HOWARD W SAMS & COMPANY Audio Libraly

Audio Technology
fundamentals
Audio students and electronic musicians alike will welcome this concise but compre-
hensive overview of the electronic circuits used in typical audio systems. Assuming only
abasic knowledge of mathematics, electricity, and other technical aspects of audio, this
book provides athorough introduction to audio concepts and electronic circuits,
featuring:

• Acomplete overview of audio systems, signals, and circuits, including digital

audio systems.
• Detailed descriptions of operational amplifiers, passive and active filters,
transformers, and semiconductors, and their uses.
• An in-depth discussion of sound measurement formulas, with sample problems
and answers, explained in terms that anonmathematician can easily understand.
• Practical audio test circuits that you can build yourself.
• Troubleshooting and maintenance tips, including acomplete step-by-step tape
recorder alignment procedure.

Clearly written, with numerous illustrations and circuit diagrams, this excellent resource
is amust for all beginning audio professionals.

Alan A. Cohen teaches multitrack recording technology at the Institute of Audio

Research. A member of the Audio Engineering Society, he holds his B.S. degree in
mechanical engineering from London University, England. He has been involved in
audio production for educational films, has worked in the CCTV studio at City University
in London, and has been an instructor in the Radio and TV Department at the Delehanty
Institute.

$19.95 US/22678

ISBN Q-672-22678-2

90000

HOWARD W SAil
Dhision of Inc.
4300 11 .
ésl 62nd Street

Indianapolis, Indiana 46268 9 780672 22678

Audio Technology
Fundamentals
AUDIO LIBRARY

Audio IC Op-Amp Applications, Third Edition

Walter G. Jung

Audio Production Techniques for Video

David Miles Huber

Handbook for Sound Engineers: The New Audio Cyclopedia

Glen Ballou, Editor

Recording Demo Tapes at Home

Bruce Bartlett (John Woram Audio Series)

How to Build Speaker Enclosures

Alexis Badmaieff and Don Davis

Introduction to Professional Recording Techniques

Bruce Bartlett (John Woram Audio Series)

John D. Lenk's Troubleshooting & Repair of Audio Equipment

John D. Lenk

Modern Recording Techniques, Third Edition

David Miles Huber and Robert E. Runstein

Musical Applications of Microprocessors, Second Edition

Hal Chamberlin

Sound System Engineering, Second Edition

Don and Carolyn Davis

Microphone Manual: Design & Application

David Miles Huber

Sound Recording Handbook

John M. Woram (John Woram Audio Series)

Principles of Digital Audio, Second Edition

Ken C. Pohlmann

For the retailer nearest you, or to order directly from the publisher,
call 800-428-SA MS. In Indiana, Alaska, and Hawaii call 317-298-5699.
Audio Technology
Fundamentals

Alan A. Cohen

e
HOWARD W SAMS &COMPANY
A Division of Macmillan, Inc.
4300 West 62nd Street
Indianapolis, Indiana 46268 USA
© 1989 by Alan A. Cohen

FIRST EDITION
FIRST PRINTING- 1989

All rights reserved. No part of this book shall be reproduced, stored

in aretrieval system, or transmitted by any means, electronic,
mechanical, photocopying, recording, or otherwise, without
written permission from the publisher. No patent liability is
assumed with respect to the use of the information contained
herein. While every precaution has been taken in the preparation
of this book, the publisher and author assume no responsibility for
errors or omissions. Neither is any liability assumed for damages
resulting from the use of the information contained herein.

International Standard Book Number: 0-672-22678-2

Library of Congress Catalog Card Number: 89-62335

Acquisitions Editor: Scott Arant

Development Editor: James Rounds
Manuscript Editor: Marie Butler-Knight
Production Coordinator: Marjorie Hopper
Illustrators: T R. Emrick & Sally Copenhaver
Cover Artist: Ron Troxel
Technical Reviewer: Bruce Bartlett
Compositor: Cromer Graphics

Printed in the United States of America

Trademarks

All terms mentioned in this book that are suspected of being

trademarks or service marks have been appropriately capitalized.
Howard W. Sams & Company cannot attest to the accuracy of this
information. Use of aterm in this book should not be regarded as
affecting the validity of any trademark or service mark.
Contents

Preface ix

1 The Audio Chain

Sound and Hearing 3
Audio Systems 6
The Audio Signal 6
The Audio Circuit 10
Inter facing 12

2 Sound Measurement 19
Intensity 21
Bel and Decibel 22
Power Level and Level Gain 24
Sound Pressure and Pressure Level 28
Voltage Level and Level Gain 30
The Standard Audio Circuit 32
dB Applications 34
Resistance Level and Level Gain 39
Summary of dB Formulas 44
Supplementary Problems and Answers 44
Problems 45
Answers 47
vi Audio Technology Fundamentals

3 Operational Amplifiers 51
Op-Amp Characteristics 53
Inverting Amplifiers 54
Non-inverting Amplifiers 57
Stepped Gain Amplifiers 59
Summing Amplifiers 60
Differential Amplifiers 63

4 Passive Filters 67
Filtering Concepts 69
Low- Pass Passive Filters 72
High-Pass Passive Filters 79
The Pole/Zero Approach 83
Bandpass Filters 88
Stopband Filters 91

5 Active Filters 93
Active Filter Characteristics 95
Inverting Active Filters 95
Non-inverting Active Filters 99
Shelving Equalizers 100

6 Transformers 107
The Voltage Changing Function 109
The Isolation Function 111
The Impedance Matching Function 113
Bandpass Characteristics 118
Avoiding Transformer Generated Distortion 124
Summary of Transformer Applications 125

7 Semiconductors 127
Diodes 129
Transistors 133
Common Emitter Amplifiers 136
Common Base Amplifiers 137
Common Collector Amplifiers 140
Class A Amplifiers 142
Contents Vii

Class B Amplifiers 144

Field Effect Transistors 146

8 The Tape Recorder 151

Tape Recording Concepts 153
Theory of Operation 154
Mechanical and Electrical Requirements 155
Electromagnetic Requirements 157
Record/Reproduce Head Requirements 158
Tape Transport Requirements 159
Internal Equalization 161
Electronic Circuit Requirements 162

9 Digital Audio 167

Digital Systems 169
Analog/Digital Interfacing 171
Sampling Requirements 172
Secondary Sampling Requirements 174
D/A and A/D Conversion 175
Digital Audio Applications 178
Advantages of Digital Audio 181

10 Practical Audio Circuits 183

Power Supplies 185
The Transformer 186
The Rectifier 186
The Smoothing and Regulating Stage 188
Power Supply Construction 191
List of Components 192
Construction Procedure 192
Signal Tracer Construction 193
List of Components 196
Construction Procedure 197
Mixer Construction 198
List of Components 201
Construction Procedure 201
Additional Pre-amplifier Stages 202
Technical Considerations 204
viii Audio Technology Fundamentals

11 Troubleshooting and Maintenance 207

Audio Chain Troubleshooting 209
Troubleshooting Left to Right 209
Troubleshooting Right to Left 210
Cable Testing 211
Electronic Circuit Troubleshooting 211
Intermittent Faults 212
Power Supply Troubleshooting 212
Tape Recorder Alignment 214
Cleaning and Demagnetizing 214
Basic Alignment Procedure 214
Professional Alignment Procedure 217

A Formulas and Derivations 221

Output Voltage of Differential Amplifiers 223
Output Level of High-Pass Shelving Filters 225
Output Level of Bandpass Filters 230
Transfer Function of LF Equivalent Circuit 237
Hysteresis 239

Glossary 243

Index 251
Preface

This text is designed for music students and others interested in audio technology.
It assumes that the reader has some background knowledge of electronics. This
book bridges the gap between this basic electronics understanding and its appli-
cation to the field of audio engineering. (For those who do not yet have a
background in electronics, there are many excellent texts on the subject.)
The first chapter, entitled The Audio Chain, presents an overview of atypical
audio system. It includes an explanation of component interfacing and impedance
matching.
Chapter 2, Sound Measurement, covers the decibel. This concept is described
in simple language that abeginning mathematician can understand. The chapter
covers the derivation of the unit and gives practical examples of its application.
In Chapter 3, basic information is given on op-amps. This provides the
necessary background to an understanding of active components such as active
filters, equalizers, mixers, and digital-to-analog converters.
Chapter 4presents the pole/zero approach to analyzing the filtering effect of
reactive components. Once this has been mastered, the concept of filtering
becomes simple. Every upward slope can then be seen as aclimb from azero, and
every downward slope as adescent from the high point of apole. Passive filters
are examined in Chapter 4, and active filters in Chapter 5.
Chapter 6, Transformers, discusses voltage changing, isolation, and imped-
ance matching functions, and offers advice on avoiding transformer generated
distortion. In the following chapter, semiconductor theory is explained, including
the principles behind transistor amplification.
Chapter 8, The Tape Recorder, gives theoretical and practical information on
tape recorders. The purpose of this chapter is to help you obtain good recordings
even without professional equipment. This is followed by achapter on the subject
of digital audio. Chapter 9includes asection on analog-to-digital interfacing and

ix
x Audio Technology Fundamentals

the process of digital recording. It also discusses the characteristic performance

of digital systems compared to analog systems.
Chapter 10, Practical Audio Circuits, then takes aclose look at power supply,
signal tracer, and mixer circuits. This chapter is designed to enable you to
construct these useful devices. The final chapter offers valuable information on
troubleshooting and maintenance of audio equipment, including a detailed tape
recorder alignment procedure.
You will notice that diagrams have been included where it was thought
helpful to do so. If an unfamiliar word or concept arises, its meaning can be
discovered by reference to the glossary at the end of the book. And every effort
has been made to clarify the concepts covered by using simple language and
analogies taken from well understood situations.
Sound and Hearing, 3

Audio Systems, 6

The Audio Signal, 6

The Audio Circuit, 10

Interfacing, 12

1
 1 The Audio Chain

Sound and Hearing

Sound consists of alternate high and low pressure waves moving through an
elastic medium such as air. Sound can also be transmitted through any other
elastic medium, such as water, steel, or rock. When sound waves in air strike a
light surface, such as an eardrum or microphone diaphragm, the pressure
fluctuations cause the diaphragm to fluctuate in proportion to the changing
pressure. This enables the ear or microphone to convert the sound— in the case of
the ear, into nerve impulses, and in the case of the microphone, into an electrical
audio signal.
The sound source can be avibrating or resonating object, such as aplucked
guitar string, a struck tuning fork, or a cymbal. Each elastic object has its own
fundamental resonant frequency; hence, a small bell resonates at a higher
frequency than a large bell, and short piano strings at a higher frequency than
long strings. The frequency is measured in cycles per second, sometimes called
hertz (Hz). A hertz is the same as acycle per second.
A simple object, such as atuning fork, emits apure sound at its fundamental
(resonant) frequency. More complex objects, such as a bell or a drum, resonate
not only at their own fundamental frequencies, but in some parts at whole
number multiples of the fundamental. These higher frequencies are called
harmonics. The relationship between frequency and harmonics is simple. The
fundamental is called the first harmonic. The second harmonic is twice the
fundamental frequency, the third harmonic is three times the fundamental, and
so on. The word high is perfectly applicable to both frequency and pitch. If we
compare the objective measurement of frequency with the subjective experience

3
4 Chapter 1

of the pitch of atone, there is nearly a100 percent correlation between them. A
high frequency is experienced as a high pitched tone, alow frequency as alow
pitched tone. The tonal parameters of human hearing are as follows:

• The lowest audible frequency is about 16 Hz.

• The highest audible frequency is about 16 kHz. (This varies. Some people
can hear up to nearly 20 kHz; others only up to 10 kHz.)

Below the lowest audible frequency of 16 Hz, sound is experienced as avibration

or as individual thumps.
Another subjective relationship between pitch and frequency involves the
recognition of the octave. It happens that double the frequency is perceived as
one octave higher. So, an octave higher is twice the frequency and an octave
lower is half the frequency of the original note. I
There are two variables within the physical concept of sound. One that has
already been mentioned is frequency. The other is the intensity or power of the
sound wave. This depends on the sound pressure. The greater the pressure
difference between the wave peaks and normal atmospheric pressure, the grea tr
the intensity. Subjectively, this is experienced as loudness. Loudness can e
measured as a pressure difference or as intensity, which is the correspondi g
acoustical power in watts per square centimeter.
The quietest audible sound occurs at an intensity of 10-16 W/cm 2.T e
loudest safe intensity is 10 - aWicm 2.A louder sound causes hearing dama e.
Under these conditions, very fine sensory hairs in the inner ear get broken by te
excessively loud sounds. The result is high frequency deafness. Unfortunat y,
during sound mixing and recording sessions, audio engineers tend to turn up th ir
monitors to produce high sound intensities in the control room. Excessi e
loudness can also be experienced at some concerts. Under these conditio s,
hearing discomfort occurs or there is ringing in the ears. This is awarning sign. It
means your hearing is being damaged and you are losing your high frequency
hearing ability.
While it may not be acceptable to ask along-time audio engineer to turn the
monitors down, you can use earplugs or stuff apiece of tissue into each ear. The
tissue will reduce the sound level by about 20 dB. (A dB is aunit of sound pressure
or sound intensity level. It is described in Chapter 2.) Remember that if you
misuse and destroy an amplifier or tape deck, you can purchase another (al a
price). But if you destroy your hearing, you are stuck with it for the rest of your
life. So be careful of your most sensitive audio equipment.
Now, let's return to consideration of some of the characteristics of sound. In
the absence of harmonics, apure tone corresponds to pressure fluctuations that
oscillate with simple harmonic motion. This type of motion can be illustrated 4
viewing apoint rotating around the circumference of acircle, as seen on edge o
that only one dimension of movement is visible. Any purely vibrating elastic bo y,
such as a tuning fork blade or guitar string, oscillates in this way. A graphi al
 The Audio Chain 5

representation of such pressure changes on atime (or angular displacement) base

is shown in Figure 1-1.

Fig. 1-1 Sine wave ROTATION V Inst

characteristics

AMPLITUDE OR
PEAK VALUE

180° 270° 360 °

This characteristically shaped curve is called asine wave, the reason being
that (from the circular diagram on the left)

[sin 0 = Vinst
—
A

where
Instantaneous value
Vinst =

A = Amplitude
O = Phase angle in degrees

From this equation, we arrive at

Vinst = A sin

So the value is always proportional to the sine of the angle.

As was mentioned earlier, musical instruments produce not only a pure
fundamental note, but also harmonics. It is the relative proportional intensity of
the different harmonics that causes what is known as the timbre of the
instrument. This is what makes the difference between a Stradivarius and a
cheap, mass-produced violin. However, most acoustical instruments produce a
combination of harmonics that is pleasant to the ear.
An amplifier circuit also produces electrically generated harmonics. This is
unfortunate, however, because an amplifier should reproduce exactly what is
present and not add anything of its own. In addition, the electronically produced
harmonics (unlike those produced by musical instruments) are objectionable and
unpleasant to the ear. This effect is called distortion. Fortunately, present day
amplifiers are being produced with a total harmonic distortion of less than 0.1
percent. You should try to avoid using equipment with total harmonic distortion of
1percent or more. It is also important to avoid creating distortion by incorrect
 Chapter 1

design or use of technical equipment. And this returns us to the main topic of
this book.

Audio Systems
In arecording studio or in the sound reinforcement system for aperforming group
or orchestra, the sound isn't transferred from asingle microphone directly to a
tape recorder or power amplifier. There are usually a number of intermediate
stages. This system forms achain of audio components linked one after the other
in series. The chain always starts and ends with atransducer.
A transducer is a component that converts nonelectrical energy into
electrical energy or vice versa. At the start of the audio chain, there can be ia
microphone, atape replay head, or arecord playing cartridge. All of these are
transducers. A microphone converts acoustical energy (sound pressure waves in
air) into electrical energy in the form of a small fluctuating voltage. A record
playing cartridge converts mechanical energy (due to movements of the stylus)
into electrical energy. A magnetic replay head converts the combined effect of a
fluctuating magnetic field and the mechanical energy of the moving tape into
electrical energy.
At the other end of the audio chain, there are one or more loudspeakers.
These transducers convert comparatively large quantities of electrical energy into
acoustical energy, sufficient to fill a room or hall with sound. In this way,
transducers start and terminate the audio chain. Briefly then, the audio chain is a
series of audio components that process and amplify an audio signal.

The Audio Signal

Sound radiation consists of alternating high and low pressure waves. Figure 12
illustrates the instantaneous position of these waves as they travel outward from
their source at the center. Their arrangement is similar to ripples on the surface of
apond into which astone has been thrown, except that on apond, these ripplr
lie on atwo-dimensional surface to form expanding concentric circles, whereas
sound waves form expanding concentric spheres moving outward from their
source in all directions.
A microphone in the path of these waves would convert the changing air
pressures to correspondingly changing voltages. A graph of the air pressure
changes of a pure sound that passes a microphone could be plotted vertically
against atime base, and would look like Figure 1-3(A). If the output voltage of this
microphone were also plotted vertically on a time base, as in Figure 1-3(B),
it would be seen that the voltage exactly corresponds to the air pressure at
all times.
 The Audio Chain 7

Fig. 1-2
Alternating high
and low pressure
waves

Fig. 1-3
Correspondence
between changing
pressure and
ATMOSPHERIC

voltage
PRESSURE

(A) Pressure wave (8) Voltage wave

Similarly, for acomplex wave, there would be equal correspondence between

the pressure graph and the voltage graph, as seen in Figure 1-4.

Fig. 1-4
Correspondence
between pressure

\vL
and voltage in a
complex wave
o

(A) Pressure wave (8) Voltage wave

Such a fluctuating voltage produced by a microphone is called an audio

signal. It is always an AC voltage because it fluctuates in both positive and
negative directions. It is called asignal because, by definition, asignal transmits
information. The audio signal transmits information about the air pressure
changes of atraveling sound wave, and the information is encoded in the form of
voltage. This voltage can, of course, be both processed and amplified under the
skilled control of an audio engineer. This basic process is described in detail in the
following chapters.
Back at the start of the audio chain, the microphone, tape head, or record
playing cartridge all output very small voltages in the order of from 2to 15 mV. It
8 Chapter I

is necessary to amplify these voltages before they can be combined or processed.

If this is not done, they will be swamped by electrical background noise generatéd
in the system. Therefore, there is apre-amplifier (designed to amplify very small
signal voltages) immediately after the first transducer. Let us assume that the
system starts with amicrophone. In this case, it must be followed by amicrophone
pre-amplifier. Then comes the processing stage or stages. These could consist of
an equalizer, compressor, or expander, mixing of other channels from other
microphones, addition of echo or reverberation, and so forth. Finally, when the
audio signal has been processed to the satisfaction of the audio engineer, comes
the power amplification stage. This must give the completed audio signal
sufficient electrical power to drive all of the speakers necessary to fill the rooth,
hall, or theater.
Using the schematic of atriangular symbol for an amplifier and arectangle
for any other type of audio component, atypical audio chain might look like that
shown in Figure 1-5.

Fig. 1-5 Typical mIc

audio chain
PRE-
AMP

EQUALIZER MIXER REVERB

The only difference that occurs when digital systems are used is that the
signal processing is done digitally. However, we inhabit an analog world, so the
microphone input and the loudspeaker output have to remain in analog form.
analog-to-digital converter is therefore needed at the start of the processing part
of the chain, and adigital-to-analog converter is needed at its end (see Figure 1-g).
To avoid confusion, Ishould mention that all active electronic devices a e
powered by DC sources. They are not powered by the audio signal. In Figure 1-5,
the DC power that drives each component is not shown. It is assumed that tie
reader knows it exists. All that is illustrated is the audio signal path, froin
microphone to loudspeaker. In a detailed schematic of an amplifier or oth r
electronic circuit (as in Figure 1-7), the DC source would be shown as a DC
subcircuit running vertically, whereas the signal path, by convention, would be
shown running horizontally.
To distinguish between the DC power supply and the AC audio signal,
remember that the DC supply powers the electronic circuit and the AC aud o
signal controls its performance. In the case of asmall, battery operated radio, the
power supply is the battery. In the case of a60 W amplifier, abattery would te
 The Audio Chain 9

Fig. 1 6Typical
-
MIC
digital audio
chain \ PRE- A/D DIGITAL DIGITAL
AMP CONVERTER EQUALIZER MIXER

DIGITAL D/A
REVERB CONVERTER LOUDSPEAKER

Fig. 1-7 DC

Schematic of
electronic circuit,
showing DC
subcircuit and • -- DC SUBCIRCUIT
signal path

AUDIO AUDIO
SIGNAL SIGNAL
INPUT _ OUTPUT

SIGNAL PATH

insufficient, so it is powered by the 115 V, 60 Hz AC power line. In this case, the

115 V line supply is converted by an internal power supply circuit within the
amplifier into apowerful DC source that can energize the amplifier to give out
the necessary 60 W of sound.
As an analogy to this relationship, consider acar. When you sit in your car, it
is not the movement of your foot on the gas pedal that powers the car. This
merely controls the power output of the engine. The engine powers the car.
Similarly, the audio signal supplies information to the amplifier and controls it,
whereas the DC power supply produces the power.
Having said that, the fact is that there is asmall, but important amount of
electrical power at all stages of the audio signal. And it is this audio signal
waveform, voltage, and power that we will be analyzing in detail in the future,
because these are the characteristics that the audio engineer must control. The
power supplied will then look after itself. In exactly the same way, the engine of a
10 Chapter 1

car looks after itself, while the driver's skill is in controlling the gas pedal, brake
pedal, and steering wheel. Just as the driver's foot on the gas pedal requires law
power but precise operation, so the audio signal is of comparatively low power,
but must be controlled precisely to achieve good results.
We can now usefully examine an audio commponent in detail. The common
factor in each component is that it acts as an electrical circuit. By examining this
circuit, we will be able to understand component interfacing and how the audio
chain works as awhole.

