY/kHz)

FIGURE 3.7.5 Typical Application of the LM3089

150
125
100
I
75
z
50
0:
25
0
0
!!O
-25
iii
-50
a:
-75
-100
-125

The AFC output at pin 7 can serve a dual purpose. In
Figure 3.7.6 AFC sensitivity, expressed as mV/kHz, is
programmed externally with a resistor from pin 7 to pin 10.
A voltage reference other than pin 10 may be used as long
as the pin 7 voltage stays less than 2V from the supply and
greater than 2 V from ground. The voltage change for a
5kn resistor will be"" 7.5mV/kHz or "" 1.5I.1A/kHz. The
AFC output can also be used to drive a center tune meter.
The full scale sensitivity is also programmed externally.
The wide band characteristics of the detector and audio
stage make the LM3089 particularly suited for stereo
receivers. The detector bandwidth extends greater than
1 MHz, therefore the phase delay of the composite stereo
signal, especially the 38 kHz side bands, is essentially zero.
This wide bandwidth will become more important in the
future when four channel stereo transmissions become a
reality.

1

"""~
~
"'
<-1~

/

...
...

1

1/

/

B

/

-150
-100

-50

+50

+100

CHANGE IN FREQUENCY 1M) - kHz

FIGURE 3.7.6 AFC (Pin 7) Characteristics vs_ IF Input Frequency
Change

- '"

The audio stage can be muted by an input voltage to pin 5.
Figure 3.7.8 shows this attenuation characteristic. The
voltage for pin 5 is derived from the mute logic detector
pin 12. Figure 3.7.7 shows how the pin 12 voltage rises
when the IF input is below 1001.lV. The 470n resistor and
0.331.1F capacitor filter out noise spikes and allow a smooth
mute transition. The pot is used to set or disable the mute
threshold. When the pot is set for maximum mute sensitivity
some competitors' versions of the LM3089 would cause a
latch-up condition, which results in pin 12 staying high for
all IF input levels. National's LM3089 has been designed
such that this latch-up condition cannot occur.

\.

\

z

0:

2

3

5

10

\

20 30

'>--

50 100

IF INPUT VOLTAGE -"v

The signal strength meter is driven by a voltage source at
pin 13 (Figure 3.7.9). The value of the series resistor is
determ ined by the meter used:

FIGURE 3.7_7 Mute Control Output (Pin 12) vs. IF Input Signal

3-22

3.8 FM STEREO MULTIPLEX
3.8.1 Introduction

m

.,z
I

10

~

20

\

:§ 30

~
z

\
\

40

~

\

50

"'"

60

'"

70

§i

The LM 131 0/1800 is a phase locked loop FM stereo
demodulator. In addition to separating left (L) and right (R)
signal information from the detected I F output, this IC
family features automatic stereo/monaural switching and a
100 mA stereo indicator lamp driver. The L.,M 1800 has the
additional advantage of 45dB power supply rejection.
Particularly attractive is the low external part count and
total elimination of coils. A single inexpensive potentiometer' performs all tuning. The resulting FM stereo system
delivers high fidelity sound while still meeting the cost
requirements of inexpensive stereo receivers.

1\

\

0.5

1.0

1.5

2.0

2.5

3.0

MUTE INPUT VOLTAGE ,PIN 51- V

FIGURE 3.7.8 Typical Audio Attenuation (Pin 6) vs. Mute Input
Voltage (Pin 5)

Pin 15

Pin 13

/

>

Figures 3.8.1 and 3.8.2 outline the role played by the
LM1310/1800 in the FM stereo receiver. The frequency
domain plot shows that the composite input waveform
contains L+R information in the audio band and L-R
information suppressed carrier modulated on 38 kHz. A
19kHz pilot tone, locked to the 38kHz subcarrier at the
transmitter, is also included. SCA information occupies a
higher band but is of no importance in the consumer FM

lX'

I

....
~

2

c

receiver .

/ \
\

">

u

/

The block diagram (Figure 3.8.2) of the LM 1800 shows
the composite input signal appl ied to the audio frequency
amplifier, which acts as a unity gain buffer to the decoder
section. A second amplified signal is capacitively coupled to
two phase detectors, one in the phase locked loop and the
other in the stereo switching circuitry. In the phase
locked loop, the output of the 76kHz voltage controlled
oscillator (VCO) is frequency divided twice (to 38, then
19kHz), forming the other inputto the loop phase detector.
The output of the loop phase detector adjusts the VCO to
precisely 76kHz. The 38kHz output of the first frequency
divider becomes the regenerated subcarrier which demodulates L-R information in the decoder section. The amplified
composite and an "in phase" 19kHz signal, generated in
the phase locked loop, drive the "in phase" phase detector.
When the loop is locked, the DC output voltage of this
phase detector measures pilot amplitude. For pilot signals
sufficiently strong to enable good stereo reception the
trigger latches, applying regenerated subcarrier to the
decoder and powering the stereo indicator lamp. Hysteresis,
built into the trigger, protects against erratic stereo/
monaural switching and the attendant lamp flicker.

V,
25

1

25

10

25

100

25

lk

25

10k

lOOk

IF INPUT VOLTAGE - fJ.V

FIGURE 3.7.9 Typical AGe (Pin 15) and Meter Output (Pin 13) vs.
IF Input Signal

AUDIO OUTPUT

10
20

~
....

30

It

±75kHz

r- "" 4DOmVRMs_DEVIATION

V

"

~ 40

;; 50 t--60

~~OISE OUTPUT

,'IHP334i DISTORTlONANALYZERI

\.

70
2 5

1

2 5

10

2 5

100

2!i

lk

2 5

10k

lOOk

IF INPUT VOLTAGE - pV

In the monaural mode (electronic switch open) the decoder
outputs duplicate the composite input signal except that
the de-emphasis capacitors (from pins 3 and 6 to ground)
roll off with the load resistors at 2 kHz. In the stereo mode
(electronic switch closed), the decoder demodulates the
L-R information, matrixes it with the L+R information,
then delivers buffered separated Land R signals to output
pins 4 and 5 respectively.

FIGURE 3.7.10 Typical (S + N)/N and IF Limiting Sensitivity vs.
I F Input Signal

RS = VMAX(13)
IFS

5V
150pA

33k

The maximum current from pin 13 should be limited to
approximately 2mA. Short circuit protection has been
included on the chip.

Figure 3.8.3 is an equivalent schematic of an LM 1800. The
LM1310 is identical except the output turnaround circuitry
(035-038) is eliAlinated and the output pins are connected
to the collectors of 039-042. Thus the LM1310 is essentially
a 14 pin version of the LM1800, with load resistors
returned to the power supply instead of ground. The
National LM1800 is a pin-for-pin replacement for the
UA758, while the LM1310 is a direct replacement for the
MC1310.

The delayed AGC (pin 15) is also a voltage source (Figure
3.7.9). The maximum current should also be limited to
approximately 2mA.
Figure 3.7.10 shows the typical limiting sensitivity (measured at pin 1) of the LM3089 when configured per Figure
3.7.3b and using PC layout of Figure 3.7.3a.
3-23

COMPOSITE INPUT SIGNAL TO LM1800,

Vc '" (L + RI + (l- RI COS wst+ kCOSwpt

L+R
SCA

7S

lS

FREQUENCY
X 1000 Hz

19 kHz PilOT
Vc

FM
FRONT
END

FM/IF

lM1800
STEREO
DEMODULATOR

AMPLIFIER
DETECTOR

POWER AMP & TONE CONTROL
LEFT
RIGHT

FIGURE 3.8.1 FM Receiver Block Diagram and Frequency Spectrum of LM1800 Input Signal

21K

5K

STEREO
INDICATOR
LAMP

3900
COMPOSITE
INPUT

3900
RIGHT

LEFT

OUTPUTS

FIGURE 3.8.2 LM1800 Block Diagram.

3-24

V(lLTAGE
RECULATOR

LOOP
PfiASE
OtTHTOR

."

AUOIO

INPHASl
PHASE
DETECTOR

FIGURE 3.8.3 LM1800 Equivalent Schematic

The capture range of the LM 1800 can be changed by
altering the external RC product on the VCO pin. The loop
gain can be shown to decrease for a decrease in VCO
resistance (R4 + R5 in Figure 3.8.4). Maintaining a constant
RC product, while increasing the capacity from 390pF to
510pF, narrows the capture range by about 25%. Although
the resulting system has slightly improved channel separa·
tion, it is more sensitive to VCO tuning.

3.8.2 LM1800 Typical Application
The circuit in Figure 3.8.4 illustrates the simplicity of
designing an FM stereo demodulation system using the
LM1800. R3 and C3 establish an adequate loop capture
range and a low frequency well damped natural loop
resonance. C8 has the effect of shunting phase jitter, a
dominant cause of high frequency channel separation
problems. Recall that the 38kHz subcarrier regenerates by
phase locking the output of a 19 kHz divider to the pilot
tone. Time delays through the divider result in the 38 kHz
waveform leading the transmitted subcarrier. Addition of
capacitor C9 (O.0025!1F) at pin 2 introduces a lag at the
input to the phase lock loop, compensating for these
frequency divider delays. The output resistance of the
audio amplifier is designed at 500n to facilitate this

When the circuits so far described are connected in an
actual FM receiver, channel separation often suffers due to
imperfect frequency response of the IF stage. The input
lead network of Figure 3.8.5 can be used to compensate for
roll off in the IF and will restore high quality stereo sound.
Should a receiver designer prefer a stereo/monaural switch·
ing point different from those programmed into the

connection.