The Audio Circuit

Every circuit consists of avoltage source and aload. It is from the voltage source
that the power, in the form of electrical energy, is derived. The load is that part of
the circuit that uses the power and that the whole circuit is designed to operate.
The load can be anything that is used and is driven by electricity. It can be an
electric motor, a light bulb, a coffee maker, or an amplifier. However, in the
circuit diagram, it can always be represented by a resistor, because it acts and
absorbs power just as aresistor would. So, for calculation purposes, we think of it
as aresistor of agiven value.
All circuits fall into two natural parts:

1. The voltage source, or generator part of the circuit

2. The load

The simplest possible circuit would look like that in Figure 1-8.

Fig. 1-8 The

simplest possible GENERATOR i LOAD
circuit

In practice, even the simplest circuit can never be this simple. The reason is
that there is no generator, however large and powerful, that can supply all
possible loads with a 100 percent constant voltage. Due to internal resistani e
within the generator, there is some voltage drop in the generating process when a
load is connected. The more current that is drawn, the lower the generator output
voltage becomes. In the case of mechanically driven generators powered by
 The Audio Chain II

steam turbines, various factors and the actual resistance of the wiring of the rotor
cause the output voltage to drop when alarge amout of current is drawn. In the
case of abattery, the speed at which the electrochemical reaction can take place
depends on the voltage drop that the load produces at the terminals. So the
greater the current drain, the greater this voltage drop has to be. Of course, a
large battery can produce more current output than a small one for a given
voltage drop. This simply means that there is less voltage drop due to internal
resistance. But however large the battery, there will be some drop. And the drop
will increase in proportion to the current.
This situation, in which the voltage drop is proportional to the current, is
exactly what happens when current flows through aresistor. The voltage drop
can be calculated by Ohm's Law, from the relationship

V = IR

where
V = Voltage drop
I = Current drawn
R = Resistance

As a result, any generator has to be considered as having an internal

resistance (sometimes called its output resistance). The larger the generator, the
lower will be its output resistance, because it can output more current for agiven
voltage drop. But in all cases, some internal resistance must be included for
calculation purposes. So the simplest practical circuit becomes avoltage source
with two resistors in series. The internal resistance of the generator is included in
the generator part of the circuit, as shown in Figure 1-9.

Fig. 1-9 Simplest

practical circuit GENERATOR LOAD

The internal resistance of the generator (signifying output resistance) is

labeled R., because we will shortly be talking about the input resistance of an
amplifier. If the internal resistance and the input resistance were both labeled 12 1,
there would be ambiguity. This circuit represents the fundamental behavior of all
audio circuits.
12 Chapter 1

If we consider asingle audio component, isolated from all others, its output
terminals being left open (with no load connected), then we can think of this
audio component as entirely made up of three parts. This fact applies to all audio
components except transducers. It applies to mixers, amplifiers, filters,
equalizers, tape recorders, and so on. The three parts are

1. The input resistance, R1(sometimes called the input impedance). This forms
the effective resistance between the input terminal and ground. It is not
made up of asingle resistor, but it acts as one. So it can be considered a
single resistance for calculation purposes.
2. The signal voltage, V. This forms the voltage source of the generator part of
the circuit.
3. The output resistance of the generator part of the circuit, R. (sometimes
called the output impedance of the component). Again, it is not asingle
resistor, but it acts as one and can be so considered for calculation purposés.

The three parts of an audio component are shown in Figure 1-10. The
realization that all audio components are made up of these three parts makes it
simple to achieve correct impedance matching when interfacing an audio chain,
or in any other required situation.

Fig. 1-10 The

three parts of an
audio component

INPUT OUTPUT
TERMINAL TERMINAL

R, = Input resistance
R, = Output resistance
V = Signal voltage
= Schematic symbol
for an AC voltage

Interfacing
Let us consider asimple audio chain consisting of amic, pre-amplifier, equalizer,
power amplifier, and loudspeaker. The three parts of each component fit together
as illustrated in Figure 1-11.
 The Audio Chain 13

Fig. 1-11 The PRE-AMP EQUALIZER

relationship
between the parts
of each
component in an MIC
audio chain

POWER AMP

••••11100.000,

LOUDSPEAKER

The key to understanding the interfacing situation is to think of each audio

component in terms of these three essential parts:

• The input resistance

• The audio signal
• The output resistance

Notice that the input resistance of the pre-amplifier acts as the load for the
microphone. The input resistance of the equalizer acts as the load for the pre-
amplifier. The input resistance of the power amplifier acts as the load for the
equalizer. This relationship is valid for the interface between any two
components in an audio chain. In general, the input resistance of one component
acts as the load for the previous component. You could say that the load for any
component is the input resistance of the following stage; thus, two adjoining
stages form a single audio circuit, linking them in an interrelated unity. As an
example, let us examine the interface between component A and component B in
Figure 1-12. We can assume that all ground symbols are connected. (All being at 0
volts, they act as though they are connected.) So the equivalent interfacing audio
circuit consists of that shown in Figure 1-12, where:

RoA = Output resistance of component A

RB = Input resistance of component B
14 Chapter 1

Fig. 142 Audio

circuit created by
interfacing

As you know, the current is the same in all parts of aseries circuit. Call this
value I. So by Ohm's law, the voltage lost across RoA, namely VA, is given by

VA = IR0A (1-1)

and the signal voltage that remains at the input terminal of the next stage, VB,

which would be across RA, is given by

VB = IRiB (1•2)

For best impedance matching, less than one-tenth of the signal voltage
should be lost at the interface. This means that the signal voltage that remains at
the input of the second stage, VB, should be at least ten times that which is lOst
across the output resistance of the first stage, VA. Thus,

VB ≥ 10 VA (. 3 )

Substituting in Equation 1-3 for VA and VB from Equations 1-1 and 1-2,

114 ≥ 10 R O A

But Icancels, so

R IB ≥ lo R O A

In other words, for good impedance matching, the input resistance of ail)/
stage should be at least ten times that of the output resistance of the previous
stage. Very often the quantities are specified as impedances rather than
resistances, but the same rule applies:

For good impedance matching, the input impedance of one stage should
be at least ten times that of the output impedance of the previous stage.
 The Audio Chain 15

Problem 1-1: What minimum input impedance would be needed in a

microphone pre-amplifier, if the mic's output impedance is 200 9?
Answer: Because It; has to be ten times the value of R. of the previous
component, pre-amp 11 1 = at least 10 x200 û = 2 kfl. An input
impedance of 2kiZ or more would be suitable.

Ihave just described what is called constant voltage coupling, forming a

constant voltage circuit. It is called constant voltage because, if RL ≥ 10 R., the
voltage across the load will not have dropped more than 10 percent from its
maximum value. So the voltage across the load will be between 90 and 100
percent of its maximum value. Maximum voltage occurs when RL = 00, or under
open circuit conditions. Then there is no current through R. and no voltage drop
across it, so the terminal voltage equals the source voltage.
When audio component (2), the load in Figure 1-13, is connected across the
output terminals of component (1) in an audio chain, the input resistance of
component (2) acts as the load for component (1). And because current now flows
through the circuit, there is avoltage drop across R. and the resultant voltage
across the terminals drops. Thus, VL is always less than V. The amount of the
voltage drop due to the insertion of this component is called the insertion loss.
Using constant voltage coupling impedances ( RL ≥ 10 R.), this insertion loss is
kept small.

Fig. 1-13
Component
interfacing

(
i2 (= RL)

CURRENT FLOW

/
16 Chapter I

There is another type of coupling called power matched. This occurs when
the load resistance equals the generator output resistance (
RL = R.). Then, for a
given source voltage, Vs,the maximum possible amount of power is transferred to
the load. In general, it can be stated that:

Maximum power is transferred to the load when the load resistance

equals the source resistance.

This is called power matched coupling. The power in the load could be
calculated from

PL = I
2RL

But by Ohm's Law,

V
I= R. + RL

SO,

p = Vs2RL
L (R0 + R32

If we take, as an example, asource voltage of 1V,

RL
PL = (
Ro + RL)
2

Fig. 1-14 PL
Variation of
PL as RL
changes with 0.3/R o
aconstant
voltage
source
0.2/R o

0.1/R 0

RL
1/100R 0 1/10R 0 R. 10R e 100R.
The Audio Chain 17

A graph, shown in Figure 1-14, is plotted on a logarithmic base of PL for

various values of RL as aproportion of Ro,using a1V source. It can be seen that
precisely when RL = Ro,the power reaches a maximum. So coupling between
components in an audio chain can be specified in two ways:

1. Constant voltage coupling, when RL ≥ 10 xRo

2. Power matched coupling, when RL = Ro

This chapter has been devoted to ageneral view of audio systems. In the next
chapter, the basic audio measurement unit, the decibel, will be defined and its
relationship to other units will be described.
Intensity, 21

Bel and Decibel, 22

Power Level and Level Gain, 24

Sound Pressure and Pressure Level, 28

Voltage Level and Level Gain, 30

The Standard Audio Circuit, 32

dB Applications, 34

Resistance Level and Level Gain, 39

Summary of dB Formulas, 44

Supplementary Problems and Answers, 44

19
ep..;kt
 2 Sound Measurement

Intensity
Consider apoint source of sound in the form of asmall, vibrating object such as a
tuning fork, bell, or small loudspeaker cone. The mechanical vibrations produce
pressure waves in the air, which travel outward in all directions. The amount of
mechanical work done per unit of time (in one second) in generating these
pressure waves is the total acoustical power being generated.
At a given distance d from the sound source, there is aspherical surface (of
radius d) through which all of the sound energy passes. It is clear that only asmall
part of the total sound passes through one square centimeter of this area (see
Figure 2-1). This is the intensity of the sound at this distance.

Fig. 2-1
Proportion of
sound radiation
that strikes 1cm 2
of area

21
22 Chapter 2

Sound intensity is defined as the acoustical power per unit area of surface
onto which the sound strikes. Thus,

I = P/A

where
I = Intensity in watts per sq cm
P = Total acoustical power in watts
A = Area in cm 2

Numerically, this amounts to the total acoustical power being radiated,

divided by the total surface area of this sphere. (The surface of asphere of raslus
R is given by 471-R2.) Therefore, the intensity at distance dis given by

I -
47rd 2

For a steady-state, continuously generated sound, P remains constant, alnd

both 4 and ir are also constants. So it can be seen that the sound intensity at
distance dfrom the source is proportional to 1/d2.Thus, the intensity of agi en
sound diminishes as the square of its distance from the source. This is called the
Inverse Square Law. It applies to any quantity that radiates in all directions froii a
point source—to radio transmissions, light intensity, gravitational field stren h,
and so forth.

Bel and Decibel

Dynamic range is the ratio of the loudest to quietest sound intensity that can be
produced or accepted without distortion or damage to the system. In an electro ic
component such as an amplifier, the high end of this parameter is the lou st
sound that can be produced, just below overload distortion level. The low end is
the noise level. Any quieter sound is swallowed up by the internally genera ed
noise. In a good quality analog tape recorder, the dynamic range is about ne
million (10 6)to one.
The human ear is amore marvelous instrument. It has adynamic range t at
extends from the quietest audible sound intensity of 10 -16 W/cm2 to the loud st
sound just below the level that causes pain and damage to the ear. This maxim m
loudness occurs at an intensity of 10 -4 W/cm 2.So the dynamic range of he
human ear is 10 -4 to 10 -16 ,namely 10 12 .This quantity, one trillion, is amili n
times greater than that of anormal man-made machine. It exceeds the dyna ic
range of the best possible amplifier, which is 10 10 .The ear has ahundred ti es
the dynamic range of agood amplifier.
Because of the vast dynamic range of the human ear, it is more convenien i to
measure sound intensity by exponents of 10 (logarithms) than by numbers o a
 HOWARD W SAMS & COMPANY

DEAR VALUED CUSTOMER:

e
neIBIUMO011

Howard W. Sams & Company is dedicated to bringing you timely and authoritative
books for your personal and professional library. Our goal is to provide you with
excellent technical books written by the most qualified authors. You can assist us in
this endeavor by checking the box next to your particular areas of interest.
We appreciate your comments and will use the information to provide you with a
more comprehensive selection of titles.
Thank you,
Vice President, Book Publishing
Howard W. Sams & Company

COMPUTER TITLES:
Hardware Programming Languages
E Apple 140 ill Macintosh toi El CLO3 CI Pascal Los
E Commodore no E Prolog L12 E Assembly Loi
III IBM & Compatibles 114 BASIC LO2 0 HyperTalk L14
Business Applications Troubleshooting & Repair
E Word Processing JOI E Computers sos
El Data Base J04 CI Peripherals sio
El Spreadsheets J02
Operating Systems Other
E MS-DOS KO5 El OS/2 Klo Communications/Networking MO3
E CP/M E UNIX KO3 E AI/Expert Systems T18

ELECTRONICS TITLES:
E Amateur Radio TOI E Instrumentation 105
E Audio 103 E Digital Electronics 111
E Basic Electronics 120 Troubleshooting & Repair
E Basic Electricity 121 E Audio Sil E Television SO4
E Electronics Design 112 E VCR soi E Compact Disc SO2
El Electronics Projects TO4 E Automotive SO6
E Satellites 109 E Microwave Oven SO3

Other interests or comments .

Name
Title
Company
Address
City
State/Zip
Daytime Telephone No.
A Division of Macmillan, Inc.
4300 West 62nd Street Indianapolis, indiana 46268 22678
 HOWARD W SAMS
Bookmark &COMPANY

1111 1 NO POSTAGE
NECESSARY
IF MAILED
IN THE
UNITED STATES

BUSINESS REPLY CARD

FIRST CLASS PERMIT NO 1076 INDIANAPOLIS, IND

POSTAGE WILL BE PAID BY ADDRESSEE 1111111M111

HOWARD W. SAMS & CO.

ATTN: Public Relations Department
P.O. BOX 7092
Indianapolis, IN 46209-9921
Sound Measurement 23

linear scale. Therefore, the scientists at the Bell Laboratories, earlier this century,
created aunit of sound measurement based on the use of exponents (logarithms).
This unit of sound measurement is the bel which is equal to the exponent (log) of
the ratio of the intensity of a sound, to a reference intensity. The reference
intensity is the lowest threshold of human hearing, which is 10 -16 W/cm 2.So if a
sound intensity of 10 - 11 W/cm 2 were to be measured, its level with respect to the
reference value would be given by

10_11
Intensity level, L = log
16

10 -11
= 10- 11 x10+ 16 = 10 5
10 -16

So,

L = log 10 5
= 5bels

A level is usually designated by the uppercase letter L with appropriate

subscript. Intensity level would, therefore, be given by

L1 = log I bels
iref

where
= Intensity level in bels
I = Intensity in W/cm 2
'
ref = Reference intensity in W/cm 2

In practice, asmaller unit was needed, so the bel was divided into tenths. A
tenth of abel is called adecibel (dB). Using this unit, the intensity level would be
given by

L, = 10 log 1 dB
iref

Problem 2-1: What is the intensity level of asound of 10 -7 W/cm 2?

Answer: L1 = 10 log —I
'ref
10 -7
= 10 log io - 16

= 10 log 109
24 Chapter 2

= 10 x9

= 90 dB

Problem 2-2: What is the maximum safe intensity level that can be accepted
by the human ear? Intensity above 10 -4 W/cm2 causes hearing
damage.
10
Answer: L1 = 10 log -
10 - 16

= 10 log 1012

= 10 x12

= 120 dB

Power Level and Level Gain

Because sound intensity is a measure of acoustical power per unit area, and
because the unit of area is constant, it follows that intensity is directly
proportional to power. Thus, I = KxP and ' ref = K xPref' where K is the constrit
of proportionality. So,

I KP
lref = K 'ref Pref

This means that power can be evaluated as alevel, in the same way as so nd
intensity (and sound pressure). In practice, this is more useful to audio engine rs,
because modern technology utilizes the power from an amplifier to achieve he
required acoustical effect.
Not only power, but also voltage can be evaluated in dB units; in fact, the dB
unit is the most useful way for signal levels to be specified. The manufacturer fa
microphone or some other piece of audio equipment is likely to supply agraph' al
representation of the equipment's performance by plotting the output level in dB
units against a frequency base plotted horizontally. Thus, the user can see at
which frequency the output has fallen by 3dB, and also between which frequency
parameters the output is flat within, say, 2dB. 1mention 2dB because 2dB is the
smallest level change that can be detected by the human ear. A change of less
than 2dB is inaudible.
Any quantity can be evaluated in dB units by taking the logarithm of the ratio
of the value to a reference value. But we will concentrate on the evaluation of
electrical power, voltage, and pressure in dB, because these are the quantities
most often used in audio technology. To evaluate power in terms of alevel, we
start with the equation for determining intensity level.
Sound Measurement 25

= 10 log I dB
iref

We then replace the intensity ratio with the corresponding and numerically equal
power ratio.

Lp = 10 log0 — dB
r ref

This gives the power level. ' is a reference power that does not necessarily
ref

correspond to the reference intensity, because various sound transducers work at

different efficiencies. All that is required is that the reference power be
standardized. This is taken almost universally as 1mW. Therefore,

(2-1)
Lp = 10 log mw dBm

The unit is specified as dBm to indicate that the reference value is 1mW. Note
that anegative value of dBm does not mean anegative power. It means that the
level is below the reference level of 1mW.

Problem 2-3: What is the power level of 2.8 W?

Answer: Lp = 10 log
ref

= 10 log . 2.8
001

Note that the numerator and denominator of the fraction must

both be stated in the same units—in this case, watts. So,
Lp = 10 log 2800
= 10 x3.45
= 34.5 dBm

Problem 2-4: Convert apower of 0.25 mW to apower level.

.25
Answer: Lp = 10 log

Note that mW is now being used; therefore, the reference power

must also be in mW (1, not . 001 as in the previous problem).

Lp = 10 log 0.25
= 10 x (— . 602)
= — 6.02 dBm

(See supplementary problem 2-1A at the end of this chapter.)

26 Chapter 2

You may wonder why electrical power and voltage are not specified in watts
and volts. There are two reasons. First, it is much easier to calculate the effect of
passing the audio signal through anumber of audio components, if amplificatién
of each is specified in decibels. This is because decibels, being exponents, can be
added, whereas asignal voltage must be multiplied by the corresponding voltage
gains. Mathematical examples of this are given later in this chapter. The other
and perhaps most important reason why the use of dB units is preferred in audio
technology relates to the characteristics of human hearing. It happens that the
ear's response to sound intensity change is logarithmic. Because decibels are also
logarithmic, aperson's experience of changes in loudness exactly corresponds to
the dB changes; but they do not correspond linearly to the changes of acoustical
power or voltage. So asignal voltage change is misleading as to the audible result,
whereas the numerical value of the dB change precisely evaluates the subjective
effect of this change in loudness.
As mentioned, the subjective experience of a change of loudness is
logarithmic. This means that at low intensity listening levels, avery small increase
in acoustical power produces agreater experience of loudness increase than the
same power increase would produce at high intensity levels. Hence, there is a
saying that at low listening levels you can hear apin drop. The same increase of
sound would be undetectable at high listening levels.
Examples have already been given of the conversion of power in watts to a
power level in dBm units. Power gain can also be converted to acorresponding
level gain in dB units.
The power gain of an amplifier is defined as the ratio of its output power to its
input power. Because all of the amplifier's output power goes into the load, the
output power of an amplifier is often called the load power, specified as PL.Its
power gain is given by the formula

G ID L
p =
1
31

or

Po
Gp = —
Pi

where
Gp = Power gain
PL or P. = Power in the load, or output power
Pi = Input power

When you divide logarithmic numbers, their logs are subtracted. Decibels
are logarithmic values; consequently, the corresponding power level gains are
given by
Sound Measurement 27

LpG = LPL p1 (2-2)

where
LPG = Power level gain
LPL = Output power level
Lpi = Input power level

However,

PL
LPL = 10 log D
ref

and

Pi
Lpi = 10 log ' ref

Substituting these two expressions in Equation 2-2,

LPG = LET — Lpi

becomes

LpG = 10 log D L — 10 log -

D

ref ref

So,

LpG = 10 logk — log Pi )

pif

PL

Pref
= 10 log
Pi
Pref

= 10 log ( pPL f x Pre)

Pi

Therefore,

(2-3)
LpG = 10 log AF

Thus the power level gain can be found in two ways, depending on the data
available.
28 Chapter 2

Lp G = 10 log ;
IL dB

or

LPG = 10 log Gp dB

These methods require that the input and output power (or power gain) be
known. Alternatively, it can be found from the difference in levels. In this case,
the formula LpG = Lin dB can be used, provided the input and output power
levels are known.

Sound Pressure and Pressure Level

The loudness of sound can be measured not only by its intensity, but also by wilat
is called its pressure. Sound pressure is the difference between normal
atmospheric pressure and the average instantaneous magnitude of the pressurel of
the sound wave. The louder the sound, the greater the difference in pressure
between the peaks and troughs of the sound wave. The sound pressure level ( S L)
is this value converted to dB units. The lowest threshold of human hearing is
taken as the reference pressure. It is also taken as the reference intensity. The
result is that pressure level is always numerically equal to intensity level. We cl
an
state, for instance, that the lowest level of human hearing is both an intensity
level and apressure level of 0dB. The loudest safe audible sound is an intensity
level and pressure level of 120 dB.
The relationship between power and pressure is not linear. A power gain is
proportional to the square of the corresponding pressure gain. This is because
power is the product of force times distance moved per unit time. Not only dnFes
pressure gain increase the force, it also causes molecular movement to incre e.
(This is what forms the pressure wave.) So an increase in pressure causes an
increase in both force and distance moved. This is why a power increas is
proportional to the square of the pressure increase.
An intensity ratio then equals the square of the corresponding pressure ra io.
If sound pressure level is to be calculated, we start by writing the formula for
intensity level.

L1 = 10 log—, I
'ref

We then replace the intensity ratio with the numerically equal square of he
pressure ratio.

SPL (sound pressure level) = 10 log (--EL

I 2
Pref
Sound Measurement 29

But, by the laws of exponents,

log a2 = 2log a

Therefore,

SPL = 10 x2log
Pref

Consequently,

(2-4)
SPL = 20 log — dB
Pref

Note that lowercase p is used to denote pressure, while uppercase P is used to

denote power. The pressure that corresponds to the lowest threshold of human
hearing is 20 micropascals. (A pascal, Pa, is 1newton per square meter.) And the
highest safe audible sound pressure is 20 Pa. So the maximum safe sound pressure
level (SPL) is given by

20
SPLma. = 20 log 20 x10 -6

= 20 log 10 6
= 20 x6
= 120 dB

A different unit of pressure sometimes used in microphone specifications is

the bar (
1bar = 10 5 Pa). Using this unit, the lowest threshold of human hearing is
a sound pressure of 2x10 - ' 0 bar (that is, 2x10 -4 µbar). The maximum safe

sound pressure is 2x10 -4 bar, and the maximum safe SPL can be found from

4
SPL max = 20 log 2x10 -
2x10 -10

= 20 log 10 6
= 20 x6
= 120 dB

Because intensity reference and sound pressure reference (in whatever units)
are both taken as the lowest threshold of human hearing, the maximum safe levels
are identical, namely 120 dB. Similarly, the lowest threshold of audible sound is
identical in each case, namely 0dB. The advantage of working with dB units shows
itself here, in that it is only necessary to remember that audible sound parameters
extend from 0dB to 120 dB, whether the original units are in intensity or pressure.
30 Chapter 2

Most professional microphones are specified as having amaximum permissible SPL

of 135 dB. This covers the audible range of 120 dB with 15 dB clearance.
Sometimes it is helpful to use as areference point the SPL that corresponds to
1Pa or 1 ¡ bar. These can be found from the fact that SPL is calculated as

SPL = 20 log P
Pref

Therefore, 1Pa gives avalue of

SPL = 20 log 1 — 94 dB
2x10 -5

And 1µbar gives avalue of

10 -6
SPL = 20 log 2x10 - lo — 74 dB

Both of these values approximate the actual SPL occurring in a studio during
sound recording.
To summarize, the lowest threshold of human hearing is evaluated as

1. An intensity of 10 - 16 W/Cn1 2 = Liof 0dB

2. A pressure of 20 µPa = SPL of 0dB
3. A pressure of 2x10 -10 bar = SPL of 0dB

The highest safe threshold of human hearing is evaluated as

1. An intensity of 10 -4 W/cm 2 = LIof 120 dB

2. A pressure of 20 Pa = SPL of 120 dB
3. A pressure of 2x10 -4 bar = SPL of 120 dB

Voltage Level and Level Gain

Most electronic measurements are made in voltages for convenience. The
relationship between voltage and power in any load R is given by

V2 (2 5)
P = R

Thus, the power gain experienced by aload (such as aloudspeaker), in which the
power increases from P2 to P1,is given by the equation
Sound Measurement 31

(2-6)

Substituting in Equation 2-6 for P1 and P2 from the relationship in Equation

2-5, the power ratio in Equation 2-6 can be written

V12
P1
P2 = Vf

V12 R
R x V22

V12
V22

(VI) 2

V2)

Thus,

p( V I 2
2
(2-7)

P2 V2

So it can be stated that, for a constant load, the power ratio equals the voltage
ratio squared. That is,

Power Ratio = (Voltage Ratio) 2

It is now not only possible to express intensity, power, and pressure as alevel
or level gain, it is equally possible to express voltage as a voltage level or level
gain. To do so, we merely have to replace the power ratio with the square of the
voltage ratio. We start by rewriting the two equations for power level and level
gain.

Lp = 10 log -
D
LPG = 10 log —
PL
I ref

Replacing the power ratio in each equation with the corresponding square of the
voltage ratio,

Lv = 10 log \„./ LvG = 10 log (—I

V )2
vref V;
32 Chapter 2

These can be written

Lv = 20 log V Lv G = 20 log V-
V„ f Vi

Just as power level gain can be found from either

P,
10 log fL or LPL — Lp1

so voltage level can be found from either

V
20 log v L
, or LVL L Vi

The Standard Audio Circuit

In order to determine the most suitable value for it was decided to use a
V„f ,,

standard audio circuit, shown in Figure 2-2, with astandard load of 600 SI Then
the reference voltage would be the voltage that produces the reference power of
1mW in this standard load.

Fig. 2-2 The Ro

standard audio
circuit
RL
600 1
-2LOAD
1 mW POWER

Using the same mathematical relationship between voltage, power, and

resistance,

V2

P=R

we see that

V2 = PR

Therefore,

V = VPR
Sound Measurement 33

Substituting the values of P and R in the standard audio circuit load gives us

V„f = x600
V.001

= NAU
= . 775 V

Thus, V„f is usually taken as 0.775 V, and the voltage level can be found from the
formula

V
v =
L 20 log .775 dBv

(The unit is specified as dBv to indicate that the value is related to a reference
voltage of 0.775 V.)
Levels can be thought of as relating to areference value. Just as we think of a
river level being above or below its normal level, in the same way we can think of
a voltage level or power level as a number of decibels above or below the
reference level. So avoltage level of — 6dBv does not mean anegative voltage. It
means 6 dB below the reference value. Similarly, 0 dBm does not mean zero
power. It means that the power level is at the reference power level (which is
1mW).

Problem 2-5: What voltage level corresponds to 24 V?

V
Answer: L., = 20 log .775

24
20 log .775

20 log 31

= 20 x1.49

29.8 dBv

Problem 2-6: Convert 15 mV to avoltage level.

V
Answer: L
v = 20 log .775

.015
= 20 log .775

= 20 log . 0194

= 20 x(— 1.71)

= — 34.3 dBv

(See supplementary problem 2-4A.)

34 Chapter 2

dB Applications
The reference voltage is chosen to correspond to the reference power across a
standard load of 600 9. It is only in a load of this value that the voltage level is
numerically equal to the power level. It is important to remember this.

The voltage level numerically equals the power level when, and only
when, the load resistor is 600 ohms.

With any other load resistance, these two values are no longer numerically
equal. This is because the power produced by a given voltage is not the same
across different resistors. A given voltage across low resistance produces more
power than it does across high resistance. The level difference that results can be
calculated by aformula that will be derived later. Note that powers, voltages, and
gains are always multiplied or divided as shown below.

V
Power gain, Gp = and Voltage gain, Gy =
Vi

Or

PL = P xGP and VL = V;xGy

or

and

Levels, however, are added or subtracted (never multiplied or divided) as follows:

LPG = LPL - Lpi and LvG = LVL LVi

or

LPL = Lpi LPG and LVL = LVi L VG

or

Lpi = LPL — LPG and Lvi= LVL - Lv G

This is because levels are exponents. When mathematical powers of agiven base
are multiplied or divided, the exponents (logs) are added or subtracted. (See
supplementary problems 2-9A, 2-10A, and 2-11A.)
Sound Measurement 35

Problem 2-7: Find the power level gain of this amplifier:

Answer: LPG = 10 log T

PL

9.5
= 10 log

= 10 log 19

= 10 x1.28

= 12.8 dB

(See supplementary problem 2-2A.)

Problem 2-8: (a) Find the input and output voltage levels of the following
amplifier.
(b) Find the voltage level gain of the amplifier.

623 mV 8.3 V

V VL
Answer: (
a) L 1 20 log .7.¡ 5 Lv, = 20 log .75
7

.623 8.3
= 20 log .775 = 20 log .775

= 20 log 0.804 = 20 log 10.7

= 20 x (— . 095) = 20 x1.03

= — 1.9 dBy = 20.6 dBy

(b) Because both the input and output voltages and also the
input and output voltage levels are known, the voltage level
gain can be found in two ways.

Using V, and VL: Using Lvi and Lv L :

VL
LvG = 20 log — LvG = Lv L -
V,

8.3
= 20 log .623 = 20.6 — (— 1.9)
36 Chapter 2

= 20 log 13.4 = 22.5 dB

= 22.5 dB

(See supplementary problems 2-3A and 2-10A.)

If it should be necessary to convert apower level or voltage level to apower

or voltage, the relationship can be manipulated as follows:

V
Lp = 10 log . Lv = 20 log .775
001
Divide both sides by 10. Divide both sides by 20.

= log Lv = 10 „, V
10 . 001 20 . 775
Antilog both sides. Antilog both sides.
Lp p
LV V
107) = 10 20 =
.001 .775
Multiply both sides by . 001. Multiply both sides by . 775.
Lp Lv
P = . 001 X1071 V = . 775 x 11:P
where P = power in watts. where V = voltage in volts

Lp Le Lv Lv

(Note that 10 10 is the antilog of 10 ,and 10 20 is the antilog of 20 .

(See supplementary problems 2-5A and 2-7A.)

Where apower level gain or voltage level gain has to be converted to apower
gain (power ratio) or voltage gain (voltage ratio), the derivation of the required
equation is:

LPG = 10 logi L LVG = 20 log 4-

\
\
1
3,

Divide both sides by 10. Divide both sides by 20.

pG P L L vG lo V L

10 = log 71, 20 g V,
Antilog both sides. Antilog both sides.

PL LPG VL LVG
= 107-5- = 0-217
P; Vi

LPG LVG
or Gp = 10 or Gy = 10-217

(See supplementary problem 2-8A.)

Sound Measurement 37

Problem 2-9: Find the voltage that would be represented by 17 dBv.

Lv

Answer: V = . 775 x 10 2°
17
= . 775 x 10 25
85
= . 775 x 10

= . 775 x 7.08

= 5.49 V

Problem 2-10: What amplifier power gain corresponds to apower level gain of
14.5 dB?
Lp

Answer: Gp = 10 17
14.5
= io -ro-
1.45
= 10

= 28.2

(See supplementary problems 2-6A and 2-7A.)

Although the mathematical derivation of dB units appears complex, their use

is simple. It is only necessary to add them where level gains are being added, or
subtract them to find the difference between two levels. Mathematically, nothing
could be easier.

Problem 2-11: What is the voltage level at the output of this amplifier?

Answer: Lvi, = 2 + 7 = 9dBv

Problem 2-12: What input power level would produce 23 dBm at the output of
this amplifier?

Answer: 4, = 23 — 16 = 7dBm
Cheek: 7 + 16 = 23

(See supplementary problem 2-11A.)

38 Chapter 2

Problem 2-13: (
a) What is the output voltage level of the following system?
(b) What is the total voltage level gain of the system?

2.5 dBv LvL = ?

23 dB 10 dB 6 dB

Answer: (
a) LVL = 2.5 + 23 — 10 + 6 = 21.5 dBv
(b) LvG = 23 — 10 + 6 = 19 dB

Problem 2-14: What are the voltage levels at points A, B, and C in this audio
chain? The input voltage level is — 35 dBv.

— 35 dBv
30 dB 12 dB 8 dB

Answer: LvA = — 35 + 30 = — 5dBv

LvB = — 5 + 12 = 7dBv
Lvc = 7 + 8 = 15 dBv

(See supplementary problems 2-9A and 2-12A.)

Of course, it is not permissible to add decibels to volts or watts, only to dB

units such as dBv or dBm. If you must work in decibels and one of the values is
given in other units, start by converting the volts or watts to dB units.

Problem 2-15: Find the output voltage level of this amplifier.

.346
Answer: .
346 V = 20 log .,75 dBv

= — 7dBv
Therefore,
LvL = — 7 + 28 = 21 dBv
(See supplementary problems 2-14A and 2-15A.)

Sometimes it is asked, " Why are voltage level gains measured in dB rather
than dBv, and power level gains in dB rather than dBm?" The reason is that the
level gain refers to the number of decibels above any level, not just above a
 Sound Measurement 39

reference level such as . 775 V or 1mW. The unit dBv, however, specifies the level
above the reference voltage (. 775 V), and the unit dBm specifies the level above
the reference power (1mW). When level gain is given in dB, the reference level
must also be included.

Resistance Level and Level Gain

This section should be treated as a follow-up to the study of decibels. It is
suggested that the reader first review the dB formulas and work through the
supplementary problems at the end of this chapter, returning to this section only
after the concepts of power level and voltage level have been thoroughly
understood_
The power at any point in an audio chain is the power absorbed by the load
at that point. This depends on two factors— the voltage and the load resistance—
and is expressed as V2/R.
As was mentioned earlier, the power level at any point is numerically equal
to the voltage level when the load resistance is the standard value of 600 9. A
smaller value of load resistance absorbs more power for a given voltage, and a
higher value of resistance absorbs less power for the same voltage. Figure 2-3
shows the variation in power level for various loads, the voltage being held
constant at 0 dBv.

Fig. 2-3 Variation 8 -

in power level for _

loads of various
6-
resistances

4-
POWER LEVEL
LEVEL IN dB

-2

4 '
200 400 600 800 1kHz 1.2 kHz 1.4 kHz
RL IN OHMS

There is thus adifference between the numerical value of Lv and Lp for any
load other than 600 9. The difference between these levels is called the resistance
level, and is designated by LR.It is found from the equation

LR = Lv — Lp dBr (
2-8)
40 Chapter 2

In similar fashion, the power level can be found from the equation

Lp = Lv — LR dBm (2-9)

The resistance level, LR,is given by

R
LR = 10 log D
,.. ref

and

R„f = 600 9 (the standard load resistance)

Therefore,

RL (2-10)
LR = 10 log -6w dBr

The formula for evaluating LR is established as follows: Resistance level is

defined as the difference between the voltage level and power level, so it is found
from Equation 2-8 by definition. However, Lv is numerically equal to the power
level produced by this voltage across a 600 12 load. Therefore,

V2
L, = 10 log 600

and

V2
Lp = 10 log R—L

assuming Mw units are used for the power in both cases. Therefore, Equation 2-8
can be written as

V2 V2
LR = 10 log 10 log i.,
600

V2
600
= 10 log
V2

)
RL

V2 R
= 10 log x L
(600 V2

RL
= 10 log dBr
600
Sound Measurement 41

This illustrates the derivation of Equation 2-10.

To demonstrate a typical application of resistance level, let's consider the
following problem. The input resistance of an amplifier is known and its input
voltage is known. The requirement is to find the voltage level and power level at
this point.

Problem 2-16: What is the voltage level and power level at an amplifier's input
terminal? The input impedance of the amplifier is 2,400 1and
the input voltage is 5.5 V.
5.5
Answer: Voltage level, Lv = 20 log .775 = 17 dBv

Next we find LR.

LR = 10 log 2400
600 = 6dBr

From Equation 2-9, we see that

Lp = L
v- LR = 17 — 6 = 11 dBm

As acheck, we can find the power level applied to this 2,400

load as follows. First, we calculate the power absorbed by this
load when the 5.5 volts are applied.

V 2

L RL
5 .52

2400

= 12.6 mW

Then the power level that corresponds to this power is given by

Lp = 10 log
1mW

12.6
= 10 log
-i
-

= 11 dBm

Just as the term resistance level is needed to reconcile the difference between
voltage level and power level, so another term called resistance level gain is
needed to reconcile voltage level gain with power level gain. This need occurs
when the two powers develop in load resistances of different values.
42 Chapter 2

As the audio signal travels from point A to point B in an audio chain, the
voltage level rises by the amount of the voltage level that has been added. This is
called the voltage level gain between points A and B. There will also be apower
level gain between these points. Only if the load at point A equals the value of the
load resistance at point B will the power level gain equal the voltage level gain. If
the load resistances at points A and B are different, the power level gain will not
equal the voltage level gain. The difference, which is due to the different values of
the load resistances, is called the resistance level gain.
In the case of two different voltages being applied to two different resistive
loads, such as that at the input and output of an amplifier, the power at these
points is given by the following equations:

V.
2
P. =
' Ri

and

p = V L
2
•L

Then the power level difference between them is given by

LPG = 10 log —
PL
Pi

VL2

= 10 log RL
V,
2

Ri

= 10 log (VL2
RL X Vi2

x Ri
= 10 log
Vi RL

= 10 log (-\9-•2 + 10 log -1È

Vi

So,

V
LPG = 20 log —I"- 10 log
Vi rçi

Inverting the fraction changes the exponent sign. Thus,

LPG = LVG LRG dB (211)

 Sound Measurement 43

where

(2-12)
LRG = 10 log RL
— dB

LRG is called the resistance level gain.

Problem 2-17: An amplifier with an input resistance of 2,500 is feeding a

load of only 500 a The input voltage of the amplifier is 0.976 V
The load voltage is 4.89 V (see Figure 2-4). Find:
(a) The voltage level gain of the amplifier
(b) The power level gain of the amplifier

Fig. 2-4
Schematic
representation of
Problem 2-17

Answer: (
a) LvG = 20 log \
\7
.-.

= 20 log 4'
89
.976

= 14 dB

Before we can find the power level gain, it is necessary to find

the resistance level gain, and subtract that from the voltage
level gain.

LRG = 10 log
Ri

500
= 10 log 2500

= — 7dB
44 Chapter 2

(b) Now it is possible to find the power level gain from

LPG = LVG LRG

LPG = 14 — (— 7) = 21 dB

To summarize, when the load voltage is anything other than 600 9, the
power level at that point can be found by subtracting the resistance level from the
voltage level. When it is required to find the power level gain between two points
at which the load resistances are unequal, the resistance level gain must be
subtracted from the voltage level gain to obtain the correct value of power level
gain.

Summary of dB Formulas

V
Lp = 10 log 1 mW dBm Lv = 20 log .775 dBv

LPG = 10 log dB LvG = 20 log -V

\-• dB

LPG = LPL — Lp1 dB LvG = LvL — Lvi dB

Lp LV

P (in watts) = . 001 x 10 7T V (in volts) = . 775 x 10 2°

L.VG
Gp = = 1041-1 Gv =VL = 10 20
Vi

Lp — L
v - Lít where LR = 10 logRL— dBr
600

RL
LPG = LVG LRG where LRG = 10 log — dB
R;

Supplementary Problems and Answers

(Answers to alternate parts of each supplementary problem are given at the end
of this section.)
 Sound Measurement 45

Problems

2-1A. Convert the following to power levels:

(a) 3.5 W (b) 186 mW
(c) 120 W (d) 0.43 mW

2-2A. What are the power level gains of the following amplifiers?

1.3 W 25 W 520 mW 7.8 W

(a) (b)

6 mW 92 mW

(c) (d)

2-3A. What are the voltage level gains of the following amplifiers?