Cl0
O.lIolFT

~~~?:~0~25~-e~-----------------------+--------~
":'"COMPOSITE ~ +
INPUT

I

C6

1DOmA

R5

STEREO
lAMP

5K
VCO

LM1800

ADJUST

10
C5
O.J3IJF

-=
TOP VIEW

FIGURE 3.8.4 LM1800 Typical Application

3·25

in series to limit current to a safe value for the LED. The
lamp or LED can be powered from any source (up to 18 V),
and need not necessarily be driven from the same supply as
the LM1800.

C
0.0022

OU~~~~~:
r-3h
RECElVER~""""OFlMI"O

3.8.3 LM1310 Typical Application

'"LTOPINI

Figure 3.8.7 shows a typical stereo demodulator design
using the LM1310. Capture range, lamp sensitivity adjustment and input lead compensation are all accomplished in
the same manner as for the LM 1800.

10K

FIGURE 3.8.5 Compensation for Receiver IF RoUoff

LM1800 (pilot: 15mVRMS on, 6.0mVRMS off typical),
the circuit of Figure 3.8.6 provides the desired flexibility.
The user who wants slightly increased voltage gain through
the demodulator can increase the size of the load resistors
(Rl and R2 of Figure 3.8.4), being sure to correspondingly
change the de'emphasis capacitors (Cl and C2). Loads as
high as 5600n may be used (gain of 1.4), Performance of
the LM1800 is virtually independent of the supply voltage
used (from 10 to 16 V) due to the on·chip regulator.

16k

OSCillATOR
ADJUST 5k

lOOk

.,

10k
lOOk

6.2k

CENTER

R1 POT SETIING

FIGURE 3.B.7 LM1310 Typical Application
FIGURE 3.B.6 Stereo/Monaural Switch Point Adjustment

3.8.4 Special Considerations of National's LM1310/1800
Although the circuit diagrams show a 100 mA indicator
lamp, the user may desire an LED. This presents no
problem for the LM1800 so long as a resistor is connected

A growing number of FM stereo systems use the industry
standard IF (LM3089) with an industry standard demodu·
lator (LM1310/1800) as in Figure 3.8.8.

elJ

+12V

02

A"

19kHl

TP

22'

A"

t-'o"'sce..:'''''''--:1f----+

~.~~

e2l

~.~k

11.22

C"
.033
c"

,
"=r300PF

*
+12V

FIGURE 3.B.B LM30B9/LM1BOO Application

3·26

The optional 300pF capacitor on pin 6 of the LM3089 is
often used to limit the bandwidth presented to the
demodulator's input terminals. As the I F input level
decreases and the limiting stages begin to come out of
limiting, the detector noise bandwidth increases. Most
competitive versions of the LM1310 would inadvertently
AM detect this noise in their input "audio amplifier,"
resulting in decreased system signal-to-noise. They therefore
require the 300pF capacitor, which serves to eliminate this
noise from the demodulator's input by decreasing bandwidth, and thus the system maintains adequate SIN.

where V 1 is the RMS amplitude of the fundamental and
V2, V3, V4, ... are the RMS amplitudes of the individual
harmonics.

The National LM1310 has been designed to eliminate the
AM noise detection phenomenon, giving excellent SIN
performance either with or without a bandlimited detected
IF. Channel separation also is improved by elimination of
the 300pF capacitor since it introduces undesirable phase
shift. The National LM1800 has the same feature, as do
competitive 16 pin versions.

Input Bias Current: The average of the two input currents.

IF Bandwidth: The range of frequencies centered about the
I F frequency limited by the -3dB amplitude points.
IF Selectivity: The ability of the IF stages to accept the
signal from one station while rejecting the signal of the
adjacent stations; it is the ratio of desired to undesired
signal required for 30dB suppression of the undesired signal
(lHF Std.).
Input Resistance: The ratio of the change in input voltage
to the change in input current on either input with the
other grounded.
Input Sensitivity: The minimum level of input signal at a
specified frequency required to produce a specified signalto-noise ratio at the recovered audio output.

For systems demanding superior THO performance, the
LM 1800A is offered with a guaranteed maximum of 0.3%.
Representing the industry's lowest THO value available in
stereo demodulators, the LM 1800A meets the tough
requirements of the top-of-the-line stereo receiver market.

Input Voltage Range: The range of voltages on the input
terminals for which the amplifier operates within specifications.
Large-Signal Voltage Gain: The ratio of the output voltage
swing to the change in input voltage required to drive the
output from zero to this voltage.

Utilization of the phase locked loop principle enables the
LM131011800 to demodulate FM stereo signals without
the use of troublesome and expensive coils. The numerous
features available on the demodulator make it extremely
attractive in a variety of home and automotive receivers.
Indeed the LM 131 0/1800 represents today's standard in
integrated stereo FM demodulators.

Limiting Sensitivity: In FM the input signal level which
causes the recovered audio output level to drop 3dB from
the output level with a specified large signal input.
Limiting Threshold: See Limiting Sensitivity.
Monaural Channel Unbalance: The ratio of the outputs
from the right and left channels with a monaural signal
applied to the input.

3.9 DEFINITION OF TERMS

Noise Figure: The common logarithm of the ratio of the
input signal-to-noise ratio to the output signal-to-noise
ratio.

AGC DC Output Shift: The shift of the quiescent IC output
voltage of the AGC section for a given change in AGC
central voltage.

Output Resistance: The ratio of the change in output
voltage to the change in output current with the output
around zero.

AGC Figure of Merit (AGC Range): The widest possible
range of input signal level required to make the output
drop by a specified amount from the specified maximum
output level.

Output Voltage Swing: The peak output voltage swing,
referred to zero, that can be obtained without clipping.

AGC Input Current: The current required to bias the
central voltage input of the AGC section.

Power Bandwidth: That frequency at which the voltage
gain reduces to 11..[2 with respect to the flat band voltage
gain specified for a given load and output power.

AM Rejection Ratio: The ratio of the recovered audio
output produced by a desired FM signal of specified level
and duration to the recovered audio output produced by an
unwanted AM signal of specified amplitude and modulating
index.

Power Supply Rejection: The ratio of the change in input
offset voltage to the change in power supply voltages
producing it.
Recovered Audio: The value of the audio voltage measured
at the output under the specified circuit conditions.

AM Suppression: See AM Rejection Ratio.
Capture Ratio: A measure of an FM tuner's ability to reject
an interfering signal of the same frequency as the desired
signal (i.e., operating on the same carrier frequency); it is
the ratio of desired to undesired signal required for 30dB
suppression of the undesired signal (IHF Std.).

RF Noise Voltage: The equivalent input noise voltage of
the R F stage.
RF Transconductance: The ratio of the RF output current
to the RF input voltage.

Channel Separation: The level of output signal of an
undriven amplifier with respect to the output level of an
adjacent driven amplifier.

SCA Rejection: The ratio of the 67 kHz SCA signal at the
output to the desired output with the standard FCC signal
input.

Harmonic Distortion: That percentage of harmonic distortion being defined as one hundred times the ratio of the
root-mean-square (RMS) sum of the harmonics to the
fundamental. Percent harmonic distortion equals:

Sensitivity: See Limiting Sensitivity.

'(V22+ V32+ V42 + ...

Slew Rate: The internally limited rate of change in output
voltage with a large amplitude step function applied to the
input.

»)1, (100%)

Supply Current: The current required from the power
supply to operate the amplifier with no load and the output
at zero.

V1
3-27

4.0 Power Amplifiers
4.1

INSIDE POWER INTEGRATED CIRCUITS

Consider for a moment the problem in audio designs with
distortion (THDI. The buffer of Figure 4.1.1 is essentially
an emitter follower (NPN during positive half cycles and
PNP during negative halves due to class B operationl. As a
result the load presented to the collector of the gain
transistor is different depending on which half cycle the
output is in. The buffer amplifier itself often contributes in
the form of crossover distortion. Suppose for a moment
that the amplifier were to be used open loop (i.e., without
any AC feedbackl and that the result was an output signal
distorted 10% at 10 kHz_ Further assume the open loop
gain-frequency is as in Figure 4.1.2 so that the amplifier is
running at 60dB of gain. Now add negative feedback around
the amplifier to set its gain at 40dB and note that its
voltage gain remains flat with frequency throughout the
audio band. In this configuration there is 20dB of loop
gain (the difference between open loop gain and closed
loop gainl which works to correct the distortion in the
output waveform by about 20dB, reducing it from the 10%
open loop value to 1 %. Further study of Figure 4.1.2 shows
that there is more loop gain at lower frequencies wh ich
should, and does, help the THD at lower frequencies. The
reduction in loop gain at high frequencies likewise allows
more of the open loop distortion to show.

Audio power amplifiers manufactured using integrated
circuit technology do not differ significantly in circuit
design from traditional operational amplifiers. Use of
current sources. active loads and balanced differential techniques predominate, allowing creation of high-gain, wide
bandwidth, low distortion devices. Major design differences
appear only in the class AB high current output stages
where unique geometries are required and special layout
techniques are employed to guarantee thermal stability
across the chip.
The material presented in the following sections serves as a
brief introduction to the design techniques used in audio
power integrated circuits. Hopefully, a clearer understanding of the internal "workings" will result from reading the
discussion, thus making application of the devices easier.

4.1.1 Frequency Response and Distortion
Most audio amplifier designs are similar to Figure 4_1_1. An
input transconductance block (gm ; io/v11 drives a high
gain inverting amplifier with capacitive feedback. To this is
added an output buffer with high current gain but unity
voltage gain. The resulting output signal is defined by:
Vo ; v1 gm Xc

(4.1.11

or, 'ewriting in terms of gain:
Av ;

~ ; gm Xc
v1

gm
sC

gm
jwC

(4.1.21

"

Setting Equation (4.1.21 equal to unity allows solution for
the amplifier unity gain cross frequency:
Av = 1 ; ~ = gm
jwC
j2nfC
fUNITY;

~

(4.1.31
Av

(4_1.41

20dB

2nfC

DECADE

Equation {4.1.21 indicates a single pole response resulting in
a 20dB/decade slope of the gain-frequency plot in Figure
4.1.1. There is, of course, a low frequency pole which
is determined by the compensation capacitor and the
resistance to ground seen at the input of the inverting
.amplifier. Usually this pole is below 100Hz so it plays
only a small role in determining amplifier performance in
usual feedback arrangements.

Jl"'.
2rrfC

FIGURE 4_1.1 Audio Amp Small Signal Model

Av

For an amplifier of this type to be stable in unity gain
feedback circuits, it is necessary to arrange gm and C so
that the unity gain crossover frequency is about 1 MHz_ This
iS"in short, due to a few other undesirable phase shifts that
are difficult to avoid when using lateral PNP transistors in
monolithic realizations of the transconductance as well as
the buffer blocks. Figure 4.1.1 shows that if fUNITY
is 1 MHz then only 34dB of gain is available at 20kHz!
Since most audio circuits require more gain, most IC audios
are not compensated to unity. Evaluation of the LM380 or
LM377 will show stability troubles in loops fed back for
less than 20dB closed loop gain.

60 dB

40 dB

+ ......

1-;---......

'ClOSEO'lOOP
AMPLIFIER GAIN

I

I

'-----'~----~-

f(kHd

1020

FIGURE 4.1.2 Feedback and "Loop Gain"

4-1

4.1.2 Slew Rate

in proximity to an RF receiver. Among the stabilization
techniques that are in use, with varying degrees of success

Not only must IC audio amplifiers have more bandwidth
than "garden variety" op amps, they must also have higher
slew rates. Slew rate is a measure of the ability of an
amplifier's large signal characteristics to match its own
small signal responses. The transconductance block of Figure
4.1.1 delivers a current out for a given small signal input
voltage. Figure 4.1.3 shows an input stage typically used in
audio amplifiers. Even for large differential input voltage
drives to the PNP bases, the current available can never
surpass I. And this constant current (I) charging the com·
pensation capacitor (C) results in a ramp at Q1 's collector.
The slope of this ramp is defined as slew rate and usually is
expressed in terms of volts per microsecond. Increasing the
value of the current source does increase slew rate, but at
the expense of increased input bias current and gm. Large
gm values demand larger compensation capacitors which
are costly in IC designs. The optimum compromise is to
use large enough I to achieve adequate slew rate and then
add emitter degeneration resistors to the PNPs to lower gm.

are:

1. Placing an external RC from the output pin to ground
to lower the gain of the NPN. This works pretty well and
appears on numerous data sheets as an external cure.
2. Utilizing device geometry methods to improve the PNP's
frequency response. This has been done successfully in
the LM377, LM378, LM379. The only problem with
this scheme is that biasing the improved PNP reduces the
usable output swing slightly, thereby lowering output
power capability.
3. Addition of resistance in series with either the emitter
or base of Q3.
4. Making Q3 a controlled gain PNP of unity, which has
the added advantage of keeping gain more nearly equal
for each half cycle.
5. Adding capacitance to ground from Q3'S collector.
These last three work sometimes to some degree at most
current levels.

-IN

(a)
FIGURE 4.1.3 Typical gm Block

Slew rate can be calculated knowing only I and C:

flV

I

flt

C

(4.1.5)

To more clearly understand why slew rate is significant in
audio amplifiers, consider a 20kHz sine wave swinging
40 V p.p, a worst case need for most of today's audios. The
rate of change of voltage that this demands is maximum at
zero crossing and is 2.5 V//1s. Equation (4.1.6) is a general
expression for solving required slew rate for a given
sinusoid. (See Section 1.2.1.)
Slew rate ~

flV
At

~ 1T

f Vp.p

(b)

(4.1.6)

4.1.3 Output Stages
In the final analysis a buffer stage that delivers amperes of
load current is the main distinction between audio and op
amp designs. The classic class B is merely a PNP and NPN
capable of huge currents, but since the IC designer lacks
good quality PNPs, a number of compromises results. Figure
4.1.4b shows the bottom side PNP replaced with a com·
posite PNP/NPN arrangement. Unfortunately, Q2/Q3 form
a feedback loop which is quite inclined to oscillate in the
2·5MHz range. Although the oscillation frequency is well
above the audible range, it can be troublesome when placed

(c)
FIGURE 4.1.4 Basic Class B Output Drivers

4-2

Figure 4.1.5 illustrates crossover distortion such as would
result from the circuit in Figure 4.1.4b. Beginning with 01
"on" and the amplifier output coming down from the top
half cycle towards zero crossing, it is clear that the emitter
of 01 can track its base until the emitter reaches zero volts.
However, as the base voltage continues below 0.7V, 01
must turn off; but 02/03 cannot turn on until the input
generator gets all the way to -0.7V. Thus, there is a l.4V
of dead zone where the output cannot respond to the input.
And since the size of the dead zone is independent of
output amplitude, the effect is more pronounced at low
levels. Of course feedback works to correct this, but the
result is still a somewhat distorted waveform - one which
has an unfortunately distasteful sound. Indeed the feedback
loop or the composite PNP sometimes rings as it tries to
overcome the nonlinearity, generating harmonics that may
disturb the receiver in radio applications. The circuit of
Figure 4.1.4c adds "AB bias." By running current through
Dl and D2, the output transistors are turned slightly "on"
to allow the amplifier to traverse the zero volts region
smoothly. Normally much of the power supply current in
audio amplifiers is this AB bias current, running anywhere
from 1 to 15mA per amplifier.

The distortion components discussed so far have all been
in terms of circuit nonlinearities and the loop gain covering

them up. However, at low frequencies (below 100HzI
thermal problems due to chip layout can cause distortion.
In the audio Ie, large amounts of power are dissipated in
the output driver transistors causing thermal gradients across
the die. Since a sensitive input amplifier shares the same
piece of silicon, much care must be taken to preserve
thermal symmetry to minimize thermal feedback.
Despite the many restrictions on audio Ie designs, today's
devices do a credible job, many boasting less than 1% THD
from 20 Hz to 20 kHz - not at all a bad feat!

v+----~~----------

__---

DC
BIAS

01

02

,--_+-;C~

FIGURE 4.1.6 Simple Current Limit

4.1.4 Output Protection Circuitry
By the very nature of audio systems the amplifier often
drives a transducer - or speaker - remote from the
electronic components. To protect against inadvertent
shorting of the speaker some audio ICs are designed to self
limit their output current at a safe value. Figure 4.1.6 is a
simple approach to current limiting: here 05 or 06 turns
"on" to limit base drive to either of the output transistors
(01 or 021 when the current through the emitter resistors
is sufficient to threshold an emitter base junction. Limiting
is sharp on the top side since 05 has to sink only the
current source (I). However, the current that 06 must sink
is more nebulous, depending on the alpha holdup of 03,
resulting in soft or mushy negative side limiting. Other
connections can be used to sharpen the limiting action, but
they usually result in a marginally stable loop that must be
frequency compensated to avoid oscillation during limiting.
The major disadvantage to the circuit of Figure 4.1.6 is that
as much as 1.4 V is lost from loaded output swing due to
voltage dropped across the two RES.

FIGURE 4.1.5 Crossover Distortion

Some amplifiers at high frequencies (say 10kHzI exhibit
slightly more crossover distortion when negative going than
when positive going through zero. This is explained by the
slow composite PNPs' (02/031 delay in turning "on." If
the amplifier delivers any appreciable load current in the
top half cycle, the emitter current of 01 causes its baseemitter voltage to rise and shut "off" 03 (since the voltage
across Dl and D2 is fixed by II. Thus, fast negative going
signals demand the composite to go from full "off" to full
"on" - and they respond too slowly. As one might imagine,
compensating the loop (02 and 031 for stability even slows
the switching time more. This problem makes very low
distortion Ie amplifiers « 0.2%1 difficult at the high end
of the audio (20kHz).
Another interesting phenomenon occurs when some Ie
amplifiers oscillate at high frequencies - their power supply
current goes up and they die! This usually can be explained
by positive going output signals where the fast top NPN
transistor (011 turns "on" before the sluggish composite
turns "off," resulting in large currents passing straight down
through the amplifier (01 and 021.

The improved circuit of Figure 4.1.7 reduces the values of
RE for limiting at the same current but is usable only in
Darlington configurations. It suffers from the same negative
side sott"ness but only consumes about 0.4 V of output
swing. There are a few other methods employed, some even
consuming less than 0.4 V. Indeed it is further possible to
4·3

add voltage information to the current limit transistor's
base and achieve safe operating area protection. Care must
be taken in such designs, however, to allow for a leading or
lagging current of up to 60° to accommodate the variety of
speakers on the market. However, the circuitry shown in
Figures 4.1.6 and 4.1.7 is representative of the vast majority
of audio ICs in today's marketplace.

v+--~t_----_I~--

ZI

v+-----e----------~~--~t_--

DC
BIAS

FIGURE 4.1.8 Typical Thermal Shutdown

The addition of thermal shutdowns in audio ICs has done
much to improve field reliability. If the heat sinking is
inadequate in a discrete design, the devices burn up. In a
thermally protected IC the amplifier merely reduces drive
to the load to maintain chip temperature at a safe value.

R,

4.2 DESIGN TIPS ON LAYOUT, GROUND LOOPS AND
SUPPL Y BYPASSING
Layout, grounding and power supply decoupling of audio
power integrated circuits require the same careful attention
to details as preamplifier ICs. All of the points discussed in
Section 2.2 of this handbook apply directly to the use of
power amplifiers and should be consulted before use.
The relevant sections are reproduced here for cross-reference
and convenience:
Section
Section
Section
Section

2.2.1 Layout
2.2.2 Ground Loops
2.2.3 Supply Bypassing
2.2.4 Additional Stabilizing Tips

FIGURE 4.1_7 Improved Current Limit

4.3 POWER AMPLIFIER SELECTION
National Semiconductor's line of audio power amplifiers
consists of two major families: the "Duals," represented by
the LM377, LM378 and LM379 family, and the "Monos,"
represented by six products. Available power output ranges
from miniscule 320 mW battery operated devices to hefty
7W line operated systems. Designed for single supply
operation, all devices may be operated from split supplies
where required. Tables 4.3.1 and 4.3.2 summarize the dual
family for ease of selection, while Table 4.3.3 compares the
six mono devices. Figures 4.3.1-4.3.3 provide graphical
comparison of power output versus supply voltage for loads
of 4, 8 and 16 ohms.

Large amounts of power dissipation on the die cause chip
temperatures to rise far above ambient. In audio ICs it is
popular to include circuitry to sense chip temperature and
shut down the amplifier if it begins to overheat. Figure
4.1.8 is typical of such circuits. The voltage at the emitter
of Q1 rises with temperature due both to the TC of the
zener (Z1) and Q(s base-emitter Voltage. Thus, the voltage
at the junction of R1 and R2 rises while the voltage
required to threshold Q2's emitter-base junction falls with
temperature. In most designs the resistor ratio is set to
threshold Q2 at about 165°C. The collector current of Q2
is then used to disable the amplifier.

4-4

TABLE 4.3.1 Dual Power Amplifier CharacteristiC$

LM377N
(14 Pin DIP)

PARAMETER

Supply Voltage

LM379 2
UNITS

MIN

TYP

MAX

MIN

TYP

MAX

MIN

TYP

MAX

10

20

26

10

24

35

10

28

35

V

15

50

15

50

15

65

mA

Quiescent Supply Current
(POUT =OW)
Output Power 3
THD <;;; 5%
THD = 10%

LM378N
(14 Pin DIP)

2.5

2

4

5
6

Total Harmonic Distortion
POUT = 1 W/CH, f = 1 kHz
POUT = 2W/CH, f = 1 kHz
POUT = 4 W/CH, f = 1 kHz

0.07
0.10

1

0.07
0.10

1

6
7
0.07
0.20

W
W

1

%
%
%

Input Impedance

3

Open Loop Gain
(R s = On)

66

90

66

90

66

90

dBV

Channel Separation
(CF = 2501lF, f = 1 kHz)

50

70

50

70

50

70

dBV

Ripple Rejection
(f = 120Hz, CF = 2501lF,
input referred)

60

70

60

70

70

dBV
V Ills

3

3

Mn

Slew Rate

1.4

1.4

1.4

Equivalent Input Noise Voltage
(Rs = 600n, 100Hz·l0kHz)

3

3

3

1. Specification, apply forTTAB
wise specified.

=

25°C, RL = 8n, Av = 50 (34dBl.

IlVRMS

v, = 20V (LM377), V, = 24V (LM37Bl. V, = 28V (LM3791. unless other·

2. LM379S = 14 Pin "S" Type Power DIP.
3. For operation at ambient temperatures greater than 25°C the IC must be derated based on a maximum 150°C junction temperature using a
thermal resistance obtained from device data sheet.
4. Output protection included on all devices.

TABLE 4.3.2 Dual Audio Amplifier Typical Po

Supply
12
16
18
20
22
24
26
28
30

1. LM379.
2. LM378 (thermal limit).

4·5

@

10% THO

TABLE 4.3.3 Mono Power Amplifier Characteristics

Min

0.32

0.18

9

0.5

0.5

12

0.3

0.9

0.6

0.4

0.2

9

1.2

1.0

0.6

12

2.0

1.5

1.1

1.0

0.6

0.34

2.0

1.2

0.77

0.32

0.45

6

Typ

Max

4, [6'J

8, [12'J

Min

Typ

Max

Output
Protection

YES

(46)

No

1.0

10

23

26

30

23

26

30

32

34

36

YES

(46)

No

20

10
0.8

6

8

10

YES

(46)

No

20

20
2.5

1.5

0.5

14

3.3

2.2

1.0

18

4.2

4.0

2,2

4.0

2.2

5.7

3.5

12

LM380
(14 Pin DIP)

2.5 3

7

No

YES

No

YES

25

26

12

I

Min

Gain
Control
(Typ dB)

124

9

LM384
(14 Pin DIP)

Max

26

4

m

Min

Fixed Gain
(dB)

(rnA)

124

4

LM390
(14 Pin DIP)

Max

Max

Typ

0.25

Min

Typ

Quiescent Current
RL = 1602

Typ

Max

6

LM388
(14 Pin DIP)

RL=Sn

RL =402
Min

Typ

4
LM386
(S Pin DIP)
[LM389']

.,..

Output Power (W) at 10% THD

Supply Voltage
(V)

Device

18

4.2

22

3.5

1. Specifications apply for T A

=

5.0

25°C. For operation at ambient temperatures greater than 25°C the

Ie must be derated

8.5

25

based on a maximum 150°C junction temperature using a thermal resistance obtained from device data sheet.

2. LM389 identical to LM386, but includes three additional NPN transistors pinned out separately for customer use. 18 Pin DIP.

3. THO == 3%.
4. Parts selected for higher absolute maximum supply voltage available on special request.

34

10.0

10.0

I I I II I I II I I I I1IIII1 I I

tt RL

= 4[2
I I I I I

II

LM380 UPPER
I""""~
V, LIMIT

~J3~Or UP~E~
4.0

:c
....
~

....

2.0 r;-H+-I-I-..J,;i,bJ..-I

"#.

=

@I

u;
....
I;[
;;::: 1.0

.......J..J..J'

1 I I I I 1III1 I I

II

~
1J:g
~~
;; 5
~
1/

....
....

~ 1.0

....

~

~

c

a:
w

a:

0.5
0.4

rl17fititl-+++++lnItt
/rutttttt-++-J+jW-UillU

;;: 0.5
c
... 0.4

0.3

U'III 1\:1 I I I I I I I IIIII I I

0.3

II

0.2

~~
~

0.2

0.1

~~i IN I I 11111111111

~n 11111 N

11111111111

67891012

20

30

40

~II
V
'If

¥s.

Vs for RL

~

UPPER~

3.0

LlMIT,Vs,

I~~

~"II
g

2.0

@I

"LM384 LOWE R
V, LIMIT

u;
....
I;[

~

....

II

~J

.., II
1.0

~

C

~

!»

~ !!!

I'

~

;;: 0.5

~

\

~ ~Il

0.4

-,"'''' ill

o.3

]j

-

LM384 LOWER
LIMIT V,

g~1;t~ §rgg

~I

0.2

-/_

."

. .,

"";1

II

0.1
4

0.1

5 6 7 8 910

20

30

40

V, (VOLTS)

= 4 Ohms

LM380+.!.lttt1j

~

'"

w

V, (VOLTS)

FIGURE 4.3.1 Po

c

II

@I

c

~

2.0

u;

~

.:..,

11

!!!
~

c
:c

~

5.0
4.0

~

3.0
c

....,~

V,LlMIT

5.0

5.0 1 I I II I 1 I I I I I I II11111 I 1

III" 1111111

10.0U~~!I~d II

~-

RL =an

FIGURE 4.3.2 Po vs. Vs for RL = 8 Ohms

I;

I

4

I.'

I

I

I I I

I

!

t ! I I ! ! I'

5 6 7 8910

20

30

40

V, (VOLTS)

FIGURE 4.3.3 Po

¥s.

Vs for RL = 16 Ohms

4.4 LM377, LM378, AND LM379 DUAL TWO, FOUR,
AND SIX WATT POWER AMPLIFIERS

Further decPease of transconductance is provided by
degeneration caused by resistors at 02 and 03 emitters,
which also allow better large signal slew rate. The second
collector provides bias current to the input emitter follower
for increased frequency response and slew rate. Full differential input stage gain is provided by the "turnaround"
differential to single-ended current source loads 05 and 06.
The input common-mode voltage does not extend below
about 0.5 V above ground as might otherwise be expected
from initial examination of the input circuit. This is because
07 is actually preceded by an emitter follower transistor
not shown in the simplified circuit.

4.4.1 Introduction
The LM377, LM378 and LM379 are two·channel power
amplifiers capable of delivering 2,4, and 6 watts respectively
into 8 or 16n loads. They feature on·chip frequency com·
pensation, output current limiting, thermal shutdown
protection, fast turn·on and turn-off without "pops" or
pulses of active gain, an output which is self-centering at
VCC/2, and a 5 to 20MHz gain-bandwidth product. Applications include stereo or multi-channel audio power output
for phono, tape or radio use over a supply range of 10 to
35 V, as well as servo amplifier, power oscillator and various
instrument system circuits. Normal supply is single-ended;
however, split supplies may be used without difficulty or
degradation in power supply rejection.

The second stage 07 operates common-emitter with a
current source load for high gain. Pole splitting compen·
sation is provided by Cl to achieve unity gain bandwidth of
about 10MHz. Internal compensation is sufficient with
closed-loop gain down to about Av = 10.
The output stage is a complementary common-collector
class AS composite. The upper, or current sourcing section,
is a Darlington emitter follower 012 and 013. The lower,
or current sinking, section is a composite PNP made up of
014, 015, and 09. Normally, this type of PNP composite
has low ft and excessive delay caused by the lateral PNP
transistor 09. The usual result is poor unity gain bandwidth
and probable oscillation on the negative half of the output
waveform. The traditional fix has been to add an external
series RC network from output to ground to reduce loop
gain of the composite PNP and so prevent the oscillation.
In the LM377 series amplifiers, 09 is made a field·aided
lateral PNP to overcome these performance limitations and
so reduce external parts count. There is no need for the
external RC network, no oscillation is present on the
negative half cycle, and bandwidth is better with this output
stage. 010 and 011 provide output current limiting at

4.4.2 Circuit Description
The simplified schematic of Figure 4.4.1 shows the important
design features of the amplifier. The differential input
stage made up of 01-(4 uses a double (split) collector
PNP Darlington pair having several advantages. The high
base-emitter breakdown of the lateral PNP transistor is
about 60V, which affords significant input over·voltage
protection. The double collector allows operation at high
emitter current to achieve good first stage f t and minimum
phase shift while simultaneously operating at low transconductance to allow internal compensation with a physically small capacitor Cl. (Unity gain bandwidth of an
amplifier with pole·splitting compensation occurs where the
first stage transconductance equals WC1.)

BIAS = Vee

2

Vee

...- -....- .....-0 OUT

-IN'

+IN

GNO

FIGURE 4.4.1 Simplified Schematic Diagram

4-8

about 1.3A, and there is internal thermal limiting protection at 150°C junction temperature. The output may be AC
shorted without problem; and, although not guaranteed
performance, DC shorts to ground are acceptable. A DC
short to supply is destructive due to the thermal protection
circuit which pulls the output to ground.

operating point. To achieve good supply rejection XC2 is
normally made much smaller than a series resistor from the
bias divider circuit (RS in Figure 4.4.3). Where a supply
rejection of 40dB is required with 40dB closed-loop gain,
SOdB ripple attenuation is required of RSC2. The turn-on
time can be calculated as follows:

To achieve a stable DC operating point, it is desirable to
close the feedback loop with unity DC gain. To achieve
this simultaneously with a high AC gain normally requires
a fairly large bypass capacitor, C1, in Figure 4.4.2.

PSRR

T
C2

P
T

C5

01~

R3

tON

wRC

PSRR

SOdB

104

w

21T 120 Hz

754

T

"'-3

wT

13.3sec

4.5 seconds to small signal operation

tON'" 3T = 40 seconds to full output voltage swing
The 3T delay might normally be considered excessive! The
LM377 series amplifiers incorporate active turn-on circuitry
to eliminate the long turn-on time. This circuitry appeared
in Figure 4.4.1 as 016 and an accompanying SCR; it is
repeated and elaborated in Figure 4.4.3. In operation, the
turn-on circuitry charges the external capacitors, bringing
output and input levels to VCC/2, and then disconnects
itself leaving only the VCC/2 divider RB/RB in the circuit.
The turn-on circuit operation is as follows. When power is
applied, approximately VCC/2 appears at the base of 016,
rapidly charging C1 and C2 via a low emitter-follower
output impedance and series resistors bf 3k and 1 k. This
causes the emitters of the differential input pair to rise to
VCC/2, bringing the differential amp 03 and 04 into
balance. This, in turn, drives 03 into conduction. Transistors

FIGURE 4.4.2 Non~lnverting Amplifier Connection

Establishing the initial charge on this capacitor results in a
turn-on delay. An additional capacitor, C2, is normally
required to supply a ripple-free reference to set the DC

r--------------------.
v"

5_6k

R,
5k

R,
30k

1k

Rs
5k

3k

3k

TO
AMP

4.5k

'-------------6

B

L ___

13

1

BIAS ..._ _ _ _ _.....NIr-_ _ _ _ _ _-.;;IN+_ _ _ _-'O;.;U;.;T+-_ _ _'.;;IN.
R3

T

R1

C2

FIGURE 4.4.3 Internal Turn-On Circuitry

4-9

R2

02 and 03 form an SCR latch which then triggers and
clamps the base of 016 to ground, thus disabling the
charging circuit. Once the capacitors are charged, the
internal voltage divider RB/RB maintains the operating
point at VCCI2. Using C2 = 250f..(F, the tON = 3T '" 0.3s
and PSRR '" 75dB at 120Hz due to the 30k resistor RS.
Using C2 = 1000f..(F, PSRR would be 86dB. The internal
turn·on circuit prevents the usual "pop" from the speaker
at turn·on. The turn·off period is also pop·free, as there is
no series of pulses of active gain often seen in other similar
amplifiers.

1M

r-----l

0.4M

INAo-1~~~

I

Note that the base of 04 is tied to the emitters of only one
of the two input circuits. Should only one amplifier be in
use, it is important that it be that with input at pins 8 and 9.
4.3.3 External Biasing Connection

O.47pF

The internal biasing is complete for the inverting gain
connection of Figure 4.4.4 except for the external C2 which
provides power supply rejection. The bias terminal 1 may
be connected directly to C2 and the non·inverting input
terminals 6 and 9. Normal gain·set feedback connections to
the inverting inputs plus input and output coupling capaci·
tors complete the circuitry. The output will Q up to VCC/2
in a fraction of one second.

INo

o-1l-'VII'Ir....'-O-'-'--I

L ~M~l~/~9:.... J
A,
1M
*(LM379S pin nos. in parentheses)

lM377

Po

P

lM319

3W/CH

4W/CH

98mV MAX

113mVMAX

50

50
24V

50
28V

A,

C2

lM377/lM378

2W/CH
80mV MAX

Vee '"

18V

C,

Ol~

FIGURE 4.4.5 Inverting Stereo Amplifier

output signal to the + input. Bypass capacitors could be
added at + inputs to prevent such instability, but this
increases the parts count equal to that of the non·inverting
circuit of Figure 4.4.6, which has a superior input imped·
ance. For applications utilizing high impedance tone and
volume controls, the non-inverting connection will normally
be used.

A2

Av

=

Rl

FIGURE 4.4.4 Inverting Amplifier Connection
lOOk

The non·inverting circuit of Figure 4.4.2 is only slightly
more complex, requiring the input return resistor R3 from
input to the bias terminal and additional input capacitor
C3. Cl must remain in the circuit at the same or larger value
than in Figure 4.4.4.
4.4.4 2/4/6 Watt Stereo Amplifier Applications
The obvious and primary intended application is as an audio
frequency power amplifier for stereo or quadraphonic music
systems. The amplifier may be operated in either the non·
inverting or the inverting modes of Figures 4.4.2 and 4.4.4.
The inverting circuit has the lowest parts count so is most
economical when driven by relatively low·impedance cir·
cuitry. Figure 4.4.5 shows the total parts count for such a
stereo amplifier. The feedback resistor value of 1 meg in
Figure 4.4.5 is about the largest practical value due to an
input bias current max of approximately 1/2f..(A (100nA
typ). This will cause a -0.1 to 0.5V shift in DC output
level, thus limiting peak negative signal swing. This output
voltage shift can be corrected by the addition of series
resistors (equal to the RF in value) in the + input lines.
However, when this is done, a potential exists for high
frequency instability due to capacitive coupling of the

13
-=

4.1"

I."v-+-'-o-i

F-V+

O.lp

INA
liN

=

0-11----"
~0~:___-1~~

1M

0.1/.1

INo

0-11---+,

;Jr
F

+J\J"".........,..;

"

60
40

I
0

'"

S (Rl + R2) C2 + 1

R1

1 +---R2 +_1_
SC2

S R2 C2 + 1

ClIOnllll\O~PI

20

IIIIIIIL...H
100

1k

10k

I

lOOk

fz

I

1M

fp

f - FREQUENCY IHzl

(4.4.1 )

Zero at fz

FIGURE 4.4.17 Frequency Response of Non-Inverting Unity Gain
Amplifier

(4.4.2)

=

21T R1 C2

(4.4.3)
4.4.12 Bridge Amplifiers
Examination of Equation (4.4.1) shows it to have a
frequency response zero at fz (Equation (4.4.2)) and a pole
at f p (Equation (4.4.3)). By selecting f z to fall at the edge of
the audio spectrum (20 kHz as shown) and fp prior to hitting
the open loop response (340 kHz as shown) the frequency
response of Figure 4.4.17 is obtained. This response satisfies
the unity gain requirements, while allowing the gain to
raise beyond audio to insure stable operation.

The LM377 series amplifiers are equally useful in the bridge
configuration to drive floating loads, which may be loud·
speakers, servo motors or whatever. Double the power
output can be obtained in this connection, and output
coupling capacitors are n~t required. Load impedance may
be either 8 or 16£1 in the bridge circuit of Figure 4.4.18.
Response of this circuit is 20Hz to 160kHz as shown in
Figure 4.4.19 and distortion is 0.1 % midband at 4 W, rising
to 0.5% at 10kHz and 50mW output (Figure 4.4.20). The
higher distortion at low power is due to a small amount of
crossover notch distortion which becomes more apparent
at low powers and high frequencies. The circuit of Figure
4.4.21 is similar except for higher input impedance. In
Figure 4.4.21 the signal drive for the inverting amplifier is
derived from the feedback voltage of the non-inverting
amplifier. Resistors Rl and R3 are the input and feedback
resistors for A2, whereas R 1 and R2 are the feedback net·
work for Al. So far as Al is concerned, R2 sees a virtual
ground at the (-) input to A2; therefore, the gain of Al is
(1 + R2/Rl). So far as A2 is concerned, its input signal is
the voltage appearing at the (-) input to Al. This equals
that at the (+) input to Al. The driving point impedance at
the (-) input to Al is very low even though R2 is lOOk.
Al can be considered a unity gain amplifier with internal
R = R2 = lOOk and RL = R1 = 2k. Then the effective
output resistance of the unity gain amplifier is:

FIGURE 4.4.15 Inverting Unity Gain Amplifier

ROUT

RINTERNAL

lOOk

AOL/AfJ

600/1

167£1

Layout is critical if output oscillation is to be avoided. Even
with careful layout, capacitors Cl and C2 may be required
to prevent oscillation. With the values shown, the amplifier
will drive a 16£1 load to 4 W with less than 0.2% distortion
midband, rising to 1% at 20kHz (Figure 4.4.22). Frequency
response is 27 Hz to 60 kHz as shown in Figure 4.4.23. The
low frequency roll off is due to the double poles C3 R3 and
C4 R l·

75k

R2
4.7k

C2
.Il00PF

FIGURE 4.4.16 Non·lnverting Unity Gain Amplifier

4·15

lOOk

98k
2k

+
1'5"F

FIGURE 4.4.18 4·Watt Bridge Amplifier
55

3.5

50

3.0

45

2.5

Vs :::20V

~

«>

40
Po'" 4W

35

~

2.0

...""

1.5

3D

1.0

25

0.5

AL =16n

1[~[O ~ 50 ~~.7
'11111

o

20
10

100

lk

10k

lOOk

100

10

lk

10k

4W

lOOk

FREOUENCY (Hz)

FREQUENCY (Hz)

FIGURE 4.4.20 Distortion for Bridge Amp of
Figure 4.4.18

FIGURE 4.4.19 Frequency Response, Bridge Amp of
Figure 4.4.18

v,
R3

lOOk
C3
0.1

INPUT~

C2
0.00471'

FIGURE 4.4.21 4·Watt Bridge Amplifier with High Input Impedance
3.5

55
Vs - 20V
RL = 16.11

3.0

Vs

50
45

2.5

~

...""

~1~~I~

RL = 16n

2.0

"'J

:s

1.5

1.0

tti.~~Iltttr=ttil
4W

il

25

o
10

Po'" 4\111

35
30

Po'" 50 rnW

0.5

40

20
100

1k

10k

lOOk

10

100

lk

10k

lOOk

FREQUENCY (Hz)

FREQUENCY (Hz)

FIGURE 4.4.23 Frequency Response, Bridge Amp of
Figure 4.4.21

FIGURE 4.4.22 Distortion for Bridge Amp of
Figure 4.4.21

4·16

level with the values shown is 5.3VRMS at 60Hz. C7 and
the attenuator R 7 and RS couple 1/2 the signal of the F ET
drain to the gate for improved FET linearity and low
distortion. The amplitude control loop could be replaced
by an incandescent lamp in non·critical circuits (Figure
4.4.25), although DC offset will suffer by a factor of about
3 (DC gain of the oscillator). RIO matches R3 for improved
DC stability, and the network Rl1, Cg increases high
frequency gain for improved stability. Without this RC,
oscillation may occur on the negative half cycle of output
waveform. A low inductance capacitor, C5, located directly
at the supply leads on the package is important to maintain
stability and prevent high frequency oscillation on negative
half cycle of the output waveform. C5 may be 0.1 pF
ceramic, or 0.47 pF mylar. Layout is important; especially
take care to avoid ground loops as discussed in the section
on amplifiers. If high frequency instability still occurs, add
the R 12, Cl 0 network to the output.

4.4.13 Power Oscillator
One half of an LM377 may be connected as an oscillator to
deliver up to 2W to a load. Figure 4.4.24 shows a Wi en
bridge type of oscillator with FET amplitude stabilization
in the negative feedback path. The circuit employs internal
biasing and operates from a single supply. C3 and C6 allow
unity gain DC feedback and isolate the bias from ground.
Total harmonic distortion is under 1% to 10kHz, and could
possibly be improved with careful adjustment of R5. The
FET acts as the variable element in the feedback attenuator
R4 to R6. Minimum negative feedback gain is set by the
resistors R4 to R6, while the FET shunts R6 to increase
gain in the absence of adequate output signal. The peak
detector 02 and Cs senses output level to apply control
bias to the FET. Zener diode 01 sets the output level
although adjustment could be made if Rg were a poten·
tiometer with RS connected to the slider. Maximum output
+ 20V

C2
O.IM..F_ _• _ _ _+--I
Cl
O.IMF

01---.--.. . .
D
-

lMF

Rl

+

>---4~-U--+--'-O-

10k

-,

t", "

Rll
lOOk

R5
lk

C9
O.OOIMFr

C7

R7
lOOk

R8
lOOk

R9
lOOk

R6

180

FIGURE 4.4.24 Wien Bridge Power Oscillator

4.4.14 Two-Phase Motor Drive
Figure 4.4.25 shows the use of the LM377 to drive a small
60 Hz two phase servo motor up to 3W per phase. Applica·
tions such as a constant (or selectable) speed phonograph
turntable drive are adequately met by this circuit. A split
supply is used to simplify the circuit, reduce parts count,

and eliminate several large bypass capacitors. An incandes·
cent lamp is used in a simple amplitude stabilization loop.
Input DC is minimized by balancing DC resistance at (+)
and H amplifier inputs (Rl = R3 and R6 = RS). High
frequency stability is assured by increasing closed· loop gain
4·17

C2
O.lpF

NC
C4

Q.

C6

0.0022pF

C5

Q
R4
2700
C3
0.04pF

R3
27k

T

C7

T5

pF

FIGURE 4.4.25 Two·Phase Motor Drive

4.4.15 Proportional Speed Controller

from approximately 3 at 60 Hz to about 30 above 40kHz
with the network consisting of R3, R4 and C3. The interstage coupling C6 R6 network shifts phase by 85" at 60 Hz
to provide the necessary two phase motor drive signal. The
gain of the phase shift network is purposely low so that the
buffer amplifier will operate at a gain of 10 for adequate
high frequency stability. As in other circuits, the importance
of supply bypassing, careful layout, and prevention of
output ground loops is to be stressed. The motor windings
are tuned to 60Hz with shunt capacitors. This circuit will
drive 8n loads to 3W each.

A low cost proportional speed controller may be simply
designed using a LM378 amplifier. For use with 12-24 VDC
motors at continuous currents up to several hundred milliamps, this circuit allows remote adjustment of angular
displacements in a drive shaft. Typical applications include
rooftop rotary antennas and motor-controlled valves.
Proportional control (Figure 4.4.26) results from an error
signal developed across the Wheatstone bridge comprised of
resistors R 1, R2 and potentiometers P1, P2. Control P1 is

+28V
R3
10k
R4
10k

Pl

\

NC

\

P2
lOOk

MOTOR

lM378

24VOC

I
I

\

\

R5

\

\

10k

\

\

RS

\

-=-

91

\

\

10k

510k

R8

\

\

\
\
L-

---------

FIGURE 4.4.26 Proportional Speed Controller

4-18

_

\

-l

mechanically coupled to the motor shaft as depicted by the
dotted line and acts as a continuously variable feedback
sensor. Setting position control P2 creates an error voltage
between the two inputs which is amplified by the LM378
(wired as a difference bridge amplifier); the magnitude and
polarity of the output signal of the LM378 determines the
speed and direction of the motor. As the motor turns,
potentiometer P1 tracks the movement, and the error
signal, i.e., difference in positions between P1 and P2,
becomes smaller and smaller until ultimately the system
stops when the error voltage reaches zero volts.

ness, balance, and tone controls. The tone controls allow
boost or cut of bass and/or treble. Transistors 01 and 02
act as input line amplifiers with the triple function of
(1) presenting a high input impedance to the inputs,
especially ceramic phono; (2) providing an amplified output
signal to a tape recorder; and (3) providing gain to make up
for the loss in the tone controls. Feedback tone controls of
the Baxandall type employing transistor gain could be used;
but then, with the same transistor count, the first two
listed functions of °1°2 would be lost. It is believed that
this circuit represents the lowest parts count for the
complete system. Figure 4.4.28 is the additional circuitry
for input switching and tape playback amplifiers. The
LM382 with capacitors as shown provides for NAB tape
playback compensation. For further information on the
LM382 or the similar LM381 and LM387, refer to Section
2.0.

Actual gain requirements of the system are determined by
the motor selected and the required range. Figure 4.4.26
demonstrates the principle involved in proportional speed
control and is not intended to specify final resistor values.

Figure 4.4.29 shows the relationship between signal source
impedance and gain or input impedance for the amplifier
stage °1°2. Stage gain may be set at a desired value by
choice of either the source impedance or insertion of
resistors in series with the inputs (as R 1 to R4 in Figure
4.4.28). Gain is variable from -15 to +24dB by choice of
series R from 0 to 10 meg. Gain required for elN ~ 100 to
200mV (approximate value of recovered audio from FM
stereo or AM radio) is about 18 to 21dB overall for 2W into
an 8n speaker at 1 kHz or 21 to 24dB for 4 W.

4.4.16 Complete Systems
The LM377 to LM379 dual power amplifiers are useful in
table or console radios, phonographs, tape players, intercoms, or any low to medium power music systems.
Figures 4.4.27 through 4.4.29 describe the complete elec·
tronic section of a 2-channel sound system with inputs for
AM radio, stereo FM radio, phono, and tape playback.
Figure 4.4.27 combines the power amplifier pair with loud-

Av 0 TO +26 dB
depending on RSOURCE
:0

Av =-26dB

Av

=

34 dB

+
3k

8200

150k

220k

OUT:J~: o-..JIN'v-"'-"'-i 1-....-----1
1.8M
50k
100k

>,~--e~V\~

__-~••CTREB
(LOGI
O.l/"F
3300

lOOk
BAL'

-:r::-

300MF

(LiNI

OUT:J~~ o--"INy-+--+-t 1--41------1 1-------'

SPKR

+

OUTPUT L

FIGURE 4.4.27 Two-Channel Power Amplifier and Control Circuits

4·19

CERAMIC
PHONO

AUX

FM
STEREO

AM
RADIO

v+

Rl

R2

OUTPUT R
TO FIGURE 4.4.27

R3

FIGURE 4.4.28 Two-Channel Tape-Playback Amplifier and Signal Switching
25

106M

20

1.4M

15 I -

102M

Av

100M

10

;

4.4.17 Rear Channel Ambience Amplifier
The rear channel "ambience" circuit of Figure 4.4.30 can be
added to an existing stereo system to extract a difference
signal (R - L or L - R) which, when combined with some
direct signal (R or Ll, adds some fullness, or "concert hall
realism" to reproduction of recorded music. Very little
power is required at the rear channels, hence an LM377 will
suffice for most "ambience" applications. The inputs are
merely connected to the existing speaker output terminals
of a stereo set, and two more speakers are connected to the
ambience circuit outputs. Note that the rear speakers should
be connected in opposite phase to those of the front
speakers, as indicated by the +/- signs on the diagram of
Figure 4.4.30.

BOOk

>

'"

"
~
2

GUOk

400k

-5

I-

R'N

-10

200k

o

-15
10k

1M

lOOk

10M

FIGURE 4.4.29 AV and RIN for Input Stage of Figure 4.4.26

300k

r-----..,
I

0.22pF

240
FROM {LF
FRONT

SPKR

TERM

300pF
25V

>_-<~H~~KR
8n

I

0-,

....L.
_

"I

~+

~
240

RF ~--~~~--~~~4t

¢14~
3.4,5
10.11.12

OV+

I
H

....l-O.lpF
CERAMIC

I ':"
>_-<>-13",-1 ~ ~KR

I
I

I
0.22pF

L.. _

~37~~ _
300k

FIGURE 4.4.30 Rear Speaker Ambience (4-Channell Amplifier

4-20

.J

300pF
25V

8n

~+

4.5 LM380 AUDIO POWER AMPLIFIER

The output is biased to half the supply voltage by resistor
ratio R2/R1. Simplifying Figure 4.5.1 still further to show
the DC biasing of the output stage results in Figure 4.5.2,
where resistors R1 and R2 are labeled R. Since the transistor
operates with effectively zero volts base to collector, the
circuit acts as a DC amplifier with a gain of one half
(i.e., Av = R/[R + RJ) and an input of V+; therefore, the
output equals V+ /2.

4.5.1 Introduction
All of the mono power amplifiers listed in Table 4.3.2 derive
from the LM380 design; therefore, a detailed discussion of
the internal circuitry will be presented as a basis for under·
standing each of the devices. Subsequent sections wi II
describe only the variations on the LM380 design respon·
sible for each unique part.

The amplifier AC gain is internally fixed to 34dB (or
50V!V). Figure 4.5.3 shows this to be accomplished by the
internal feedback network R2·R3. The gain is twice that of
the ratio R21R3 due to the slave current·source (05, 06)
which provides the full differential gain of the input stage.

The LM380 is a power audio amplifier intended for
consumer applications. It features an internally fixed gain
of 50 (34dB) and an output which automatically centers
itself at one half of the supply voltage. A unique input stage
allows inputs to be ground referenced or AC coupled as
required. The.output stage of the LM380 is protected with
both short circuit current limiting and thermal shutdown
circuitry. All of these internally provided features result in
a minimum external parts count integrated circuit for audio
appl ications.

v+

4.5.2 Circuit Description
Figure 4.5.1 shows a simplified circuit schematic of the
LM380. The input stage is a PNP emitter·follower driving a
PNP differential pair with a slave current·source load. The
PNP input is chosen to reference the input to ground, thus
enabling the input transducer to be directly coupled.
The second stage is a common emitter voltage gain amplifier
with a current·source load. Internal compensation is pro'
vided by the pole'splitting capacitor C'. Pole·splitting
compensation is used to preserve wide power bandwidth
(100kHz at 2W, 8Q). The output is a quasi·complementary
pair emitter· follower.

FIGURE 4.5.2 LM380 DC Equivalent Circuit

.------------------<> (2)
Rs

150k

(3.4,5,10,11,121

GNO

FIGURE 4.5.1 LM380 Simplified Schematic

4·21

v+

lk

25k

FIGU R E 4.5.3 LM380 AC Equivalent Circuit

A gain difference of one exists between the negative and
positive inputs, analogous to inverting and non-inverting
amplifiers. For example, an inverting amplifier with input
resistor equal to 1 k and a 50k feedback resistor has a
gain of 50VIV, while a non-inverting amplifier constructed
from the same resistors has a gain of 51 VIV. Driving the
inverting terminal of the LM380, therefore, results in a gain
of 50, while driving the non-inverting will give a gain of 51.

12.0

~

'"a

10.0
8.0

>=
~

~

6.0

i5

4_5.3 General Operating Characteristics

u

4.0

~

2.0

10

20

30

40

50

60

70

80

TA - AMBIENT TEMPERATURE (Ge)

The output current of the LM380 is rated at 1.3A peak.
The 14 pin dual-in-line package is rated at 35°C/W when
soldered into a printed circuit board with 6 square inches
of 2 ounce copper foil (Figure 4.5.4). Since the device
junction temperature is limited to 150°C via the thermal
shutdown circuitry, the package will support 2.9W dissipation at 50°C ambient or 3.6W at 25°C ambient.

FIGURE 4.5.4a Device Dissipation vs. Maximum Ambient
Temperature

Figure 4.5.4a shows the maximum package dissipation vs.
ambient temperature for various amounts of heat sinking.
(Dimensions of the Staver V7 heat sink appear as Figure
4.5.4b.)
Figures 4.5.5a, -b, and -c show device dissipation versus
output power for various supply voltages and loads.
The maximum device dissipation is obtained from Figure
4.5.4 for the heat sink and ambient temperature conditions
under which the device will be operating. With this maximum allowed dissipation, Figures 4.5.5a, -b, and -c show the
maximum power supply allowed (to stay within dissipation
limits) and the output power delivered into 4, 8 or 16.11
loads." The three percent total harmonic distortion line is
approximately the onset of clipping.

*-Staver Co.
Bayshore, N.Y.

FIGURE 4.5.4b Staver' "'V7" Heat Sink

4-22

3.'

2.0

3% o 1ST.
LEVEL

~
r

'"~
~

~

ill

"

~

~

2.5
2.0
1.5
1.0

V

Vs

;:~V I-"'"

/

/

10Vf-- 1--0 j7'

f;

10%
OIST.
LEVEL

'V

f--

0.5 1.0

1.5 2.0

1.4 f--

'"'
~

1.2

;l

'"~I-

§

~f--

o

~

II

Ti
II

1.0

~'lo0Ti I-- IjW

0.4

L!

~
i=

:

ill

lW'

"
w

'"'

~

v

Vs
2.0 22

I/: V

~~vV'

-

1-1-.

-z

~

'"

~>

3D

Vs
1.5 22

:

ille;

20':f""
1.0

~

~

I-

:vl-

0.5

l- I-

?'

i:'-<

~~ ~

I>

10

240

POUT" 2W

100

lk

10k

lOOk

'"~

0

300'
360

1M

0

10M

FIGURE 4.5.7 Output Voltage Gain vs. Frequency

40

~

3D

II
lOlFI

20

kYf

'"'"

I

~±:I!r

i

LE1VElI

II
o

~

180 e

RL =an I

FREQUENCY (Hz)

an Load

~rl
~

120"

11111

3% DlST.
LEVEl

2.0

60·

TiEl

15

10

II

0

i=

O·

I I I

OUTPUT POWER (WATTS)

~

20k

vee =18V

=~

20

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

2.5

10k

TIT
Rl

25

10%
OIST.
LEVEl

3.0

5k

III
III

J5

\~II:F

12V~ !"='"

0.5

2k

FIGURE 4.5.6 Total Harmonic Distortion vs. Frequency