0.5 V I 4.5 V 250 mV 4.9 V

(a) (b)

2.5 dBv 7.5 dBv

(c) (d)

2-4A. Convert the following to voltage levels:

(a) 4.5 V (b) 120 V
(c) 0.083 V (d) 77.5 mV

2-5A. Find the power in watts that corresponds to:

(a) 25 dBm (b) 7.4 dBm
(c) — 8dBm (d) OdBm

2-6A. Find the power gain that corresponds to apower level gain of:
(a) 22 dB (b) 6dB
(c) 10 dB (d) — 3dB
46 Chapter 2

2-7A. Find the voltage that corresponds to:

(a) 35 dBv (b) 14 dBv
(c) OdBv (d) — 12 dBv

2-8A. Find the voltage gain of an amplifier that has avoltage level gain of:
(a) 7dB (b) 20 dB
(c) 26 dB (d) — 6dB
(e) 0dB

2-9A. Find the output power level of the following amplifiers:

(a) (b)

2-10A. Find the voltage level gain of the following amplifiers:

12 dBv 45 dBv 2.5 dBv 16 dBv

(a) (b)

2-11A. Find the input power level of the following amplifiers:

(a) (b)

2-12A. Find the output voltage level of the following audio chains:

—18 dBv ? dBv

22 dB 6 dB 8 dB

(a)

2 dBv ? dBv
8 dB 0 dB 26 dB

(b)
 Sound Measurement 47

2-13A. Find the output power level of the following amplifiers:

(a) (b)

2-14A. Find the input voltage level of the following amplifiers:

(a) (b)

2-15A. Find the input and output voltage levels of the following audio chains:

3.9 V

(a)

(b)

Answers

186
2-1A. (b) Lp = 10 log = 22.7 dBm
1

(d) Lp = 10 log 1
0.43 = — 3.7 dBm

7.8
2-2A. (b) LPG = 10 log 0.520 — 11.8 dB

3.4
(d) LPG = 10 log — 13.3 dB
0.160
48 Chapter 2

4.9
2-3A. (b) LvG = 20 log = 25.8 dB

(d) LyG = 16 — (- 2) = 18 dB

120
2-4A. (b) L
v = 20 log = 44 dBv
.775

(d) L
v = 20 log 0.0775
0.775 = 20 dBv

7.4
2-5A. (b) P = . 001 x10 1° = 5.50 mW

o
(d) P = . 001 x 107) = 1mW

6
2-6A. (b) Gp = 10 7) = 4

-3

(d) Gp = 10 = 0.5

14
2-7A. (b) V = . 775 x 10 20 = 3.88 V

- 12

(d) V = . 775 x 10 2° = 0.195 V

20
2-8A. (b) Gy = 10 79 = 10

-6
(d) Gy = 10 20 = 0.5

2-9A. (b) LrL = —7 + 10 = 3dBm

2-10A. (b) L vG = 16 — 2.5 = 13.5 dB

2-11A. (b) Lp, = 38 — 42 = — 4dBm

2-12A. (b) LvL = 2 + 8 + O + 26 = 36 dBv

Sound Measurement 49

2-13 A. (b) = 0.24 W

0.24
SO , = 10 log . 001 dB m

LF,
1 = 23.8 dBm

0.24 W
12 dB
Lpi = 23.8 dBm LPL = 35.8 dBm

LPL = 23.8 + 12 = 35.8 dBm

2-14A. (b) VL = 2.8V

2.8
SO , LvL = 20 log

Lv L = 11.2 dBv

— 3.8 dBv 2.8 V

15 dB
LVL = 11.2 dBv

= 11.2 — 15 = — 3.8 dBv

.0155
2-15A. (b) 15.5 mV = A level of 20 log = 34 dBv
.775

LvL = —34 + 9 = — 25 dBv

L
v, = —34 — 5 = — 39 dBv
fflamirassara----
Op-Amp Characteristics, 53

Inverting Amplifiers, 54

Non-inverting Amplifiers, 57

Stepped Gain Amplifiers, 59

Summing Amplifiers, 60

Differential Amplifiers, 63

51
 3 Operational Amplifiers

Op-Amp Characteristics
Op-amps were originally developed to perform mathematical operations—such as
multiplying, dividing, adding, and subtracting data—in the form of analog
voltages. The amplification characteristics needed for this purpose turned out to
be ideally suitable for the processing of audio and video signals. Consequently, a
new field of op-amp applications suddenly opened up, which boosted the
economic efficiency and quality of most aspects of electronics. When one
considers that adual op-amp (two amplifiers in an integrated circuit chip about
half an inch square) can be obtained for about the cost of afuse, the imagination
boggles.
The amplifier characteristics that enabled these mathematical operations to
take place are

• Extremely high open-circuit gain

• Extremely high input impedance
• Extremely low output impedance
• Wide bandwidth (including DC amplification)

In describing these characteristics, when Isay "extremely high," Imean the value
is so great that we can take the figure as infinitely high for calculation purposes.
Audio and video signal processing requires many low-power, high-quality
amplifiers such as these. You will see how these characteristics lend themselves to
signal processing in the following sections.

53
54 Chapter 3

Inverting Amplifiers
The schematic symbol for an op-amp consists of a triangular shape (used for all
types of amplifiers) with two inputs at the left and asingle output on the right (see
Figure 3-1). The DC power supply runs vertically, according to normal convention.

Fig. 3-1 DC +

Schematic symbol
for an op-amp

INVERTING INPUT

OUTPUT

NON- INVERTING INPUT

DC—

The minus sign at the inverting input does not mean that the input is of
negative voltage. Nor does this input have anything to do with the power supply.
It simply means that this is the inverting input, and that the output will be
amplified but of opposite polarity to this input. Thus, apositive input voltage here
will produce an amplified negative output. A negative voltage here will produce
an amplified positive output. And an AC input here will produce an amplified
output 180° out of phase with the input. Similarly, the plus sign at the non-
inverting input means that the resulting output will be in phase with this input.
By the addition of only three external resistors, this op-amp can be made into
an inverting amplifier having any required gain and input resistance. The
schematic for this is shown in Figure 3-2A. (The DC power supply inputs are not
shown. It is assumed that they exist.)
Due to extremely high input impedance, no measurable current enters the
input terminals. Consequently, there is no measurable current through Rg.
Because Rg conducts zero current, there is no voltage drop across it, so the non-
inverting input is at the same voltage as ground.
The open circuit gain of the differential amplifier is almost infinitely high.
Because adifferential amplifier amplifies the voltage difference between the two
input terminals, this differential voltage has to be kept extremely small to avoid
overloading. In fact, it never exceeds about 1microvolt (almost immeasurably
small), so you can always assume that the two input terminals are at the same
voltage. And because the non-inverting input terminal is at ground potential, the
inverting input terminal must also be at ground potential (within 1microvolt).
This inverting input terminal is, therefore, said to be at virtual ground. This
means ground for all calculation purposes.
 Operational Amplifiers 55

Fig. 3-2 CURRENT FLOW

Inverting op- amp

,r

VL
VIRTUAL
GROUND

V, = Input signal voltage

VL = Output signal voltage
R, = Input resistance ( or impedance)
Rf = Feedback resistance
R9 = Ground resistance (This should equal Rf.Its purpose
is to cancel any DC imbalance in the output.)

(A) Schematic

V,

TO
VIRTUAL INVERTING
GROUND INPUT

(B) Equivalent circuit

The other consequence of the extremely high input impedance is that no current
can flow in or out at the inverting input. In fact, the current flows through 11;and then
continues through Rf During the next half cycle, when the signal voltage is reversed,
the current flows in the opposite direction, back through and then back through Ili.
Rf

In effect, II; andRf form two resistors in series with each other; so the equivalent
circuit looks like that shown in Figure 3-2B. During the first half of acycle, if Viis
positive, VI,is negative, and conventional current flows downward from Vito VL.In
the second half of the cycle, when the potentials are reversed, current flow is upward.
Because the current is the same in all parts of aseries circuit, it can be stated that
56 Chapter 3

I, = I
f

where
I, = Current in R,
If = Current in Rf

By Ohm's Law,

V VL
Ri Rf

From this equation, we arrive at

VL R
Vi Ri

But by definition,

V
= Voltage gain, Gv
V,

Therefore,

G Rf
Ri

Thus, the voltage gain of an inverting op-amp is controlled entirely by the

resistance ratio between Rfand Ri.
In addition to being able to determine the voltage gain of an op-amp, the
input impedance can also be predetermined. The input impedance, as you know,
is the effective impedance (or resistance) between the input terminal and ground.
However, the inverting input of this amplifier is at virtual ground. Because R, is
the effective resistance between the input of the circuit and this virtual ground, R,
acts as the input impedance. Consequently, you can choose whatever input
impedance you desire simply by making R, that value. It is easy, therefore, to
construct an op-amp of any required input impedance and gain, as shown in
Problem 3-1.

Problem 3-1: It is required to make a pre-amplifier with a voltage gain of 50

and an input impedance of 8lat.
Answer: R, is fixed at 8kl by the input impedance requirement.
Because

Gv = TFR,
 Operational Amplifiers 57

then

Rf Gy Ri

and

Rf 50 x8000
= 400 ke

The final circuit is shown in Figure 3-3.

Fig. 3-3 400 k52

Pre-amplifier with
voltage gain of 50
and input
impedance of
8Id/

Non-inverting Amplifiers
By applying the incoming signal to the non-inverting input, anon-inverting amplifier
can be made. This circuit is illustrated in Figure 3-4A. In this type of amplifier, the
non-inverting input is not held at ground potential as was the case with the inverting
amplifier. In addition, there is no virtual ground. In fact, both inputs (and the whole
amplifier) are at the AC voltage of the incoming signal, oscillating with the amplitude
and frequency of the signal. This can be seen from the equivalent circuit shown in
Figure 3-4B.
Again, the current flows through Riand Rfin series. Because the amplifier is non-
inverting, the input signal is in phase with the output. So both signal voltages are
positive together or negative together. The gain of this amplifier can be found by
considering the equivalent circuit shown in Figure 3-4B. The current being the same
in all parts of aseries circuit,

=I
T

where
= Current in Ri
I
T = Total circuit current
58 Chapter 3

CURRENT FLOW
Fig. 3-4
Non-inverting
op-amp

(A) Schematic

TO
NON- INVERTING
INPUT

(B) Equivalent circuit

V
Writing Ias R'
— the above equation becomes

V; VL
Ri R; + Rf

So,

VL _ R, + Rf
V1 — R,
 Operational Amplifiers 59

or,

V Rf
= 1 +
V,

However,

VL
V, =
— Gain, Gv

So the gain of anon-inverting op-amp can be found from

As in the previous amplifier, both gain and input impedance are determined
by the external resistance values. Because the input impedance is the effective
resistance between the incoming signal and ground, and because the incoming
signal voltage occurs at both the inverting and non-inverting inputs, it follows
that the input impedance is made up of the two resistors R, and Rg in parallel. (If
the gain is high, Rg is large compared with R„ so the input resistance
approximates R1.) In general, the input impedance of anon-inverting op-amp can
be found from

= R, Rg
Z,
R. + R

Note that the gain of anon-inverting op-amp can never be less than unity. So
this amplifier configuration can only amplify and never attenuate, whereas an
inverting op-amp can both amplify and attenuate asignal.

Stepped Gain Amplifiers

Because the gain of an inverting op-amp can be entirely determined by the ratio
of the two resistors (Rf/R,) and because this amplifier can both amplify and
attenuate, an inverting op-amp is often used where stepped gain amplification is
required—for example, in situations such as the vertical input of an oscilloscope
or the range control of aVTVM. In both of these situations, it is required to alter
the voltage gain at the input terminal by predetermined steps, as opposed to the
continuously variable requirement needed by avolume control.
Take as an example the vertical input of an oscilloscope. The vertical
amplifier of this equipment is designed to supply a voltage to the vertical
deflection plates of the cathode ray tube, which causes the trace to move up or
60 Chapter 3

down, depending on the applied input voltage. Suppose this vertical amplifier
were calibrated to produce asensitivity of 0.01 V per centimeter deflection. In this
case, an acceptable trace would appear from input voltages between — . 05 V and
+ . 05 V. (Each would produce a5cm deflection either upward or downward.)
But suppose the input voltages were very small, in the region of 5mV Then
the deflection would be hardly visible. In this case, arange control switch would
be needed. When set to the 1mV range, it would amplify the input by 10, thus
converting the 5 mV input to . 05 V at the input of the vertical amplifier, and
producing an acceptable trace of 5 cm deflection. Similarly, if the input voltage
happened to be large, in the region of 500 V, the range control would have to
attenuate this down to . 05 V, or most of the trace would be lost above and below
the screen.
Assume that the sensitivity of the vertical amplifier produces adeflection of 1
cm for an input of . 01 V, its sensitivity being . 01 V/cm (or 10 mV/cm). Assume also
that the required input ranges are 1mV, 10 mV, 100 mV, 1V, 10 V, and 100 V each
per centimeter. This could be achieved quite simply. The required gains would be,
from the lowest to the highest, 10, 1, . 1, . 01, . 001, and . 0001. The low input voltage
obviously requires gain, and the high input voltage requires attenuation. Now
assume that the input impedance of this oscilloscope has to be high, say 2 MI-
2.
This would be the value of Rf in our stepped gain inverting op-amp. Then the
required circuit resistances could be found in the following way:

Rf

UV =

So,

Rf = 12 1Gy

But R, is already determined by the input impedance requirement of 2Me.

So,

Rf = 2MO x Gy

For each gain listed, it is therefore possible to calculate the required feedback
resistor from this formula simply by multiplying the gain by 2 MO. The resulting
range setting, required gain, and feedback resistors are listed in Table 3-1. The
first value of is given by
Rf

Rf = 2MO x 10 = 20 MO

The schematic of astepped gain op-amp that could be used in this situation is
shown in Figure 3-5. It can be seen that as the moving contact of the switch is
connected to the various feedback resistors, the gain is changed in discrete steps.
At each setting, the gain consists of the actual value of R1/R 1.R, is constant. The
 Operational Amplifiers 61

six different values of Rf produce the required six different gains needed for the
various input range settings.

Table 3-1
Feedback resistors Input Range Required Gain Feedback Resistance
needed for 1mV 10 20 MO
vertical sensitivity
10 mV 1 2M11
control of an
oscilloscope 100 mV .1 200 k9
1V . 01 20 1(0
10 V . 001 2klt
100 V . 0001 200 1
/
.

20 MO,
Fig. 3-5 Stepped
gain op-amp

Summing Amplifiers
One of the situations where the audio and video application of op-amps is most
often used is in the summing amplifier. This forms the basic mixer circuit. It is
particularly advantageous because at the inverting input (which is used as the
summing point), the voltage is at virtual ground. Consequently, there is no
feedback to other channels. The summing amplifier circuit is shown in Figure 3-6.
62 Chapter 3

Fig. 3-6 SUMMING Rf

Summing op-amp R1
POINT

R2

R3

The theoretical basis for the summing action of this amplifier can be seen by
application of Kirchhoff's Current Law. Kirchhoff states that at any point in a
circuit, the total current entering equals the total current leaving. Applying this
principle to the summing point in Figure 3-6, the current entering this point is
through the input resistors RI,R2, and R3. All current leaving travels through R.
Therefore,

11+ 12 + 13 = If

By Ohm's Law,

V I + V2 + V 3 VL
R1 R2 R3 Rf

Thus,

VL = VI (Pi) + V2 (P 2) + V 3 (R
1 )
1 3

For purely mathematical summing, the input resistors are all of the same value.
Call this value Ri.Then the equation becomes

VL =v. () + v2 + v3 -
ffy (
R

Factoring, this produces the following equation:

(3-1)
VL = Ri(V 1 + V2 + V3)

This illustrates the summing function. Of course, any number of inputs can be
 Operational Amplifiers 63

faders. A typical four-input mixer circuit is shown in Figure 3-7. R1through R4 are
input faders. R„, is the master gain.
Rf is the feedback resistor. And Riis the value
of the input resistors needed to limit the gain to the required maximum level.

Fig. 3-7 Mixer V1 • Ri

circuit
R1

Ft,

V2 RI

R2

V3 Ri VL

R3
R,

f R,
V4 RI

R4

Differential Amplifiers
The fact that the op-amp has an inverting and anon-inverting input characterizes
it as a differential amplifier. This means that it must amplify the voltage
differential between the two inputs. For example, suppose the gain for both inputs
were 5. An input of + 1V at the inverting input would then produce an output of
—5V. A similar input of + 1V at the non-inverting input would produce an output
of + 5V. These two outputs, namely — 5V and + 5V, would cancel each other out.
So it is clear that an identical voltage at both inputs produces no output voltage.
Only adifference between input voltages produces any output.
In the case of an op-amp, its extremely high open-circuit gain necessitates a
reduction in gain by means of negative feedback. If this is not done, the amplifier
will become unstable and give out acontinuous howl. Thus, the basic differential
amplifier circuit looks like that in Figure 3-8.
If this amplifier is designed to give equal weight to both inputs, the two input
resistors Rishould be of equal value. For optimum DC balance, Rg should equal R.
Then the resulting output voltage will be given by

(3-2)
VL = -17!(
V2 — V1)
64 Chapter 3

Fig. 3-8 Rf

Differential
amplifier

Because Rf/Ri represents the gain factor, it can be seen that the output voltage is
equal to the gain times the difference between the input voltages. This illustrates the
amplifier's differential characteristic. The mathematical derivation of Equation 3-2 is
given in the appendix at the end of this book.
The differential amplifier has many useful applications. Perhaps one of the
most useful is its ability to filter noise from a balanced line input. Low-level
signals, such microphone signals, are most susceptible to noise. This is because
noise voltages are at nearly the same level as the mic signal. The audio signal
being fed to aloudspeaker is normally in the range of 5to 10 volts; anoise voltage
of 7 millivolts represents a negligible proportion of this signal. However, 7
millivolts of noise will cause a disastrous amount of interference with a 15 mV
microphone signal.
An unbalanced line uses two conductors. The balanced line system uses three
conductors. One carries the ground of 0 volts. The other two conductors carry
opposing polarities equally balanced on each side of ground potential. Figure 3-9
illustrates the difference between an unbalanced and abalanced line. Note that
both carry the same signal. The two balanced conductors each carry half of the
waveform amplitude. These add across the differential input terminals to give a
full waveform input.
Let us assume that a microphone line is receiving noise interference in the
form of low-frequency hum. The required signal and noise can be graphed
individually as shown in Figure 3-10A and B. When combined in an unbalanced
line and passed through a non-inverting amplifier, the results are as shown in
Figure 3-10C. Note that the signal and noise have both been amplified equally. If
this same microphone signal were carried by abalanced line (which is achieved
by keeping the ground conductor separate from either of the signal conductors)
and if the amplifier were a differential amplifier, then the waveforms would
appear as in Figure 3-10D.
You can see that the low-frequency noise is in phase in both conductors, so it
produces no differential voltage across the input terminals. Therefore, it produces
no output voltage. Only the required signal (opposite in phase on each conductor)
 Operational Amplifiers 65

Fig. 3-9 Signal HOT LEAD

voltages applied VOLTAGE
to an unbalanced
and abalanced
line
GROUND

(A) Unbalanced line (using two conductors)

NON- INVERTING
CONDUCTOR
VOLTAGE

GROUND
VOLTAGE

INVERTING
CONDUCTOR
VOLTAGE

GROUND
VOLTAGE

(B) Balanced line (using three conductors)

Fig. 3-10 Noise

amplification vs.
noise filtering
(A) Required signal

A.AAsevvvi
(B) Noise interference

(C) Unbalanced signal and noise amplification

(D) Balanced line—signal amplification and noise elimination

66 Chapter 3

has been amplified. This demonstrates one of the great advantages that can be
obtained from adifferential amplifier. It is called common mode rejection, and
amounts to auseful ability to eliminate noise interference.
Comparatively few of the many applications of op-amps have been discussed
so far. This is because it is necessary first to clarify the principle characteristics of
op-amps. In fact, op-amps are used in audio and video technology for many
aspects of signal processing. For instance, they are used as pre-amplifiers, line
amplifiers, mixers, equalizers, compressors, expanders, active filters, tone
controls, analog comparators and buffers in digital systems, digital-to-analog
converters, and so forth. The list goes on and on. Some of these applications will
be mentioned later. At this stage, it is only necessary to understand the
fundamentals of op-amp characteristics and behavior. Once this has been
understood, it is then possible to troubleshoot or construct audio components with
awareness of the principles behind their applications.
Filtering Concepts, 69

Low- Pass Passive Filters, 72

High-Pass Passive Filters, 79

The Pole/Zero Approach, 83

Bandpass Filters, 88

Stopband Filters, 91

67
 4 Passive Filters

Filtering Concepts
The action of a filter is to exclude certain objects or characteristics, while
allowing others to pass. In the realm of electronics, afilter discriminates between
frequencies. A low-pass filter passes the low frequencies, while it filters out the
high. A high-pass filter does the reverse. With a little ingenuity, more
sophisticated filters can be made that filter out one band of frequencies or allow
only alimited band to pass. Others, called active filters, can not only attenuate,
but also amplify. A tone control or equalizer is of this nature.
However, there is an altogether different group of filters that have to be dealt
with by the audio engineer. These are the ones that we don't want, but can't
avoid. They are the troublemakers. And it is the need to be able to deal with these
that requires us to understand something about the theory of filtering.
Unintentional filters can result from such things as the reactance of the coil in
amagnetic tape head or the capacitance in along microphone cable. They can
destroy the quality of arecording or sound reinforcement system by cutting out
frequencies that should be present. It is important to be able to counteract the
effect of these filters or at least to confine it to arange of frequencies where it will
be harmless.
All filtering results from the action of reactive components such as inductors
(coils) or capacitors. The reactance of these components to AC voltages is
frequency dependent. This means that their opposition to current varies,
depending on its frequency. Remember that capacitive reactance (a capacitor's
opposition to current) is inversely proportional to frequency. The higher the
frequency, the less the capacitor opposes current, in accordance with the formula,

69
70 Chapter 4

1
Xe 27rfC

The reactance of an inductor, on the other hand, is exactly opposite, as given

by the formula,

XL = 27rfL

This means that the higher the frequency, the more it opposes current. These two
facts hold the whole key to filtering theory, so they are important to remember.

• A capacitor passes high frequencies, but blocks low.

• An inductor passes low frequencies, but blocks high.