~~~ ~

FIGURE 4.S.5b Device Dissipation vs. Output Power -

'"z

lk

FREQUENCY (Hz)

.....
\;:-- .....

~

16V
1.0 14V

o 0.5

~

500

40

2.'

1.5

'J

I

100 200

3.'

'"

I--

II

O.B

2.5 3.0 3.5 4.0

FIGURE 4.5.5a Device Dissipation vs. Output Power - 4,n Load

~

"t,,, HIEA~ L~K

0.6

0.2

RL"Bn

I I I 'BV

SWER

OUTPUT POWER (WATTS)

~

I I II

III

to

"a:

Bf-X

0.5

I.B
1.6

e;

/

/

z
0

~ f-- f--

3.0

1pF

10

/'

....

V
O.47pF

III I

NO CAP

III 1IIIIili
10

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.04.55.0

100

1k

10k

FREQUENCY (Hd

OUTPUT POWER (WATTS)

FIGURE 4.S.5c Device Dissipation vs. Output Power - 160 Load

FIGURE 4.5.8 Supply Oecoupling vs. Frequency

Figure 4.5.6 shows total harmonic distortion VS. frequency
for various output levels, while Figure 4.5.7 shows the
power bandwidth of the LM380.

to ground to be direct·coupled to either the inverting or
non·inverting inputs of the amplifier. The unused input
may be either: (1) left floating, (2) returned to ground
through a resistor or capacitor, or (3) shorted to ground. In
most applications where the non·inverting input is used, the
inverting input is left floating. When the inverting input is
used and the non-inverting input is left floating, the
amplifier may be found to be sensitive to board layout
since stray coupling to the floating input is positive feedback. This can be avoided by employing one of three
alternatives: (1) AC grounding the unused input with a
small capacitor. This is preferred when using high source
impedance transducers. (2) Returning the unused input to
ground through a resistor. This is preferred when using
moderate to low DC source impedance transducers and

Power supply decoupling is achieved through the AC
divider formed by R 1 (Figure 4.5.1 ) and an external bypass
capacitor. Resistor R 1 is spl it into two 25 kD halves
providing a high source impedance for the integrator.
Figure 4.5.8 shows supply decoupling vs. frequency for
various bypass capacitors.
4.5.4 Biasing
The simplified schematic of Figure 4.5.1 shows that the
LM380 is internally biased with the 150kD resistance to
ground. This enables input transducers which are referenced
4·23

when output offset from half supply voltage is critical. The
resistor is made equal to the resistance of the input transducer, thus maintaining balance in the input differential
amplifier and minimizing output offset. (3) Shorting the
unused input to ground. This is used with low DC source
impedance transducers or when output offset voltage is
non-critical.

Vs'= l8V

0.1

~
-8

.~ Rc*

~

~ 2.70
Cc *

_ ..... -

0.1 pF "T,

4.5.5 Oscillation
The normal power supply decoupling precautions should be
taken when installing the LM380. If Vs is more than 2" to
3" from the power supply filter capacitor it should be
decoupled with a O.lpF disc ceramic capacitor at the Vs
terminal of the IC.

8n

-t:.-

'FOR STABILITY WITH
HIGH CURRENT LOADS

FIGURE 4.5.11 Ceramic Phono Amp

The Rc and Cc shown as dotted line components on figure
4.5.9 and throughout this section suppresses a 5 to 10MHz
small amplitude oscillation which can occur during the
negative swing into a load which draws high current. The
oscillation is of course at too high a frequency to pass
through a speaker, but it should be guarded against when
operating in an RF sensitive environment.

4.5.9 Common Mode Volume and Tone Controls
When maximum input impedance is required or the signal
attenuation of the voltage divider volume control is undesirable, a "common mode" volume control may be used
as seen in Figure 4.5.12.

+18V

v"

*FOR STABILITY WITH
HIGH CURRENT LOADS

*FOR STABILITY WITH
HIGH CURRENT LOADS

FIGURE 4.5.12 "Common Mode" Volume Control

FIGURE 4.5.9 Oscillation Suppression Components

With this volume control the source loading impedance is
only the input impedance of the amplifier when in the fullvolume position. This reduces to one half the amplifier
input impedance at the zero volume position. Equation
(4.5.1) describes the output voltage as a function of the
potentiometer setting.

4.5.6 R F Precautions - See section 2.3.10.
4.5.7 Inverting Amplifier Application
With the internal biasing and compensation of the LM380,
the simplest and most basic circuit configuration requires
only an output coupling capacitor as seen in Figure 4.5.10.

VOUT = 50VIN ( 1 -

Vs

150x 103

)
(4.5.1)
k1Rv+150x1030';;k1';;1

This "common mode" volume control can be combined
with a "common mode" tone control as seen in figure
4.5.13.
+18V

Co

~DD1nP~
FIGURE 4.5.10 Minimum Component Configuration

.~ Rc*
:,.. 2.70 80
Cc• .......
0.1 pF

-:1:-

..

4.5.8 Ceramic Phono Amplifier
An application of this basic configuration is the phonograph amplifier where the addition of volume and tone
controls is required. Figure 4.5.11 shows the LM380 with a
voltage divider volume control and high frequency roll-off
tone control.

'FOR STABILITY WITH
HIGH CURRENT LOADS
'-AUDIO TAPE POTENTIOMETER
(10% OF RT AT 50% ROTATION)

FIGURE 4.5.13 "Common Mode" Volume and Tone Control

4-24

This circuit has a distinct advantage over the circuit of
Figure 4.5.10 when transducers of high source impedance
are used, in that the full input impedance of the amplifier is
realized. It also has an advantage with transducers of low
source impedance, since the signal attenuation of the input
voltage divider is eliminated. The transfer function of the
circuit of Figure 4.5.13 is given by:

This provides twice the voltage swing across the load for a
given supply, thereby increasing the power capability by a
factor of four over the single amplifier. However, in most
cases the package dissipation will be the first parameter
limiting power delivered to the load. When this is the case,
the power capability of the bridge will be only twice that of
the single amplifier. Figures 4.5.16a and ·b show output
power vs. device package dissipation for both 8 and 16n
loads in the bridge configuration. The 3% and 10% harmonic
distortion contours double back due to the thermal limiting
of the LM380. Different amounts of heat sinking will
change the point at which the distortion contours bend.

VOUT
VIN

E

4.0

 3") filter capacitor it should be decoupled with a
1 >IF tantalum capacitor.

'FOR STABILITY WITH
HIGH CURRENT LOADS
FIGURE 4.5.15 Bridge Configuration

4·25

"FOR STABILITY WITH
HIGH CURRENT LOADS
FIGURE 4.5.17 Quiescent Balance Control

Vs

Re*

Cc*

2.7H

O.lJ.lF

,...-vvv--i~-,

I
I

R1
15k

~.

I
I

Vs

,
...L

~

PBM

i

'"'FOR STABILITY WITH
HIGH CURRENT LOADS

FIGURE 4.5.18 Voltage Divider Input

4.5.11 Intercom
The circuit of Figure 4.5.19 provides a minimum component
intercom. With switch S1 in the talk position, the speaker
of the master station acts as the microphone with the aid of
step·up transformer T 1.

A turns ratio of 25 and a device gain of 50 allows a
maximum loop gain of 1250. Rv provides a "common
mode" volume control. Switching S1 to the listen position
reverses the role of the master and remote speakers.

v,

*FOR STABILITY WITH
HIGH CURRENT LOADS

FIGURE 4.5.19 Intercom

4·26

4.5.12 Low Cost Dual Supply

At 20 kHz the reactance of this capacitor is approximately
-j4MQ, giving a net input impedance magnitude of 3.9MQ.
The values chosen for R 1, R2 and C1 provide an overall
circuit gain of at least 45 for the complete range of para·
meters specified for the KE4221.

The circuit shown in Figure 4.5.20 demonstrates a minimum
parts count method of symmetrically splitting a supply
voltage. Unlike the normal R, C, and power zener diode
technique the LM380 circuit does not require a high
standby current and power dissipation to maintain
regulation.

I
I
I

'I

I

R1

•I

I
I

(,

VGS)

(4.5.4)

= gmo~-v;

gm

v'

I

(4.5.31

(501

gm

I

I

~~::

( ~-)
+-

+
v'

..

~

When using another FET device the relevant design equa·
tions are as follows:

(4.5.5)

IDS

FIGURE 4.5.20 Dual Supply

With a 20V input voltage (±10V output) the circuit exhibits
a change in output voltage of approximately 2% per 100mA
of unbalanced load change. Any balanced load change will
reflect only the regulation of the source voltage VIN.

(,

= IDSS~ -

VGS)2

V;

(4.5.61

The maximum value of R2 is determined by the product of
the gate reverse leakage IGSS and R2. This voltage should
be 10 to 100 times smaller than Vp. The output impedance
of the FET source follower is:

The theoretical plus and minus output tracking ability is
100% since the device will provide an output voltage at one
half of the instantaneous supply voltage in the absence of a
capacitor on the bypass terminal. The actual error in
tracking will be directly proportional to the imbalance in
the quiescent output voltage. An optional potentiometer
may be placed at pin 1 as shown in Figure 4.5.20 to null
output offset. The unbalanced current output for the
circuit of Figure 4.5.20 is limited by the power dissipation
of the package.

(4.5.7)
so that the determining resistance tor the interstage RC
time constant is the input resistance of the LM380.
4.5.14 Power Voltage-to·Current Converter
The LM380 makes a low cost, simple voltage·to·current
converter capable of supplying constant AC currents up to
1 A over variable loads using the circuit shown in Figure
4.5.22.

In the case of sustained unbalanced excess loads, the device
will go into thermal limiting as the temperature sensing
circuit begins to function. For instantaneous high current
loads or short circuits the device limits the output current
to approximately 1.3A until thermal shutdown takes over
or until the fault is removed.

lOOk

11-1

Rl

VIN 0-11 !-"V'Irv-...,,..,
10k

4.5.13 High Input Impedance Circuit
The junction FET isolation circuit shown in Figure 4.5.21
raises the input impedance to 22 MQ for low frequency
input signals. The gate to drain capacitance (2pF maximum
for the KE4221 shown) of the FET limits the input
impedance as frequency increases.

R,
5n,2W

v,

lOOk

~ID'

-

v,. 0--+--+1

R2

KE4221
S

v,

-=-

lass

FIGURE 4.5.22 Power VOltage-to-Current Converter

R2

Rl

22M

20K

Current through the load is fixed by the gain setting
resistors R1·R3, input voltage, and R5 per Equation (4.5.8).

I

I

Rl

10k

o

Il

IL

=

R3VIN

(4.5.81

R1 R5

'::"

For AC signals the minus sign of Equation (4.5.8) merely

~RE.4__.5_.2_1__H_i9_h_l_n_p_ut__lm_p_e_d_a_nc_e_________________S_h_O_W_s_p_h_a_s_e_i_n_ve_r_s_io_n_._A_s__Sh_O_W__n_,_F_ig_U_r_e_4_._5._2_2_W__il_l_d_e_liv_e_r~
4-27

1/2ARMS to the load from an input signal of 250mVRMS,
with THO less than 0.5%. Maximum current variation is
typically 0.5% with a load change from 1-5n.

v,
'OOk

Flowmeters, or other similar uses of electromagnets,
exemplify application of Figure 4.5.22. Interchangeable
electromagnets often have different impedances but require
the same constant AC current for proper magnetization.
The low distortion, high current capabilities of the LM380
make'such applications quite easy.

4.5.15 Muting
FIGURE 4.5.23 Muting the lM380

Muting, or operating in a squelched mode may be done with
the LM380 by pulling the bypass pin high during the mute,
or squelch period. Any inexpensive, general purpose PNP
transistor can be used to do this function as diagrammed in
Figure 4.5.23.

During the mute cycle, the output stage will be switched
off and will remain off until the PNP transistor is turned off
again. Muting attach and release action is smooth and fast.

R,
RATE
250k

'NO'4

.....- I - - - -...--')
i1"'
:
>,
~",. ~
22V t,.c
flY

20VT;"
8V ;1":

~...!:

~'10%loJ l L l _

~J% OIST. LEVEl

iSTAVEJ
,
OUTPUT POWER (W)

0

>=

:l:

iii0;
u

~

4

"V~" JEA+ SI~K

5

6

7

B

9 10

4n

FIGURE 4.6.2 Device Dissipation vs. Output Power - 8n Load

load

2.4
2.2
2.0
24V
1.B
22V""'"
1.6
1.4
........
Iv
1.2 -lBV
1.0
16V", .>'
O.B
0% O'fT. lEyEl_
0.6
0.4
STAVER "V,..IHEATISINK
0.2 -

-

z

3

OUTPUT POWER (WI

FIGURE 4.6.1 Device Dissipation vs. Output Power -

~

2

~

-f..
~~
V

"'."

r----

0.1
OUTPUT POWER (WI

1.0

FIGURE 4.6.3 Device Dissipation vs. Output Power - 16 n load

FIGURE 4.6.4 Total Harmonic Distortion vs. Output Power

0.4

III
III

g
z
0

~

0.3

14~

in

IT I

0;

u

~

0.2

2W

~
~

0.1

~
....

,.,

lW
Vee

~22V

RL '" 8
STAVER "V7" HEAT SINK

100

lk

10k

lOOk

FREQUENCY (Hz)

FIGURE 4.6.5 Total Harmonic Distortion vs. Frequency

+22V

V'N

10k

10

OUTPUT POWER (W)

>

I'T

20
10

o
100

0.1 0.20.3 0.4 0.5 0.6 0.1 O.B 0.9 1.0

1k

10k

lOOk

1M

FREOUENCY (Hz)

OUTPUT POWER (WATTS)

FIGURE 4.7.4 Device Dissipation vs. Output Power - 1651 Load

FIGURE 4.7.5 Voltage Gain vs. Frequency

4.7.3 Input Biasing

4.7.4 Gain Control

The schematic (Figure 4.7.1) shows that both inputs are
biased to ground with a 50krl resistor. The base current of
the input transistors is about 250nA, so the inputs are at
about 12.5mV when left open. If the DC source resistance
driving the LM386 is higher than 250krl it will contribute
very little additional offset (about 2.5mV at the input,
50mV at the output). If the DC source resistance is less
than 10)(rl, then shorting the unused input to ground will
keep the offset low (about 2.5mV at the input, 50mV at
the output). For DC source resistances between these
values we can eliminate excess offset by putting a resistor
from the unused input to ground, equal in value to the DC
source resistance. Of course all offset problems are elimi·
nated if the input is capacitively coupled.

Figure 4.7.6 shows an AC equivalent circuit of the LM386,
highlighting the gain control feature. To make the LM386 a
more versatile amplifier, two pins (1 and 8) are provided for
gain control. With pins 1 and 8 open the 1.35 krl resistor
sets the gain at 20 {26dB). If a capacitor is put from pin 1
to 8, bypassing the 1.35krl resistor, the gain will go up to
200 (46dB).

When using the LM386 with higher gains (bypassing the
1.35 krl resistor between pins 1 and 8) it is necessary to
bypass the unused input, preventing degradation of gain
and possible instabilities. This is done with a 0.1 fJF
capacitor or a short to ground depending on the DC source
resistance on the driven input.

Gains less than 20dB should not be attempted since the
LM386 compensation does not extend below 9 VIV (19dB).

If a resistor (R3) is placed in series with the capacitor, the
gain can be set to any value from 20 to 200. Gain control
can also be done by capacitively coupling a resistor (or
FET) from pin 1 to ground. When adding gain control with
components from pin 1 to ground, the positive input (pin
3) should always be driven, with the negative input (pin 2)
appropriately terminated per Section 4.7.3.

4.7.5 Muting
Similar to the LM380 (Section 4.5.15). the LM386 may be
muted by shorting pin 7 (bypass) to the supply voltage.
The LM386 may also be muted by shorting pin 1 (gain) to
ground. Either procedure will turn the amplifier off without
affecting the input signal.

v+

4.7.6 Typical Applications
Three possible variations of the LM386 as a standard audio
power amplifier appear as Figures 4.7.7·4.7.9. Possible gains
of 20, 50 and 200VIV are shown as examples of various
gain control methods. The addition of the optional 0.05fJF
capacitor and 10rl resistor is for suppression of the
"bottom side fuzzies" (i.e., bottom side oscillation occurring
during the negative swing into a load drawing high current
- see Section 4.5.5).

...J...

I
150

1.l5k

15k

0.1

v,

r1

V'N

10k~t--1

FIGURE 4.7.7 Amplifier with Gain = 20V/V (26dB) Minimum
Parts

FIGURE 4.7.6 lM386 AC Equivalent Circuit

4·31

4.7.8 Square Wave Oscillator
A square wave oscillator capable of driving an 8n speaker
with 0.5W from a 9V supply appears as Figure 4.7.11.
Altering either R 1 or C1 will change the frequency of
oscillation per the equation given in the figure. A reference
voltage determined by the ratio of R3 to R2 is applied to
the positive input from the LM386 output. Capacitor C1
alternately charges and discharges about this reference value,
causing the output to switch states. A triangle output may
be taken from pin 2 if desired. Since DC offset voltages are
not relevant to the circuit operation, the gain is increased to
200VN by a short circuit between pins 1 and 8, thus
saving one capacitor.
FIGURE 4.7.8 Amplifier with Gain = 50V/v (34dB)

Vs

1

250l-lF

v,.

1+

I

Cl

O.l MF

T

....l.-O.OM

%"~l ~

R2
l'

f~_l_

0.36Rl Cl

FIGURE 4.7.9 Amplifier with Gain = 200V/V (46dB)

f " 1 kHz AS SHOWN

FIGURE 4.7.11 Square Wave Oscillator

4.7.7 Bass Boost Circuit

4.7.9 Power Wien Bridge Oscillator

Additional external components can be placed in parallel
with the internal feedback resistors (Figure 4.7.10) to tailor
the gain and frequency response for individual applications.
For example, we can compensate poor speaker bass response
by frequency shaping the feedback path. This is done with
a series RC from pin 1 to 5 (paralleling the internal 15kn
resistor). For 6dB effective bass boost: R '" 15kn, the
lowest value for good stable operation is R = 10 kn if pin 8
is open. If pins 1 and 8 are bypassed then R as low as 2kn
can be used. This restriction is because the amplifier is only
compensated for closed·loop gains greater than 9.

The LM386 makes a low cost, low distortion audio frequency oscillator when wired into a Wien bridge configuration (Figure 4.7.12). Capacitor C2 raises the "open-loop"
gain to 200VN. Closed-loop gain is fixed at approximately
ten by the ratio of R 1 to R2. A gain of ten is necessary to
guard against spurious oscillations which may occur at
lower gains since the LM386 is not stable below 9VN. The
frequency of oscillation is given by the equation in the
figure and may be changed easily by altering capacitors C1.

r

0.1

v,

27
26
25

-z

2.

w

22

'"'"

21

~

V'N~

~>

lOkI

I

23

I

II

20
19

Vr\
1\

\

"

~

18
17
20

50 100200 500 l' 2.

5' 10k 20.

FREQUENCY (H,I

(b) Frequency Response with Bass Boost

(a) Amplifier with Bass Boost

FIGURE 4.7.10 LM386 with Bass Boost

4·32

The amplifier inputs are ground referenced while the output
is automatically biased to one half the supply voltage. The
gain is internally set at 20 to minimize external parts, but
the addition of an external resistor and capacitor between
pins 4 and 12 will increase the gain to any value up to 200.
Gain control is identical to the LM386 (see Section 4.7.4).

R3

390

vs

The three transistors have high gain and excellent matching
characteristics. They are well suited to a wide variety of
applications in DC through VHF systems.

L,
ElOEMA
CF-S-2158

4.8.2 Supplies and Grounds
The LM389 has excellent supply rejection and does not
require a well regulated supply. However, to eliminate
possible high frequency stability problems, the supply
should be decoupled to ground with a O.lJ1F capacitor. The
high current ground of the output transistor, pin 18, is
brought out separately from small signal ground, pin 17. If
the two ground leads are returned separately to supply, the
parasitic resistance in the power ground lead will not cause
stability problems. The parasitic resistance in the signal
ground can cause stability problems and it should be
minimized. Care should also be taken to insure that the
power dissipation does not exceed the maximum dissipation
(825mW) of the package for a given temperature.

R,
41k

RZ
4.7k

c,

O.OM"*"

f~--'2rr C, ";f:f,-R-2

f "" 1kHz AS SHOWN

FIGURE 4.7.12 Low Distortion Power Wien Bridge Oscillator

4.8.3 Muting
Resistor "R3 provides amplitude stabilizing negative feedback in conjunction with lamp Ll. Almost any 3V, 15mA
lamp will work.

Muting is accomplished in the same manner as for the
LM386 (Section 4.7.5), with the exception of applying to
different pin numbers.
4.8.4 Transistors
The three transistors on the LM389 are general purpose
devices that can be used the same as other small signal
transistors. As long as the currents and voltages are kept
within the absolute maximum limitations, and the collectors
are never at a negative potential with respect to pin 17,
there is no limit on the way they can be used.

4.8 LM389 LOW VOLTAGE AUDIO POWER AMPLIFIER
WITH NPN TRANSISTOR ARRAY
4.8.1 Introduction

For example, the emitter-base breakdown voltage of 7.1 V
can be used as a zener diode at currents from lJ1A to 5mA.
These transistors make good LED driver devices; VSAT is
only 150mV when sinking 10mA.

The LM389 is an array of three NPN transistors on the
same substrate with an audio power amplifier similar to the
LM386 (Figure 4.8.1).

r------------------------------t-------i~Ovs

'3

SUBSTRATE

FIGURE 4.8.1 LM389 Simplified Schematic

4-33

In the linear region, these transistors have been used in AM
and FM radios, tape recorders, phonographs, and many
other applications. Using the characteristic curves on noise
voltage and noise current, the level of the collector current
can be set to optimize noise performance for a given source
impedance (Figures 4.8.2-4.8.4).

zo
18

~

16

w

12

"!:;"

10

~

1\

14

cJ

Ie· 10 rnA
_Ie "'1 rnA

">
w

4.8.5 Typical Applications
Ie'" 10,uA

The possible applications of three NPN transistors and a
O.5W power amplifier seem limited only by the designer's
imagination. Many existing designs consist of three transis·
tors plus a small discrete power amplifier; redesign with the
LM389 is an attractive alternative - typical of these are
battery powered AM radios. The LM389 makes a costsaving single IC AM radio possible as shown in Figure 4.8.5.

i5

z

z
o

Ie

11 ololl~1

10

III III

100

lk

10k

FREQUENCY 1Hz}

FIGURE 4.8.2 Noise Voltage vs. Frequency

Several appl ications of the LM389 follow as examples of
practical circuits and also as idea joggers.

100

4.8.6 Tape Recorder

I~

~

:!

A complete record/playback cassette tape machine ampli·
fier appears as Figure 4.8.6. Two of the transistors act as
signal amplifiers, with the third used for automatic level
control during the "record" mode. The complete circuit
consists of only the LM389 plus one diode and the passive
components.

10

>-

I
w

!1

""

4.8.7 Ceramic Phono Amplifier with Tone Controls

0.1
10

100

10k

1k

For proper frequency response (particularly at the low end),
ceramic cartridges require a high termination impedance.
Figure 4.8.7 shows a low-cost single IC phono ampli fier
where one of the LM389 transistors is used as a high input
impedance emitter follower to provide the required cartridge load. The remaining transistors form a high-gain
Darlington pair, used as the active element in a low distortion 8axandall tone control circuit (see Section 2.14.7).

FREQUENCY 1Hz)

FIGURE 4.8.3 Noise Current vs. Frequency

10k
7k

S

4k

u

'"
"";;

lk
700

I

400

~

Zk

~

~

<£

ZOO
100

4d~=~

IVCE!5~dB

6'

BW",2kHz
f-1 MHz

i~B

4.8.8 Siren

'\

The siren circuit of Figure 4.8.8 uses one of the LM389
transistors to gate the power ampl ifier on and off by
applying one of the muting techniques discussed in Section
4.8.3. The other transistors form a cross-coupled multivibrator circuit that controls the rate of the square wave
oscillator. The power amplifier is used as the square wave
oscillator with individual frequency adjust provided by
potentiometer R2B.

~4d'

'"

161d~

I"- 6 dB

~~

N1;a-' H.
0.1

0.3

1.0

3.0

10

Ie - COLLECTOR CURRENT (mA)

FIGURE 4.8.4 Contours of Constant Noise Figure

+~

~-{~}-fg-{:g-. *
LOCAL OSC
& MIXER

1ST
IF

2ND
IF

DETECTOR

VOLUME
OUTPUT AMPLIFIER & SPEAKER

FIGURE 4.8.5 AM Radio

4-34

6.Bk

FIGURE 4.8.6 Tape Recorder
+12V

1.Jk

T

+
50}JF

-=-=

Ceramic Phono A mpllfler
.. with T
ON RATE
V
one Controls
(1-7 Hz)

S

FREO

1250-1500 ",I

FIGURE 4.8.8 Siren

4·35

+12V

lk

+10V

+

O.1.uF

50"F-r

120k

"*

10k

vc

GAIN
lOOk

1.2k
":'

10k

*OPTIONAL

TREMOLO
INPUT
2V
VON
SIGNAL

2k

2.7k

*TREMOLO FREa.