With this knowledge, it is possible to look at any filtering circuit and tell
instantly what sort of filter it is and what it is likely to do at different frequencies.
Of course, it requires mathematical analysis to pinpoint the quantitative effect.
But it is easy to obtain ageneral picture, and that is helpful.
Both acapacitor and an inductor can be used in conjunction to produce an
increased filtering effect. However, at a certain frequency, when both of their
reactances are numerically equal, they will go into electronic resonance. This can
be used under certain circumstances—for example, to achieve atuned circuit—
because at their resonant frequency they can greatly amplify an applied signal.
This is how aradio or TV is tuned. But they do not form what is called aprecision
filter, such as alow- or high-pass filter, in which the effect is linearly proportional
to frequency. And this is what is required for abass and treble tone control.
In fact, a precision filter can be made out of a single reactive component,
either a capacitor or an inductor in conjunction with resistors. The reactive
component chosen is almost always acapacitor. Hence, precision audio filters are
mostly RC filters containing only resistors and capacitors. The reason why
capacitors are chosen is two-fold. First, they are cheaper than inductors. But more
important, an inductor progressively becomes less efficient as frequency drops,
due to the fact that the resistance of the winding becomes more significant than
the reactance of the coil at lower frequencies. Having said this, we will now
ignore any imperfections in filtering components and look at the theoretical basis
of filtering. This is best seen by examination of simple low-pass and high-pass
circuits.
Most RC filters can be thought of as forming avoltage divider made up of two
impedances, Z1and Z2 ( see Figure 4-1). The output (filtered) voltage is that which
develops across Z2. The ratio of output voltage to input (source) voltage, Vo/V i,is
called the transfer function, designated by the letters H(f). The voltage gain, Gv ,
of an amplifier is also the ratio Vo/Vi.The difference is that the transfer function
is frequency dependent, whereas the gain of an amplifier is constant at all
frequencies.
 Passive Filters 71

• Amplifier gain is given by

• Transfer function is given by

H(f) =

The suffix (f) indicates that the value of H is afunction of f, the frequency. That
means it will be different at different frequencies.

Fig. 4-1 Voltage

divider

In avoltage divider, as in Figure 4-1, the ratio of the voltages is equal to the
ratio of the impedances. (This results from the Voltage Proportionality Law.) So
the transfer function, from Figure 4-1, is given by

Vo Z2

El(f) =Vi = 274--Z2

The half arrow over Z1 + Z2 means that these two impedances must be
added vectorially, taking their magnitude and phase (relative direction) into
account. Because they are at right angles in an RC circuit, they cannot be added
algebraically.
To simplify any phase angle problems, we will use the complex frequency
variable S when calculating capacitive reactance. This takes phase angle into
account automatically and solves all phase related problems. In this way, the
phase angle will be included in the frequency terms. Sis defined as follows:

S = jcir
72 Chapter 4

where
S = Complex frequency variable
j = The imaginary number (= 1), indicating change of dimension or
phase angle of + 90°
= Radial frequency, the number of radians turned per second (= 27j-f,
where fis the number of cycles per second)

So, the radial frequency phasor Sis given by

S =

But,

w = 27rf

So,

S = j27rf

Now, capacitive reactance is given by

—
X e = 27rfC

(being in the — jdirection with respect to resistance).

So,

1
X = . (because — j = 1/j).
c j27rfC

But,

j2rf = S

So,

We will use 1/SC for X, in future, to simplify all calculations.

Low-Pass Passive Filters

A single element LP (low-pass) or HP (high-pass) filter is made up of a single
resistor and capacitor forming a voltage divider. It is possible to visualize the
 Passive Filters 73

action of these filters by remembering that capacitive reactance increases as

frequency falls, and that voltage across the capacitor follows the capacitive
reactance. Thus, Figure 4-2A represents an LP filter, because its output voltage
increases at low frequencies. Similarly, Figure 4-2B represents an HP filter,
because the voltage lost across the capacitor falls as frequency rises. This leaves
more voltage across the resistor at high frequencies.

Fig. 4-2 Single

element filters

(A)Low-pass filter (B) High-pass filter

The transfer function of the LP filter is given by

Vo Xc 1 /SC
H(S) = = R + ( 1/SC)
v R + Xc =

Multiplying numerator and denominator by SC,

1 (4 -
1)
H(S) —
SRC + 1

But in an RC series circuit, the time constant, designated by the Greek letter T
(pronounced Tau) is given by

T = RC

Substituting this in Equation 4-1, the standard form for the transfer function of an
LP filter becomes

1 (4-2)
H(S) —
ST + 1

It is useful to note that the term (ST + 1) is the mathematical representation

of all single element filters. It demonstrates that the time constant of the circuit is
the key element in determining the frequency at which it takes effect. But more
about this later.
74 Chapter 4

From the standard form of an LP filter, namely H(S) = 1/(ST + 1), it can he
seen that at very low frequencies, when the radial frequency S is very low, the
term ST becomes negligible. Then the transfer function becomes

H(0) — 1 1
+ 1 — 1 — .1

Thus, there is no loss of signal voltage and the graph of output plotted on a
frequency base is a horizontal line.
However, when we consider the high-frequency response, when S is very
large, the term ST becomes so large compared with the added 1 in the
denominator that the 1 becomes negligible. So the high-frequency transfer
function becomes

1 1
H(co) = ST + O = ST

This represents agraph in which the value is inversely proportional to frequency.

If S doubles, 1/ST halves, and so on. The low- and high-frequency responses of
this filter, as alevel change (in dB units), are illustrated in Figure 4-3.

Fig. 4-3 Limiting LVG LVG

response of an LP
filter

(A) LF response (B) HF response

Of course, the actual filter response doesn't suddenly change from aperfectly
straight horizontal line to a straight downward sloping line. At intermediate
frequencies, there is a curve joining these two limiting responses. However, the
two lines illustrated in Figure 4-3 show exactly what happens at the low- and high-
frequency limits. They also form what are called asymptotes. An asymptote is a
straight line which acurve ever more closely approaches, but never quite reaches.
Asymptotes are useful because they form agraphical structure within which the
filter curve fits. The downward slope can be found from the fact that the output
voltage halves its value at each higher octave (double the frequency). The
corresponding voltage level change per octave (since the voltage ratio is 1
2 )is
/
given by
Passive Filters 75

Lv G = 20 log '/2

= 20 x (— . 3)

= — 6dB/octave

It is of fundamental significance that the output of this filter tends to a — 6

dB/octave slope as the frequency rises. But this is only valid at high frequencies.
At low frequencies, the graph becomes a horizontal line.
The next, and equally important, characteristic of this filter is the frequency
at which the transition occurs between the horizontal and sloping lines. This can
be found by extending the two lines until they meet in acorner. At that point, the
filter is said to break, so this is called the break frequency, or corner frequency.
Filters with different value components can be made to break at different
frequencies. Another way to look at it is this: It is only when the frequency has
risen to ahigh enough level to reduce X, to avalue comparable to that of R, in the
filter circuit, that filtering takes place. In fact, this frequency, the break frequency,
is specified as occurring when X, is numerically equal to R. Then,

R = X, =1—
SIC

SI being the frequency at which this equality occurs). And so,

(

RC is the time constant of an RC series circuit, namely T. So,

1
=

But in general, Sis numerically equal to 27rf. Thus,

1
27rf i =

Therefore,

1 (4-3)

fi = 27rTi

where
f1 = Break frequency in hertz

T1 = Time constant of the circuit in seconds

76 Chapter 4

The break frequency is found from the circuit time constant, and the circuit
time constant from the appropriate combination of the circuit component values.
Thus, analyzing filtering circuits simplifies itself into finding the time constants of
the filtering elements. From each time constant we can find the corresponding
break frequency from

1 1
f f2 = etc.
271-7
-1'

There is a problem in evaluating the transfer function of a filter at any

frequency f, because the frequency variable in our standard form is a complex
frequency term. As stated earlier, S = jw. Also, the time constant is defined as the
period that corresponds to the radial frequency at which R = X. Therefore,

1
7 1= -
071

Substituting these identities—S = jo.) and T1 = 1/(0 1—we obtain

(ST 1 + 1) = Lt) + 1
co l

Because the jterm is at right angles to the number term, we can evaluate by
Pythagoras' theorem. Or, we can convert from rectangular form to polar form, as
follows:

w
(ST, + 1) = ,\/(— + 1

It is more convenient to use cyclical rather than radial frequencies; however, the
frequency ratios are interchangeable, the 2r conversion factors canceling out.
Thus,

(ST 1 + 1) = f +1

where
f = A given frequency in Hz
f1 = Break frequency in Hz

Any filtering term can, therefore, be evaluated by converting each (STr, + 1)

term to

,‘N,)2
Passive Filters 77

It is often more convenient to find the effect of afilter in terms of voltage level
change, rather than as voltage gain. This can easily be achieved because H(f) is a
voltage ratio; therefore, the corresponding level gain is given by

LvG = 20 log H(f) (

4-4)

Our LP filter transfer function from Equation 4-2 can now be written as a
level change.

(4-5)
LvG = 20 log 1 dB
+ 1

As mentioned earlier, between the horizontal asymptote in Figure 4-3 and

the sloping asymptote, there is acurve that bridges the transition. Thus, the level
at break frequency will be somewhat below its maximum value. To find out how
much this level will drop, we use Equation 4-5 to evaluate the level change at
break frequency. At this frequency, fis specified as equal to f l,the break

frequency. Then, Equation 4-5 becomes

1
Lv G = 20 log

1
= 20 log
NÍF

= 20 x(— 0.1505)

= — 3dB

This 3dB drop at break frequency is characteristic of all single element filters.
It is now possible to plot the output response level of asingle element LP filter,
and compare the actual response with the LF and HF asymptotes (see Figure 4-4).
The vertical axis is the voltage level gain. The frequency base is scaled in octaves.
This is logarithmic base 2, so each equal increment (representing one octave
increase) is twice the frequency of the previous increment.
Notice that the point at which the horizontal and sloping lines meet identifies
the break frequency. The 3 dB down level has special significance. It is
conventionally taken as the cutoff point. By this Imean that any signal that has
fallen by 3dB or more is said to have been cut off. (While — 3dB is not all that
much of an attenuation, the line has to be drawn somewhere, and this is where it
78 Chapter 4

Fig. 4-4 Typical LVG

LP filter response

0 dB

ACTUAL
FILTER
RESPONSE

TENDS TOWARD
—6 dB/OCTAVE SLOPE

fi
BREAK
FREQUENCY

is drawn.) So, in audio technology, output signals at 3 dB or more below the

maximum value are said to have been cut off. In this LP filter, all frequencies
above the break frequency are cut off. In this type of filter, the break frequency is
often called the cutoff frequency or, as noted previously, the corner frequency.
It is now possible to calculate the response of an actual filter at various
frequencies, as shown in Problem 4-1.

Problem 4-1: In the filter circuit shown in Figure 4-5, find:

(a) The break frequency
(b) The response level at break frequency
(c) The response level at 23 kHz

Fig. 4-5 Filtering

circuit specified in
Problem 4-1

o o

Answer: (
a) To find the break frequency, use Equation 4-3.

1
2.7rT i

In an RC series circuit,

= RC
Passive Filters 79

From these component values,

r1 = (6x10 3)x (0.0044 x10 -6 ) = 24 µsec

So,

1
f1 = 2ir x24 x10 -6

Thus, the break frequency is given by

f
1 = 6.63 kHz

(b) To find the response level at break frequency (when f =

6.63 kHz), use Equation 4-5.

1
LvG = 20 log
6.63) 2
6.63 + 1
-\/
1
= 20 log
V2

= - 3dB

(e) To find the response level at 23 kHz, use Equation 4-5 again.

1
Lv G = 20 log

-\.
/(::3) 2 + I

= - 11.15 dB

Summary of RC Low- Pass Filter Characteristics

1. The break frequency (above which the filtering action becomes apparent) is
inversely proportional to the time constant, which equals the product of RC. It is
given by

1
f1 -
27r7,
or
80 Chapter 4

1
f - 27RC

where
f, = Break frequency in Hz
R = Resistance in ohms
C = Capacitance in farads
71 = Time constant in seconds

2. At break frequency, the level is 3dB below its maximum value. All frequencies
above this are cut off.
3. At high frequencies, the filter's response falls off at arate approaching — 6dB/
octave.

High-Pass Passive Filters

An HP filter circuit is shown in Figure 4-6. To understand how this circuit passes
high frequencies while filtering out low, it is only necessary to remember that, at
very high frequencies, the capacitive reactance becomes extremely small; there-
fore, the capacitor acts as a shorting link, then all input voltage passes to the
output. At very low frequencies, the capacitive reactance becomes very large,
much larger than the resistor; therefore, the majority of the voltage is lost across
the capacitor. Only the small portion that develops across the resistor reaches the
output. It can, therefore, be seen that this filter passes the high frequencies and
filters out the low.

Fig. 4-6 HP
passive filter

The transfer function, by definition, is equal to the voltage ratio, Vo/V i.This,
in turn, is equal to the impedence ratio, which is given by

H(S) —

R
Passive Filters 81

Multiplying numerator and denominator by SC,

SRC
H(S) —
1 + SRC

But, in an RC series circuit, RC = T; therefore,

H(S) — ST (4-6)
1 + ST

This is the standard form for an HP filter transfer function. For calculation
purposes it is convenient to divide the numerator and denominator by ST. Then,

As we did before, we can substitute jw for S, and co i for 1/Ti.The function then
becomes

1
H(c.0) —
+1
1(0

We then convert from rectangular form to polar form, to obtain the numerical
value.

1
H(w) —
j(01 )2 + 1

Replacing the radial frequency ratio with the cyclical frequency ratio,

H(f) — 1

gives amethod for calculating the transfer function at any frequency, f. As alevel
change, this can be found by taking 20 times the log of the voltage ratio. So,

(4-7)
1
Lv G = 20 log

,\k -fi
f +
82 Chapter 4

The level at break frequency can be found, just as it was for the LP filter. Then f =
f
1 and the equation becomes

LvG = 20 lo g 1 — 3dB
V2

So the level at break frequency, LvG = — 3dB.

The limiting slope of this filter occurs at the lowest frequency (when S
approaches zero). Then, in the equation

ST
H(S) —
(1 + ST)

the ST term in the denominator becomes negligible compared to the added 1. So

the effective LF transfer function becomes

H(0) =ST
— = ST
1

This means that at each higher octave (double the frequency), the value of the
transfer function doubles. This gives avoltage increase per octave of 2to 1, again
of 2. The corresponding level gain is given by

LvG = 20 log 2 = 20 x0.3 = + 6dB/octave

Just as in the LP filter, the break frequency is defined as the frequency at

which the resistance equals the reactance. Looked at as a concept, this means
that the filter has no effect, while the capacitor acts as ashorting link. Only when
the frequency has fallen sufficiently for the reactance to be comparable (in fact,
equal) to the resistance, does this circuit start to filter out the lower frequencies.
This happens when

R = Xc

or

1
R=
27rf

So,

1
f — 271-RC
 Passive Filters 83

But,

RC = T

Therefore,

A typical HP filter response plotted on a frequency base scaled in octaves is

shown in Figure 4-7.

Fig. 4-7 Typical -VG

HP filter response

0 dB

3 dB

ACTUAL FILTER
RESPONSE

TENDS TOWARD
+ 6 dB/OCTAVE SLOPE

fi
BREAK
FREQUENCY

Summary of RC High- Pass Filter Characteristics

1. The break frequency (below which the filtering action becomes apparent) is
inversely proportional to the time constant, which equals the product of RC. It is
given by

fi - 2rT,
or
1
= 2rRC

2. At break frequency, the level is 3 dB below its maximum value. All frequencies
below this are cut off.
S.1 Chapter 4

3. At low frequencies, the filter's response increases at arate approaching + 6dB/

octave as frequency rises.
4. As ageneral rule, asingle element filter has alimiting slope of 6dB/octave. This
is the same as 20 dB/decade (decade means 10 times the frequency).

The Pole/Zero Approach

It is clear that the reactive component (the capacitor in an RC filter or the inductor
in an RL filter) is the one that produces the filtering effect. The pole/zero
approach derives from the reactive response to frequency change. This response
starts from zero frequency and rises if it is an upward slope, or falls if it is a
downward slope, at arate of ± 6dB/octave. However, because each lower octave
is half the frequency of the previous one, and because you can go on halving a
value forever before reaching zero, an interesting situation arises.
The HP filtering effect has to be considered as starting at — co dB (called a
zero, being at the lowest possible level) at zero frequency, and rising by 6dB at
each higher octave. The LP filtering effect has to be thought of as starting at + co
dB (called apole, being at the highest level) at zero frequency and dropping by
6dB at each higher octave. The filtering effects of asingle filtering element can,
therefore, be visualized from the graphs in Figure 4-8.

Fig. 4-8 Reactive LVG L

VG
effects of filtering

(A) LP filter reactive effect (8) HP filter reactive effect

Now let us consider separately the frequency dependent effect of the resistive
and reactive components of these filters. The resistive component is unaffected by
frequency; hence, it can be represented graphically by a horizontal line. The
reactive component produces either an upward or downward sloping straight line
of 6dB/octave, as previously indicated. These two aspects of afilter can be seen
in Figure 4-9, separated into their respective components. The frequency at which
 Passive Filters 85

Fig. 4-9 Resistive L„

and reactive
aspects of afilter RESISTIVE EFFECT RESISTIVE EFFECT

v•

1 t
BREAK BREAK
FREQUENCY FREQUENCY

(A) LP filter (B) HP filter

the horizontal and sloping lines intersect, as has already been mentioned, is
called the break frequency.
Any filter, however complex, can be graphically represented by a number of
horizontal and sloping lines. The horizontal lines represent purely resistive effects and
the sloping lines purely reactive effects. A graph plotted in this way is called aBode
plot, after the name of the man who invented it. The advantage of this construction is
that it clearly identifies the break frequencies. It also forms a simplified linear
structure into which the actual graph of the filter fits. The relationship between the
Bode plot and the actual filter response is shown in Figure 4-10.

Fig. 4-10 Bode LVG I

-VG
plot vs. actual filter
response BODE PLOT BODE PLOT
\I I

/ / e\
ACTUAL
/ FILTER
/

/
i
fl f2

(A) LP filter (B) HP filter

In a more complex filter with many filtering elements, such as a bandpass

filter, the relationship between the Bode plot and the actual filter looks like
Figure 4-11.
At break frequencies, there tends to be a3dB difference between the actual
filter response and the Bode plot. Elsewhere, the graphs become progressively
closer. Another advantage of the Bode plot is that it clearly isolates the reactive
86 Chapter 4

Fig. 4-11 Complex Lv G

filter showing
actual response BODE PLOT
and Bode plot
••••••••

ACTUAL
FILTER

.0e

effect of the filter from the resistive effect. Consequently, it is easy to understand
all filtering concepts. It is only necessary to visualize each filtering element as a
reactive effect causing a slope of 6 dB/octave, starting from zero frequency.
However, these do not manifest until their corresponding break frequencies
occur. We can see how this takes place by adding to the graph the full reactive
effect of each filtering element, starting from zero frequency (see Figure 4-12).
The effect of a pole is to reduce the slope of the Bode plot by — 6 dB/octave
where it strikes the graph. The effect of azero is to increase the slope of the Bode
plot by + 6dB/octave where it strikes the graph.
A steeper cutoff could be produced by having two or more poles or two or
more zeros at the same break frequency. Thus, two poles with identical break
frequencies would turn down the graph by — 12 dB/octave. Similarly, two identi-
cal zeros would turn the graph up by + 12 dB/octave at their break frequency.
This effect is illustrated in Figure 4-13.
Any filter, however complex, can also be represented by a combination of
four filtering terms and aconstant. The constant, often labeled K, is also the DC
gain, because it has apurely resistive effect. It is independent of frequency and is,
therefore, represented graphically by a horizontal straight line. The reactive
filtering terms are as follows:

ST An upward sloping straight line of + 6dB/octave

1/ST A downward sloping straight line of — 6dB/octave
(ST, + 1) A zero producing achange in slope of + 6dB/octave
1/(S7d + 1) A pole producing achange in slope of — 6dB/octave

From this list, anumerator term of the form (S7, + 1) is called azero of the
function. At its break frequency (which can be found from its time constant 7,,
 Passive Filters 87

Fig. 4-12 Reactive \ — 6 dB/OCTAVE SLOPES STARTING FROM

effect of each \ + co dB ( A POLE )AT ZERO FREQUENCY
filtering element

LVG

POLES

ZEROS /

/ f
/
/

/
+6 dB/OCTAVE SLOPES STARTING FROM
— co dB (A ZERO)AT ZERO FREQUENCY

the coefficient of S), the slope of the Bode increases by + 6 dB/octave. A

denominator term of the form 1/(STd + 1) is called apole of the function. At its
break frequency (found from rd, the coefficient of S), the Bode turns more
downward by — 6dB/octave.
It is comparatively easy to write the standard form of afilter from the Bode
plot. Let us take a notch ( stopband) filter for example (Figure 4-14). The break
frequencies are f i,f2,f
3, and f4.It can be seen that f
3 and f
4 are poles (denomina-

tor terms) because they reduce the slope of the Bode by — 6dB/octave. Also, f 1

and f
2 are zeros (numerator terms) because they increase the slope of the Bode by

+6dB/octave. So the standard form would be

H(S) = K (ST 1 + 1)(ST2 + 1)

(ST3 + 1)(ST4 + 1)

Knowledge of filtering is needed not only to design or construct filters, but

also because many incidental filtering effects occur within audio systems, and it is
necessary to understand filtering in order to be able to deal with them. Also, you
may come across technical documents in which the author writes of acoupling
capacitor acting as a zero, or a transistor acting as a pole. It is useful to
understand what is meant by this. Finally, it has been stated that the break
frequencies of afilter and the output level can be calculated simply by finding the
88 Chapter 4

Fig. 4-13 Effect of LVG

two or more poles
at the same break
frequency

Fig. 4-14 LVG

Stopband filter POLES BODE

/
LEVEL K ..--

),
--......,,

..
ACTUAL
FILTER

ZEROS

-
1 1
I I
I I
I I
f3 f
i f2 f4

time constants and deriving the transfer function. An example of this procedure is
given in the appendix at the end of this book.

Bandpass Filters
A bandpass filter can be made from ahigh-pass filter followed by alow-pass filter
(see Figure 4-15A). Provided the components values are such that the break
 Passive Filters 89

Fig. 4-15
Bandpass filter

(A) Filter schematic

BANDWIDTH-

LC fHC

(B) Response curve

frequency of the HP filter is lower than that of the LP filter, the response will be
as shown in Figure 4-15B.
The frequencies that are said to be passed are those within 3 dB of the
maximum value. Hence, the bandwidth of a filter includes only the frequencies
within this band. The two frequencies at which the curve has fallen by 3dB are
known as cutoff frequencies. The low cutoff frequency is fLc, and the high cutoff
frequency is fHc .Any frequencies outside this band are said to have been cut off.
In the simple HP or LP filters illustrated earlier in this chapter, the cutoff
frequency coincides with the break frequency. In a bandpass filter, especially
where the passband is narrow, there is an interference effect between the two
break frequencies. As a result, the cutoff frequencies do not coincide with the
break frequencies. Because of the interference effect, the levels at the break
frequencies can be considerably more than 3 dB below the Bode plot (see Fig-
ure 4-16).
In this case, f
1and f
2 are the break frequencies. The cutoff frequencies are f
i,
c

and f ie .The bandwidth occupies the frequency range between these, namely

(flic — f
ix). It can be seen that the cutoff frequencies do not coincide with the

break frequencies here.

90 Chapter 4

It is useful to be able to look at a filtering schematic and tell immediately

what type of filter it is. For this purpose, we simply consider three things:

• The high-frequency (HF) response

• The low-frequency (LF) response
• The mid-frequency response

Fig. 4-16 LVG

Interference effect BODE PLOT
in abandpass
filter

f2 fHC

Applying this method to Figure 4-15A:

• HF response: At the highest frequencies, the capacitor reactances fall to

zero; therefore, they can be considered shorting links. Then it can be seen
that C2 shorts the output signal to ground and there is no output voltage.
• LF response: At the lowest frequencies, the capacitive reactances become
infinitely high, like an open circuit. Consequently, C1 prevents any signal
voltage from passing. So, this filter passes neither the HF nor the LF signals.
• Mid-frequency response: At this frequency, the capacitors have some
reactance, so some signal passes through C1.Also, some output signal
develops across C2.
 Passive Filters 91

If the filter passes mid-range signals, but cuts out both high and low frequen-
cies, it must be abandpass filter. An analysis of abandpass filter is given in the
appendix at the end of this book.

Stopband Filters
These are designed to cut out acertain band of frequencies. An example is the
bias trap in atape recorder's replay circuit. A schematic for this type of filter is
shown in Figure 4-17.

Fig. 4-17
Stopband filter v, o O Vo
schematic
R1 R2

_
rc,
O O

To verify that this acts as a stopband filter, we will apply the method
described in the previous section for deriving the filter type from an inspection of
the schematic (see Figure 4-17).

• HF response: At the highest frequencies, the capacitors have zero

reactance, so they act as shorting links. Consequently, the signal passes
directly through C1.There is no attenuation.
• LF response: At the lowest frequencies, the capacitors have infinite
reactance and act as open circuits. The entire signal passes through RI,and
there is no loss of signal across C2. So again, there is no level loss. This filter
passes both the high and low frequencies without loss.
• Mid-frequency response: Here the capacitors have some reactance, so the
filter acts as avoltage divider, causing some loss of signal. If afilter reduces
the output only at mid frequencies, it must act as we intend, namely as a
stopband filter. Sometimes this filter is called anotch, because of the
appearance of its frequency response curve (see Figure 4-18).

As can be seen from the Bode plot, there are two poles (where the slope is
reduced by 6dB/octave) and two zeros (where the slope increases by 6dB/octave).
The zeros correspond to the break frequencies marked f 1 and f 2 on the graph, the

poles to f3 and f 4.Remembering that zeros are numerator terms and poles are

denominator terms, and that the DC gain is unity in this case, it is possible to write
the transfer function in standard form from inspection of the Bode plot
92 Chapter 4

Fig. 4-18 LVG

Stopband filter
BODE PLOT
response

ACTUAL FILTER

f
3 f
i 2 f4

(ST, + 1)(ST2 + 1)
Transfer function, H(S) = (sT3 u(s7-4 + 1)

For calculation purposes, the output response can be derived from this
formula simply by replacing each (ST n + 1) term with

to obtain its numerical value. f

n is the break frequency that corresponds to atime

constant T. SO,

Lv G = 20 log dB

To find the break frequencies, it is necessary mathematically to find the time

constants by obtaining the transfer function Vo/V, in terms of the component
values. Examples of this procedure are included in the appendix.
Active Filter Characteristics, 95

Inverting Active Filters, 95

Non-inverting Active Filters, 99

Shelving Equalizers, ioo

93
 5 Active Filters

Active Filter Characteristics

An active filter is distinguished from a passive filter in that it incorporates an
amplifier (usually an op-amp) in its circuit. The result is that it can amplify as well
as attenuate asignal. An active high-pass filter can attenuate the low frequencies
and amplify the high frequencies. Tone controls and equalizers are usually active
for this reason.
In order to make an op-amp into an active filter, it is only necessary to
replace either the input resistor or the feedback resistor with a capacitor. This
makes the amplifier gain frequency dependent, so it becomes atransfer function.
You may remember that the gain of an amplifier is Vo/V,. When it is frequency
dependent, it is called atransfer function.
There are two types of active filters:

• Inverting active filters, in which the input signal is fed into the inverting
input
• Non-inverting active filters, in which the input signal is fed into the non-
inverting input

We will look at inverting active filters first.

Inverting Active Filters

As was shown in Chapter 3, the gain of an inverting op-amp is given by

95
96 Chapter 5

Vo Rf
uv — —
V1 rsi

When acapacitor is used instead of one of these resistors, we have to write RfIR,
as Zf/Z,. So, for any filter using an inverting op-amp, the inverting active filter
transfer function becomes

Zf (5-1)
H(S) =

A low-pass active filter is constructed as shown in Figure 5-1A, and its frequency
response is shown in Figure 5-1B. In this circuit, Zf = 1/SC and Z, = R; so the
transfer function is

Zf 1/SC
H(S) —
Z, R

Multiplying numerator and denominator by SC,

1
H(S) =
SRC

But RC = 7 for these two components; so,

1 (5-2)
H(S) =

This forms adownward sloping line, because the transfer function is inversely
proportional to the frequency S. The slope is — 6dB/octave because Sdoubles at
each higher octave, and 1/S halves (half of the voltage corresponds to — 6dB).
For calculation purposes, we can use the fact that

w1 f
1

So,

1
H(S) = sTI

can be written

H(f) = ft
—
 Active Filters 97

Fig. 5-1Low-pass
(inverting) active
filter

(A) Schematic diagram

LVG

— 6 dB/OCTAVE SLOPE

0 dB f

(B) Frequency response

At break frequency (when f = f

i),

H(f) = f—1 = 1 = 0dB

fi

Thus, the downward sloping line cuts the 0dB level at break frequency. The level
at any frequency, f, can be found from

LvG = 20 log (f
-I) dB
f
where
1
fi = 277-1

and

71 = RC
98 Chapter 5

A high-pass active filter is constructed as shown in Figure 5-2A, and its

frequency response is shown in Figure 5-2B. In this filter,

Zf

El(S) Z, ( 1/SC)
R

Multiplying numerator and denominator by SC produces

SRC
H(S) = — = SRC
1

But RC = T, so

H(S) = ST (
5-3)

This causes an upward sloping line of + 6dB/octave.

Fig. 5-2
High-pass
(inverting) active
filter

(A) Schematic diagram

LVG

+6 dB/OCTAVE
SLOPE

0 dB

(B) Frequency response

 Active Filters 99

The transfer function of this HP filter is the inverse of the transfer function of
the LP filter. Again, at break frequency (when f = f i), the gain becomes 1, which
equals 0 dB. So, the rising straight line of + 6dB/octave cuts the 0 dB level at
break frequency.

Non-inverting Active Filters

If the same set of components were used with the non-inverting input of an op-
amp, the resulting frequency response would be different. It was demonstrated in
Chapter 3that the gain of anon-inverting op-amp is given by

R.
Gv = F'
1- Ri

In anon- inverting active filter,

Z‘
H(S) = 1
Zi

The 1in this function signifies that, even at the lowest gain, when Z1/Z = 0, the
transfer function can never be less than 1. Therefore, this type of active filter
cannot attenuate; it can only amplify. In some situations, such as obtaining a
treble or bass boost, this is just what is required. An LP (non-inverting) active filter
is shown in Figure 5-3, together with its frequency response curve.
An HP (non-inverting) active filter is constructed according to the schematic
in Figure 5-4A; its frequency response is shown in Figure 5-4B.
One of the applications of non-inverting active filters is in equalizing a
magnetic replay head. The combined effect of the narrow head gap and increased
wavelength of LF signals causes areplay head to act as azero. In other words, its
output falls off by 6dB/octave as the signal frequency drops. To counteract this, a
reproduce equalizer is incorporated, which produces the effect of apole. This
boosts the LF by 6 dB/octave as the frequency falls, and so achieves level
response (aprocess called post emphasis). However, ashelving effect is required,
to prevent increasing amplification of infrasonic frequencies, because we don't
want to amplify signals that are too low in frequency to be heard. To do so would
increase noise and distortion. The shape of the response curve we require is like
that shown in Figure 5-5B; the filter schematic is illustrated in Figure 5-5A.
Under DC conditions (when the frequency is zero), the capacitor has an
infinitely high reactance and can be considered an open circuit. Under these
conditions, it virtually doesn't exist. It is clear that the LF gain (at level K) is
produced only by the ratio of the two resistors, Rf/Ri.
When the frequency rises to 15 Hz, the capacitive reactance starts to take
effect. At that point, its value has dropped to equality with Rf. The response curve
100 Chapter 5

Fig. 5-3 LP
(non-inverting)
active filter

(A) Schematic diagram

LVG

TENDS TOWARD
—6dB/OCTAVE

0dB

(8) Frequency response curve

starts the downward slope from this frequency. Because anon-inverting op-amp is
being used, the gain cannot fall below unity, so the curve flattens out at 15 kHz.
One of the most useful applications of active filters is in the construction of
shelving equalizers or tone controls. We will look at these in more detail next.

Shelving Equalizers
A shelving equalizer can either boost or cut the high or low end of the frequency .
spectrum. In addition, the response curve shelves (flattens out) at the limits of the
audio spectrum. This produces the shelving effect that gives rise to its name (see
Figure 5-6). Normally there are two boost/cut controls, one affecting the HF end
of the range, the other affecting the LF end. This type of equalizer is often called
atone control, because it can boost or cut the treble range or the bass range, each
independently of the other. Figure 5-6A illustrates the possible shelving effects
 Active Filters 101

Fig. 5-4 HP
(non-inverting)
active filter

(A) Schematic diagram

LVG

TENDS TOWARD
+6dB/OCTAVE

0dB

(B) Frequency response curve

available at the high and low ends of the frequency range, while Figures 5-6B and
C show the LF and HF filtering circuit schematics.
A single linear potentiometer is used to produce the boost/cut effect by
changing the gain of the op-amp. The most effective way to do this is to use the
potentiometer as adifferential gain control, so that as it increases the feedback
resistance, it also reduces the input resistance, or vice versa. Consider the low-
frequency stage of the shelving equalizer schematic illustrated in Figure 5-6B.
Ignoring the capacitor for now, notice that as the sliding contact of the
potentiometer moves toward point A, the total feedback resistance is reduced,
while the input resistance is increased by the same amount. Because the gain of
the op-amp is R1/R 1,this reduces the gain to less than unity, producing LF cut. If
the potentiometer slider is moved toward point B, the value of Rf/Riis increased,
increasing the gain and producing LF boost. Because resistance R1 equals R2,
unity gain is produced when the potentiometer is in the mid position. This gives
0dB boost or cut.
102 Chapter 5

Fig. 5-5 LP
shelving filter

(A) Filter schematic

LVG

LEVEL K

0 dB

(B) Filter response curve

The next requirement is to limit the effect of this boost/cut facility to low
frequencies only. This is achieved by the capacitor. As the frequency rises, the
reactance of the capacitor falls. The capacitance must, therefore, be chosen so
that, at low frequencies, the reactance is high (as if it were an open circuit), giving
full boost/cut effect to the potentiometer. At mid and high frequencies, the
capacitive reactance falls, so that the potentiometer is effectively short circuited
by the capacitor. Then, the potentiometer has no effect on the gain of the op-amp,
which is held at unity (R1being equal to R2)for all positions of the potentiometer.
The opposite frequency limitations are needed in the high-frequency stage of
the shelving equalizer, illustrated in Figure 5-6C. In place of the capacitor, a
component is required that has the effect of shorting out the potentiometer at low
and mid frequencies, but has a high enough reactance at high frequencies to
 Active Filters 103

Fig. 5-6 Shelving LvG

equalizer

0dB

(A) High- and low-frequency response curves

(B) LF filter schematic

(C) HF filter schematic

104 Chapter 5

allow the potentiometer to take effect. An inductor is the obvious answer, because
its frequency response is exactly opposite to that of acapacitor. (Equalizers can
be made with more elaborate resistive networks, so that both high- and low-
frequency stages can be made entirely with RC circuits. However, we will use an
inductor for the HF stage in order to keep the circuitry simple.) The next
requirement is to decide on the values of the components, so that the required
effect can be produced.
Let us say that we need amaximum boost or cut of 20 dB. This corresponds
to avoltage ratio of 10 to 1. In round figures, it is convenient to give R1 and R2
values of 1Id/ each, and R3 a value of 10 Id/. The maximum gain (when the
potentiometer slider is at position B) is Rf/Ri,which is

(R 1+R3)
R2

These resistance values produce aratio of 11,000/1,000, which is approximately

10. The level gain can be calculated from

11,000
Max boost = 20 log — = 20.8 dB
1,000

When the potentiometer contact is in position A, the gain, Rf/Ri,is equal to

R1
(R2 +R3)

This is 1,000/11,000, the inverse of the previous gain. And the level is now given by

Max cut = 20 log

1,000 _ 20.8 dB
11,000 —

In round figures then, with these resistance values, we achieve the required
maximum boost or cut of ± 20 dB.
Now it is necessary to decide on the values of the reactive components. Let us
take the LF equalizer stage first. At extremely low frequencies, the capacitor acts
as an open circuit and maximum equalization control is achieved. Let us establish
that the rolloff starts when the frequency has risen to 20 Hz, so that the filtering
effect shelves (becomes flattened) at all frequencies below this. It follows that only
at or above this frequency is the reactance of the capacitor low enough to be
comparable to the 10 ku potentiometer. In fact, following the break frequency
concepts stated earlier, the capacitive reactance is then equal to the resistance.
When this happens, the actual filter response is 3dB from the maximum value.
Therefore, the rolloff frequency is said to occur at 3dB from shelf value. This is f 1

in Figure 5-6A. Then Xc = R3 and f = 20 Hz. The required capacitance can now
be found from
Active Filters 105

R = Xc

or

1
R - —
27rfC

So,

1
C=
27rfR

Substituting 20 Hz for f, and 10 I<S2 for R,

1
C -
27- x20 x10 x10 3

= 7.96 x10- 7 farads

.,F.
= 0.796 1

In practice, we would use a0.8 tiF value for the capacitor.

We now know all of the component values in the LF equalizer stage. It would
be useful, however, to know the higher break frequency. This is the frequency at
which maximum boost or cut makes only a ± 3dB level change from the 0dB
level. This is f2 in Figure 5-6A.

Because we have decided that the maximum boost or cut should be 20 dB,
and because asingle element filter response rises or falls at 6dB/octave (which is
20 dB/decade), it follows that f 2 must be one decade (10 times the frequency)

above f i.But f 1 is 20 Hz, so f

2 must be about 200 Hz. This sounds reasonable. It

means that only low frequencies below 200 Hz can be affected by the bass
control. The control becomes progressively more effective as the frequency falls
from 200 to 20 Hz. As far as the HF equalizer stage is concerned, the resistance
values are the same. It only remains to find the value of the inductor.
Let us decide that the high shelving frequency should occur at about 15 kHz.
This is f 4 in Figure 5-6A. All frequencies above this become flattened out. Then

the inductive reactance, which falls as the frequency falls, is equal to the
potentiometer resistance at 15 kHz. Below this frequency, the inductor more
completely shorts out the potentiometer, so that it has progressively less effect. If
XL = R3 at 15 kHz, we can find the value of the inductor from

R = XL

or

R = 27rfL
106 Chapter 5

So,

R
= —
2irf

Substituting 15 kHz for f, and 10 kn for R,

10 x10 3
L
— 27rx 15 x10 3

= 0.106 henrys

So, we use a0.1 H inductor in the HF equalizer stage.

Now we need to find out what the lower break frequency is for this HF stage.
This is f
3 in Figure 5-6A. At this frequency, the maximum boost or cut does not

exceed 3dB above or below the 0dB level. At frequencies lower than this, the
filter has negligible effect. Again, we can use the fact that the shelf level is ± 20
dB and the filter slope is 20 dB/decade. It follows that the lower break frequency
must be one-tenth of the rolloff frequency. One-tenth of 15 kHz is 1.5 kHz. So this
HF control acts only above 1.5 kHz and becomes progressively more effective up
to 15 kHz. Above that frequency, its effect remains constant.
These two circuits, consisting of the treble and bass controls, are connected in
series in the complete equalizer. It doesn't matter in which sequence they are
connected; the equalizer will work just as well either way.
These filtering concepts are intended to illustrate how capacitive and
inductive reactance can be used in filtering circuits. When these frequency
dependent components are inserted into the input or feedback loop of an op-amp,
auseful active filter can be made. Just as atop-quality racing driver can improve
his or her driving technique by understanding the mechanical principles
incorporated in the car, so atop-quality audio engineer can improve his or her
ability by understanding the theoretical principles behind filtering techniques.
The Voltage Changing Function, /
os

The Isolation Function, iii

The Impedance Matching Function, 113

Bandpass Characteristics, 118

Avoiding Transformer Generated Distortion, 124

Summary of Transformer Applications, 125

107
 6 Transformers

The Voltage Changing Function

To understand how atransformer works, it is necessary to realize that an electric
current always surrounds itself with a magnetic field. Current cannot exist
without a corresponding magnetic field. The field that develops around a single
conductor is comparatively small. When this conductor forms a closely wound
coil, however, the field is magnified by the number of turns in the coil. The reason
is illustrated by the enlarged cross section through a wire and a coil, shown in
Figure 6-1. From this we can see that the fields from the individual turns of acoil
combine to produce an amplified field through the core.

Fig. 6-1 Magnetic

field surrounding
single conductor
and coil

(A) Cross section of asingle conductor (B) Cross section through four turns of acoil.
Each turn adds to the field.

When electric current begins to flow through a coil, some of its energy is
converted to magnetic field energy within and around the coil. This extraction of
electrical energy causes an induced voltage that opposes the current increase. It

109
110 Chapter 6

is the work done in overcoming this induced opposing voltage that supplies the
energy needed to generate the magnetic field.
Conversely, when a magnetic field surrounding a coil degenerates due to a
reduction in current flow, some of its energy is transferred back to the coil in the
form of electrical energy, inducing a voltage in the direction of the current
(opposing its reduction). The magnitude of this induced voltage depends on the
rate at which the magnetic field changes and on the number of turns in the coil. It
is given by

(6-1)
dc/3
V = N dt

where
V = Induced voltage
N = Number of turns in the coil
dedt = Rate of change of magnetic field with respect to time

A transformer consists essentially of two coils, a primary and a secondary.

The two coils are wound close together so that the changing magnetic field
interacts equally with both. A primary coil connected to an AC source produces a
fluctuating magnetic field common to both coils. If we call the induced voltage
across the primary Vp,and that across the secondary Vs,the ratio of these
induced voltages is found from Equation 6-1 by

where
N = Number of primary turns
N, = Number of secondary turns
dO/dt = Rate of change of magnetic field

But the quantity dedt in this equation cancels; therefore,

Vs Ns
Vp = Np

This demonstrates that the voltage ratio equals the turns ratio in a
transformer. This effect can be achieved only under AC conditions. Under DC
conditions, the magnetic field remains constant, so dedt is zero and no voltage
can be induced in the secondary.
The voltage changing function is most commonly used in the construction of
power supplies. A transistorized circuit requires only about 30 volts. So a line
 Transformers 111

voltage power supply would have, as its first stage, a step-down transformer
whose primary at 115 AC volts energizes asecondary coil, to produce just over 30
AC volts. However, avacuum tube amplifier might requiire 300 volts. In this case,
a step-up transformer would be used to convert 115 AC line volts to 300 AC
secondary volts.