0;;;;

211 (R ~ 10k) C '" 160Hz AS SHOWN

":'

FIGURE 4.8.9 Voltage..controlled Amplifier or Tremolo Circuit

+50

range of 50dB (Figure 4.8.10). VIN signal levels should be
restricted to less than 100mV for good distortion perfor·
mance. The output of the differential gain stage is
capacitively fed to the power amplifier via the R·C network
shown, where it is used to drive the speaker .

<40

+30
~ +20

.1

+10

Tremolo (amplitude modulation of an audio frequency by a
sub·audio oscillator - normally 5·15 Hz) applications reo
quire feeding the low frequency oscillator signal into the
optional input shown. The gain control pot may be set for
optimum "depth." Note that the interstage R·C network
forms a high pass filter (160Hz as shown). thus requiring
the tremolo frequency to be less than this time constant for
proper operation.

4 4.5
CONTROL VOLTAGE, Vc (VOLTS)

FIGURE 4.8.10 VCA Gain

YS.

Control Voltage

4.8.9 Voltage-Controlled Amplifier or Tremolo Circuit

4.8.10 Noise Generator

A voltage·controlled amplifier constructed from the LM389
appears as Figure 4.8.9. Here the transistors form a differential pair with an active current-source tail. This configuration, known technically as a variable-transconductance
multiplier, has an output proportional to the product of the
two input signals. Multiplication occurs due to the depen·
dence of the transistor transconductance on the emitter
current bias. As shown, the emitter current is set up to a
quiescent value of 1 mA by the resistive string. Gain control
voltage, Vc, varies from OV (minimum gain = -20dB) to
4.5V (maximum gain = +30dB), giving a total dynamic

By applying reverse voltage to the emitter of a grounded
base transistor, the emitter·base junction will break down in
an avalanche mode to form a handy zener diode. The
reverse voltage characteristic is typically 7.1 V and may be
used as a voltage reference, or a noise source as shown in
Figure 4.8.11. The noise voltage is amplified by the second
transistor and delivered to the power amplifier stage where
further amplification takes place before being used to drive
the speaker. The third transistor (not shown) may be used
to gate the noise generator similar to Section 4.8.8 if
required.
12V

vs

lOOk

6.2k

Nt
510k

16k

lk

FIGURE 4.8.11 Noise Generator Using Zener Diode

4·36

4.8.11 Logic Controlled Mute

Vs

Various logic functions are possible with the three NPN
transistors, making logic control of the mute function
possible. Figures 4.8.12-4.8.14 show standard AND, OR
and Exciusive·OR circuits for controlling the muting
transistor. Using the optional mute scheme of shorting pin
12 to ground gives NAND, NOR and Exclusive NOR