The Isolation Function

There is no electrical connection between the primary and secondary coils of a
transformer. The power transfer takes place by means of magnetic field energy. In
the case of a one-to-one transformer, which has a turns ratio of unity, its AC
secondary voltage is equal to the AC voltage applied to the primary. It is,
therefore, useful not for voltage changing, but for isolation purposes. It can enable
afluctuating DC voltage to flow in the primary while passing only the AC audio
component through to the secondary. The DC component, producing no rate of
change of magnetic flux, induces no voltage in the secondary.
A typical application consists of transformer coupling aClass A amplifier to
its following stage. As described later in Chapter 7, aClass A amplifier requires a
direct current through its output transistor at all times. The fluctuations of this
current produce the audio signal (see Figure 6-2A). When an audio signal is being
carried, the primary circuit current can be thought of as an AC audio signal riding
on the shoulders of aDC component. The current is shown graphically in Figure
6-2C.
An isolating transformer can also achieve useful results when installed in the
electrical power line that energizes a console. In the resulting balanced line
power cable, both conductors are at equal and opposite voltages, instead of one
conductor carrying all of the voltage, while the other is effectively grounded. This
cuts out noise interference caused by feedback from other equipment and
fluorescent lights, and avoids formation of ground loops, which frequently cause
hum interference. The method is illustrated in Figure 6-3.
The same method can be used to achieve balanced line transmission between
any two audio components, such as a microphone and console input. In these
cases, impedance matching can be achieved at the same time. This is because a
transformer can not only isolate, it can also match impedance.
To understand how atransformer converts an unbalanced line to abalanced
line, consider atransformer with 10 volts induced across the secondary. The only
physical requirement is that there must be 10 volts across the secondary
terminals. If one end of the secondary is grounded, the other end will be at 10
volts above or below ground potential.
Now consider the transformer illustrated in Figure 6-3. The effect of the
grounded center tap is to hold the center of the secondary at 0volts. Therefore,
one terminal of the secondary must be at 5volts more positive and the other 5
volts more negative. In this way, the requirement of 10 secondary volts is
112 Chapter 6

Fig. 6-2Coupling DC +
transformer

ív AC AUDIO
COMPONENT
FORMED
BY INDUCED
VOLTAGE

(
A)Coupling transformer blocks DC and passes the AC audio signal

l
o I
Q

DC COMPONENT

(B) Primary current, i, under quiescent (C) Primary current consisting of an AC audio
conditions, when no audio signal is present component riding on aDC quiescent comporzent

(D) Audio signal voltage induced in the secondary No DC component is present

Fig. 6-3Isolating
transformer LINE TO
CONSOLE
supplying power SUPPLY
to a console

obtained. Of course, the voltages alternate, but they are equally spaced above
and below ground potential. This is what is required for balanced line conduction.
It is achieved simply by holding the center tap of the secondary at ground
potential.
 Transformers 113

Another use for an isolating transformer is in the breaking of aground loop.

Sometimes, after anew audio system is installed, it is discovered that a60 Hz hum
is superimposed on the audio output. Likewise, when a piece of outboard
equipment is patched into a system, it might generate unacceptable line hum,
which wasn't there before.
These problems are very often created by the formation of aground loop. A
ground loop occurs when there are two paths to ground from any given point in a
system. It can be created by two connections to acommon metal chassis or by
grounding both ends of a shield surrounding a cable. (Only one end should be
grounded.) A ground loop looks schematically like the circuit in Figure 6-4.

Fig. 6-4 Creation

of aground loop

GROUNDED SHIELDING OR CHASSIS

The ground loop acts as asingle turn of acoil. In other words, it acts as an
inductor. The result is that it creates an induced voltage caused by its interaction
with electromagnetic radiation from other equipment. Because most equipment
contains a line transformer that leaks electromagnetic radiation, ground loops
tend to produce a60 Hz frequency hum. If you take hold of an audio patch cord or
cable and move it slightly, and if this movement alters the intensity of the hum,
this is asure sign of aground loop.
Often a one-to-one isolating transformer is needed to break such a ground
loop and eliminate the line hum. An example of this use of transformer isolation is
shown in Figure 6-5. Compare this with Figure 6-4.

The Impedance Matching Function

It is now necessary to say afew words about impedance matching in relation to
coupling. Direct coupling involves direct connection between one stage and the
next. This is only permissible if the DC voltages and impedances are known to be
compatible.
The method more often used is capacitive coupling. This blocks DC voltages,
but enables the AC signal to pass. As can be seen from Figure 6-6, the coupling
114 Chapter 6

capacitor forms an RC series circuit in conjunction with the input resistance of the
following stage, Ri.

Fig. 6-5 Ground

loop broken by
isolating
transformer

GROUND LOOP
BROKEN HERE

Fig. 6-6Filtering
effect of
capacitive
coupling
COUPLING
CAPACITOR

The size of the capacitor must be large enough so that its reactance to the
lowest audio frequency is only about one-tenth of the impedance value of Ri.In
this way, not more than one-tenth of the signal voltage is lost across the coupling
capacitor. The size of the capacitor can be calculated from

X, =

Therefore,

1 R,
27rfC 10

From this equation, we arrive at

10
C = 2rfR, farads (F)

Or,
 Transformers 115

10 x10 6 r.
C —
27rx 15 xR;141

(at the lowest audible frequency, when f = 15 Hz).

Manufacturers often include acoupling capacitor at the output terminals of
their products. The user can then directly connect the output to the input of the
next stage, without fear of a DC voltage unbalancing the signal. However, this
requires that the output impedance of the previous stage, Ro,be compatible with
the input impedance of the following stage, RL.If these stages are not already
impedance matched (see Chapter 1, The Audio Circuit), then capacitance
coupling will not be acceptable, because any impedance mismatch will remain.
However, the use of asuitable transformer can solve many impedance matching
problems. Transformer coupling may, therefore, be preferable to capacitance
coupling. A transformer blocks DC, allows the AC audio signal to pass, and
corrects any impedance mismatch at the same time. The only disadvantage is the
cost; still, in some cases, atransformer is well worth its price.
When we think of an inductor, we assume that its impedance is frequency
dependent. This is because the reactance of acoil is given by the product of 271-fL.
However, if a transformer is correctly impedance matched, its secondary current
produces what is called reflected impedance in the primary. (Correct impedance
matching requires that the primary coil impedance equal the source output
impedance, and that the secondary coil impedance equal the load impedance. See
Figure 6-7.) Under these impedance matched conditions, the reflected impedance
produces aprimary circuit phase angle, which cancels out the phase angle that would
normally be produced by the primary current. Thus, the current in the primary
becomes in phase at all frequencies. The result is that the primary coil is no longer
seen as an inductor by the source. It is seen as aresistor, and its impedance ceases to
be frequency dependent. It now acts as aresistive load, whose value is constant at all
frequencies. This is how an audio transformer maintains linear response over awide
band of frequencies.

Fig. 6-7 For

correct impedance
matching, Zr,=
and Zs =

When using acoupling transformer, the primary coil impedance should equal
the source output resistance. The secondary coil impedance should equal the
load resistance. This method of impedance matching is called power matching.
The values are not critical; amatch within 20 percent is satisfactory.
As described in Chapter 1, maximum power is transferred from asource to a
load when RL = Ro at the interface. Due to reflected impedance, this source sees
the primary coil as aresistive load equal to itself. Therefore, it transfers maximum
power to the primary. By means of magnetic energy, all of this power is
116 Chapter 6

transferred to the secondary. But Zsacts as the source impedance to the load, and
because its value is numerically equal to RL,it transfers maximum power to the
load. The transformer, then, transforms impedances. It enables R. to see aload
equal to itself, while enabling the load to see asource impedance equal to itself.
Under these power matched conditions, there is a definite relationship
between the voltage ratio of a transformer and the impedance ratio. At audio
frequencies, there is very little energy loss. For calculation purposes, we can
assume that an audio transformer is 100 percent efficient. Thus, the power
absorbed by the primary equals the secondary power supplied to the load. But
P = V2/R or, at zero phase angle, R = V2/Z. Also, P = IV. From these
relationships we can make two useful deductions.

Primary Power = Secondary Power

So,

V 2 V 2
--E— —
Z p — Zs

or

(Vs)
_ 2 ._
Zs
Vp Zp

Thus,

Vs = ,\,/
VP Zp

This shows that the voltage ratio, or turns ratio, equals the square root of the
impedance ratio. Also, because Primary Power = Secondary Power,

IV
P P
— IV
— s s

So,
V 1P
VsP
V Is

(Notice that the current ratio is the inverse of the voltage ratio.) A complete
summary of transformer relationships, showing the turns ratio (secondary/
primary), is as follows:

Turns ratio, A = 1 = Vs = 1p = Zs
NP VP Is ZP
 Transformers 117

When purchasing audio transformers, the impedance values of the primary and
secondary coils are supplied by the manufacturer. Often this data is printed on the
side of the metal shielding. You can calculate the corresponding voltage ratio, but it is
not necessary to do so. In using an audio transformer for coupling, you need to
choose the correct impedances and let the voltage ratio fall where it may.
Many audio transformers used for matching have two or more secondary
coils. This gives an opportunity for various possible impedance matches, as
shown in Figure 6-8.

Fig. 6-8 Typical

matching
transformer
9 k52

This transformer could match a600 11 source circuit (terminals 1and 2) to the
following loads:

1kl/ (terminals 3and 4)

4kfl (terminals 5and 6)
9MI (short 4and 5, and use 3and 6)

Optionally, the 600 SI could be used as the secondary, and any of the other coils
as the primary.
You might be surprised to learn that both secondary coils in series produce a
total impedance of 9 1(11, and not 5 kg. Because the turns are added and the
voltages are directly proportional to the turns, the two secondary voltages are
added. However, the voltages are proportional to the square root of the
impedances, so the roots of the two impedances must be added. Their sums must
then be squared to convert the resultant voltage back to impedance. The total
secondary impedance is given by

(.14 +NZ2

or

1,#`400 + V473717W1)2

Another option is available for matching asource to aprimary. If the primary

coil impedance is too great to be correctly matched, abuilding-out resistor can be
added in series with the source. This resistor should be of sufficient value to raise
the resistance of the source to that of the primary coil; then, good match and
linear frequency response will be achieved. For example, if the source R. = 600 I/
118 Chapter 6

and the transformer ; = 1kg, then a400 I/ building-out resistor, Rp,is needed.
(Rp+Rp = ;, or 600 + 400 = 1kg. See Figure 6-9.)

Fig. 6-9 A
building-out = 600 Rp = 400 D

resistor used for Zp 1 k≤≥ RL =

primary 3 kQ
impedance •

matching

Some consoles have abuilt-in 600 load resistor, which can be inserted by a
switch as either a building-out resistor (in series with the source) or as a
terminating resistor (shunted across the output terminals, effectively paralleling it
with the console's output impedance). For transformer coupling this is a useful
facility, because it gives the console three possible effective output impedances.
Let us assume that the console has anatural output impedance of 600 IL By using
the building-out resistor, this can be increased to 1,200 By using the
terminating resistor, this can be reduced to 300 IL
If in doubt, it is a good idea to run a quick frequency response test. Put a
constant level signal through the system at, say, 20 Hz, 1kHz, and 20 kHz. If there
is loss of low or high frequencies, try using the additional resistor, first in building-
out position, then in terminating position, and see which gives the most level
response.

Bandpass Characteristics
Transformers designed for audio frequency work are constructed around an iron
core. This concentrates the magnetic field, resulting in increased effectiveness
and reduced production cost. At radio frequencies, an iron core is not necessary,
and many RF transformers are made with air cores. This is acceptable because
the rate of change of magnetic flux is greater at higher frequencies.
Although the efficiency of an iron cored transformer is nearly 100 percent,
there are two significant sources of energy loss. One consists of hysteresis loss in
the iron core. The other consists of coupling loss. (See the appendix for adetailed
description of hysteresis.)
Hysteresis loss results from internal friction in the iron as the magnetic flux
reverses. This draws additional magnetizing current from the source. Because the
required magnetic field intensity increases at low frequencies (to maintain the rate
of change of flux), this current loss becomes significant at low frequencies.
Coupling loss, on the other hand, occurs significantly at high frequencies.
This is because the fluctuating magnetic field extends farther at high frequencies;
consequently, some of the magnetic field energy fails to couple with the
secondary. This unused energy is dissipated as electromagnetic radiation.
 Transformers 119

The result of these losses causes an audio transformer to act as abandpass

filter. The output remains flat over the audio frequency range, but falls off at very
low and very high frequencies.
We will ignore the DC resistance of the coils, because they are small
compared to the source and load resistances, and also because they are not
frequency dependent. Taking hysteresis and coupling losses into account, the
equivalent circuit of astep-down transformer would look like that in Figure 6-10,
where:

R, = Output resistance of generator (source)

RL = Load resistance
1.14 = Equivalent inductance, drawing extra magnetizing current due to
hysteresis loss
Ls = Equivalent inductance, representing coupling loss leakage

Fig. 6-10
Transformer
equivalent circuit

Lm is shunted across the generator terminals because its effect is to draw

additional current from the source. Ls is in series with the load because it reduces
the voltage across the load. The low- and high-frequency equivalent circuits are
illustrated in Figure 6-11.

Fig. 6-11
Low- and
high-frequency
equivalent circuits

(A) Low frequency (B) High frequency

At low frequencies, the reactance of Ls becomes negligible, so it can be

ignored. But when the reactance of Lm becomes small, it draws far more current
from the source. Consequently, there is agreater voltage drop across R., and this
effect reduces the voltage across 12L.
At high frequencies, the reactance of Lm is so high that it acts as an open
circuit; therefore, it can be ignored. But the high reactance of Ls reduces the
voltage across RL,so this effect is now significant. The low-frequency equivalent
circuit consequently acts as ahigh-pass filter, cutting off the low frequencies. The
120 Chapter 6

high-frequency equivalent circuit acts as a low-pass filter, cutting off the high
frequencies. The effect of both is to produce awide-band bandpass filter.
The characteristic structure built into these transformers by the manufacturer
is designed to ensure that the full range of audio frequencies is included within the
bandwidth. This only happens, however, when both transformer coils are
correctly impedance matched. If either coil is severely mismatched, the passband
moves either up or down the audio spectrum, causing loss of low or high
frequencies. To illustrate this, we will derive an expression for the cutoff
frequencies of the equivalent circuits in Figure 6-11.
At low frequencies, the break frequency of the circuit in Figure 6-11A can be
found using the filter analysis method given in Chapter 4. The derivation of the
transfer function that results is given in the appendix at the end of the book. By
this method, the transfer function appears as

RL
( ÍR. +N.
K J R. RL
] ) (6-2)

H(S) — R. + RL sL m l
R. + Ri + 1
-

[ R. RL

This is astandard HP filter of the form

ST
H(S) = Ksr + 1

It is clear that the coefficient of Sis T. So, by comparison of forms,

=L
m[R° R I
RL

However, the break frequency (and in this case, the low cutoff frequency) is
always found from 1/27T. So, the low cutoff frequency is

1
f
LC = 22-Lm (R. + RL)
R. RL

Thus,

RL (
6-3)
fie - 2Iftm (
R. + R3

It is comparatively easy to find the break frequency of the equivalent HF

circuit shown in Figure 6-11B, because this is a simple RL series circuit. In this
case, the time constant is simply L/R. So,
 Transformers 121

Ls
T —
Ro + RL

Thus,

fHC = 27r Ls
Ro + RL

Therefore,

R. + RL (6-4)
f
HC - 2/rLs

These transfer functions and the resulting time constants will shortly be used
to examine the effect on bandwidth of atransformer impedance mismatch. Due
to the physical problems inherent in transformer construction, a large ratio
transformer is highly sensitive to impedance mismatching. A low ratio
transformer has amuch larger bandwidth; therefore, impedance matching is not
critical.
One situation in which amatching transformer must be precisely matched is at
the output from a vacuum tube power amplifier. Here, the tubes, with an output
impedance of about 6 kfl, are feeding a speaker of about 8 a A high ratio
transformer is, therefore, needed. The construction of such atransformer, with the
necessary bandwidth, is so difficult that the manufacturers have to use amultitapped
secondary. The schematic of such atransformer is shown in Figure 6-12.

Fig. 6-12 16 SI
Loudspeaker
matching 8S
2
transformer
41
2
-

COMMON

When connecting a loudspeaker to the secondary, the common terminal is

always used. The other output terminal is chosen to match the speaker
impedance. Therefore, a411 speaker is connected between the 4 9 output and
common. Of course, two 8SI speakers could be connected in parallel between the
41/ output and common, or in series between the 16 St output and common. Series
connection requires apositive (+) terminal from one speaker to be connected to
a negative (—) terminal on the other. In parallel, the two + terminals are
connected, and the two — terminals are connected (see Figure 6-13). If the
speaker polarities are wrongly connected, phase cancellation results in loss of low
frequencies.
122 Chapter 6

Fig. 6-13 Phase

matching two 16 Q
loudspeakers

8Q

8Q
8Q 8n

(A) Series connected speakers (B) Parallel connected speakers

Because impedance matching is so critical with high ratio transformers, you

have to watch out for this type of situation. lf, for instance, you need to record
from the loudspeaker terminals of atube amplifier, it is important not to connect
the 16 fi output terminals directly to a 600 2 (or more) recorder input, without
using a 16 CI load. Calculating the cutoff frequency from Equations 6-3 and 6-4,
and using 1.14 = 0.085 H and Ls = 0.060 H, the indication is that all low
frequencies below 1kHz will be cut off. For correct results, it is necessary to
connect adummy 16 0 (10 W) load resistor across the amplifier output, in parallel
with the output cable. Then, the full frequency response will be restored.
However, if an output transformer is built onto a transistorized amplifier,
there will be no problem. The low turns ratio needed to match a 12 power
transistor to a42 speaker has such awide bandwidth that practically any load
above 4 2 is acceptable. In fact, manufacturers often specify the load as being
"4 2 or more." In this case, you can connect the power amplifier directly to a
recorder input, without aspeaker load, and still obtain full frequency bandwidth.
Another situation in which a high ratio matching transformer is needed
occurs when using ahigh impedance microphone or guitar pickup with along mic
cable. A microphone matching transformer is needed then to change the high
output impedance of the mic to the low impedance that must be fed to along
cable. (The necessity for this will be discussed later.) At the other end of the cable,
another transformer is needed to match the low impedance source to the high
impedance pre-amplifier input. Without this input transformer, incorrect loading
of the microphone transformer could restrict the available bandwidth. A
schematic of the system illustrates the arrangement in Figure 6-14. (The letter Z
refers to impedance.)
On the other hand, alow impedance microphone can be directly connected
to a long microphone cable. A step-up transformer might be used at the
 Transformers 123

Fig. 6-14 MIC OR PICKUP PRE- AMP

Transformer
matching a
high-impedance LOW
LONG CABLE
mic to along Z,
transmission cable

pre-amplifier input (this would slightly reduce the noise level), or it need not be
used. In this case, transformer matching is optional.
Now Ishould explain why a long transmission cable must be fed by a low
impedance source. When two conductors in acable run close together, there is a
small amount of capacitance between them. The capacitance increases in
proportion to the length of the cable, so a long cable produces significant
capacitance between the two conductors. When this cable is fed from a
microphone (or any audio generator), the equivalent circuit looks like that in
Figure 6-15.

Fig. 6-15 Filtering

effect of
transmission line,
where R. is
source output
impedance and C
is cable
capacitance

By looking back at Chapter 4, we can see that this is, in fact, the circuit of an
LP filter. It is not that we want to include an LP filter; we can't help it. So we must
ensure that the resulting cutoff frequency is above the highest audible frequency
of about 15 kHz. In any capacitor, the reactance drops as the frequency rises; the
cutoff point is reached when the reactance is low enough to equal the resistance
of R.. The level at this frequency is 3dB down. All higher frequencies are cut off.
In practice, we do not wish to lose even 3 dB from the response of our
microphone at 15 kHz. One-tenth of the signal voltage is the maximum acceptable
HF loss. If fed by alow-impedance microphone of 100 S2, the total cable reactance
at 15 kHz should, therefore, be ten times this value (namely, 1k12). Using typical
microphone cable, it works out that the maximum acceptable cable length would
be 200 feet. (The cable capacitance would then be so high that the reactance
across the cable would be reduced to 1kS2.)
However, if the output impedance of the microphone were 11(12 (ten times
higher), then the frequency at which one-tenth of the signal is lost would be ten
times lower (namely, 1.5 kHz). This is obviously unacceptable because all higher
audio frequencies would experience excessive loss. (Remember that frequency is
inversely proportional to capacitive reactance. Therefore, ten times the reactance
would occur at one-tenth of the frequency.) This is why it is necessary to use alow
124 Chapter 6

impedance source for long audio cable transmission. Matching transformers

provide the solution, where high impedance sources have to be used.
In many situations, where no serious matching problems exist, the use of
transformer or capacitance coupling is optional. The need to control cost favors
capacitance coupling; however, some advantages of transformers are

1. The bandpass characteristic limits frequencies to within fixed parameters.

2. Isolation from other circuits is complete. This helps to prevent ground loops.
3. Although capacitors block most of the DC, electrolytic capacitors often leak
acertain amount of DC voltage. Transformers block DC voltages
completely.