10k

13

vs
FIGURE 4.8.13 OR Muting

10k
V1 G--f\J\."-_ _C

10k

V2o-----.........I
10k

FIGURE 4.8.12 AND Muting

10k

FIGURE 4.8.14 Exclusive-OR Muting

4.9 LM388 BOOTSTRAPPED AUDIO POWER
AMPLIFIER

14
~---------------------'--'--ovs

13

-INPUT

~~~~~------~-~-~----~---~-oGND
FIGURE 4.9.1 LM388 Simplified Schematic

LM380) extends maximum package dissipation to values
where heatsinking is eliminated for most designs.

4.9.1 Introduction
The LM388 audio power amplifier, designed for low voltage,
medium power consumer applications, extends the LM386
design concept one step further by incorporating a bootstrapped output stage (Figure 4.9.1). Bootstrapping allows
power levels in excess of 1 W to be obtained from battery
powered products (Figures 4.9.2-4.9.4). Packaging the
LM388 into National's 14-pin copper lead-frame (same as

4.9.2 General Operating Characteristics
The gain, internally set to 20V IV, is externally controlled
in the same manner as the LM386. Consult Section 4.7.4 for
details. Input biasing follows LM386 procedures outlined in
Section 4.7.3; likewise, muting is the same as Section 4.7.5.
4-37

2.0
1.8

~

1.6

is

1.4

;::

::
ili
C

<.>

~

"

IA"

If"

1.2
1.0
0.8
0.6
0.4
0.2

r-------~-----9~vs

1
Vs =:12V

Vs =

v

""
,;-...,

IX /

II) I

14

,/

II-: L

10% DISTORTION

3% DISTORTION

'I ...[.L.
VVs"6V
I

I

I

I

I

I

I

0.20.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

OUTPUT POWER (WI

FIGURE 4.9.2 Device Dissipation vs. Output Power - 4U Load

2.0

~

1.8
1.6

'"
;::
::

1.4

C

0.8

e

ili

<.>

~

1.2
1.0
0.6

D.'

vs::;Vs

II.

=:

/

/'-"

%f. 1

0.2 Joys

=

6V

FIGURE 4.9.5 LM388 Output Stage

9v

10% DISTORTION

"off

12

3 DltiRTIIOO

9'

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

FIGURE 4.9.3 Device Dissipation vs. Output Power -

10

?

OUTPUT POWER (WI

w

'"«c;
e

an Load

>
~

~
1.0

~ 0.8

'"'"
;::

0.6

ili
c

0.4

::

10

12

FIGURE 4.9.6 Peak-to-Peak Output Voltage Swing
Voltage

<.>

~

11

SUPPl Y VOL TAGE (VI

YS.

Supply

0.2

The stored charge converts to a current with time and
supplies the necessary base drive to keep the top transistor
saturated during the critical peak period. The net effect
allows higher positive voltage swings than can be achieved
without bootstrapping. (See Figure 4.9.6.)

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

OUTPUT POWE R (WI

FIGURE 4.9.4 Device Dissipation vs. Output Power - 16n load

For design purposes, resistors (R) and bootstrap capacitor
(CB) can be determined from the following.
IB

4.9.3 Bootstrapping (See also section 4.1.5.)
The base of the top side output transistor is brought out to
pin 9 for bootstrapping. The term "bootstrapping" (derived
from the expression, " . . . pull oneself up by one's bootstraps") aptly describes the effect. Figure 4.9.5 shows the
output stage with the external parts necessary for standard
bootstrapping operation. Capacitor CB charges to approximately Vs/4 during the quiescent state of the amplifier and
then acts to pull the base of the top transistor up ("by the
bootstraps") as the output stage goes through its positive
swing - actually raising pin 9 to a higher potential than the
supply at the top of the swing. This occurs since the
voltage on a capacitor cannot change instantaneously, but
must decay at a rate fixed by the resistive discharge path.

IL

Vs12 - VBE

-----

=-

2R

{l

Vs

'"

4R

(lVs
IL

4R

also, I L(max) =

so,

{3 Vs

Vs

4R

2 RL

or, R
4-38

{3RL
2

V s/2

RL

(4.9.1)

To preserve low frequency performance the pole due to CB
and R/2 (parallel result of R·R) is set equal to the pole due
to Cc and RL:
(4.9.2)
Substituting Equation (4.9.1) into (4.9.2) yields:
CB =

4CC

(4.9.3)

!l
Letting!l = 100 (nominal) gives:
(4.9.4)

R = 50 RL
CB

=

Cc
25

(4.9.5)

FIGURE 4.9.8 Load Returned to Ground (Amplifier with Gain = 20)

For reduced component count the load can replace the
upper resistor, R (Figure 4.9.7). The value of bootstrap
resistors R+R must remain the same, so the lower R is
increased to 2 R (assuming speaker resistance to be negli·
gible). Output capacitor (CC) now serves the dual function
of bootstrapping and coupling. It is sized about 5% larger
since it now supplies base drive to the upper transistor.

vs

V,N

v'

14

FIGURE 4.9.9 Load Returned to Vs (Amplifier with Gain = 20)

IJ

FIGURE 4.9.7 Bootstrapping with Load to Supply

FIGURE 4.9.10 Amplifier with Gain =200 and Minimum Cs

Examples of both bootstrapping methods appear as Figures
4.9.S and 4.9.9. Note that the resistor values are slightly
larger than Equation (4.9.4) would dictate. This recognizes
that I L(max) is, in fact, always less than [V s/21/R L due to
saturation and VBE losses.

4.9.4 Bridge Amplifier
For low voltage applications requiring high power outputs,
the bridge connected circuit of Figure 4.9.11 can be used.
Output power levels of 1.0W into 4n from 6V and 3.5W
into sn from 12V are typical. Coupling capacitors are not
necessary since the output DC levels will be within a few
tenths of a volt of each other. Where critical matching is
required the 500k potentiometer is added a~d adjusted for
zero DC current flow through the load.

A third bootstrapping method appears as Figure 4.9.10,
where the upper resistor is replaced by a diode (with a
subsequent increase in the resistance value of the lower
resistor). Addition of the diode allows capacitor CB to be
decreased by about a factor of four, since no stored charge
is allowed to discharge back into the supply line.
4·39

270

Jl -,

V,N

F

2.7

10k

... ""\Mr- - -l~

>0.-. .---1

I

RL

I

500k

Vs=6V

RL=4n

Vs =12V Rl

Po "'1.0W

",an Po" 3.SW

FIGURE 4.9.11 Bridge Amp

vsO-......~...- - - - .

TALK

TALK

REMOTE

MASTER

FIGURE 4.9.12 Intercom

1. Low cost FM scanners; Vs = 6V, Po = 0.25W'
2. Consumer walkie talkie (including CB); Vs = 12V,
Po = 0.5W
3. High quality hand·held portables; Vs = 7.5V, Po = 0.5W

4.9.5 Intercom
A minimum parts count intercom circuit (Figure 4.9.12) is
made possible by the high gain of the LM388. Using the
gain control pin to set the AC gain to approximately
300VIV (Av"" 15k/51 n) allows elimination of the step'up
transformer normally used in intercom designs (e.g., Figure
4.5.22). The optional 2.7 n·0.05!lF R·C network suppresses
spurious oscillations as described for the LM380 (Section
4.5.5).

Since all equipment is battery operated, current consump·
tion is important; also, the amplifier must be squelchable,
i.e., turned off with a control signal. The LM388 meets
both of these requirements. When squelched, the LM388
draws only 0.8mA from a 7.5V power supply.

4.9.6 FM Scanners and Two Way Walkie Talkies

A typical high quality hand held portable application with
noise squelch appears as Figure 4.9.13. Diodes D1 and D2
rectify noise from the limiter or the discriminator of the
receiver, producing a DC current to turn on 01, which
clamps the LM388 in an off condition.

Designed for the high volume consumer market, the LM388
ideally suits applications in FM scanners and two way
walkie talkie radios. Requirements for this market generally
fall into three areas:
4·40

In all other respects (including pin-out) the LM390 is
identical to the LM388 (Section 4.9). Gain control, input
biasing, muting, and bootstrapping are all as explained
previously for the LM386 and LM388.

v,

o---1l---+................'VIIV--t:
NOISE
INPUT
(FOR
SQUELCH)

~~--------

__.--v+

2.2H
O.l.uF

I
CB

FIGURE 4.9.13 LM3BB Squelch Circuit for FM Scanners and
Walkie Talkies

As shown, the following performance is obtained:
•

Voltage gain equals 20 to 200 (selectable with Rl).

•

Noise (output squelched) equals 20/lV.

•

Po

THD

= 5%)

= 0.19W (V s = 4.5V, RL = 8Q, THD
• Current consumption (V s = 7.5V):

= 5%)

•

= 0.53W (V s = 7.5V,

Po

RL

= 8Q,

squelched - 0.8mA
Po = 0.5W - 110mA
4.10

LM390 1 WATT BATTERY OPERATED AUDIO
POWER AMPLIFIER

FIGURE 4.10.2 LM390 Output Stage

Battery operated consumer products often employ 4Q
speaker loads for increased power output. The LM390
meets the stringent output voltage swings and higher
currents demanded by low impedance loads. Bootstrapping
of the upper output stage (Figure 4.10.1) maximizes positive
swing, while a unique biasing scheme (Figure 4.10.2) used
on the lower half allows negative swings down to within
one saturation drop above ground. Special processing
techniques are employed to reduce saturation voltages to a
minimum. The result is a monolithic solution to the
difficulties of obtaining higher power levels from low
voltage supplies. The LM390 delivers 1 W into 4Q (6V)
at a lower cost than any competing approach, discrete or IC
Figure 4.10.3).

VI"1

7
-

10k

3,4 5
10,11,12

FIGURE 4.10.3 1 Watt Power Amplifier for 6 Volt Systems

14
~----------------------------------~--~~vs

13

- INPUT

L---~--~-------------t---t~~~------~------~~GND

FIGURE 4.10.1 LM390 Simplified Schematic

4-41

4.11 BOOSTED POWER AMPLIFIERS
4.11.1 Introduction

35

When output power requirements exceed the limits of
available monolithic devices, boosting of the output with
two external transistors may be done to obtain higher
power levels. The simplest approach involves adding a
complementary emitter follower output stage within the
feedback loop. The limiting factor is the limitation upon
output voltage swing imposed by the 8·E drop from the
driver's output. Such designs cannot swing closer to the rail
voltages than about one volt less than the IC's swing.

30

Po'" lOW

25
~

20

«>

15

10
Vs" 26V

o

l~r~'I'1
10

100

4.11.2 Output Boost with Emitter Followers

lk

10k

FREQUENCY (Hz)

The simple booster circuit of Figure 4.11.1 allows power
output of 1OW/channel when driven from the LM378. The
circuit is exceptionally simple, and the output exhibits
lower levels of crossover distortion than does the LM378
alone. This is due to the inclusion of the booster transistors
within the feedback loop. At signal levels below 20mW, the
LM378 supplies the load directly through the 5.Q resistor
to about 100mA peak current. Above this level, the booster
transistors are biased ON by the load current through the
same 5.Q resistor.

FIGURE 4.11.2 10 Watt Boosted Amplifier, Frequency Response

+13V

N.C.

4n
SPKR

C3

-flV O.41JlF

82

MYlAR

2k

FIGURE 4.11.3 12 Watt Low·Distortion Power Amplifier

lOOk

35
Po ""11W

30
25

FIGURE 4.11.1 10 Watt Power Amplifier
CD

'"J
The response of the lOW boosted amplifier is indicated in
Figure 4.11.2 for power levels below clipping. Distortion is
below 2% from about 50Hz to 30kHz. 15W RMS power is
available at 10% distortion; however, this represents ex·
treme clipping. Although the LM378 delivers little power,
its heat sink must be adequate for about 3W package
dissipation. The output transistors must also have an
adequate heat sink.

20
15

10
Vs "'±13V

11~1\1I"~nlll

o
10

100

lk

10k

lOOk

FREQUENCY (Hz)

FIGURE 4.11.4 Response for Amplifier of Figure 4.11.3

The circuit of Figure 4.11.3 achieves about 12W/channel
output prior to clipping. Power output is increased because
there is no power loss due to effective series resistance and
capacitive reactance of the output coupling capacitor
required in the single supply circuit. At power up to lOW/
channel, the output is extremely clean, containing less than
0.2% THD midband at lOW. The bandwidth is also im·
proved due to absence of the output coupling capacitor.
The frequency response and distortion are plotted in Figures
4.11.4 and 4.11.5 for low and high power levels. Note that
the input coupling capacitor is still required, even though
the input may be ground referenced, in order to isolate and
balance the DC input offset due to input bias current. The

feedback coupling capacitor, Cl, maintains DC loop gain
at unity to insure zero DC output voltage and zero DC load
current. Capacitors Cl and C2 both contribute to decreasing
gain at low frequencies. Either or both may be increased for
better low frequency bandwidth. C3 and the 27k resistor
provide increased high frequency feedback for improved
high frequency distortion characteristics. C4 and C5 are low
inductance mylar capacitors connected within 2 inches of
the IC terminals to ensure high frequency stability. Rl and
Rf are made equal to maintain VOUTDC = O. The output
should be within 10 to 20mV of zero volts DC. The internal
4-42

I.'

v++

Vs - +13V
RL =4n

1.2
1.0

wi

Po" 750 m

0.8

~
0
~
~

5W
0.6
10W
0.4

(a)
0.2

v+

o
10

100

1k

10k

lOOk

FREQUENCY (Hz)

FIGURE 4.11.5 Distortion for Amplifier of Figure 4.11.3

(b)

bias is unused; pin 1 should be open circuit. When experimenting with this circuit, use the amplifier connected to
terminals 8, 9 and 13_ If using only the amplifier on
terminals 6, 7 and 2, connect terminals 8 and 9 to ground
(split supply) to cause the internal bias circuits to disconnect.

FIGURE 4.12.1 Simple Audio Circuits

where :

Po = power output
Vo = RMS output voltage
10

4.11.3 LM391 Power Driver
Coming in late 1976 will be National's LM391 power driver
IC designed to provide complementary output drive for
external transistors. Power amplifiers up to 50W will be
possible with complete SOA protection provided on-chip,
allowing for simple, low parts-count designs. User gain
control, set externally, offers maximum flexibility, while
special internal techniques allow for the high supply
voltages required by high power amplifiers, thus eliminating
the expense and inconvenience of two power supplies.
Optimized for the top-of-the-line medium power amplifiers,
the LM391 promises to simplify and cut costs of these
designs while retaining true high quality performance.

= RMS output current

Transforming Equation (4.12.1) into peak-to-peak quantities
gives:
RL IOpp2

_._.._-8

(4.12.2)

VS--":",C.....- -

For high power, battery operated audio products, work is
being finalized on a new low voltage driver IC designed to
complement the LM391 in operation and performance, but
optimized for 6·12V, 2.Q designs. Scheduled for introduction in early 1977, this IC will greatly reduce the cost and
difficulties of obtaining the high output swing and large
currents demanded.

VCE

RL

GND------~------4>_

4.12 POWER DISSIPATION
Power dissipation within the integrated circuit package is a
very important parameter requiring a thorough understanding if optimum power output is to be obtained. An
incorrect power dissipation (PD) calculation may result in
inadequate heatsinking, causing thermal shutdown to
operate and limit the output power. All of National's line
of audio power amplifiers use class B output stages. Analysis
of a typical (ideal) output circuit results in a simple and
accurate formula for use in calculating package power
dissipation.

FIGURE 4.12.2 Class B Waveforms

Figure 4.12.2 illustrates current and voltage waveforms in a
typical class B output. Dissipation in the top transistor aT
is the product of collector-emitter voltage and current, as
shown on the top axis. Certainly aT dissipates zero power
when the output voltage is not swinging, since the collector
current is zero. On the other hand, if the output waveform
is overdriven to a square wave (delivering maximum power
to the load, R Ll aT delivers large currents, but the voltage
across it is zero - again resulting in zero power. In the
range of output powers between these extremes, aT goes
through a point of maximum dissipation. This point always
occurs when the peak-to-peak output voltage is 0.637 times

4.12.1 Class B Power Considerations
Begin by considering the simplest audio circuit as in Figure
4.12.1, where the power delivered to the load is:
(4.12.11

4-43

Equation (4.12.51 is the peak value of VL that results in
max PO; multiplying by two yields the peak-to-peak value
for max PO:

the power supply. At that level, assuming all class 8 power
is dissipated in the two output transistors, the chip
dissipation is:
max Po

V s2
--2112 RL

~

2 Vs
VLp_p ~--;- ~ 0.637 Vs

(4.12.31

(4.12.61

Substitution of Equation (4.12.51 into Equation (4.12.4)
gives the final value for max PO:

Inserting the applicable supply voltage and load impedance
into Equation (4.12.31 gives the information needed to size
the heat sink for worst case conditions.

max Po
4.12.2 Derivation of Max PD

~

V s2

(4.12.7)

- - - ""
2112 RL

Another useful form of Equation (4.12.71 is obtained by
substitution of Equation (4.12.21:

The derivation of Equation (4.12.31 for maximum power
dissipation follows from examination of Figure 4.12.2 and
application of standard power formulas:

max Po

~

4

(4.12.8)

- Po(maxl
112

Neglect XCc and let VL' ~ voltage across the load (resistive 1
4.12.3 Application of Max PD

then

Max Po determines the necessity and degree of external
heatsinking, as will be discussed in Section 4.14.

VL' ~ VL sin wt
VCE

~

Vs

_(~s + VL sin wt) ~ ~s - VL sin wt

10.0

Rl -4n
Rl '" an
RL - 16.1"2

5.0

~

IC

~

'"

~

since

PD

211

(2{:dd(W t l

1.0

/
~

Po _ Vp_pl
SRL

0.5

_

PBpAi9;:'"THD
0.1

~two

'-L

V

Po AT 10% THOUP30%

V
4

=

6

8 10

20

30 40

v p. p - PEAK·TO·PEAK OUTPUT IV)

transistors operated Class 8 (since both
transistors are in the same IC package)

FIGURE 4.12.3 Power Out

where:

Po

~

average power

Pd

~

instantaneous power

(;(~s - VL . Wj\(VL
sin wt)
JS"2-R-L- d(wtl

then 1
Po ~ :;;

Sin

Vs VL1·"
VL 21"
Sin wtd(wt) - - - (1·- cos 2wt) d(wt)
211RL
211RL
o
0
Vs VL
VL 2
--(21---(111
211RL
211RL

~ ---

~

Vs VL

VL 2

6

B 10

20

30 40

SUPPl V VOLTS IVsl

(4.12.4)

FIGURE 4.12.4 Max Chip Dissipation

Equation (4.12.41 is the average power dissipated; the

maximum average power dissipated will occur for the value

The nomographs of Figures 4.12.3 and 4.12.4 make it easy
to determine package power dissipation as well as output
VI characteristics for popular conditions. Since part of the
audio amplifier specmanship game involves juggling output
power ratings given at differing distortion levels, it is useful
to know that:

of VL that makes the first derivative of Equation (4.12.41
equal to zero:
d(Pol
d(VLI

~
11 RL

11

_ VL
RL

~

0 at maximum

Po increases by 19% at 5% TH D
Po increases by 30% at 10% THD

(4.12.5)

4·44

3.5

3.0

3.5

3% olST

Ll

LEVEL

f--f--f--f--t-t",-r--

~

~

0.5

f-+-l-+--+-f-+~
o

0.. 1.0 1.5 2.0

2.5

r-

2.5 3.0 3.5 4.0

o

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.55.0

o

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.04.5 5.0

OUTPUT POWER IWATTS)

OUTPUT POWER IWATTS)

OUTPUT POWER {WATTS)

Device Dissipation vs. Output

Device Dissipation vs. Output

Device Dissipation vs. Output

Power - 411 Load

Power - 811 Load

Power - 16n Load

FIGURE 4.12.5 Data Power Curves as Shown on Many Data Sheets

FIGURE 4.12.6 Bridge Audio

Equation (4.12.6) raises an intriguing question: If max Po
occurs at peak-to-peak output voltages equal to 0.637 times
the power supply, will Po go down if the output swing is
increased? The answer is yes - indeed if an amplifier runs
at 0.637Vs to the load, and then is driven harder, say to
0.8V s , it will cool off, a phenomenon implied in the power
curves given on many audio amplifier data sheets (Figure
4.12.5).

4.13 EFFECT OF SPEAKER LOADS
The power dissipation results found in the previous section
assumed a purely resistive load; however, real-world speakers are anything but resistive. Figure 4.13.1 shows an
impedance curve for a typical dynamic loudspeaker. As can
be seen, there is a wide variation in impedance between
20 Hz and 20kHz. The impedance at the resonant frequency
can commonly measure five times or more the rated
impedance. Indeed, many speakers will only display their
rated impedance at one frequency (typically 400 Hz). The
actual impedance is a complex value of DC resistance,
inductive reactance of the voice coil, coupling capacitor
reactance, crossover network impedance and frequency. In
general, though, loudspeakers appear inductive with a worst
case phase angle of 60 degrees. This means that the voltage
through the speaker leads the current by 60 degrees.

4.12.4 Max PD of Bridge Amplifiers
Bridge connecting two amplifiers as in Figure 4.12.6 results
in a large increase of output power. In this configuration
the amplifiers are driven antiphase so that when A1's
output voltage is at V s, A2'S output is at ground. Thus
the peak-to-peak voltage is ideally twice the supply voltage.
Since output power is the square of Voltage, four times
more power can be obtained than from one of these same
amplifiers run single. Note, however, that since the peak
voltage across the bridged load is twice that run as a single,
the amplifiers must be capable of twice the peak currents.
This, along with the fact that no real power amplifier can
swing its output completely to Vs and ground, explains why
actual bridge circuits never fu"Y realize four times their
single circuit output power.

Abandoning mathematical rigor for a more intuitive approach to what phase angle does to maximum average
power dissipation produces the realization that the worst
case load for power dissipation is purely reactive, Le., 90
degrees phase angle. This becomes clear by considering the
resistive case of zero phase angle depicted in Figure 4.13.2a,
where the maximum voltage across the load, V L resultS in
maximum current, I L; but since they are in phase there
exists zero volts across the device and no package dissipation
results. Now, holding everything constant while introducing
a phase angle causes the voltage waveforms to shift position
in time, while the current stays the same. The voltage across
the load becomes smaller and the voltage across the package
becomes larger, so with the same current flowing package
dissipation increases. At the limit of 90 degree phase

Power dissipation in a bridge is calculated by noting that
the voltage at the center of the load does not move. Thus,
Equation (4.12.3) can be applied to half the load resistor:

PA1 or A2

(4.12.8)

4-45

S

14~HH~4H~+H~IJ~~

~

12 ~~~-+~~-H~~++~

ii
cA

0.6

~~

~ 10 ~-jfJf!M-+~~-+JlI+I1rt1f-++~

!

1.0

;::

16

z

..
z

18

~

8 ~-tt1fttm-+~Hl\b.4-Ic++lIl!If-++~

0.8
0.4
0.2

l-/

6~Hffl~~~+H~4+~
4~Hffi~4H~+H~4+~
2 ~~~++~~~m-~~

1

I

:l:

I

10 20 30 40 50 60 10
LOAD .ANGLE (DEGREES)

OL....L.J..llllJlL....LJ.LW"'-Lllllilll-L.I..llIlW
20
100 400 lk
10k
lOOk

FIGURE 4.13.3 Class B Package Dissipation for Reactive Loads

FREQUENCY (Hz)

FIGURE 4.13.1 Impedance Curve for a Typical Dynamic

loudspeaker

v/Sh
VS/2

Vs
vl

:
1

VeE

\.

1\

Vs/2

1/

1

I

VS/2

maximum power output; also, most heat sinks have
adequate thermal capacity to ride through these peaks. In
any event, phase angle is real and it does increase power
dissipation and needs to be considered in heat sink design.

I

0,-1

IL :ri\
1

VS/2

veE

o

I:

~
1

ILolH\

900

4.14 HEATSINKING
Insufficient heatsinking accounts for many phone calls
made to complain about power ICs not meeting published
specs. This problem may be avoided by proper application
of the material presented in this section. Heatsinking is not
difficult, although the first time through it may seem
confusing.

90 0

Speaker Voltage and Current

Phase Angle Equal to 0 Degrees

Speaker Voltage and Current
Phase Angle Equal to 90 Degrees

(a)

(b)

If testing a breadboarded power IC results in premature
waveform clipping, or a "truncated shape," or a "melting
down" of the positive peaks, the IC is probably in thermal
shutdown and requires more heatsinking. The following
information is provided to make proper heat sink selection
easier and help take the "black magic" out of package
power dissipation.

FIGURE 4.13.2 Phase Angle Relationship Between Voltage and
Current

difference Figure 4.13.2b results, where there exists zero
volts across the load, maximum voltage across the package,
and maximum current flowing through both, producing
maximum package dissipation.
Returning to mathematics for a moment to derive a new
expression containing phase angle and plotting the results
produces the curve shown in Figure 4.13.3. The importance
of Figure 4.13.3 is seen by comparing the power ratio at
zero degrees (0.405) with that at 60 degrees (0.812) double! This means that the maximum package dissipation

4.14.1 Heat Flow

can be twice as much for a loudspeaker load as for a resistive
load. What softens this hard piece of reality is the relative

Heat can be transferred from the IC package by three
methods, as described and characterized in Table 4.14.1.

rarity and short duration of amplifiers running at (or near)

TABLE 4.14.1 Methods of Heat Flow

METHOD

DESCRIBING PARAMETERS

Conduction is the heat transfer method most effective
in moving heat from junction to case and case to heat
sink.

Thermal resistance JC and CS. Cross section, length
and temperature difference across the conducting
medium.

Convection is the effective method of heat transfer
from case to ambient and heat sink to ambient.

Thermal resistance GSA and eCA. Surface condition,
type of convecting fluid, velocity and character of
the fluid flow (e.g., turbulent or laminar), and temperature difference between surface and fluid.

Radiation is important in transferring heat from cooling fins.

Surface emissivity and area. Temperature difference
between radiating and adjacent objects or space. See
Table 4.14.2 for values of emissivity.

e

4-46

e

(a) Mechanical Diagram

Symbols and Definitions

-.-.CHIP JUNCTION

Po

e
eJL
eLS

TEMP ITJI

'Jl
-.-.lEADFRAME

OSA
OJS
0JA
TJ
TA
PD

TEMP ITL!
'lS
-.-.HEAT SINK

TEMP ITsl
'SA

-=-

-.-. AMBIENT

TEMP ITAI

Thermal Resistance (0 CIWI
Junction to Leadframe
Leadframe to Heat Sink
Heat Sink to Ambient
Junction to Heat Sink = 0 JL + 0 LS
Junction to Ambient = OJL + 0 LS + OSA
Junction Temperature (maximuml (oCI
Ambient Temperature
Power Dissipated (WI

(c) Symbols and Definitions

(b) Electrical Equivalent

FIGURE 4.14.1 Heat Flow Model

4.14.4 Where to Find Parameters

4.14.2 Thermal Resistance
Thermal resistance is nothing more than a useful figure-ofmerit for heat transfer. It is simply temperature drop
divided by power dissipated, under steady state conditions.
The units are usually °C/W and the symbol most used is
AB. (Subscripts denote heat flowing from A to B.I

PD
Package dissipation is read directly from the "Power
Dissipation vs. Power Output" curves that are found on all
of the audio amp data sheets. Most data sheets provide
separate curves for either 4, 8 or 162 loads. Figure 4.14.2
shows the 82 characteristics of the LM378.

o

The thermal resistance between two points of a conductive
system is expressed as:

(4.14.11

4_14_3 Modeling Heat Flow
An analogy may be made between thermal characteristics
and electrical characteristics which makes modeling straightforward:
T - temperature differential is analogous to V (voltagel

o - thermal resistance is analogous to R (resistancel
P - power dissipated is analogous to I (currentl
Observe that just as R = VII. so is its analog
model follows from this analog.

a = TIP. The

A simplified heat transfer circuit for a power IC and heat
sink system is shown in Figure 4.14.1. The circuit is valid
only if the system is in thermal equilibrium (constant heat
flowl and there are, indeed, single specific temperatures T J,
TL, and TS (no temperature distribution in junction, case,
or heat sinkl. Nevertheless, this is a reasonable approximation of actual performance.

POWER OUTPUT IW/CHANNElI

FIGURE 4.14.2 Power Dissipation vs. Power Output

4-47

And with the best heat sink possible, the maximum dissi·
pation is

Note: For Po = 2W and Vs = 18V, PD(max) = 4.1 W,
while the same Po with Vs = 24 V gives PD(max) =
6.5W - 50% greater! This point cannot be stressed
too strongly: For minimum PD, Vs must be selected
for the minimum value necessary to give the required
power out.

Or, for you formula lovers:
For loads other than those covered by the data sheet
curves, max power dissipation may be calculated from
Equation (4.14.2). (See Section 4.12.)

V s2
PD(max) = - - 20 RL

Max Allowable PD =

TJ(max)-TA

(4.14.3)

OJA
4.14.5 Procedure for Selecting Heat Sink

(4.14.2)

1. Determine PD(max) from curve or Equation (4.14.2).

eLS if soldering; if not, eLS must be considered.

Equation (4. 14.2) is for each channel when applied to duals.

2. Neglect

When used for bridge configurations, package dissipation
will be twice that found from Figure 4.14.2 (or four times
Equation (4.14.2).

3. Determine eJL from curve.
4. Calculate 0JA from Equation (4.14.3).
5. Calculate OSA for necessary heat sink by subtracting (2)
and (3) from (4) above, i.e., OSA = JA - 0 JL - LS.

e

OLS
Thermal resistance between lead frame and heatsink is a
function of how close the bond can be made. The method
recommended is use of 60/40 solder. When soldered, 0 LS
may be neglected or a value of 0 LS = 0.25°C/W may be
used.

e

For example, calculate heat sink required for an LM378
used with Vs = 24V, RL = 8~, Po = 4W/channel and
TA=25°C:
1. From Figure 4.14.2, PD = 7W.
2. Heat sink will be soldered, so 0 LS is neglected.

TJ(max)
Maximum junction temperature for each device is 150°C.

3. From Figure 4.14.3, OJL = 13.4°C/W.
4. From Equation (4.14.3),

OJL
eJA = 150°C - 25°C = 17.90C/W.
7W

Thermal resistance between junction to lead frame (or
junction to heat sink if 0 LS is ignored) is read, directly from
the "Maximum Dissipation vs. Ambient Temperature" curve
found on the data sheet. Figure 4.14.3 shows a typical
curve for the LM378.

10

~

.,

0

>=

;,:
~

c
u

~

!i

.,.
;;;
x

5. From Equation (4.14.4),
OSA = 17.90C/W-13.4°C/W = 4.5°C/W.
Therefore, a heat sink with a thermal resistance of 4.5°C/W
is required. Examination of Figure 4.14.3 shows this to be
substantial heatsinking, requiring forethought as to board
space, sink cost, etc.

INFINITE SINK

[""...

I
I

.......

Results modeled:

........

PC + V7

~.

r--...

r-

FREJ AIR
l~flNITE

SINK !HOC,IIN

PCBOARO·V,21°CIW
21/2SQ.INPCBOARD29'CIW

FREEAIR,8"Crw

10

20

30

40

.........

7W

---

50

60

13.4'CIW
-

LEAOFRAME TEMP' 150 _ (13;'C)7W ' 56.2'C

0.25'C/W

70

- - HEATSINK TEMP" 56

TA - AMBIENTTEMPERATURE I'C)

zoe _/O.Z5°C)7W
'"
\ W

.

545°C

.

4.5'CIW
_

-=

FIGURE 4.14.3 Maximum Dissipation vs. Ambient Temperature

0

AMBIENT TEMP , 54.Soc _(4.5 C)7W
W

0

23°C 12°C ERROR OUE TO
NEGLECTING 8LS)

FIGURE 4.14.4 Heat Flow Model for LM378 Example

Note: OJL is the slope of the curve labeled "Infinite
Sink." It is also 0 JA(best), while 0JA(worst) is the
slope of the "Free Air" curve, i.e., infinite heat sink
and no heat sink respectively.
So, what does it mean? Simply that with no heat sink you
can only dissipate

4.14.6 Custom Heat Sink Design
The required OSA was determined in Section 4.14.5. Even
though many heat sinks are commercially available, it is
sometimes more practical, more convenient, or more
economical to mount the regulator to chassis, to an
aluminum extrusion, or to a custom heat sink. In such
cases, design a simple heat sink.
4-48

Simple Rules

The procedure for use of the nomogram of Figure 4.14.6 is
as follows:

1. Mount cooling fin vertically where practical for best
conductive heat flow.

1. Specify fin height H as first approximation.

2. Anodize, oxidize, or paint the fin surface for better
radiation heat flow; see Table 4.14.2 for emissivity data.

2. Calculate h ~ hr + hc from Equations (4.14.6) and
(4.14.7).

3. Use 1/16" or thicker fins to provide low thermal
resistance at the IC mounting where total fin cross·
section is least.

4. Determine 1"/ from values of B (from Figure 4.14.5) and c<
(line b).

3. Determine c< from values of h and fin thickness x (line a).

Fin Thermal Resistance

The value of 1"/ thus determined is valid for vertically
mounted symmetrical square or round fins (with H }> d)
in still air. For other conditions, 1"/ must be modified as
follows:

The heat sink-to-ambient thermal resistance of a vertically
mounted symmetrical square or round fin (see Figure 4.4.5)
in still air is:

Horizontal mounting - multiply he by 0.7.
OSA ~

(4.14.5)

°C/W

Horizontal mounting where only one side is effective multiply 1"/ by 0.5 and hc by 0.94.

2 H21"/ (h c + h r )
where:

1"/

height of vertical plate in inches

For 2:1 rectangular fins - multiply h by 0.8.

fin effectiveness factor

For non-symmetrical fins where the IC is mounted at the
bottom of a vertical fin - mUltiply 1"/ by 0.7.

~

H

~

hc

convection heat transfer coefficient (4.14.6)

hr

radiation heat transfer coefficient

hc

TS-TA)Y.
2.21 x 10-3 ( --H--W/in2°C

hr ~ 1.47 x 10- 10 E (
where:

~

TS

TS+TA

-""2--

Fin Design

(4.14.7)

1. Establish initial conditions, T A and desired 8SA as
determined in Section 4.14.5.
2. Determine TS at contact point with the IC by rewriting
Equation (4.14.1):

,\3. °
+ 273) W/m 2 C

temperature 00f heat sink at IC
mounting, In C

TJ - TS
8JL + OLS ~ - - PD

(4.14.8)

TS ~ TJ - (8JL + 8LS) (PD)

(4.14.9)

T A ~ ambient temperature in °c
E

~

surface emissivity (see Table 4.14 ..2)
3. Select fin thickness, x

Fin effectiveness factor 1"/ includes the effects of fin thickness, shape, thermal conduction, etc. It may be determined
from the nomogram of Figure 4.14.6.

>

0.0625" and fin height, H.

4. Determine hc and hr from Equations (4.14.6) and
(4.14.7).
5. Find fin effectiveness factor 1"/ from Figure 4.14.6.

TABLE 4.14.2 Emissivity Values for Various Surface Treatments

6. Calculate 0SA from Equation (4.14.5).
7. If 8SA is too large or unnecessarily small, choose a
different height and repeat steps (3) through (6).

EMISSIVITY, E

SURFACE
Polished Aluminum
Polished Copper
Rolled Sheet Steel
Oxidized Copper
Black Anodized Aluminum
Black Air Drying Enamel
Dark Varnish
Black Oil Paint

0.05
0.07
0.66
0.70
0.7
0.85
0.89
0.92

Design Example
Design a symmetrical square vertical fin of black anodized
1/16" thick aluminum to have a thermal resistance of
4°C/W. LM379 operating conditions are:

- 0.9
- 0.91
- 0.93
- 0.96

1. TJ ~ 150°C, TA ~ 60°C, PD ~ 9.5W, OJL
neglect 8 LS.
2. TS ~ 150°C - 6°C/W (9.5W) ~ 93°C.
3. x ~ 0.0625" from initial conditions. E ~ 0.9 from Table
4.14.2.
Select H ~ 3.5" for first trial (experience will simplify
this step).

! ±fo\ ! ! Ic;l
_I_t_'::;J _1_' LJ
B"'1i=.!t
1

4. hc

~

2.21 x

10-3(~L::.~~\Y.
\

B '" O.564H-t

d, uSIng B = HI2 "a satIsfactory approximatIon
for either square Or round fins

Note: For H»

3.5 }

3.86 x 1O-3 WtCin 2
1.47 x 10-10 x 0.9(93; 60 +

FIGURE 4.14.5 Symmetrical Fin Shapes

4-49

273Y

B:". H/2

40
30

0.05

20

0.1

0.2

x: FIN THICKNESS
FOR
ALUMINUM

h=hr+h e

FOR
COPPER

--rr

"%
FIN EFFECTIVENESS

~~'0

1.0
0.8
0.7

05

~'.0

1.0

0.1

1.0

0.6
05
04
2.0
0.3

3.0

0.01

0.01

94

90
88
88

0.001

0.001

84

40
5.0

0.2

,a

O.,~

0.0001

82

80

75

INCHES
INCHES

70

10.0

65

l/rNCH

0.001

60
55
50
45
40

K: Thermal Conductivity of the Fm

35
%

FIGURE 4.14.6 Fin Effectiveness Nomogram for Symmetrical, Flat, Uniformly-Thick, Vertically Mounted Fins

5.6

5.

X

1O-3 wfc in 2

National Semiconductor's use of copper leadframes in
packaging power ICs, where the center three pins on either
side of the device are used for heatsinking, allows for
economical heat sinks via the copper foil that exists on the
printed circuit board. Adequate heatsinking may be
obtained for many designs from single-sided boards con·
structed with 2 oz. copper. Other, more stringent, designs
may require two-sided boards, where the top side is used
entirely for heatsinking. Figure 4.14.7 allows easy design
of PC board heat sinks once the desired thermal resistance
has been calculated from Section 4.14.5.

1O- 3 wfc in 2

h

hC + hr = 9.46

1)

0.84 from figure 4.14.6.

X

4.14.7 Heatsinking with PC Board Foil

103
6. 8SA = - - - - - - 2 X 12.3 x 0.84 x 9.46
which is too large.
7. A larger fin is required, probably by about 40% in area.
Accordingly, using a fin of 4.25" square, a new cal cui ati on is made.
4: hc = 2.21 x 1 0-3

(E_)

70

%

3.7 x 10-3

,.
'"~~

4.2

W
u

60

I-

hr = 5.6 x 10-3 as before

Wu

50

"'''-

~'"

h = 9.3 x 10-3

5:

1)

"'~

40

~

30

"'~

= 0.75 from Figure 4.14.6.
103

6. iJSA = - - - - - 2 x 18 x 0.75 x 9.3

SO. INCHES COPPER P.C. FOIL. SINGLE SIDE
(3 MILLS THICK DR 2 02/50. FT)

which is satisfactory.

FIGURE 4.14.7 Thermal Resistance vs. Square Inches of Copper

Foil

4·50

5.0 noobydusl
5.1 BIAMPLIFICATION

effects. The first results from the consequence of bass
transient clipping. Low frequency signals tend to have much
higher transient amplitudes than do high frequencies, so
amplifier overloading normally occurs for bass signals.
By separating the spectrum one immediately cleans up half
of it and greatly improves the other half, in that the low
frequency speaker will not allow high frequency compon·
ents generated by transient clipping of the bass amplifier to
pass, resulting in cleaner sound. Second is a high frequency
masking effect, where the low level high frequency distor·
tion components of a clipped low frequency signal are
"covered up" (i.e., masked) by high level undistorted high
frequencies. The final advantage of biamping is allowing the
use of smaller power amplifiers to achieve the same sound
pressure levels.

The most common method of amplifying the output of a
preamplifier into the large signal required to drive a speaker
system is with one large wideband amplifier having a flat
frequency response over the entire audio band. An alternate
method is to employ two amplifiers, or biamplification,
where each amplifier is committed to amplifying only one
part of the frequency spectrum. Biamping requires splitting
up the audio band into two sections and routing these
signals to each ampl ifier. Th is process is accompl ished by
using an active crossover network as discussed in the next
section.
The most common application of biamping is found in con·
junction with speaker systems. Due to the difficulty of
manufacturing a single speaker capable of reproducing the
entire audio band, multiple speakers are used, where each
speaker is designed only to reproduce one section of
frequencies. In conventional systems using one power
amplifier the separation of the audio signal is done by
passive high and low pass filters located within the speaker
enclosure as diagrammed in Figure 5.1.1. These filters must
be capable of processing high power signals and are there·
fore troublesome to design, requiring large inductors and
capacitors.

5.2 ACTIVE CROSSOVER NETWORKS
An active crossover network is a system of active filters
(usually two) used to divide the audio frequency band into
separate sections for individual signal processing by biamped
systems. Active crossovers are audibly desirable because
they give better speaker damping and improved transient
response, and minimize midrange modulation distortion.
5.2.1 Filter Choice
The choice of filter type is based upon the need for good
transient and frequency response. Bessel filters offer excel·
lent phase and transient response but suffer from frequency
response change in the crossover region, being too slow for

easy speaker reproduction. Chebyshev filters have excellent
frequency division but possess unacceptable instabilities in
their transient response. Butterworth characteristics fall
between Bessel and Chebyshev and offer the best compro·
mise for active crossover design.

TWEETER

SIGNAL

WOOFER

5.2.2 Number of Poles (Filter Order)
Intuitively it is reasonable that if the audio spectrum is
split into two sections, their sum should exactly equal the
original signal, i.e., without change in phase or magnitude
(vector sum must equal unity). This is known as a constant
voltage design. Also it is reasonable to want the same power
delivered to each of the drivers (speakers). This is known as
constant power design. What is required, therefore, is a
filter that exhibits constant voltage and constant power.
Having decided upon a Butterworth filter, it remains to

FIGURE 5.1.1 Passive Crossover, Single Amp System

Biamping with active crossover networks (Figure 5.1.2)
allows a more flexibl€ and easier design. It also sounds
better. Listening tests demonstrate that biamped systems
have audibly lower distortion. 4 This is due chiefly to two

TWEETER

SIGNAL

WODFER

FIGURE 5.1.2 Active Crossover, Biamp System

5·1

determine an optimum order of the filter (the number of
poles found in its transfer function) satisfying constant
voltage and constant power.

Applying Equation (5.2.3) yields:

TL(S) =

TH(S)

S
S+l

which at S = -j Wo gives
(5.2.10)
Equation (5.2.9) shows that there is a gradual phase shift
power with one nagging annoyance - the phase has been
inverted. Examination of the phase characteristics of
Equation (5.2.9) shows that there is a gradual phase shift
from 0 0 to _360 0 as the frequency is swept through the
filter sections, being -180 0 at woo Is it audible? Ashley2
demonstrated that the ear cannot detect this gradual phase
shift when it is not accompanied by ripple in the magnitude
characteristic. (It turns out that all odd ordered Butter·
worth filters exhibit this effect with increasing amounts of
phase shift, e.g., 5th order gives 0 to _.720 0 , etc.)

(5.2.2)

=-

where TL(S) equals low pass transfer function and TH(S)
equals high pass transfer function. This filter exhibits
constant voltage (hence, constant power) as follows:
require TL(S) + TH(S) = 1

(5.2.9)

S3 + 2S2 + 2S + 1

(5.2.1 )

S+ 1

S3 + 1

TL(S) + TH(S) =

Both active and passive real izations of a Butterworth filter
have identical transfer functions, so a good place to start is
with conventional passive crossover networks. Passive cross·
overs exhibit a single pole (1st order) response and have a
transfer function given by Equations (5.2.1) and (5.2.2)
(normalized to Wo = 1).

(5.2.3)

The conclusion is that the best compromise is to use a 3rd
order Butterworth filter. It will exhibit maximally flat
magnitude response, i.e., no peaking (which minimizes the
work required by the speakers); it has sharp cutoff charac·
teristics of -18dB/octave (which minimizes speakers being
required to reproduce beyond the crossover point); and it
has flat voltage and power frequency response with a
gradual change in phase across the band.

Inspection of Equations (5.2.1) and (5.2.2) shows this to be
true.
The problem with a single pole system. is that the roll off
beyond the crossover point is only -6dB/octave and requires
the speakers to operate linearly for two additional octaves
if distortion is to be avoided. 6
The 2nd order system exhibits transfer functions:

hIS) = .. _ _
1 __
S2+y'2S+1
S2

(5.2.4)
5.2.3 Design Procedure for 3rd Order Butterworth Active
Crossovers
(5.2.5)

Many circuit topologies are possible to yield a 3rd order
Butterworth response. Out of these the infinite·gain,
multiple· feedback approach offers the best tradeoffs in
circuit complexity, component spread and sensitivities.
Figure 5.2.1 shows the general admittance form for any 3rd
order active filter. The general transfer function is given by
Equation (5.2.11).

These transfer functions exhibit constant power but not
constant voltage. This is demonstrated by applying Equation
(5.2.3), yielding:
(5.2.6)

At crossover, S = -jwo = -j (since Wo = 1); substitution
into Equation (5.2.6) equals zero. This means that at the
crossover frequency there exists a "hole," or a frequency
that is not reproduced by either speaker. Ashley! demon·
strated that this hole is audible. A commonly seen
solution to this problem is to invert the polarity of one
speaker in the system. Mathematically this changes the
sign of the transfer function and effectively subtracts the
two terms rather than adds them. This does eliminate the
hole, but it creates a new problem of severe phase shifting
at the crossover point which Ashley also demonstrated to
be audible, making consideration of 3rd order Butterworth
filters necessary.

';
'0

The transfer functions for 3 pole Butterworth filters are
given as Equations (5.2.7) and (5.2.8).

FIGURE 5.2.1 General Admittance Form for 3rd Order Filter

(5.2.7)

hIS)
S3

(5.2.8)

By substituting resistors and capacitors for the admittances
per Figures 5.2.2 and 5.2.3, low and high pass active filters
are created.
5·2

(5.2.11)

ei

Low Pass:

ei

S3+(R5 R6 + R3 R6 + R3 R5 + R1 + R3)S2
R3R5R6C4

R1 R3C2

+( __

Rl R3R5C2C4C7

1_ _

+

R5 R6 + R3 R6 + R3 R5 + R1 R5 + R1 R6)S +

R5R6C4C7

R1 R3R5R6C2C4

R1

+ R3

R1 R3R5R6C2C4C7

(5.2.12)

High Pass:
eoH

ei

Cl IC3 + C5 + C61 + C31 C5 + C61
1
)
(1
C3 + C5 + C6
)
_ _ _ _1:......._ __
S3 + (
+
S2 + ------- +
S+ R7C5C61C1 + C31
IC1 + C31 R2
C5C6 R4 R7 C5C61C1 + C31 R2 R7
C5C61C1 + C31 R2 R4 R7

(5.2.13)

By letting R1 = R3 = R5 = Rand R6 = 2 R and equating
coefficients between Equations (5.2.12) and (5.2.71. it is
possible to solve for the capacitor values in terms of R.
Doing so yields the relationships shown in Figure 5.2.4. For
the high pass section, let C1 = C3 = C5 = C and C6 = C/2
and equate coefficients to get the resistor values in terms of
C. The high pass results also appear in Figure 5.2.4, which
shows the complete 3rd order Butterworth crossover network.

Substitution of the appropriate admittances shown in
Figures 5.2.2 and 5.2.3 into Equation 5.2.11 gives the
general equation for a 3rd order low pass (Equation (5.2.12))
and for a 3rd order high pass (Equation (5.2.13)):

'.L

Example 5.2.1
Design an active crossover network with -18dB/octave
rolloff (3rd order), maximally flat (Butterworth) charac·
teristics having an input impedance of 20 kD. and a crossover
frequency equal to 500 Hz.

FIGURE 5.2.2 General 3rd Order Low Pass Active Filter

1. Select R for low pass section to set the required input
impedance:
let:

R = 10K (1%)

for RIN = 20K, since RIN = 2 R, then 2R = 20K = 1%.

'.H

2. Calculate C2, C4 and C7 from Figure 5.2.4:
C

FIGURE 5.2.3 General 3rd Order High Pass Active Filter

= .

x

10 8

-

6.71 x 10-8

Use C4 = 0.068pF, 2%.
C

=

7 82

C 2.1089
4 - (2IT)(50Q)(1O K)

K w03

K

2.4553

Use C2 = 0.082pF, 2%.

Equation (5.2.12) is of form

where:

-

2 - (2IT)(500)(10 K)

passband gain

=

1

Letting a = b = 2 and normalizing w0 3 = 1 gives the 3rd
order Butterworth response of Equation (5.2.7).

0.1931
7 - (2IT) (500)(10 K)

6.51 x 10-9

Use C7 = 0.0056pF, 2%.
3. Select C for high pass section to have same impedance
level as RIN for low pass, i.e., 20K ohms:

Similarly, Equation (5.2.13) is of form

C =

(2IT) (500) (20 K)

= 1.592

x 10-8

Use C = 0.015pF, 2%, and use C/2 = 0.0082pF, 2%.

and corresponds to Equation (5.2.8).
5·3

4. Calculate R2, R4 and R7 from Figure 5.2.4:
R2 ~

The completed design is shown in Figure 5.2.5 using LF356
op amps for the active devices. LF356 devices were chosen
for their very high input impedances, fast slew and extremely
stable operation into capacitive loads. A buffer is used to
drive the crossover network for two reasons: it guarantees
low driving impedance which active filters require, and it
gives another phase inversion so that the outputs are in
phase with the inputs. Power supplies are ±15 V, decoupled
with 0.1 ceramic capacitors located close to the integrated
circuits (not shown). Figure 5.2.6 gives the frequency
response of Figure 5.2.5.

0.4074
~ 8148
(2rr)(500)(1.592 x 10-8

Use R2 = 8;06K, 1%.
R4 ~

0.4 742
(2rr)(500)(1.592 x 10-8 )

9484

Use R4 = 9.53 K, 1%.
R7 ~

5.1766
(2rr) (500)(1.592 x 10- 8 )

Figure 5.2.7 can be used to "look up" values for standard
crossover frequencies of 100Hz to 5kHz.

103532

Use R7 = 102K, 1%.

5.2.4 Alternate Design for Active Crossovers
The example of Figure 5.2.5 is known as a symmetrical
filter since both high and low pass sections are symmetrical
about the crossover point (see Figure 5.2.6), An interesting
alternate design is known as the asymmetrical filter (since
the high and low pass sections are asymmetrical about the
crossover point). This design is based upon the simple
realization that if the output of a high pass filter is sub·
tracted from the original signal then the result is a low
pass. 3 Constant voltage is assured since the sum of low and
high pass are always equal to unity (with no phase funnies).
But, as always, there are tradeoffs and this time they are
not obvious.

e,"
-S3
elN
t

S3+2S2 +2S+1

1

_

OH ~ 217Ciflfi R4 R]-

Q '"

0.707, Av

:0

-1

o H-H#HH.tHttlli;;;
-5

H-ttHtIH-HftIlIII-

-10 1-+++H1IIl--~.1I1';

e,l

~ - t 5 H+++H!II--fI!+Hl1I1-,

"

-20

.. -25

Cz

=

z~::~3R

RZ = Z~·::~4C

C4

=

Z~:~~9R

R4 = z~:~:zc

C, = Z~:~~lR

1-+++H1IIl--++I1Hr+.

H+++H!II--IH+ttI!lt.ffi

-30 H-t!+ltllH'tHftttII-\T
eOl _

-1

;jN-

83 +2s2 +28+1

tOl =

n '"

-35

1

'hR$C2C4C,

0.707, Av

=

H-tHfIilllt-HtHl/ll--\H-

-40 H+++\llIII--H-lllI~1+
10

100

lk

lOOk

FIGURE 5.2.6 Active Crossover Frequency Response for Typical
Example of Figure 5.2.5

FIGURE 5.2.4 Complete 3rd Order Butterworth Crossover Network

R,

00~1E"3:rI:.
RZ

-

e,"

R4

a.06k

9.53k

-

-::-

lOOk

fc = 500Hz
GAIN = OdBV

10k

FREUUENCY (Hz)

-1

10k

e,l

FIGURE 5.2.5 Typical Active Crossover Network Example

5-4

fe

C

R2

R4

R7

C2

C4

C7

Hz

J.lF

n

n

n

J.lF

J.lF

J.lF

8148

9484

103532

0.391
0.195
0.130
0.0977
0.0782
0.0651
0.0558
0.0488
0.0434
0.0391
0.0195
0.0130
0.00977
0.00782

0.336
0.168
0.112
0.0839
0.0671
0.0559
0.0479
0.0420
0.0373
0.0336
0.0168
0.0112
0.00839
0.00671

0.0307
0.0154
0.0102
0.00768
0.00615
0.00512
0.00439
0.00384
0.00341
0.00307
0.00154
0.00102
768pF
615pF

100
200
300
400
500
600
700
800
900
lk
2k
3k
4k
5k

0.080
0.040
0.027
0.020
0.016
0.013
0.011
0.010
0.0088
0.008
0.004
0.0027
0.002
0.0016

• Assumes R ~ 10k, 2 R = 20k for Rin ~ 20 kn.
FIGURE 5.2.7 Precomputed Values for Active Crossover Circuit Shown in Figure 5.2.4 (Use nearest available value.)

Referring back to Equation (5.2.8) for the transfer function
of a 3rd order high pass and subtracting it from the original
signal yields the following:
(5.2.14)

TL(5)

1 - TH(5)

TL(5)

53
1 - ._-----5 3 +25 2 +25+1

TL(5)

~

Figure 5.2.8 shows the circuit design for an asymmetrical
filter, and Figure 5.2.9 gives its frequency response.

Will
Will
-10

E

25 2 + 25 + 1

~

(5.2.15)

..l1
~

111111

II

-18
d8! OCT~~E

-6
dBf2-CTAVE

-20

.l

5 3 + 25 2 +25 + 1

-30

Analysis of Equation (5.2.15) shows it has two zeros and
three poles. The two zeros are in close proximity to two of
the poles and near cancellation occurs. The net result is a
low pass filter that exhibits only -6dB rolloff and rather
severe peaking (~ +4dB) at the crossover point. For low
frequency drivers with extended frequency response, this is
an attractive design offering lower parts count, easy adjustment, no crossover hole and without gradual phase shift.

II

-40
10

10k

lk

100

lOOk

FREQUENCY IHd

FIGURE 5.2.9 Frequency Response of Asymmetrical Filter Shown

in Figure 5.2.8

R,

rp:::

o.:tpl'
..
R2

8.06k

'OH

R4
9.53k

-=-

"N--VVV_",
lOOk

Rg'
lOOk

127k
• MISMATCH BETWEEN R8 AND A9 CORRECTS
FOR GAIN ERROR OF HIGH PASS DUE TO
CAPACITOR TOLERANCES.

'OL

FIGURE 5.2.8 Asymmetrical 3rd Order Butterworth Active Crossover Network

5·5

5.2.5 Use of Crossover Networks and Bial1lping
Symbolically, Figure 5.2.5 can be represented as shown in
Figure 5.2.10:

-OH
B- BUffER AMPLIFIER
HP- HIGH PASS FILTER
LP- lOW PASS FIL TER

-IN

Figures 5.2.11-5.2.14 use Figure 5.2.10 to show several
speaker systems employing active crossover networks and
biamping.

-OL

FIGURE {;_2_10 Symbolic Representation of Figure 5_2.5

TWEETER }

LEFT CHANNel

LEFT
WOOFER

A' POWER AMPLIFIER

TWEETER }

RIGHT CHANNEL

RIGHT
WOOFER

FIGURE 5.2_11 Stereo 2-Way System (Typical crossover poi"t. from 800 to 1600Hz)

Cascading low pass (LP) and high pass (HP) active filters
creates a bandpass and allows system triamping as follows:

TweETER

MIDRANGE

INPUT

WOOFER

FIGURE 5_2_12 Single Channel 3-Way System
(Duplicate for Stereo)
(Typical crossover points: LP = 200 Hz, HP = 1200 Hz)

LEFT TWEETER

LEFT

COMMON WOOFER

RIGHTTWEETER

RIGHT

FIGURE 5.2.13 Common Woofer 2-Way Stereo SystemS
(Stereo-to-mono crossover point typically 150 Hz)

5-6

TWEETER

1
LEFT CHANNEL

MIDRANGE

LEFT

COMMON WOOFER

RIGHT

MIDRANGE

1
RIGHT CHANNel

TWEETER

FIGURE 5.2.14 Common Woofer 3·Way Stereo System (Typically LP1 ~ HPl ~ 150 Hz, LP2 ~ HP2 ~ 2500 Hz)

spring slowly propagates along the length of the unit until
it arrives at the other end, where similar magnets convert it
back into an electrical signal. (Reflection also occurs, which
creates the long decay time, relative to the delay time.)

REFERENCES

1. Ashley, J. R., "On the Transient Response of Ideal
Crossover Networks," Jour. Aud. Eng. Soc., vol. 10, no.
3, July 1962, pp. 241-244.

5.3.1

2. Ashley, J. R. and Henne, L. M., "Operational Amplifier
Implementation of Ideal Electronic Crossover Networks,"
Jour. Aud. Eng. Soc., vol. 19, no. 1, January 1971,
pp.7-11.

Design Considerations for Driver and Recovery
Amplifiers

Since the reverb driver is applying an electrical signal to a
coil, its load is essentially inductive and as such has a rising
impedance vs. frequency characteristic of +6 dB/octave.
Further, since the spring assembly operates best at a fixed
value of ampere/turns (independent of frequency), it becomes desirable to drive the transducer with constant
current. Constant current can be achieved in two ways:
(1) by incorporating the input transducer as part of the
negative feedback network, or (2) by creating a riSing
output voltage response as a function of frequency to
follow the corresponding rise in output impedance. Method
(1) precludes the use of grounded input transducers, which
tend to be quieter and less susceptible to noise transients.
(While grounded load, constant current sources exist, they
require more parts to implement.) For this reason method
(2) is preferred and will be used as a typical design example.

3. Ashley, J. R. and Kaminsky, A. L., "Active and Passive
Filters as Loudspeaker Crossover Networks," Jour. Aud.
Eng. Soc., vol. 19, no. 6, June 1971, pp. 494-501.
4. Lovda, J. M. and Muchow, S., "Bi-Amplification Power vs. Program Material vs. Crossover Frequency,"
AUDIO, vol. 59, no. 9, September 1975, pp. 20-28.
5. Read, D. C., "Active Crossover Networks," Wireless
World, vol. 80, no. 1467, November 1974, pp. 443-448.
6. Small, R. H., "Constant-Voltage Crossover Network
Design," Jour. Aud_ Eng. Soc., vol. 19, no. 1, January
1971, pp. 12-19.

A high slew rate (~ 2V/ps) amplifier should be used since
the rising amplitude characteristic necessitates full output
swing at the maximum frequency of interest (typical spring
assemblies have a frequency response of 100Hz-5kHz),
thereby allowing enough headroom to prevent transient
clipping. It is also advisable to roll the amplifier off at high
frequencies as a further aid in headroom. "Booming" at low
frequencies is controlled by rolling off low frequencies
below 100Hz.

5.3 REVERBERATION
Reverberation is the name applied to the echo effect
associated with a sound after it has stopped being generated.
It is due to the reflection and re-reflection of the sound off
the walls, floor and ceiling of a listening environment and
under certain conditions will act to enhance the sound. It is
the main ingredient of concert hall ambient sound and
accounts for the richness of "live" versus "canned" music.
By using electro-mechanical devices, it is possible to add
artificial reverberation to existing music systems and
enhance their performance. The most common reverberation
units use two precise springs that act as mechanical delay
lines, each delaying the audio signal at slightly different
rates. (Typical delay times are ~ 30 milliseconds for
one spring and ~ 40 milliseconds for the other, with
total decay times being around 2 seconds.) The electrical
signal is applied to the input transducer where it is translated
into a torsional force via two small cylindrical magnets
attached to the springs. This "twisting" of one end of each

The requirements of the recovery amplifier are determined
by the recovered signal. Typical voltage levels at the transducer output are in the range of 1-5mV, therefore requiring
a low noise, high gain preamp. Hum and noise need to be
minimized by using shielding cable, mounting the reverb
assembly and preamp away from the power supply transformer, and using good single point ground techniques to
avoid ground loops. Equalization is not necessary if a
constant current drive amplifier is used since the output
voltage is constant with frequency.
5-7

C"

0.01

R.

C2

"3

10k

"6

220k

RIO
22k

C6

220k

C6

14

R,

2.2M

C6

'24V~~
L ,

'24V~~
.... -

51 Ok

"s

"I
lOOk

"

R12

C7
10pF

22Dk

160pF

'24Vo-,---lOh-

. . --t- . .

-...,

r-+-L-,
I

I

I

I
C3
LEFT

>.:--0-4--0 LEFT

0.02

>._o-.--oRIGHT

RIGHT

I
I

C3
0.02

IL ____ ..JI

",

R3
10k

.,

"s

lOOk

220k

2.2M

C2
160pF

I-= 0.06~

"6
220k

C•

A8

C7

220k

10pF

RlO

A12

22k

51Dk

-=

C'3
0.01
MIXING AMPLIFIER

RECOVERY AMPLIFIER

DRIVER AMPLIFIER

FIGURE 5.3.1 Stereo Reverb System

The +6dB/octave response is achieved by proper selection
of R 1, R2 and Cl as follows:

5.3.2 Stereo Reverb System
A complete stereo reverb system is shown in Figure 5.3.1,
with its idealized "straightline" frequency response appearing as Figure 5.3.2.

1

fl =

'" 100 Hz (as shown)

(5.3.2)

21T(Rl + R2)Cl
The LM377 dual power amplifier is used as the spring
driver because of its ability to deliver large currents into
inductive loads. Some reverb assemblies have input transducer impedance as low as 8Q and require drive currents of
- 30 mAo (There is a preference among certain users of
reverbs to drive the inputs with as much as several hundred
milliamps.) The recovery amplifier is easily done by using
the LM387 low noise dual preamplifier which gives better
than 75dB signal-to-noise performance at 1 kHz (10mV
recovered signal). Mixing of the delayed signal with the
original is done with another LM387 used in an inverting
summing configuration.

f2 = _ _1___ '" 10kHz (as shown)
21T R2Cl

"10
iii

.:s.

+20

.-

'"~

Figure 5.3.2 shows the desired frequency shaping for the
driver and recovery amplifiers. The overall low frequency
response is set by fa and occurs when the reactance of the
coupling capacitors equals the input impedance of the next
stage. For example, the driver stage low frequency corner fo
is fou nd from Equation (5.3.1).
1
fo = - - - '" 80Hz (as shown)
21T R4C3

(5.3.3)

-20

l-++HIitI-+tttltHrttI!fltt--+tffitttI
10

100

Ik

10k

lOOk

FREQUENCY (Hd

(5.3.1)
FIGURE 5.3.2 Straightline Frequency Response of Reverb Driver
and Recovery Amplifiers

5-8

sum of the original signal and the delayed signal. Scaling
factors are adjusted per Equation (5.3.10).

Ultimate gain is given by the ratio of R2 and R1:
R2
Ao = 1 + (gain beyond f2 corner)
R1

(5.3.4)
-VOUT

High frequency roll off is accomplished with R3 and C2,
beginning at f2 and stopping at f3.
211 R1 C2

Vs = original signal

As shown, the output is the sum of approximately one half
of the original signal and all of the delayed signal.

f3 = __1_~ '" 100kHz (as shown)
211 R3C2

(5.3.6)

Stopping high frequency rolloff at f3 is necessary so the
gain of the amplifier does not drop lower than 20dB,
thereby preserving stability. (LM377 is not unity gain
stable.) Resistors R5 and R6 are selected to bias the output
of the LM3S7 at half-supply. (See Section 2.S.) Low
frequency corner f1 is fixed by R7 and CS:
fl = __1___ '" 100Hz (as shown)
211 R7 Cs

5.3.3 Stereo Reverb Enhancement System
The system shown in Figure 5.3.3 can be used to synthesize
a stereo effect from a monaural source such as AM radio or
FM-mono broadcast, or it can be added to an existing
stereo (or quad) system where it produces an exciting
"opening up" spacial effect that is truly impressive.
The driver and recovery sections are as in Figure 5.3.1 with
the exception that only one spring assembly is required.
The second half of the LM387 recovery amplifier is used as
an inverter and a new LM387 is added to mix both channels
together. The outputs are inverted, scaled sums of the
original and delayed signals such that the left output is
composed of LEFT minus DELAY and the right output is
composed of RIGHT plus DELAY.

(5.3.7)

High frequency rolloff is done similar to the LM377 by
RS and Cr
f4 = ___
1__
211R5C7

(5.3.10)

VD = delayed signal

(5.3.5)

'" 10kHz (as shown)

f2 =

where:

R9
R9
Vs +
VD
R12
Rll

7 kHz (as shown)

(5.3.S)

f5 = _ _
1 __ "" 70kHz (as shown)
211 RS C7

When applied to mono source material, both inputs are tied
together and the two outputs become INPUT minus D E LAY
and INPUT plus DELAY, respectively. If the outputs are to
be used to drive speakers directly (as in an automotive
application, or small home systems), then the LM387 may
be replaced by one of the LM377 /378/379 dual 2W/4 W/6W
amplifier family wired as an inverting power summer per
Figure 5.3.4.

(5.3.9)

The same stability requirements hold for the LM3S7 as for
the LM377.
Resistors R9 and RlO are used to bias the LM387 summing
amplifier. The output of the summer will be the scaled
LEFTIN

0.01

510k

+t--o

220k

LEFT
OUT'"
-(LEFT - DelAY)

DRiVER

INVERTER

RECOVERY

+24V

22Dk

MIXERS

0::11:..,0.015

~IOOk

22M
22k

t---'\A'V-it-'
220k

D.068~-=-

10pF

22Dk

220k

220k

0.01

"

S10k

+~
22k

RIGHT IN

FIGURE 5.3.3 Stereo Reverb Enhancement System

5-9

RIGHT
OUT ..
- (RIGHT + DELAY)

LEFT IN

lOOk

0-/
0.047

56k

vCC~~
0.1
- DELAY IN

r-

Vs - - ,

-

I

0-/
0.22

10k

I
LEFT

SPEAKER
4.1] OR Sn

lOOk

RIGHT
SPEAKER

4n OR 8n

+

DELAY IN

0-/
0.22

RIGHT IN

10k

0-/
0.047

56k

lOOk

FIGURE 5.3.4 Alternate Output Stage for Driving Speakers Directly Using lM377/378/379 Family of Power Amplifiers

REFERENCES

magnitude and a varying phase shift of 0-180° as a function
of the resistance between the positive input and ground.
Each stage shifts 90° at the frequency given by 1/(21T R C),
where C is the positive input capacitor and R is the resistance to ground. Six phase shift stages are used, each
spaced one octave apart, distributed about the center of the
audio spectrum (160Hz-3.2kHz). JFETs are used to shift
the frequency at which there is 90° delay by using them as
voltage adjustable resistors. As shown, the resistance varies
from lOOn (FET full ON) to 10kn (FET full OFF), allowing a wide variation of frequency shift (relative to the 90°
phase shift point)_ The gate voltage is adjusted from 5 V to
8V (optimum for the AM9709CN), either manually (via
foot operated rheostat) or automatically by the LM741
triangle wave generator. Rate is adjustable from as slow as
0.05 Hz to a maximum of 5 Hz. The output of the phase
shift stages is proportionally summed back with the input
in the output summing stage.

1. "Application of Accutronic's Reverberation Devices,"
Technical paper available from Accutronics, Geneva, III.

2. "What Is Reverberation?," Technical paper available
from Accutronics, Geneva, III.

5.4 PHASE SHIFTER
A popular musical instrument special effect Circuit called a
"phase shifter" can be designed with minimum parts by
using two quad op amps, two quad JFET devices and one
LM741 op amp (Figure 5.4.1). The sound effect produced is
similar to a rotating speaker, or Doppler phase shift
characteristic, giving a whirling, ethereal, "inside out" type
of sound. The method used by recording studios is called
"flanging," where two tape recorders playing the same
material are summed together while varying the speed of
one by pressing on the tape reel "flange." The time delay
introduceq will cause some signals to be summed out of
phase and cancellation will occur. This phase cancellation
produces the special effect and when viewed in the
frequency domain is akin to a comb filter with variable
rejection frequencies.' The phase shift stage used (Figure
5.4.1) is a standard configuration" displaying constant

REFERENCES
1. Bartlett, B., "A Scientific Explanation of Phasing
(Flanging)," Jour. Aud. Eng_ Soc_, vol. 18, no. 6,
Decel1']ber 1970, pp. 674-675.
2. Graeme, J. G., Applications of Operational Amplifiers,
McGraw-Hili, New York, 1973, pp. 102-104.
5-10

INPUT BUFFER

OUTPUT SUMMER

BYPASS
"--"'---~............o-- VOUT

36k

PHASE SHIFTERS

20k

20k

2Dk

20k

20k

2Dk

2Dk

20k

2Dk

20k

2Dk

2Dk

FIGURE 5.4.1 Phase Shifter

5.5 FUZZ

diodes limit the output swing to ±O.7V by clipping the
output waveform. The resultant square wave contains pre·
dominantly odd·ordered harmonics and sounds similar to a
clarinet. The level at which clipping begins is controlled by
the Fuzz Depth pot while the output level is determined by
Fuzz Intensity.

Two diodes in the feedback of a LM324 create the musical
instrument effect known as "fuzz" (Figure 5.5.1). The
FUZZ

DEPTH
10k

lOOk
lN914

5.6 TREMOLO
lN914

Tremolo is amplitude modulation of the incoming signal
by a low frequency oscillator. A phase shift oscillator
(Figure 5.6.1) using the LM324 operates at an adjustable
rate (5·10Hz) set by the SPEED pot. A portion of the
oscillator output is taken from the DEPTH pot and used to
modulate the "ON" resistance of two 1 N914 diodes
operating as voltage controlled attenuators. Care must be
taken to restrict the incoming signal level to less than
O.6V p. p or undesirable clipping will occur. (For signals
greater than 25mV, THO will be high but is usually
acceptable. Applications requiring low THO require the use
of a light detecting resistor (LOR) or a voltage·controlled
gain block. See Figure 4.8.9.)

>---tFUZZ
INTENSITY

;;;--If-o Vo UT

FIGURE 5.5.1 Fuzz Circuit

5·11

·t
0.33

5.1k

...--....ovs

33

Vs

0.33

SPEED

Z5k
15·10Hz)

51k

TREMOLO
FOOT SW.

lOOk

1

ZZOk

DEPTH

lOOk

+VS/Z
10k

AUOIO IN

~~;oVJi"C~~P~INGI

1k

-tE--o MODULATED

~t----"'-I4t--'N:4I9~'4':'"s-l""-...
lOp

lk

1k

~~~IO

0.1

FIGURE 5.6.1 Tremolo Circuit

an active two-band tone control block, the complete
circuit is done ,with only one 8-pin IC and requires very
little space, allowing custom built-in d~signs where desired.

5.7 ACOUSTIC PICKUP PREAMP
Contact pickups designed for detection of vibrations produced by acoustic stringed musical instruments (e.g., guitar,
violin, dulcimer, etc.) require preamplification for optimum
performance. Figure 5.7.1 shows the LM387 configured as
an acoustic pickup preamp, with Bass/Treble tone control,
volume control, and switchable ±10dB gain select. The
pickup used is the Ibanez "Bug," which is a flat response
piezo-ceramic contact unit that is easy to use, inexpensive,
and has excellent tone response. By using one half of the
LM387 as the controllable gain stage and the other half as

The tone control circuit is as described in Section 2.14.8.
Addition of the midrange tone control (Section 2.14.9) is
possible, making tone modification even more flexible.
Switchable gain control of ±10dB is achieved using a DPDT
switch to add appropriate parallelling resistors around the
main gain setting resistors R8 and R6. Resistor Rg is
capacitively coupled (C14) so as not to disturb DC
conditions set up by R8 and R1O.

O+10dB

C,
0.05

.,
lOOk

A,

A,

C,

11k

11k

15j.lF

Od8()-o,.:c~---,"",W--"

+

A5

C,

A14
620k

11k

C11

+
15/.1F

I'

C,

A,
INPUT FROM IBANEZ
"BUG" PIElQ·CERAMIC
CONTACT PICKUP

3.6k

SOOk
TREBLE

~

005

.,
l.6k

"-availablefromElgerCo"
P.O. Box 469, CornweUs Hts., PA 19020

FIGURE 5.7.1 Acoustic Pickup Preamp

5-12

-=

-=

·'1p··

._-----------

_-

--,_.
--------_
--.- .------

-.-----.
- -- ....- --....
- -..

::

-- -~.;;;-_=;: i; ,.- ~. ,~"'_:~ :~ =h~-~-§-~;-;-t-~ - _:-· ~-:- · ·~-·~- ~-: -o-·-._---

-

.. ,

...

-

+ •• _ - - ' . - , •• _ - , " - , - - , . -

..

..

..

. _ - - . _ . _ - _ •• _ - - -

- - - - _ ••• _ - -

-.---_---_._._--. ------------.--.--------.-.
. .-- .-"
--_.
.-_.----.

------,-----~
..

--

.

-----.-~--

---

-------~0· _
_
__
_ _ _ •I
_• -

.

..

.. _ _...__
----

-

---__
_

eo·
----- --_------ --.

_ _ _ .. - - - - _ . _ _ -

.-

.. -_ ...__ ._--------.. ---- - -----_.
---------------------:--+--•

__-i=:-=7r:::cl~:';- !-!:-~
1-..-..-"t

----'.---.-.-."
.....
------_._-.. -------- .. ........ --.--.
---_._--------... -----."
.
.

.

i=.-i--=--- -1.......

-------.
---.----.----------

,'

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ .___ . __ --. _____. _ ..

.......
._-----------

.

..

••• - - - - - -

o

-. ...----------

--- .
.
-------.
-.----.. ---

:.

----- --- -- --

--

.-..'.

_.-_

:.

_._--------.-,.---------_._.
.-. --------_._-------_.
_._----._-----------------._---------_
_-------------_._-------_
_..--_ ------_._--------.- -...... --- -- -._------- -- -.------ .-.
-

i i

.
..

a

...

~-

·· .. . ..

6.0 Appendices
A1.0 POWER SUPPLY DESIGN

Figure A 1.1.) Therefore, V I N and liN become the governing
conditions, where:

A1.1 Introduction

output current plus regulator quiescent cu rrent

One of the nebulous areas of power IC data sheets involves
the interpretation of "absolute maximum ratings" as
opposed to "operating conditions." The fact that parameters are specified at an operating voltage quite a few
volts below the absolute maximum is not nearly so important in "garden variety" op amps as in power amps - because a key spec of any power ampl ifier is how much
power it can deliver, a spec that is a strong function of the
supply. Indeed Po is approximately proportional to the
square of supply voltage. Since many audio ICs are powered
from a step down transformer off the 120V AC line, the
"absolute maximum voltage" is an attempt to spec the
highest value the supply might ever reach under power
company overvoltages, transformer tolerances, etc. This
spec says the IC will not die if taken to its "absolute
maximum rating." Operating voltage, on the other hand,
should be approximately what a nominal supply will sag
under load at normal power company voltages. Some audio
amplifiers· are improperly specified at their "absolute
maximum voltages" in order to give the illusion of large
output power capability. However, since few customers
regulate the supply voltage in their applications of audio
ICs, this sort of "specsmanship" can only be termed
deceptive.

IIN(MAX) "" IO(MAX),full-load operating current
IIN(MIN)";; 10,

maximum permissible instantaneous
no-load filter output voltage equal
to peak value of transformer secondary voltage at highest design line
voltage VPRI; limited by absolute
maximum regulator input voltage
nominal DC voltage input to the
regulator, usually 2 to 15V higher
than Vo
VIN(MIN) "" Vo + 2V, minimum instantaneous full-load
filter output voltage including ripple
voltage; limited by minimum regulator input voltage to insure satisfactory regulation (Va + Vdropout)
or minimum regulator input voltage
to allow regulator start-up under
full load or upon removal of a load
short circuit

A 1.2 General

RMS ripple factor at filter output
expressed as a percentage of VIN;
limited by maximum permissible
ripple at load as modified by the
ripple rejection characteristics of
the regulator

rf

This section presents supply and filter design methods and
aids for half-wave, full-wave center tap, and bridge rectifier
power supplies. The treatment is sufficiently detailed to
allow even those unfamiliar with power supply design to
specify filters, rectifier diodes and transformers for singlephase supplies. A general treatment referring to Figure A 1.1
is given, followed'by a design example. No attempt is made
to cover multi phase circuits or voltage multipliers. For
maximum applicability a regulator is included, but may be
omitted where required.

A1.4 Filter Selection, Capacitor or Inductor-Input
For power supplies using voltage regulators, the filter will
most often use capacitor input; therefore, emphasis will be
placed upon that type of filter in following discussions.
Notable differences between the two types of filters are
that the capacitor input filter exhibits:

A1.3 Load Requirements
The voltage, current, and ripple requirements of the load
must be fully described prior to filter and supply design.
Actually, so far as the filter and supply are concerned, the
load requirements are those at the regulator input. (See

1. Higher DC output voltage
2. Poorer output voltage regulation with load variation
3. Higher peak to average diode forward currents

FUll-WAVE

voc

rvY'Y"\
HALF-WAVE

VMSIN~V
\IRMS

no-load or minimum operating current; could be near zero

vocAA
loci

liN

10

(IF REaUIRED)

FIGURE A1.1 Power Supply Block Diagram, General Case

6-1

TABLE A1.1 Summary of Significant Rectifier Circuit Characteristics, Single Phase Circuits
Capacitive Data is for wC RL = 100 & RslRL = 2% (higher valuesl
and for wC RL = 10 & RS/RL = 10% (lowe, valuesl
Single Phase
Full Wave
Center Tap

Single Phase
Half Wave

Single Phase
Full Wave
Bridge

~

[DLOAD BlDLOAD

Rectifier Circuit Connection

Voltage Waveshape to Load
of Filter

( \ (\

rvYY\

rvYY\

CHARACTERISTIC LOAD

R

L

C

R

L

C

R

L

C

Average Diode Current
IF(AVGI/IO(OCI

1

1

1

0.5

0.5

0.5

0.5

0.5

0.5

Peak Diode Current
IFM/IF(AVGI

3.14

-

8
5.2

3.14

2

10
6.2

3.14

2

10
6.2

Diode Current Form
Factor, F = IF(RMSI/IF(AVGI

1.57

-

2.7
2

1.57

1.41

3
2.2

1.57

1.41

3
2.2

RMS Diode Current
IF(RMSI/IO(OCI

1.57

-

2.7
2

0.785

0.707

1.35
1.1

0.785

0.707

1.35
1.1

RMS Input Voltage per
Transformer Leg
VSEclV IN(OCI

2.22

2.22

0.707

1.11

1.11

0.707

1.11

1.11

0.707

Transformer Primary
VA Rating VA/Poc

3.49

-

-

1.23

1.11

-

1.23

1.11

-

Transformer Secondary
VA Rating VA/Poc

3.49

-

-

1.75

1.57

-

1.23

1.11

-

Total RMS Ripple %

121

-

-

48.2

-

-

48.2

-

-

Rectification Ratio
(Conversion Efficiency) %

40.6

-

-

81.2

100

-

81.2

100

-

Transformer

i Rectifier I

.
=:J11-n .
I
I
I

:
I
I

I
I
I
I

I
I
I
I
I

Filter

I

,I
--

I

C1

II

~-

I

+

Cz

l,

R,

1>--<>JT
'I

¢

"1 1

-- - - VmSINwt

R

VI"

~

tr~
\,,/

1/\

"""

(c) Voltage Across Input Capacitor C1

FIGURE Al.2

1

(bl Equivalent Circuit

(a) Actual Circuit

..

r

f\

(d) Current Through Diodes

Actual and Equivalent Circuits of Capacitor-Input Rectifier System, Together with Oscillograms of Voltage and Current for a
Typical Operating Condition

6·2

Normal load Resistance

Very large

low load
Resistance

Capacitance

(b) Equivalent Circuit

(a) Actual Circuit

/lOW Im~:dance

,

~

0>

~

"0

>

I

I

I

<

\

I
\

I

,

~Normal

\ Impedance

'- I

Increased
leakage
Inductance

(dl Current Through Diodes

Ie) Voltage Across Input Capacitor C1

FIGURE A1.3 Effects of Circuit Constants and Operating Conditions on Behavior of Rectifier Operated with Capacitor-Input Filter

Figures A 1.4 and A 1.5 show the relationship between peak
AC input voltage and DC output voltage as a relation to
load resistance R L, series circuit resistance RS, and filter
input capacitance C. Figure A 1.4 is for half·wave rectifiers
and Figure A 1.5 is for full·wave rectifiers. Note that the
horizontal axis is labeled in units of w C R L where:

4. Lower diode PIV rating requirements
5. Very high diode surge current at turn-on
6. Higher peak to average transformer currents
The voltage regulator overcomes disadvantage (2) while
semiconductor diodes of moderate price meet most of the
peak and surge requirements except in supplies handling
many amperes. Still, it may be necessary to balance increased diode and transformer cost against the alternative
of a choke-input filter. In power supply designs employing
voltage regulators, it is assumed that only moderate filter
output regulation and ripple are required. Therefore, a
capacitor input filter would exhibit peak currents considerably lower than indicated in the comparison of Table A 1.1.

w = AC line frequency in Hertz x 211
C = value of input capacitor in Farads
RL = VINIIIN '" Vallo, equivalent load resistance in
Ohms
RS

A1.5 Filter Design, Capacitor-Input
Figure A 1.2 shows a full-wave, capacitor·input (filter)
rectifier system with typical voltage and current waveforms.
Note that ripple is inevitable as the capacitor discharges
approximately linearly between voltage peaks. Figure A 1.3
shows the effects on DC voltage, ripple, and peak diode
current under varying conditions of load resistance, input
capacitance, series diode and transformer resistance RS, and
transformer leakage inductance. The most practical design
procedure for capacitor-input filters is to use the graphs of
Figures A l.4-A 1.7. Note, however, that these include the
effects of diode dynamic resistance within RS. Diode
forward drop is not included, and must be subtracted from
the transformer secondary voltage. A good rule of thumb is
to subtract 0.7 V from the transformer voltage and assume
diode dynamic resistance is insignificant (0.02£1 at IF = 1 A,
0.26£1 at IF = 100mA); ordinarily the transformer resistance will overshadow diode dynamic resistance.

=

total of diode dynamic resistance, transformer
secondary resistance, reflected transformer primary
resistance, and any added series surge limiting
resistance

The major design trade-off encountered in designing
capacitor-input filters is that between achieving good
voltage regulation with low ripple and achieving low cost.
Referring to Figures A 1.4 and A 1.5:
1. Good regulation means wCRL '" 10.
2. Low ripple may mean wC RL

> 40.

3. High efficiency means RS/R L < 0.02.
4. Low cost usually means low surge currents and small C.
5. Good transformer utilization means low VA ratings, best
with full-wave bridge FWB circuit, followed by full-wave
center tap FWCT circuit.
In most cases, a minimum capacitance accomplishing a
reasonable full-load to no-load regulation is preferable for
low cost. To achieve this, use an intercept with the upper
6·3

In~

100

f

90

V

••1'

Rs

~CI£}l

~ t..-

80

0.5
1.0

~

2.0
IFirst .AI prox.
4.0

~-

70

~

60

-

:n
8.0
10
12.5
15

r--

20
VC(OC} (%)

50

25
30
35
40
50
60
70
80
90
100

I'

Vm

'/",-

40

30

~ ~~
F=

20

10

~~1
V
VI
~
r:-V
V

0
0

0

0

1000

WCRl (C in farads, Rl in ohms)

FIGURE A 1.4 Relation of Applied Alternating Peak Voltage to Direct Output Voltage in Half-Wave Capacitor-Input Circuits

(From O. H. Schade, Proc. IRE, vol. 31, p. 356, 1943.)

100

r:P

Vm Rs

Jt;jLBRL

90

80

70
VC(OC}

---v;- (%)

~~

60

50

40

30
0.1

~~
~~
~~

d~

~F':::t ApproX.

f..-

005
0 .1
0 .5
1. o
2. o

o

--

o

o
1o

I-'"

1 2.5
15

""""

~ I1v

2o

25

:/: V'"
:::t ....... '"

3o
35
4o

/,'"

rV ....-

---

5o
6o
7o

:.--

Bo

::::::>-->-1.0

9o
1 00
10

100

1,000

WCRl (C in farads, Rl in ohms)

FIGURE A1.5 Relation of Applied Alternating Peak Voltage to Direct Output Voltage in Full-Wave

(From O. H. Schade, Proc. IRE, vol. 31, p. 356,1943.)

6-4

Capacitor~lnput

Circuits

3. Determine diode surge current requirement at turn-on of
a fully discharged supply when connected at the peak of
the highest expected AC line waveform. Surge current is:

knee of the curves in Figures A 1.4 and A 1.5. Occasionally, a
minimum value filter capacitor will not result in a lower
cost system. For example, increasing the value of C may
allow higher RS/RL to result in lower surge and RMS
currents, thus allowing lower cost transformers and diodes.
Be sure that capacitors used have adequate ripple current
ratings.

EM
ISURGE = - - RS + ESR
where ESR = effective series resistance of capacitor.

Design procedure is as follows:

4. Find required diode PIV rating from Figure Al.S.
Actually, required PIV may be considerably more than
the value thus obtained due to noise spikes on the line.
See Section A 1.9 for details on transient protection.
Remember that the PIV for the diodes in the FWB
configuration are one half that of diodes as found in
FWCT or HW rectifier circuits.

1. Assuming that Va, 10, w, and load ripple factor rf have
been established and an appropriate voltage regulator
has been selected, we know or can determine:
W

= 27Tf

= 377rad/sec for 60Hz line

rf(in) = rf(out) x ripple reduction factor of selected
regulator

The diodes may now be selected from diode manufacturers'
data sheets. If calculated surge current rating or peak
current ratings are impractically high, return to Step A 1.5(2)
and choose a higher RS/RL or lower C. Conversely, it may
be practical to choose lower RS/R L or higher C if diode
current ratings can be practically increased without adverse
effect on transformer cost; the result will be higher supply
efficiency.

VIN(PK) < Max VIN for the selected regulator; allow for
highest line voltage likely to be encountered
IIN(MIN) '" Va + 2V; allow for lowest line voltage
VIN(DC)+ = VIN, usually 2·15V above Va; if chosen
midway between VIN(PK) and VIN(MIN) or slightly
below that point, will allow for greatest ripple voltage
liN'" 10 for full load
IIN(MIN)

=

IQ for open load

A 1. 7 Transformer Specification

RL = VIN(DC)/IIN

A decision may have been made at Step A 1.5(2) as to using
half-wave or full-wave rectification. The half-wave circuit is
often all that is required for low current regulated supplies;
it is rarely used at currents over 1 A, as large capacitors
and/or high surge currents are dictated. Transformer
utilization is also quite low, meaning that higher VA rating
is required of the transformer in HW circuits than in FW
circuits. (See VA ratings of Table A 1.1.)

RL(MIN) = VIN(MIN)/IIN

< VIN(PK) and calculate VIN(DC)IVIN. Enter
the graph of Figure A 1.4 or A 1.5 at the calculated
VIN(DC)IVM to intercept one of the RS/RL = constant
lines. Either estimate RS at this time or intercept the
curve marked "First Approximation."

2. Set VM

3. Drop vertically from the intercept of Step (2) to the
horizontal axis and read wCRL. Calculate C, allowing
for usual commercial tolerance on capacitors of +100,
-50%.
If VIN(DC) is midway between VIN(PK) and VIN(MIN),
the supply can present maximum ripple to the regulator.
A low value of C is then practical. If VIN(DC) is near
VIN(MIN), regulator power dissipation is low and supply
efficiency is high; however, ripple must be low, requiring
large C.

Half-wave circuits are characterized by low VIN(DC)IVM
ratio, or very large C required (about 4 times that required
for FW circuits, high ripple, high peak to average diode and
transformer current ratios, and poor transformer utilization).
They do, however, require only one diode.

Full-wave circuits are characterized by high VIN(DC)IVM
ratio, low C value required, low ripple, low peak to average
diode and transformer current ratios, and good transformer
utilization. They do require two diodes in the center-tap
version, while the bridge configuration with its very high
transformer utilization requires four diodes.

4. Determine ripple factor rf from Figure A 1.6. Make
certain that the ripple voltage does not drop instan·
taneous VIN below VIN(MIN)·

The information necessary to specify the transformer is:
1. Half-wave, full-wave CT or full-wave bridge circuit

The ripple factor could determine minimum required C
if ripple is the limiting factor instead of voltage
regulation. Again, allow for -50% tolerance on the
capacitor.

2. Secondary VRMS per transformer leg, (VM + 0.7*)/Vi,
from Section A1.5
3. Total equivalent secondary resistance including reflected
primary resistance from Section A 1.5

rf
Vripple(pk)

=

Vi VIN(DC)
100

4. Peak, average, and RMS diode or winding currents from
Sections A 1.6(1) and -(2), and VA ratings.

A 1.6 Diode Specification

*1.4 for full-wave bridge circuit.

Find diode requirements as follows:
1. IF(AVG)

=

IIN(DC) for half·wave rectification

= IIN(DC)/2 for full·wave rectification

2. Determine peak diode current ratio from Figure A1.7;
remember to' allow for highest operating line voltage and
+100% capacitor tolerance.
IFM = IFM/IF(AVG) x IIN(DC) for half-wave
=

IFM/IF(AVG) x IIN(DC)/2 for full-wave
6-5

100
7D
50
3D

...

Half-Wave

~

~~

-..;;:

g

Rs/RL (%)

10
CC 7.0
~ 5.0
Co)

~~

r'
A

''''1Il10.

~~

20

Parameter

Circuit

A

~

Full-Wave

-'10
:: 10
-- 30

~

__ 0.1
--1.0
--10
--30

A

A

~ 3.0
~ 2.0

,'~

~
~

" ....... ~
,

t

~

i

~

t'....-: ~

t'~

....... l '

ii: 1.0
i' 0.7

i

I'

:=:

~

0.5

,~

...... ~

...........-.......: ~ [ , ~~
...... ~ ~

0.3
0.2
0.1
1.0

2.0 3.0

20

5.0 7.0 10

3D

50 70 100

""'"

::::

1'...:: ~ "........ I""t':

200 300

SOD

1,000

wCRl (C IN FARADS, RllN OHMS)
FIGURE A 1.6 Root-Mean-Square Ripple Voltage for Capacitor-Input Circuits
(From O. H. Schade, Proc. IRE, vol. 31, p. 356,1943.)

0.02
0.05
0.1
0.2 a::.....
0.5
1
.2 a::
'"
0.5 ;,<:
10
30
100

8. ..::

I.L.

1~~~-7~~~~~~~~7-~~~~~~~~~~~.
1.0
2.0 3.0 5.0 7.0 10
20 30
50 70 100
200 300
1,000
nwCRl

40
.... 30
Q
Q

C 20

~

CC

....c..

I-"
~ l=-

~ f--

:;10
ct

u::

..==:
F-_

7

~

.... ~

~ 5

3
1.0

--

2.0 3.0

Ie'

5.0 7.0 10

FIGURE A1.7 Relation of RMS and

20 30
50 70 100
nwCRl

Peak~to-Average

Diode Current in Capacitor-Input Circuits

(From O. H. Schade, Proc. IRE, vol. 31, p. 356,1943.)

6-6

200 300

0.02
0.05
0.1
0.2
0.5
1.0
2.0
5.0
10
30
100
500 700 1,000

.....

a::

..::

a::
'"

;,<:

A1.9 Transient Protection

Transformer VA rating and secondary current ratings are
determined as follows:

HW

FWCT

FWB

IRMS(SEC)

IIN(DC) F/..j'i

IIN(DC) F/2

IIN(DC) F

VASEC

VRMS IRMS

2vRMS IRMS

VRMSIRMS

VAPRI

VASEC

VASEC/-.j2

VASEC

where:

Often the PIV rating of the rectifier diodes must be
considerably greater than the minimum value determined
from Figure A 1.8. This is due to the likely presence of highvoltage transients on the line. These transients may be as
high as 400V on a 115 V line. The transients are a result of
switching inductive loads on the power line. Such loads
could be motors, transformers, or could even be caused by
SCR lamp dimmers or switching·type voltage regulators, or
the reverse recovery transients in rectifying diodes. As the
transients appearing on the transformer primary are coupled
to the secondary, the rectifier diodes may see rather high
peak voltages. A simple method of protecting against these
transients is to use diodes with very high PIV. However,
high·current diodes with very high PIV ratings can be
expensive.
There are several alternate methods of protecting the rec·
tifier diodes. All rely on the existence of some line
impedance, primary transformer resistance or secondary
circuit resistance. See Figure A 1.10 for the system circuit.

F = IR(RMS)/IIN(DC)
= form factor from Figure AI. 7

VRMS = secondary RMS voltage per leg

A1.8 Additional Filter Sections

The several methods of transient protection rely on shunting
the transient around the rectifier diodes to dissipate the
transient energy in the series circuit resistance and the
protective device. The usual protection methods are:

Occasionally, it is desirable to add an additional filter to
reduce ripple. When this is done, an LC filter section is
cascaded with the single C section filter already designed.
If the inductor is of low resistance, the effect on output
voltage is small. The additional ripple reduction may be
determined from Figure A1.9.

1. Series resistor at the primary with shunt capacitor across
the primary winding - see Figure A1.10

2.0

1.8

~10J
r--r-. r..... r..

WCRl·\O

1.6
E

~

>
;;:

r....

1.4

~

1.2

I""-r-.,
r--

1.0
0.01

0.02

0.1

0.07

0.04

0.2

0.7

0.4

1.0

Rs/Rl

FIGURE A1.S Ratio of Operating Paak Invene Voltage to Peak Applied AC for Rectifiers Used in Capacitor-Input. Single-Pha.e. Filter Circuits

.
....
~

o. 5

6.0

'\

~

~~

O. 2

>"

0->
:>0~:>

1!

~

r\

'\
'\

o. 1

'\.

'

"

0:

~

20

26

:-'\: ~~

1\"-:, 1\

,,~

0.02

"1\.

0:

ffi

..~~

{oil'

~ 0.0 5

:::z
"

14

I\.

'\.

0.01

..
~
0:

"G
:::
15

~
0:

34

1'\

'"

!:;

ffi

!:;
;;:

40

;;:

46

0.005

1.0

0.2

5.0 7.0 10

20

30

50 70

100

200

~OO

l (Henries) X C<¢ds)

FIGURE A1.9 Reduction in Ripple Voltage Produced by a Single Section Inductance-eapecitance Filter at Various Ripple Frequencies

6-7

Figure A 1.11 b is a half-wave doubler circuit wherein C2 is
partially charged on one half cycle and then on the second
half cycle the input voltage is added to provide a doubling
effect. Cl is normally considerably larger than C2. The
advantage of the half-wave circuit is that there is a common
input and output terminal; disadvantages are high ripple,
low 10 capability, and low VOUT.

=r111rnJ]
FIGURE A1.10 Transformer/Filter Circuit Showing Placement of
Transient Protection Components

RS1
C1

RS1

ic

°1

2. Series inductance at the primary, possibly with a shunt
capacitor across the primary - see Figure A 1.10
3. Shunt capacitor on the secondary - see Figure A 1.10

°2
VM sin wt

4. Capacitor shunt on the rectifier di ode - transient power
is thus dissipated in circuit series resistance.

1

f--

0----

C2
RS2

~
~

RS2
C2

!

2VM

j

(a) Conventional FuliMWave Voltage Doubling Circuit

5. Surge suppression varactor shunt on the rectifier diode this scheme is quite effective, but costly.

°1
RS
C1
~r-C2
VM sin wt

(b) Cascade (Half·Wave) Voltage Doubling Circuit

6. Dynamic clipper shunt on the rectifier diode - the
clipper consists of an R, a C and a diode.

s

FIGURE Al.ll Voltage Doubler Circuits

These rectifying circuits, being capacitively loaded, exhibit
high peak currents when energy is transferred to the
capacitors. Filter design for the doubler circuits is similar to
that of the conventional capacitor filter circuits. Figures
Al.12, Al.13 and A1.14 provide the necessary design aids
for full-wave voltage doubler circuits. They are used in the
same way as Figures A 1.5, A 1.6 and A 1. 7.

7. Zener shunt on the rectifier diode - may also include a
series resistance.

A 1.11 Design Examp)e

B. Shunt varistor (e.g., GE MOVs) on the secondary - see
Figure ALl O.

Design a 5V, 3A regulated supply using an LM123K.
Determine the filter values and transformer and diode
specifications. Ripple should be less than 7 mVRMS. Assume
60dB ripple reduction from typical curves.

Of the several protective circuits:
•

(1), (2), (3) and (4) are least costly, but are limited in
their utility to incomplete protection.

•

(4) is probably the circuit providing the most protection
for the money and is all that may be required in lowcurrent regulated supplies.

•

1. Establish operating conditions:

w

~

377 rad/sec

VIN(PK) ~ lBV and 10% high line voltage; this allows
some 2V headroom before reaching the 20V absolute
maximum VIN rating of the LM123K

(5), (6). (7) and (B) are most costly, but provide greatest
protection. Their use is most worthwhile on high current
supplies where high PIV ratings on high·current diodes is
costly, or where very high transient voltages are en·
countered.

VIN(MIN) ~ 7.5 V at 10% low line voltage including
effects of ripple voltage
VIN(DC) ~ 11 V at nominal line voltage; chosen to
exceed VIN(MIN) + peak ripple voltage

Al.l0 Voltage Doublers

Vripple(out)';; 7mVRMS

Occasionally, a voltage doubler is required to increase the
voltage output from an existing transformer. Although the
doubler circuits will provide increased output voltage, this
is accomplished at the expense of an increased component
count. Specifically, two filter capacitors are required. There
are two basic types of doubler circuits as indicated in
Figure A 1.11. Figure A 1.11 a is the conventional full·wave
doubler circuit wherein two capacitors connected in series
are charged on alternate half cycles of the line waveform.

Vripple(in)';; 7VRMS
rf(in) .;; 7 V/11 V ~ 63.5%
liN

~

3A

IIN(MIN) ~ IQ ~ 20mA
RL

~

11 V/3A

~

3.67£1

RL(MIN) = 7.5V/3A ~ 2.5£1

6-B

2. 0

V- I-I--

0.1

0.25
1. 8

1.,.-/.",

1.6

"

0.5

~~

0.75
1 .0

.....

1 .5

J. ~

IE

1.4

2.0

/~

~~

~

U
Q
::;

4.0

RL

v/ :..--

>

1.2

~

~

10

~~~

I.~r:

0.6
1.0

6.0

~

,.--/

1
1

-

A~

0.8

RS

-%

1

2

Jv
2.0

4.0 6.0

10

20

40

60

100

200

400

weRl (e in Farads. RL In Ohms)

FIGURE A 1.12 Output Voltage as a Function of Filter Constants for Full·Wave Voltage Doubler
for Full-Wave Voltage Doubler

6. Diode specifications are:

2. Set:
VM = 16.3V nominal, which is 18V -10% line variation
VIN(DC)iVM

IIN(DC)
IF(AVG) = - - - = 1.5A for FW rectifiers

= 11/16.3 = 0.67

2

Assume full·wave bridge rectification because of the high
current load. Enter the graph of Figure A 1.5 at
VIN(DC)iVM = 0.67 to intercept the "First Approxi·

IFM = 8 x 1.5A = 12A, from figure A1.7, allowing
C = 100% high, for commercial tolerances

mation" curve.

ISURGE = 18V/0.48n = 37.5A, worst case with 10%
high line, neglecting capacitor ESR

3. Drop down to the horizontal axis to find wC RL = 3.33.

IF(RMS) = 2.1 x 1.5A = 3.15A, from Figure A1.7,
allowing for 100% high tolerance on C

Thus, RS/RL "'" 13%, or RS = O.4n is allowable.
C =

7. Transformer specifications are:

3.33
= 2400!.!F
3.67 x 377

V SEC(RMS) = 16.3+1.4 = 12.6 for FWB

.Ji

(4800!.!F allowing for -50% capacitor tolerance)
4. Ripple factor is 15% from Figure A1.6. Ripple is then
Vripple(pk) =

.Ji x 0.15 x 11

(24 VCT for FWCT)

= 2.33V pk.

RS = 0.48n including reflected primary resistance, but
subtract 2 x diode resistance

5. Checking for VIN(MIN),

IAVG = IIN(DC) = 3A

VM = 16.3Vor, allowing for 10% low line voltage, 14.8V
VIN(DC) = 14.8 x 0.67 = 9.91 V

IIN(DC) x F
3Ax 21
"'2
= - - _ . = 4.45A
v£
1.414

Subtracting peak ripple, VIN(MIN) = 9.91 - 2.33 = 7.6V
which is within specifications

ISEC(RMS) =

In fact, all requirements have been met.

VA rating = 4.45A x 12.6 = 56VA, or 62VA, allowing
for 10% high line.
6·9

100
50
20
10

~

'"t;
0

!j.0

10% -

RS/Al

~ 2.0

"'"

b.o· 1%

~

I'"

1.0
0.5
0.2
0.1
1.0

5.0

2.0

10

20

50

100

200

1000

500

weRl (C IN FARADS, AllN OHMS)

FIGURE A1.13 Ripple as a Function of Filter Constants for Full-Wave Voltage Doubler

10
8.0
6.0

f--

:; 4.0

;;;-

if

1

-

,,-,:;:'-:

0.01
0.025
0.05 ~
0.1

:c

0.25 ~
0.5

3.0

I"'"

1.0
2.5
5
15
50

;....2.0

1.0
0.2

0.4

1.0

2.0

4.0

10

20

40

100

200

400

1000 2000

weRl (C IN FARADS, RllN OHMS)

RMS Rectifier Current as a Function of Filter Constants for Full-Wave Voltage Doubler

100
60

-

40

.-;:::.

20

:::::.

'>

if

l

0.01
0.025
0.05
0.1
0.25 '#.
0,5 j;I
1.0 ~
2.5 ~

10
8.0
6.0

5.0
15
50

4.0
3.0
2.0
1.0
0.2

0.4

1.0

2.0

4.0

10

20

40

100

200

400

1000 2000

weRl (C IN FARADS, AL IN OHMS)

FIGURE A1.14 Relation of RMS to Peak and Average Diode Currents

6-10

A2.0 DECIBEL CONVERSION

A3.0 WYE·DELTA TRANSFORMATION

A2.1 Definitions

Wye·delta transformation techniques (and the converse,
delta·wye) are very powerful analytical tools for use in
understanding feedback networks. Known also as tee· pi and
pi·tee transformations, their equivalencies are given below.

The decibel (dB) is the unit for comparing relative levels of
sound waves or of voltage or power signals in amplifiers.
The number of dB by which two power outputs Pl and P2
(in Watts) may differ is expressed by:

A3.1 Wye·Delta (Tee· Pi)
Pl
1010gP2

Wye or Tee

Delta or Pi

or, in terms of volts:

lIZ
Z

El
20 log E2

(Figure A2.1)

IS
ELECTRICALLY
) EQUIVALENT

TO,

or, in current:
11
20 log12
While power ratios are independent of source and load
impedance values, voltage and current ratios in these
formulas hold true only when the source and load imped·
ances Zl and Z2 are equal. In circuits where these imped·
ances differ, voltage and current ratios are expressed by:

where:

Specific reference levels, i.e., the OdB point, are denoted by
a suffix letter following the abbreviation dB. Common
suffixes and their definitions follow:

Z1 Z2
Z12 = Z1+ Z2 + - Z3

(A3.1.1)

Z23

Z2 Z3
Z2+ Z3 + - Z1

(A3.1.2)

Z31

Z3 Z1
Z3+ Z1 + - Z2

(A3.1.3)

dBm - referenced to 1 mW of power in a 600n line
(OdBm = 0.775V)
A3.2 Delta·Wye (Pi·Tee)

dBV - referenced to 1 V (independent of impedance levels)
dBW - referenced to 1 W
10.000

I11111111I

'OZ

Wye or Tee

Delta or Pi

Zz

lIZ

1000 . . . .

~ 100

~-'~_fII
o

10

ZO

30 40

50

60

70

Z23 )

l31

~SLECTRICALL

EnUlvALENT

IY

TO,

3

BO

DECIBELS (dB)

FIGURE A2.1 Gain Ratio to Decibel Conversion Graph
(Note: for negative values of decibels. i.e., gain
attenuation. simply invert the ratio number. For
example, -20dB = 1110VN.1

where:
Z1 =

Z12 Z31

(A3.2.1)

Z12 + Z23 + Z31
A2.2 Relationship Between dB/Octave and dB/Decade
dB/Octave
3
6
9

10
12
15
18

Z12 Z23

dB/Decade

(A3.2.2)

Z12 + Z23 + Z31

10
20
30
33.3
40
50
60

Z31 Z23
Z3 = - - - - - Z12 + Z23 + Z31

6·11

(A3.2.3)

A4.0 STANDARD BUILDING BLOCK CIRCUITS
General Comments:

Definitions:
Av = Closed Loop AC Gain

Power supply connections omitted for clarity.

fa = Low Frequency -3 d B Corner

Split supplies assumed.

Rin = Input Impedance

Single supply biasing per A4.9 or A4.10.

A4.1 Non-Inverting AC Amplifier

A4.4 Non-Inverting Buffer

'; 0--11-+---1
Co

'0

'0

Av'" 1

Rio'" Rt
fo=_I_
2rrRl Co

R2
Av"" 1+-

Rl

Rin'" RZ
fo =

A4.5 Inverting Buffer

21f~2CO = h~lCl

A4.2 Inverting AC Amplifier
Cl
ejo-fo-'V\rv-_-I
'0

'0

AV '" -1

RZ

Ain :: A1
fo '"

Ay =

-R'1

Rin = R1
fo •

21r~1 C,

RZ

A4.6 Difference Amplifier

h~' Co

A4.3 Inverting Summing Amplifier
Cl

ej

0-1 !--'V'oIV-_--1
'0

'1

<>-ICl

'2

o-fI-'lM
...........-I
Cz
R2
'0

••
•
'no-f~
C
Rn
n

eo

=(Rl+R2)~e2_~el
R3+ R4 Rt

-:.-

IF Rl = RZ = ••. = RN THEN

IF Rt

=:

eo '"

~(e2-el)

fo '"

2n~'Cl

R,

R3* AND R2 = R4* THEN

Rl

'" 2rr(R3+R4IC3

RZ = R4 FOR MINIMAL OFFSET ERROR
• - 0.1% MATCH FOR MAX CMRR

6-12

A4.7 Variable Gain AC Amplifier

AS.O MAGNETIC PHONO CARTRIDGE NOISE
ANALYSIS

RZ

AS.1 Introduction
Present methods of measuring signal·to·noise (SIN) ratios
do not represent the true noise performance of phono
preamps under real operating conditions. Noise measurements with the input shorted are only a measure of the
preamp noise voltage, ignoring the two other noise sources:
the preamp current noise and the noise of the phono
cartridge.

'0

AV

0'

=

(SLIDER AT GROUND)

AVmax = R;n =
fo

~
R,

(SLIDER ATPDS. INPUT)

't (MINIMUM)

Modern phono preamps have typical SIN ratios in the 70dB
range (below 2mV @ 1 kHz), which corresponds to an input
noise voltage of O.64JlV, which looks impressive but is quite
meaningless. The noise of the cartridge 1 and input network is typicallY greater than the preamp noise voltage,
ultimately limiting SIN ratios. This must be considered
when specifying preamplifier noise performance. A method
of analyzing the noise of complex networks will be presented and then used in an example problem.

=--'-

z.('t)c,

'LIMITED BY CMRR OF AMPLIFIER AND MATCH OF
R, = R3. RZ =R4 ••. g•• LF356 AND 0.'% MATCH
EQUALS> BOdB FOR AVmax = ZOdB.

M.8 Switch Hitter (Polarity Switcher, or 4·Quadrant Gain
Cantrall

AS.2 Review of Noise Basics
The noise of a passive network is thermal, generated by the
real part of the complex impedance, as given by Nyquist's
Relation:

C,

'; o-j 1'-'_v../\r"--1
'0

V n2; 4kTRe(Z)Af

+'

Av =
Av = O·

where:
(SLIDER AT Cll
(SLIDER MIDPDSITION)

V n2; mean square noise voltage
k ; Boltzmann's constant (1.38 x 10-23W-sectK)

Av = -, (SLIDER AT GROUND)

T ; absolute temperature (0 K)

't (MINIMUM)
f o =--'z.('t)c,

Af ; noise bandwidth (Hz)

Re(Z) ; real part of complex impedance (n)

R;n =

'WITHIN CMRR OF AMPLIFIER

The total noise voltage over a frequency band can be readily
calculated if it is white noise (Le., Re(Z) is frequency
independent). This is not the case with phono cartridges or
most real world noise problems. Rapidly changing cartridge
network impedance and the RIAA equalization of the preamplifier combine to complicate the issue. The total input
noise in a non-ideal case can be calculated by breaking the
noise spectrum into several small bands where the noise is
nearly white and calculating the noise of each band. The
total input noise is the RMS sum of the noise in each of the
bands N1, N2, ... , Nn .

M.9 Single Supply Biasing of Non-Inverting AC Amplifier

'0

Avo

RZ
= 1 +R,

RiR = Rz

to

(AS.2.1)

Vnoise = (VN1 2 + VN2 2 + ... + VNn 2)%

= 2'11'~2CO = h~'Cl

(A5.2.2)

This expression does not take into account gain variations
of the preamp, which will also change the character of the
noise at the preamp output. By reflecting the R IAA equalization to the preamp input and normalizing the gain to
OdB at 1 kHz, the equalized cartridge noise may then be
cal cu Iated.

M.10 Single Supply Biasing of Inverting AC Amplifier
RZ

'0
VEQ =

(I Al I2 VNl 2 + I A21 2 + ... + I An I2 VNn 2%
)
(A5.2.3)

lOOk

Rz

where:
Av =-~

I An I ; magnitude of the equalized gain at the

R,

lOOk

center of each noise band (V IV)

Rin'" R1

fo ..

VEQ = equalized preamp input noise

Z'/l'~' Co
6-13

pI, +tIH-f-ftt-+-t-ttl
20 H..J.lfI-+t+I+-H+tt-+-++-H
H-+I+t-+""t+I+-H+13

30 L.L..J.J..LLl-J...llL-L...LJ.l.LLJ..LU
10Hz
100Hz
1kHz
10kHz
100kHz

R = RAIIR.
L = L.

FREQUENCY

C'" Cs+Cc

FIGURE AS.1 Normalized RIAA Gain

A5.3 Cartridge Impedance

The impedance relations for this network are:

The simplified lumped model of a phono cartridge consists
of a series inductance and resistance shunted by a small
capacitor. Each cartridge has a recommended load consisting
of a specified shunt resistance and capacitor. A model for
the cartridge and preamp input network is shown in Figure
A5.2.

Re(Z) =
(RXL-RXcl 2 +XL 2 XC 2
(A5.3.2)

IZ I

---------1
'---''"""""T-;I-1-...,
I

I
I
I
I

I

I

I
L,
PHONO
CARTRIDGE

I
I
C, I

R,

I
I
I

_________ ...1

A5.4 Example

RA

Calculations of the R IAA equalized phono input noise are
done using Equations (A5.2.1 HA5.3.2). Center frequencies
and frequency bands must be chosen: values of Rp, Lp,
Re(Z). I Z I and noise calculated for each band, then
summed for the total noise. Octave bandwidths starting at
25 Hz will be adequate for approximating the noise.

PREAMP INPUT AND
CABLE CAPACITANCE

IT'

An ADC27 phono cartridge is used in this example, loaded
with C = 250pF and RA = 47kn, as specified by the
manufacturer, with cartridge constants of Rs = 1.13 kn and
Ls = 0.75H. (CC may be neglected.) Table A5.1 shows a
summary of the calculations required for this example.

L _________

FIGURE AS.2 Phono Cartridge and Preamp Input Network

This seemingly simple circuit is quite formidable to analyze
and needs further simplification. Through the use of Q
equations,2 a series L·R is transformed to a parallel L·R.

A5.5 Conclusions
The R IAA equalized noise of the ADC27 phono cartridge
and preamp input network was 0.751lV for the audio band.
This is the limit for SIN ratios if the preamp was noiseless,
but zero noise amplifiers do not exist. If the preamp noise
voltage was 0.641lV then the actual noise of the system is
0.991lV ([0.64 2 + 0.75 2 ]YzIlV) or -66dB SIN ratio (re 2mV
@ 1 kHz input). This is a 4dB loss and the preamp current
noise will degrade this even more.

Q :: wLS

RS

R. = Rs(' + Q2)

L.

=

RXL Xc

(A5.3.1)

' + 02)
= Ls ( Ii2

Simplifying the input network,
6-14

TABLE AS.1 Summary of Calculations

f Range (Hz)

25·50

f Center (Hz)

37.5

75

150

300

faw (Hz)

25

50

100

Q = w Ls
Rs
Q2

0.156

0.313

0.0244

1 + Q2
1 + Q2

~

6.4k . 12.8k

12.8k·20k

1.6k ·3.2k

600

1200

2400

4800

9600

16.4k

200

400

800

1600

3200

6400

7.2k

0.625

1.25

2.5

5

10

20

40

68.4

0.098

0.391

1.56

6.25

25

100

400

1600

4678.6

1.0244

1.098

1.391

2.56

7.25

26

101

401

1601

4679.6

42

11.24

3.56

1.64

1.16

1.04

1.01

1.0

1.0

1.0

Rp (,\1)

1.16k

1.24k

1.57k

2.9k

8.2k

29.4k

114k

454k

108M

5.29M

Lp (H)

31.5

8.43

2.67

1.23

0.87

0.78

0.76

0.75

0.75

0.75

32.9k

42.6k

45.8k

46.6k

0 2

0>

3.2k·6.4k

800·1.6k

50 ·100

100·200

200·400

400·800

RpllRA (.\1)

1.13k

1.21 k

1.52k

2.74k

7k

18.1 k

XL (,\1)

7.42k

3.97k

2.52k

2.32k

3.28k

5.88k

11.45k

22.6k

45.2k

77.2k

Xc (,\1)

17M

8.48M

4.24M

2.12M

1.06M

0.53M

0.265M

0.133M

66.3k

38.8k

Re(Z) (.\1)

1.11k

1.11k

1.11k

1.15k

1.26k

1.73k

3.86k

12.4k

41.5k

34k

IZI (,\1)

1.12k

1.15k

1.3k

1.77k

2.97k

5.59k

11.7k

24.4k

43.6k

40.1k

enz (nV/y'RZ)

4.1

4.1

4.1

4.1

4.3

5.1

7.3

14

26

23

VN (nVI

20.5

29

41

58

86

144.2

292

792

2080

1952

V n 2 (nV2)

420.3

840.5

1681

3362

7396

20.8k

85.3k

627.7k

4.33M

3.81M

A2

63.04

31.6

10

3.17

1.59

0.89

0.45

0.159

0.05

0.025

38.1k

99.7k

216.3k

95.2k

(J1

A2V n2 (nV2)

26.5k

26.6k

16.8k

10.7k

11.8k
-

(~Vn2)Y,

= 2.98>JV unequalized noise.

(~IAn212 V n2) y,

=

0.75>JV RIAA equalized noise.

i

.-

18.5k

AG.O GENERAL PURPOSE OP AMPS USEFUL FOR
AUDIO

Thus it is apparent that present phono preamp SIN ratio
measurement methods are inadequate for defining actual
system performance, and that a new method should be
used - one that more accurately reflects true performance.

National Semiconductor's line of integrated circuits designed specifically for audio applications consists of 4 dual
preamplifiers, 3 dual power amplifiers, and 6 mono power
amplifiers. All devices are discussed in detail through most
of this handbook; there are, however, other devices also
useful for general purpose audio design, a few of which
appear in Table A6.1. Functionally, most of these parts find
their usefulness between the preamplifier and power
amplifier, where line level signal processing may be required.
The actual selection of anyone part will be dictated by its
actual function.

REFERENCES
1. Hallgren, B. I., "On the Noise Performance of a Magnetic
Phonograph Pickup," Jour. Aud. Eng. Soc_, vol. 23,
September 1975, pp. 546-552.
2. Fristoe, H. T., "The Use of Q Equations to Solve Complex
Electrical Networks," Engineering Research Bul/etin,
Oklahoma State University, 1964.
3. Korn, G. A. and Korn, T. M., Basic Tables in Electrical
Engineering, McGraw-Hili, New York, 1965.
4. Maxwell, J., The Low Noise JFET - The Noise Problem
Solver, Application Note AN-151, National Semiconduc
tor, 1975.

TABLE A6.1 General Purpose Op Amps Useful for Audio

,}~o; ()';:'>~ 0';:'>"'~
,>-'"

Device!
LM301A

X

LM310

X

LM318

X
X
X

LM344

X

54

±3-+±18

3

Low THD.

X

30

±5-+±18

5.5

Fast unity-gain buffer.

X

50

±5 -+ ±18

10

High slew rate.

X

0.3

3 -+ 30
(±1.5 -+ ±15)

2

Low supply current quad.

X

2.5

±4 -+ ±34

5

High supply voltage.

X

LM324
LM343

General Features of
Audio Application Interest

X
X

LM348

X
X

30

±4 -+ ±34

5

Fast LM343.

0.5

±5-+±18

4.5

Quad LM741.
Fast LM348.

2

±5 -+ ±18

4.5

LF355

X

X

5

±5-+±18

4

Low supply current LF356.

LF356 s

X

X

12

±5-+±18

10

Fast, JFET input, low noise.

LF357

X

50

±5-+±18

10

Higher slew rate LF356.

X

0.3

3 -+ 30
(±1.5 -+ ±15)

1.2

Dual LM324.

X

0.5

±3-+±18

2.8

X

LM349

X
X

LM358

Supermatch low noise transistor pair.

LM394
LM741

X

Workhorse of the industry.

LM747

X

X

0.5

±3-+±18

5.6

Dual LM741 (14 pin).

LM1458

X

X

0.2

±3-+±18

5.6

Dual LM741 (8 pin).

X

0.5

4 -+ 30
(±2 -+ ±15)

10

Quad current differencing amp.

X

0.03 ±1-+±18

0.1

Micropower.

X

LM3900
LM4250

1.
2.
3.
4.
5.

X

Commercial.devices shown (O°C-700e); extended temperature ranges available.
Decompensated devices stable above a minimum gain of 5 VIV.
Av:O:; 1 V/V unless otherwise specified.

Compensation capacitor

=

3pF; Av

=

10V/V minimum.

Highly recommended as general purpose audio building block.

6-16

A7.0 FEEDBACK RESISTORS AND AMPLIFIER NOISE

FIGURE A7.1 Practical Feedback Amplifier

gmv l

R2

FIGURE A7.2 Model of First Stage of Amplifier

To see the effect of the feedback resistors on amplifier
noise, model the amplifier of Figure A7.1 as shown in
Figure A7.2.

e;;2 gives:
(A7.2)

We must now show that the intrinsic noise generators e;;2
and 1;;2 are related to the noise generators outside the
feedback loop, ii2 2 and 122. In addition, the output noise
at va can be related to v1 by the open loop gain of the
amplifier G, i.e.,

1;;2 gives:

va = v1 G
Assume Zi }> R 111 R2

Thus v1 is a direct measure of the noise behavior of the
amplifier. Open circuit the amplifier and equate the effects
of the two noise current generators. By superposition:

also

:. 1,;2

v1 = in Zi
=

(A7.3)

v1

Add Equations (A7.2) and (A7.3) and equate to Equation
(A7.1):

122

Short circuit the input of the amplifier to determine the
effect of the noise voltage generators. To do this, short the
amplifier at e:i 2 and determine the value of V1, then short
circuit the input at e;;2 and find the value of v1.

ii2 2
v1

(A7.1)

Now short the input at ;;;:;2; e;;2 and ~2 both affect v1.

6·17

AB.O RELIABILITY

Thus ICs of high quality may, in fact, be of low reliability,
while those of low quality may be of high reliability.

Consumer Plus Program

Improving the Reliability of Shipped Parts

National's Consumer Plus Program is a comprehansive
program that assures high quality and reliability of molded
integrated circuits. The C+ Program improves both the
quality and reliability of National's consumer products. It
is intended for the manufacturing user who cannot perform
100% inspection of his ICs, or does not wish to do so, yet
needs significantly-better·than·usual incoming quality and
reliability levels for his ICs.

The most important factor that affects a part's reliability is
its construction: the materials used and the method by
which they are assembled.
It's true that reliability cannot be tested into a part, but
there are tests and procedures that can be implemented
which subject the IC to stresses in excess of those that it
will endure in actual use. These will eliminate most marginal
parts.

Integrated circuit users who specify Consumer Plus proces·
sed parts will find that the program:
• eliminates 100% the need for incoming electrical inspec·

In any test for reliability the weaker parts will normally
fail first. Stress tests will accelerate the failure of the weak
parts. Because the stress tests cause weak parts to fail prior
to shipment to the user, the population of shipped parts
will in fact demonstrate a higher reliability in use.

tion

• eliminates the need for, and thus the costs of, indepen·
dent testing laboratories
• reduces the cost of reworking assembled boards
• reduces field failures
• reduces equipment downtime

Quality Improvement
When an IC vendor specifies 100% final testing of his parts,
every shipped part should be a good part. However, in any
population of mass·produced items there does exist some
small percentage of defective parts.

Reliability Saves You Money
With the increased population of integrated circuits in
modern consumer products has come an increased concern
with IC failures, and rightly so, for at least two major
reasons. First of all, the effect of component reliability on
system reliability can be quite dramatic. For example,
suppose that you, as a color TV manufacturer, were to
choose ICs that are 99% reliable. You would find that if
your TV system used only seven such ICs, the overall
reliability of IC portion would be only 50% for one out of
each ten sets produced. In other words, only nine out of
your ten systems would operate. The result? Very costly to
produce and probably very difficult to sell. Secondly,
whether the system is large or small, you cannot afford to
be hounded by the spectre of unnecessary maintenance
costs, not only because labor, repair or rework costs have
risen - and promise to continue to rise - but also because
field replacement may be prohibitively expensive.

One of the best ways to reduce the number of such faulty
parts is, simply, to retest the parts prior to shipment. Thus,
if there is a 1 % chance that a bad part will escape detection
initially, retesting the parts reduces that probability to only
0.01 %. (A comparable tightening of the QC group's
sampled-test plan ensures this.)
National's Consumer Plus Program Gets It All Together
We've stated that the C+ Program improves both the quality
and reliability of National's molded integrated circuits, and
pointed out the difference between these two concepts.
Now, how do we bring them together? The answer is in the
C+ Program processing, which is a continuum of stress and
double testing. With the exception of the final QC inspection, which is sampled, all steps of the C+ processes are
performed on 100% of the program parts. The following
flow chart shows how we do it.

Reliability vis·a-vis Quality
The words "reliability" and "quality" are often used interchangeably as though they connoted identical facets of a
product's merit. But reliability and quality are different
and IC users must understand the difference to evaluate
various vendors' programs for product improvement that
are generally available, and National's Consumer Plus
Program in particular.

Epoxy B Processing for All Molded Parts
At National, all molded semiconductors, including
ICs, have been built by this process for some time
now. All processing steps, inspections and QC
monitoring are designed to provide highly reliable
products. (A reliability report is available that
gives, in detail, the background of Epoxy B, the
reason for its selection at National and reliability
data that proves its success.)

The concept of quality gives us information about the
population of faulty IC devices among good devices, and
generally relates to the number of faulty devices that arrive
at a user's plant. But looked at in another way, quality can
instead relate to the number of faulty ICs that escape
detection at the IC vendor's plant.

Six Hour, 150°C Bake
This stress places the die bond and all wire bonds
into a tensile and shear stress mode, and helps
eliminate marginal bonds and connections.

It is the function of a vendor's Quality Control arm to
monitor the degree of success of that vendor in reducing
the number of faulty ICs that escape detection. QC does
this by testing the outgoing parts on a sampled basis. The
Acceptable Quality Level (AQL) in turn determines the
stringency ofthe sampling. As the AQL decreases it becomes
more difficult for bad parts to escape detection; thus the
quality of the shipped parts increases.

Five Temperature Cycles (O°C to 100°C)
Exercising the circuits over a 100°C temperature
range generally eliminates any marginal bonds
missed during the bake.
High Temperature (100°C) Functional Electrical
Test

The concept of reliability, on the other hand, refers to how
well a part that is initially good will withstand its environment. Reliability is measured by the percentage of parts
that fail in a given period of time.

A high-temperature test such as this with voltages
applied places the die under the most severe stress
6-18

possible. The test is actually performed at 100°C30°C higher than the commercial ambient limit.
All devices are thoroughly exercised at the 100°C
ambient. (Even though Epoxy B has virtually
eliminated thermal intermittents, we perform this
test to insure against even the remote possibility of
such a problem. Remember, the emphasis in the
C+ Program is on the elimination of those margin·
ally performing devices that would otherwise lower
field reliability of the parts.)

DC Functional and Parametric Tests
These room·temperature functional and parametric
tests are the normal, final tests through which all
National products pass.
Tighter·Than·Normal OC Inspection Plans
Most vendors sample inspect outgoing parts to a
0.65% AOL. Some use even a looser 1 % AOL.
However, not only do we sample your parts to a
0.28% AOL for all data·sheet DC parameters, but
they receive a 0.14% AOL for functionality as well.
Functional failures - not parameter shifts beyond
spec - cause most system failures. Thus, the five·
times to seven·times tightening of the sampling
procedure (from 0.65%·1 % to. 0.14% AOL) gives a
substantially higher quality to your C+ parts. And
'you can rely on the integrity of your received ICs
without incoming tests.
Ship Parts

Here are the OC sampling plans used in our Consumer Plus
test program:

Test

Temperature

AOL

Electrical Functionality
Parametric, DC
Parametric, DC
Parametric, AC
Major Mechanical
Minor Mechanical

25°C
25°C
100°C
25°C

0.14%
0.28%
1%
1%
0.25%
1%

6·19

7.0 Index
Baxandall Tone Control (see Tone Control, Active)
Biamplification: 5-1
Bias Erasure: 2-31
Bias Trap: 2-32
Boosted Power Amplifiers
Emitter Followers: 4-42
LM391: 4-43
Bootstrapped Amplifiers (see Power Amplifiers, LM388,
LM390)
Bridge Amplifiers
LM377/LM378/LM379: 4-15
LM380: 4-25
LM388: 4-39
Power Dissipation of: 4-45
Buffer Amplifier: 6-12
Butterworth Filters: 2-50, 5-1

AB Bias: 4-3
Absolute Maximum Ratings: 1-2,6-1
Acoustic Pickup Preamp: 5-12
Active Crossover Networks
Filter Choice: 5-1
Filter Order: 5-1
Table of Values: 5-5
Third-Order Butterworth: 5-2
Use of: 5-6
AGC: 3-27
AM9709: 5-11
AM97Cll: 2-62
Ambience, Rear Channel, Amplifier: 4-20
Amplifiers
AB Bias: 4-3
Bootstrapped: 4-37, 4-41
Buffer: 6-12
Class B: 4-2
Current Limit: 4-3
Difference: 6-12
Distortion: 4-1,4-3
Frequency Response: 4-1
gm: 4-1
Inverting AC: 6-12
Loop Gain: 4-1
Non-Inverting AC: 6-12
Output Stages: 4-2
Protection Circuits: 4-3
RF Oscillation in: 4-2
Single Supply Biasing: 6-13
Slew Rate: 4-2
Summing: 6-12
Thermal Shutdown: 4-4
Transconductance: 4-1
Variable Gain: 6-13
Amplitude Modulation (see AM Radio)
AM Radio
Field Strength Conversion: 3-1
LM1820: 3-4
Regenerative: 3-1
Superheterodyne: 3-1
Tuned RF: 3-1
Typical Gain Stages: 3-4
AM Rejection Ratio: 3-27
AM Suppression: 3-27
Analog Switching (see Switching, Noiseless)
Antenna Field Strength (see AM Radio)
Antennas
Capacitive: 3-2
Ferrite Rod: 3-1
AOL (Acceptable Ouality Level): 6-18
Audio Rectification: 2-10
Audio Taper Potentiometer: 2-40

Capacitive Antenna (see Antennas, Capacitive)
Captu re Ratio: 3-27
Cartridges (see Phono Cartridges)
Ceramic Phono Amplifier: 4-24, 4-34
Channel Separation: 3-27
Circuit Layout (see Layout, Circuit)
Class B Output Stage: 4-2
Closed-Loop Gain: 2-1
CMRR in Mic Preamps: 2-39
Conduction: 4-46
Constant Amplitude Disc Recording: 2-26
Constant Velocity Disc Recording: 2-26
Consumer Plus Program: 6-18
Contact Mic Preamp (see Acoustic Pickup Preamp)
Convection: 4-46
Crest Factor: 2-8
Crossover Distortion (see Distortion)
Crossover Networks (see Active Crossover Networks)
Current Amplifier: 2-61
Current Limit: 4-3
Cutover: 2-25
Decibel: 6- 11
Decompensated Op Amp: 1-2
Delta-Wye Transformation: 6-11
Difference Amplifier: 2-38,6-12
Disc (see Phono Disc)
Dissipation (see Power Dissipation)
Distortion
Harmonic: 1'2,3-27,4-1
Crossover: 4-3
Dynamic Range
Phono Disc: 2-25
Supply Voltage: 1-2
Emissivity: 4-49
Emitter Coupled RF Amplifier: 3-9
Epoxy B: 6-18
Equalization (see RIAA or NAB Equalization)
Equalizer: 2-53
Equalizing Instrument: 2-56
Excess Noise: 2-3

Balance Control: 2-44, 4-19
Balanced Mic Preamp (see Mic Preamps)
Bandwidth: 1-2
Bass Control
Active: 2-45, 2-47, 4-35, 5-12
Passive: 2-40

Feedback, Effects of
Bandwidth: 2-1
7-1

General: 2-1
Harmonic Distortion: 2-1
Input Impedance: 2-1
Inverting Amplifier: 2-1
Noise Gain: 2-1
Non-Inverting Amplifier: 2-1
Output Impedance: 2-1
Series-Shunt: 2-1
Shunt-Shunt: 2-1
Feedback Tone Control (see Tone Control, Active)
Ferrite Rod Antenna (see Antennas, Ferrite Rod)
Field Strength (see Antenna Field Strength)
Filters, Active
Bandpass: 2-51, 2-52, 2-57
High Pass: 2-49, 5-3
Low Pass: 2-49, 5-3
Parameter Definitions: 2-49
Rumble: 2-49
Scratch: 2-49
Speech: 2-51
Flanging: 5-10
Flat Response: 2-40
Fletcher and Munson (see Loudness Control)
Flicker Noise: 2-4
FM Radio
Detectors: 3-8
Gain Blocks: 3-11
I F Amplifiers: 3-8,3-13
Limiters: 3-8
LM171: 3-9
LM703: 3-9
LM1310: 3-22, 3-23
LM1351: 3-13
LM1800: 3-23
LM2111: 3-13
LM3011: 3-11
LM3065: 3-15
LM3075: 3-15
LM3089: 3-18
Narrowband: 3-14
Stereo: 3-23
FM Scanner Power Amp: 4-40
FM Stereo Multiplex (see FM Radio, Stereo)
Form Factor: 6-7
Frequency Modulation (see FM Radio)
Full-Power Bandwidth: 1-1
Fuzz: 5-11

Modelling: 4-47
PC Board Foil: 4-50
Procedure: 4-48
Staver V-7: 4-22
Thermal Resistance: 4-47
Where to Find Parameters: 4-47
IF Bandwidth: 3-27
IF Selectivity: 3-27
Input Bias Current: 3-27
Input Referred Ripple Rejection: 1-2
Input Resistance: 3-27
Input Sensitivity: 3-27
Input Voltage Range: 3-27
Instrumentation Amplifier: 2-39
Intercom: 4-26, 4-40
Inverse RIAA Response Generator: 2-30
Inverting AC Amplifiers: 6-12
JFET Switching: 2-62
Lag Compensation: 2-56
Large Signal Response: 1-1
Large Signal Voltage Gain: 3-27
Layout, Circuit: 2-1
LF356/LF357
Active Crossover Network: 5-4,5-5
Mic Preamp: 2-39
LH0002: 2-61
Limiting Sensitivity: 3-27
Limiting Threshold: 3-27
Line Driver: 2-61
LM171: 3-9
LM324: 5-11
LM348: 5-11
LM349
Active Tone Control: 2-47, 2-49
Equalizing Instrument: 2-58
Ten Band Octave Equalizer: 2-55
LM377 /LM378/LM379
Boosted: 4-42
Bridge Connection: 4-15
Characteristics: 4-5
Circuit Description: 4-8
Comparison: 4-5
Fast Turn-On Circuitry: 4-9
Heatsinking: 4-13
Inverting Amplifier: 4-10
Layout: 4-13
Non-Inverting Amplifier: 4-9, 4-10, 4-14
Power Oscillator: 4-17
Power Output: 4-11
Power Summer: 5-10
Proportional Speed Controller: 4-18
Rear Channel Ambience Amplifier: 4-20
Reverb Driver: 5-8, 5-9
Split Supply Operation: 4-13
Stabilization: 4-13
Stereo System: 4-19
Two-Phase Motor Drive: 4-17
Unity Gain Operation: 4-14

Gain-Bandwidth Product: 1-2
General Purpose Op Amps: 6-16
Graphic Equalizer: 2-53
Groove Modulation: 2-25
Ground Loops: 2-1
Harmonic Distortion (see Distortion)
Head Gap (Width): 2-31
Headroom (see Dynamic Range)
Heatsinking
Custom Design: 4-48
Heat Flow: 4-46
LM377 /LM378/LM379: 4-13
7-2

LM380
AC Equivalent Circuit: 4-22
Biasing: 4-23
Bridge: 4-25
Ceramic Phono: 4-24
Characteristics: 4-6
Circuit Description: 4-21
Common-Mode Tone Control: 4-24
Common-Mode Volume Control: 4-24
DC Equivalent Circuit: 4-21
Device Dissipation: 4-22
Dual Supply: 4-27
Heatsinking: 4-22
Intercom: 4-26
JFET Input: 4-27
Oscillation: 4-24
R F Precautions: 4-24
Siren: 4-28
Voltage-to-Current Converter: 4-27
LM381
Audio Rectification Correction: 2-10
Biasing: 2-12
Characteristics: 2-11
Circuit Description: 2-12
Equivalent Input Noise: 2-9
Inverting AC Ampl ifier: 2-15
Mic Preamp: 2-58
Non-Inverting AC Amplifier: 2-15
Split Supply Operation: 2-14
Tape Playback Preamp: 2-33
Tape Record Preamp: 2-32
LM381A
Characteristics: 2-11
Equivalent Input Noise: 2-9
General: 2-15
Mic Preamp: 2-37
Optimizing Input Current Density: 2-16
Phono Preamp: 2-44
Tape Playback Preamp: 2-36
LM382
Adjustable Gain for Non-Inverting Case: 2-22
Characteristics: 2-11
Equivalent Input Noise: 2-9
Internal Bias Override: 2-22
Inverting AC Amplifier: 2-23
Non-Inverting AC Amplifier: 2-21
Tape Preamp: 2-36,4-20
Unity Gain Inverting Amplifier: 2-24
LM384
Characteristics: 4-28
Five Watt Amplifier: 4-29
General: 4-28
LM386
Bass Boost: 4-32
Biasing: 4-31
Characteristics: 4-6
Ga i n Control: 4-31
General: 4-30
Muting: 4-31
Non-Inverting Amplifier: 4-31,4-32
Sine Wave Oscillator: 4-33
Square Wave Oscillator: 4-32

LM387 /LM387 A
Acoustic Pickup Preamp: 5-12
Active Bandpass Filter: 2-53
Active Tone Control: 2-48
Adjustable Gain: 5-12
Characteristics: 2-11
Equivalent Input Noise: 2-9
Inverse RIAA Response Generator: 2-31
Inverter: 5-9
Inverting AC Amplifier: 2-19
Line Driver: 2-61
Mic Preamp: 2-38
Mixer: 5-8, 5-9
Noise Measurement of: 2-8
Non-Inverting AC Amplifier: 2-19
Passive Tone Controls: 2-43
Reverb Recovery Amplifier: 5-8,5-9
Rumble Filter: 2-50
Scratch Filter: 2-52
Speech Filter: 2-52
Summer: 5-8, 5-9
Tape Playback Preamp: 2-33
Tape Record Preamp: 2-32
Tone Control Amplifier: 2-20, 5-12
Two Channel Panning Circuit: 2-60
Unity Gain Inverting Amplifier: 2-19
LM388
Bootstrapping: 4-38
Bridge: 4-39
Characteristics: 4-6
FM Scanner Power Amp: 4-40
General: 4-37
Intercom: 4-40
Squelch: 4-41
Walkie Talkie Power Amp: 4-40
LM389
Ceram ic Phono: 4-34
Characteristics: 4-6
General: 4-33
Logic Controlled Mute: 4-37
Muti ng: 4-33
Noise Generator: 4-36
Siren: 4-34
Tape Recorder: 4-34
Transistor Array: 4-33
Tremolo: 4-36
Voltage-Controlled Amplifier: 4-36
LM390
Characteristics: 4-6
General: 4-41
One Watt, 6 Volt Amplifier: 4-41
LM391: 4-43
LM703: 3-9
LM741: 5-11
LM1303
Characteristics: 2-11
Inverting AC Amplifier: 2-25
Non-Inverting AC Amplifier: 2-25
Tape Preamp: 2-36
LM1310: 3-23
LM1351: 3-13
7-3

Generators: 2-4
Index of Resistors: 2-3
Measurement Techniques: 2-8
Modelling: 2-4
Non-Inverting vs. Inverting Amplifiers: 2-7
Phono Disc: 2-25
Pink: 2-56
Popcorn: 2-4
Resistor Thermal Noise: 2-3
RF: 2-7
Shot: 2-3
Signal-to-Noise Ratio: 2-7
Thermal: 2-3
Total Equivalent Input Noise Voltage: 1-2,2-4
Voltage: 2-4
White: 2-3, 2-56
1If: 2-3,2-4
Non-Inverting AC Amplifier: 6-12

LM1800: 3-23
LM1800A: 3-27
LM1820
AM Radio: 3-6, 3-7
Auto Radio: 3-8
Characteristics: 3-5
Circuit Description: 3-4
Configurations: 3-5
General: 3-4
Impedance Matching: 3-5
LM2111: 3-13
LM3011: 3-11
LM3065: 3-15
LM3075: 3-15
LM3089
AFC: 3-22
AGC: 3-23
Applications: 3-21,3-26
Circuit Description: 3-18
General: 3-18
Mute Control: 3-22
PC Layout: 3-20
Quad Coil Calculations: 3-21
SIN: 3-23
Logarithmic Potentiometer: 2-4,0
Loop Gain: 2-1,4-1
Loudness Control: 2-43, 4-19

Octave Equalizer: 2-53
Op Amps (see Amplifiers)
Open Loop Gain: 1-2,2-1
Oscillations, Circuit (see Layout, Ground Loops, Supply
Bypassing, or Stabilization)
Oscillator: 4-32, 4-33
Oscillator, Power: 4-17
Output Referred Ripple' Rejection: 1-2
Output Resistance: 3-27
Output Voltage Swing: 3-27
Overmodulation (Phono): 2-25

Magnetic Phono Cartridge Noise Analysis: 6-13
Microphone Mixer: 2-59
Microphone Preamplifiers
CMRR of: 2-39
LF356: 2-39
LF357: 2-39
LM381A: 2-37
LM387 A: 2-38
Low Noise, Transformerless, Balanced: 2-39
Transformer-Input, Balanced: 2-38
Transformerless, Balanced: 2-39
Transformerless, Unbalanced: 2-37
Microphones: 2-37
Midrange Tone Control: 2-48
Mixer (see Microphone Mixer)
MM5837: 2-56, 2-58
Monaural Channel Unbalance: 3-27
Motorboating: 2-2
Motor Drive: 4-17,4-18
Multiple Bypassing: 2-2
NAB (Tape) Equalization: 2-31
Narrowband FM: 3-14
Noise
Bandwidth: 2-3
Cartridge: 6-13
Constant Spectral Density: 2-3
Crest Factor: 2-8
Current: 2-4
Differential Pair: 2-8
Effect of Ideal Feedback on: 2-4
Effect of Practical Feedback on: 2-5
Excess: 2-3
Feedback Resistors: 6-17
Figure: 2-18,3-27
Flicker: 2-4

Panning: 2-60
Passive Crossover: 5-1
Phase Shifter: 5-10
Phono Cartridges
Ceramic: 2-27
Crystal: 2-27
Magnetic: 2-27
Noise: 2-27, 6-13
Typical Output Level: 2-28
Phono Disc
Dynamic Range: 2-25
Equalization: 2-25
Noise: 2-25
Recording Process: 2-25
SIN: 2-25
Phono Equalization (see RIAA Equalization)
Phono Preamplifiers
General: 2-25
Inverse RIAA Response Generator: 2-30
LM381: 2-27
LM381A: 2-29
LM382: 2-29
LM387: 2-27
LM1303: 2-29
Pickup (see Acoustic Pickup Preamp)
Piezo-Ceramic Contact Pickup: 5-12
Pink Noise: 2-56
Pink Noise Generator: 2-56
Playback Equalization (Phono): 2-25
Playback Head Response: 2-31
Popcorn Noise: 2-4
Power Amplifiers: 4-5, 4-6, 4-7
7-4

Speed Controller, Proportional: 4-18
Square Wave Oscillator: 4-32
Stabilization of Amplifiers: 2-2
Staver Heat Sink: 4-22
Stereo IC Power Amplifiers: 4-5
Stereo IC Preamps (see Preamplifiers)
Stereo Multiplex (see FM Radio, Stereo)
Summing Amplifier: 6-12
Supply Bypassing: 2-2
Supply Current: 3-27
Supply Rejection (see Ripple Rejection)
Supply Voltage: 1-2
Sweep Generator: 5-11
Switching
Active: 2-62
Mechanical: 2-62

Power Bandwidth: 3-27
Power Dissipation
Application of: 4-44
Bridge Amps: 4-45
Calculation of: 4-44
Class B Operation: 4-43
Derivation of: 4-44
Effect of Speaker Loads: 4-45
Maximum: 4-44
Reactive Loads: 4-46
Power Supply Bypassing: 2-2
Power Supply Design
Characteristics: 6-2
Diode Specification: 6-5
Filter Design: 6-3
Filter Selection: 6-1
Load Requirements: 6-1
Transformer Specification: 6-5
Transient Protection: 6-7
Voltage Doublers: 6-8
Power Supply Rejection: 3-27
Preamplifiers (see Microphone, Phono, or Tape)
Preamplifiers, IC: 2-11
Proportional Speed Controller: 4-18
Protection Circuits: 4-3

Tape Bias Current: 2-31
Tape Equalization (see NAB Equalization)
Tape Preamplifiers
Fast Turn-on NAB Playback: 2-34
LM381: 2-33,2-32
LM381A: 2-36
LM382: 2-36
LM387: 2-32
LM387A: 2-33
LM1303: 2-36
Playback: 2-33
Record: 2-32
Ultra Low Noise Playback: 2-36
Tape Record Amplifier Response: 2-32
Tape Recorder: 4-34
Tape Record Head Response: 2-32
Thermal Noise: 2-3
Thermal Resistance: 4-47
Thermal Shutdown: 4-4
Tone Controls
Active: 2-44,4-35,5-12
Passive: 2-40,4-19,4-24
Total Harmonic Distortion: 1-2
Transconductance: 4-1
Transient Protection: 6-7
Treble Control
Active: 2-45, 4-35, 5-12
Passive: 2-41
Tremolo: 4-36, 5-11
TV Sound IF: 3-8
Two Channel Panning: 2-60
Two-Phase Motor Drive: 4-18
Two-Way Radio IF: 3-8

Quality: 6-18
Radiation: 4-46
Reactive Loads (see Power Dissipation)
Recovered Audio: 3-27
Reliability: 6-18
Reverberation
Driver and Recovery Amplifiers: 5-7
General: 5-7
Stereo: 5-8
Stereo Enhancement: 5-9
RF Interference: 2-10
RF Noise Voltage: 2-7, 3-27
RF Transconductance: 3-27
RIAA (Phono) Equalization: 2-25
RIAA Standard Response Table: 2-27
Ripple Factor: 6-1
Ripple Rejection: 1-2
Rumble Filter: 2-49
Scanners (see FM Scanners)
SCA Rejection: 3-27
Scratch Filter: 2-49
Self-Demagnetization: 2-31
Sensitivity: 3-27
Series Shunt Feedback (see Feedback)
Shot Noise: 2-3
Shunt-Shunt Feedback (see Feedback)
Signal-to-Noise of Phono Disc: 2-25
Signal-to-Noise Ratio: 2-7
Sine Wave Oscillator: 4-33
Single-Point Grounding (see Ground Loops)
Single Supply Biasing of Op Amps: 6-13
Siren: 4-28,4-34
Slew Rate: 1-1,1-2,3-27,4-2
Speaker Crossover Networks (see Active Crossover
Networks)
Speaker Loads (see Power Dissipation)
Speech Filter: 2-51

Unbalanced Mic Preamp (see Mic Preamps)
Uncompensated Op Amp: 1-2
Variable Gain AC Amplifier: 6-13
Voltage-Controlled Amplifier: 4-36
Voltage Doublers: 6-8
Voltage-to-Current Converter: 4-27
Walkie Talkie Power Amp: 4-40
White Noise: 2-3, 2-56
White Noise Generator: 4-36
Wien Bridge Oscillator: 4-33
Wien Bridge Power Oscillator: 4-17
Wye-Delta Transformation: 2-45,6-11
7-5



---

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-c041 52.342996, 2008/05/07-21:37:19
Create Date                     : 2016:08:11 19:24:44-08:00
Modify Date                     : 2016:08:11 21:43:53-07:00
Metadata Date                   : 2016:08:11 21:43:53-07:00
Producer                        : Adobe Acrobat 9.0 Paper Capture Plug-in
Format                          : application/pdf
Document ID                     : uuid:3296b959-2a57-674e-a82d-21288e423b3d
Instance ID                     : uuid:f19d0ddf-9035-154d-9b28-d583b6a78744
Page Layout                     : SinglePage
Page Mode                       : UseNone
Page Count                      : 202

EXIF Metadata provided by EXIF.tools