Avoiding Transformer Generated Distortion

One aspect of transformers hasn't been mentioned yet. That is the need to avoid
overload distortion. A piece of iron can be magnetized only up to acertain level.
It is then fully magnetized and no additional magnetizing force will increase its
magnetization. It is said to be magnetically saturated, meaning it contains the
greatest amount of magnetic energy it can hold. If more magnetic energy has to
be stored, more iron is needed. That is why high-power transformers have large
iron cores, and are heavy and expensive.
If atransformer is asked to transmit more than its rated power, the iron in the
core can become magnetically saturated. The result is that during periods when
the primary current approaches its peaks, more and more of the core saturates
and fails to transmit the corresponding energy increase to the secondary. This
gives rise to apeculiar distorted waveform. The effect is illustrated in Figure 6-16.

Fig. 6-16 MAGNETIC SATURATION

Distorted STARTS AT THIS POINT

waveform / TRUE SECONDARY WAVEFORM

resulting from
power overload
DISTORTED SECONDARY WAVEFORM

\
\
/
\
\ /
• /
..... /
.........,
 Transformers 125

An unusual characteristic of this type of distortion is that it occurs first at low

frequencies. This is because, at low frequencies, a phase angle develops and
magnetic peaks become higher in order to transmit the same amount of power.
So, if in doubt as to the power handling ability of atransformer, test at 15 Hz with
the maximum signal voltage you are likely to use. If the output wave, seen on an
oscilloscope, looks good, you can be sure that all higher frequencies will behave
correctly. Remember to test for overload distortion at 15 Hz, not at the normal test
frequency of 1kHz.

Summary of Transformer Applications

1. Transformers are used in power supplies to transform AC voltages either up or
down. Because only AC voltages can be transformed, both the primary input
and secondary output are in AC form.
2. A one-to-one transformer can be used to isolate one stage or component from
another. This can eliminate noise interference by preventing ground loop
formation.
3. Transformers can convert unbalanced lines to balanced line transmission. This
reduces electrical interference and also helps to eliminate noise.
4. Transformers are often used to couple one stage to the next. The transformer
completely blocks DC voltage, while allowing the AC audio signal to pass. A
coupling transformer can also correct an impedance mismatch; however, high
turns ratio transformers need to be impedance matched to their source and
load, to avoid loss of high or low frequencies.
5. The power handling ability of any transformer depends on the amount of iron
in its core. Test for possible overload distortion at the lowest audio frequency of
15 Hz, not the usual test frequency of 1kHz.
Diodes, 129

Transistors, 133

Common Emitter Amplifiers, 136

Common Base Amplifiers, 138

Common Collector Amplifiers, 140

Class A Amplifiers, 142

Class B Amplifiers, 144

Field Effect Transistors, 146

127
,te
 7 Semiconductors

Diodes
Atoms consist of apositively charged nucleus surrounded by orbiting negatively
charged electrons. The number of electrons equals the number of protons in the
nucleus, so the positive and negative charges balance. Electric current consists of
a flow of free electrons. Free means released from orbital constraint around a
nucleus.
Atoms prefer to have eight electrons in the outer valent orbit. Beyond that
number, they start an additional outer ring. If this contains one electron, they
have one more than the eight preferred, so they easily lose the extra electron
through thermal agitation (heat vibration). Therefore, this type of material
contains many free electrons, which act as current carriers. Such materials make
good conductors.
Atoms that have seven electrons in their outer orbit are reluctant to lose one
(their preference being to gain one, making eight). This type of material contains
practically no free electrons; thus, it has good insulation properties.
Atoms that contain four electrons in their outer valent orbit are indifferent to
losing or gaining one electron. If they lose one, they have three. This is three
above what exists in the complete inner ring of eight. If they gain one, they have
five. This is three below the preferred number of eight. These atoms, with four
valent electrons, are called semiconductors.
A FN junction diode consists of two wafers of asemiconductor element fused
together. The two halves of the diode are each impregnated with a different
element. This process is called doping. A small quantity of doping material is
evenly diffused throughout the semiconductor material.

129
 130 Chapter 7

There are two types of doping material. One contains five valent electrons
(called pentavalent doping). The other contains three valent electrons (called
trivalent doping). Where a pentavalent atom replaces a semiconductor atom,
there are five valent electrons instead of four. Consequently, one free negative
current carrier is produced by each doping atom. This type of doped
semiconductor material is called N-type material, because it carries a large
number of extra electrons, which are negative current carriers. On the other
hand, where atrivalent atom replaces asemiconductor atom, there are only three
valent electrons instead of four. This leaves ahole, which attracts electrons and
acts as a positive current carrier. This type of doped semiconductor material is
called P-type material, because it contains a large number of positive current
carriers.
Briefly then, a diode consists of a wafer of N-type semiconductor material
fused at ajunction to awafer of P-type semiconductor material. This produces a
junction that offers extremely high resistance to current flow in one direction,
and extremely low resistance in the other. In effect, it acts as aone-way street,
allowing current to flow only in one direction. When avoltage is applied to the
diode in a polarity that permits current to flow, it is called forward biased; a
polarity in which no current can flow is called reverse biased. The schematic
symbol for adiode is shown in Figure 7-1. The arrowhead indicates the direction
in which conventional current can flow, which is from positive (+) to negative
(—). The results of forward biasing are illustrated in Figure 7-2. (Note that like
charges repel and unlike charges attract.)

Fig. 7-1
Schematic symbol
for adiode
ANODE CATHODE

Fig. 7-2 Effect of

biasing a PN
junction
+ -

(A) Forward biased. Current carriers forced (8) Reverse biased. Current carriers being pulled
toward the junction experience little resistance away from the junction cannot cross over, so no
in crossing over, so current flows. current flows.
 Semiconductors 131

These current carriers produced in the semiconductor material by doping are

called majority carriers. There are also afew free electrons and holes, inherent in
the pure semiconductor material itself, that have been released by thermal
agitation. These are called minority carriers. Therefore, to say that no current
flows in reverse bias polarity is not 100 percent accurate. In fact, a very small
amount of current flows when reverse biased, due to the minority carriers only.
While under forward biased conditions, current flows as aresult of both majority
and minority carriers.
One other property of PN junction diodes needs to be mentioned. This is the
spontaneous creation of ions (charged atoms) in close proximity to the junction,
even when no external voltage is applied. Due to the attraction between opposite
charges, some electrons cross over and occupy some of the holes on the other
side of the junction. The effect is two-fold.

1. It produces asmall area on each side of the junction where the majority
carriers are missing, or cancelled out. This is called the depletion zone,
because here the number of carriers has been depleted.
2. It also causes avoltage to develop spontaneously across the junction.

To see why this voltage occurs, consider the doping material. Pentavalent
doping atoms have five electrons and five corresponding protons. Consequently,
their net charge is zero. Similarly, trivalent doping, with three valent electrons,
has only three corresponding protons, also with no resultant charge. However, as
soon as some electrons on the N side cross over the junction, negative charge is
lost and apositive charge results. When these electrons fill holes on the Pside, the
positive charges become cancelled and a negative charge results. This causes a
barrier potential to occur (see Figure 7-3).

Fig. 7-3 Enlarged DEPLETION

view of aPN ZONE
I ...1
junction

- +
BARRIER POTENTIAL CAUSED
BY TRANSFER OF IONS
ACROSS JUNCTION

The barrier effect of this spontaneously generated voltage is experienced

when the diode starts to be forward biased. In Figure 7-3, forward bias occurs
132 Chapter 7

when a + polarity is applied to the P side and a — polarity to the N side.

However, there is already areverse bias that has been spontaneously produced at
the junction. This acts as abarrier to current flow, even when forward biased. The
result is that the applied forward bias voltage must be higher than the barrier
potential before any current can flow. If it is lower (or reversed), the diode is said
to be cut off. This barrier potential is of the same magnitude as the cutoff voltage,
which depends on the semiconductor material. Silicon diodes have a cutoff
voltage of about 0.6 V and germanium diodes have acutoff voltage of about 0.2 V.
The current/voltage graph of adiode is shown in Figure 7-4. As can be seen
from the graph, current increases linearly in proportion to bias voltage once the

Fig. 7-4 Diode

characteristic
curve
SATURATION

REVERSE
BIASED

CUTOFF FORWARD
POINT BIASED
POINT OF
ZENER BREAKDOWN

cutoff point has been passed in the forward bias direction. Below cutoff voltage
and covering aconsiderable extent of reverse bias, there is practically no current
flow at all. However, if too much reverse bias is applied, there is asudden increase
in reverse current at the point of zener breakdown. This happens when the
applied reverse polarity voltage reaches a level at which the semiconductor
insulation breaks down. The diode is then said to be zenering. This property
enables azener diode to output aconstant voltage within acircuit under certain
conditions. This characteristic is used in voltage regulated power supplies.
 Semiconductors 133

Transistors
A transistor consists basically of two diodes back to back, which form asandwich
of P- and N-type material. This can be made in two ways (see Figure 7-5).

Fig. 7-5 Types of

transistors
E N N

(A) NPN transistor (

B) PNP transistor

Current flow is from the emitter through the base and out through the
collector. These terminals are labeled E, B, and C, respectively. The two types of
transistors are distinguishable by the direction of the arrow in the schematic
symbols (see Figure 7-6). This indicates the direction of conventional current flow
(holes flow). Of course, electron flow is in the opposite direction. It is useful to
remember that electrons always flow in the opposite direction from the arrows, in
the symbols both for diodes and transistors.

Fig. 7-6
Schematic symbol
for atransistor

(A) NPN transistor (B) PNP transistor

In the case of an NPN transistor, forward bias for the E/B junction requires
that the P material of the base be more positive than the N material of the
emitter. Electrons can then flow through this junction in the opposite direction
from the arrow. Because electron flow has to be in this direction, certain DC
voltages must be present. The required polarities and current paths are shown in
Figure 7-7. The double polarity sign at the collector is intended to show that the
collector voltage must be greater than the same polarity at the base. Also
indicated is asmall (approximately 2percent) current leakage from the base.

Fig. 7-7 Electron LEAKAGE MAIN LEAKAGE MAIN

flow and CURRENT c t CURRENT CURRENT - CURRENT
polarities in \ + •++ FLOW \ - FLOW
transistors

I _I_
y

(A) NPN transistor PNP transistor

134 Chapter 7

It is the junction between the emitter and base that is the key to controlling
the current flow through a transistor. The bias voltage across this junction is
called the bias of the transistor. When this bias, VEB ,is below cutoff, no current
flows and the transistor is cut off. As can be seen from the graph of diode current
in Figure 7-4, once bias voltage has risen to cutoff point, any increase produces a
corresponding increase in current. It is the same with the transistor. Above cutoff
point, the current flow is linearly proportional to bias voltage.
When Ispeak of current flow in this analysis, Iam referring to electron flow
(as opposed to holes flow, which is in the opposite direction). Consider an NPN
transistor in which the bias voltage W EB )is below cutoff. No current can flow from
E to B; however, there is aquiescent voltage between B and C that could cause
current to flow between these two points. This doesn't happen because the
polarities are such that the B/C junction is reverse biased. Thus, no current can
flow through either junction while VEB is below cutoff.
Now consider an NPN transistor in which the bias voltage is above cutoff.
Electrons immediately flow from the emitter into the base (through the now
forward biased E/B junction). Because the base is extremely thin and only lightly
doped, aminute fraction of this initial charge of electrons is enough to neutralize
the majority carriers (holes) in the base. This cancels the effect of the PN junction
between the base and collector. The result is that the rest of the incoming current
is conducted out through the collector by its higher positive polarity. So, in effect,
the base/collector junction ceases to be adiode junction after electrons from the
emitter flow into the base. This junction returns to reverse bias only when no
more current flows, which happens when VEB falls below cutoff.
So, the collector current, Ic,is determined by the current that flows through
the E/B junction, 1 E.The base, being wafer thin, offers high resistance to current
flow through its edge connection. Only about 2 percent leaks out this way. The
remaining 98 percent goes out through the collector.
A transistor can conduct current only if the correct quiescent voltages are
present at its terminals. As an analogy, it would be useless to press the accelerator
of acar whose engine is switched off. It wouldn't matter how hard the accelerator
were pressed; the car wouldn't move. Similarly, with a transistor, if the correct
quiescent voltages are not present, the transistor won't amplify a signal.
(Therefore, one of the first things to do when troubleshooting is to check that the
quiescent voltages are correct.)
In an NPN transistor, the collector voltage should be from 5to 30 V more positive
than the emitter. In aPNP, it should be more negative by the same amount. If the
collector voltage is 0, there is abreak in the power supply to the collector.
The bias voltage, VEB ,is far more critical and more difficult to measure. Whereas
it is possible to use an ordinary multimeter to measure the voltage between the
emitter and ground, between the collector and ground, or between the collector and
emitter, it is not possible to use amultimeter to measure the voltage between the base
and anything. This is because there is high resistance between the base and other
parts of the circuit. Although amultimeter has moderately high resistance, it is not
high enough. Consequently, the act of measuring the base voltage changes it, and a
 Semiconductors 135

completely false reading is obtained. Either aVTVM (vacuum tube voltmeter) with a
DC voltage range or a FET meter must be used. (It would be useful for anyone
wanting to do serious audio work to own aFET meter.)
Both the VTVM and FET meters use abuilt-in amplifier to power the meter
movement, so they draw practically no current and have almost infinitely high
input resistance. The advantage of a FET meter is that it is as portable as a
multimeter and cheaper than aVTVM. It contains its own field effect transistor
(FET) amplifier, which is powered by an internal battery.
If either aVTVM or FET meter is available, the bias voltage between base
and emitter can be measured. In an amplifier, the quiescent bias voltage should
be at or just above cutoff point. In an NPN transistor, the base should be 0.6 V (for
silicon) or 0.2 V (for germanium) more positive than the emitter. In a PNP
transistor, the base should be more negative than the emitter by this amount.
From consideration of these quiescent voltages, it is possible to tell if asemicon-
ductor is conducting. For example, in Figure 7-8, can you tell if these semiconduc-
tors are conducting? The answers are in the box below. (The elemental abbrevia-
tions of Si for silicon and Ge for germanium are used.)

Fig. 7-8 Can you

tell if these
semiconductors
are conducting?
2.5 V
0.9 V 0.3 V 1.0 V 1.2 V

(.4) (B) (C)

9.4 V

1.8 V
1.2 V

(F)

Answers to Figure 7-8

(A) Yes. Bias is 0.6 V forward.

(B) No. Bias is 0.2 V, which is numerically above cutoff, but in the reverse bias
polarity. (Electrons can flow only in the opposite direction from the arrow.)
(C) No. Bias is 0.2 V. Although it is in the forward polarity, it is below cutoff. 0.6
V is required for silicon.
(D) No. Bias is 0.2 V in the reverse polarity.
(E) Yes. Bias is 0.2 V in the forward polarity.
(F) Yes. Bias is 0.6 V in the forward polarity.
136 Chapter 7

Common Emitter Amplifiers

There are three basic configurations in which atransistor amplifier can be made.
Each has its own characteristics, and these determine the type of application for
which it is suitable.
The configuration most often used for general purpose amplification is the
common emitter. This circuit diagram (using an NPN transistor) is shown in Figure
7-9, where:

R1 is the load resistor, across which the output voltage develops.

R2 is the emitter, or stabilizing resistor. It is needed to prevent self-
destruction of the transistor by thermal runaway.
R3 and R 4 form avoltage divider, producing avoltage at the base that

ensures correct forward bias.

C1and C2 are coupling capacitors. These pass the audio signal while
blocking any DC voltage.
C3 is aby-pass capacitor. It by-passes any audio signal that might develop
across R2 to ground, thus cutting out any voltage fluctuations at the emitter.

Fig. 7-9 Common

emitter amplifier

VI

In order to understand how this circuit amplifies, look first at the quiescent
DC voltages at the three terminals of the transistor. The bias voltage between the
base and emitter is 0.22 V, so this must be agermanium transistor. The transistor
conducts about half its maximum current under quiescent conditions. The nega-
tive (electron) current flow is shown as adotted line.
The audio signal enters through C1.If we assume that the amplitude of this
wave is about .02 V, then the first half-cycle, being positive, will raise the voltage
Semiconductors 137

at the base by . 02 V, from 2.72 V to 2.74 V. This increases the forward bias from
.22 V to . 24 V, which in turn increases the current flow through the transistor.
This increased current, which flows through the load resistor RI,produces an
increased voltage drop across RI.Notice that under quiescent conditions, the
voltage drop across R1 is 4.5 V; however, when this increased current flows, it
produces agreater voltage drop.
Because the 12 volts at the top of R1 is abattery or regulated power supply
voltage, it cannot change, so the 7.5 V voltage at the bottom of R1drops. In fact, it
drops to about 3.5 V (a4V drop) at the peak of the first half-cycle.
In the second half-cycle, the signal voltage applied through C1is negative, so
the base voltage now drops by .02 V, to 2.70 V. This reduces the forward bias from
0.22 V to 0.20 V, and this, in turn, reduces the current flow. The consequently
reduced current reduces the voltage across R1 by the same amount as the
increase of the first half-cycle. So the voltage at the bottom of R1goes up toward
the 12 V supply by 4V, to 11.5 V.
Now the output waveform at the collector is 180° out of phase with the input
signal. There is also considerable voltage amplification. An input peak-to-peak
signal amplitude of . 04 V produces an output peak-to-peak signal amplitude of 8
V. So the voltage gain is in the ratio of 8to . 04, which is 200.
Although R2 has been described as the emitter resistor, or stabilizing resistor,
its function hasn't been fully described. The need for its presence derives from the
fact that semiconductor material becomes less resistant to current as its tempera-
ture rises. When current flows through a transistor, its temperature inevitably
rises. Because of the consequent reduction in its internal resistance, more current
flows. This, in turn, leads to a higher temperature, a further reduction in resis-
tance, and afurther increase in current. This process creates an unstable cycle
that leads to eventual burnout of the transistor due to thermal runaway. The
emitter resistor is able to prevent this.
When the current flow through the transistor increases, there is, of course, an
equal increase in current through the emitter resistor, R2. This causes an increase
in voltage across it, which pushes the voltage at the emitter toward that of the
base, thus reducing the forward bias. This forms afail-safe system that automat-
ically reduces the forward bias to cutoff, if the current should reach ahigh enough
level. This prevents thermal runaway, because it ensures that the current through
the transistor cannot exceed afixed amount.
There is one other component in this circuit whose function hasn't been
described. This is the by-pass capacitor, C3. As you know, the output signal from
this amplifier is produced by fluctuations in the current passing through the load
resistor, RI.But the same current fluctuations occur in the emitter resistor, R2.
Fortunately, R2 is very small compared to RI,so a current fluctuation that
produces a4V change in R1produces only about a4mV change across R2. Still,
even this 4mV change tends to reduce the gain of the amplifier. This is because
the 4mV voltage is in phase with the incoming signal and thus reduces the bias
change by 4mV. Remember that the peak signal voltage of . 02 V caused achange
138 Chapter 7

of this amount in forward bias. However, if the emitter voltage also changes in the
same direction by 4 mV, the net change in bias falls from .02 V to .02 — . 004,
which equals .016 V
This loss is prevented by the by-pass capacitor, which acts as asort of "shock
absorber," flattening out the fluctuations in voltage at the emitter. In this way, it
holds the emitter voltage constant. As a result, the gain of the amplifier is not
degraded. Some amplifier designers don't find this component necessary; there-
fore, amplifiers may be made without aby-pass capacitor.
Notice that, in this amplifier, only the emitter voltage is not fluctuated by the
audio signal. The base voltage is fluctuated by the incoming signal, and the output
signal is produced by fluctuations in the collector voltage. The emitter voltage
does not fluctuate. This remains constant and acts as acommon reference point to
the audio signal; hence, its name—common emitter configuration. The charac-
teristics of this type of amplifier are as follows:

Common Emitter Amplifier Characteristics

Input impedance Moderate (500 II to 12 kl1)
Output impedance Moderate ( 12 kg to 50 kg)
Current gain (= I
c/I B) Moderate
Voltage gain Moderate
Power gain Moderate
Phase reversal

It follows from these characteristics that the common emitter can be used for
general purpose amplification, because it gives moderate amplification of both
voltage and power. It can also be used as a middle stage in amplifying low to
medium power signals in a signal generator or other signal processing
equipment.

Common Base Amplifiers

The circuit diagram of atypical NPN common base amplifier is shown in Figure
7-10. The theory of operation for this amplifier is similar to that for the common
emitter configuration. In all of these amplifiers, R3 and R4 act as avoltage divider,
producing the required forward bias at the base. Also, C1 and C2 are coupling
capacitors, passing the AC audio signal and blocking the DC.
The only difference between the common base and common emitter is that,
in the common base circuit, the incoming signal fluctuates the emitter voltage
instead of the base voltage. Thus, in this configuration, the signal fluctuates the
forward bias by bringing the emitter voltage closer or farther from the base
voltage, instead of bringing the base voltage closer or farther from the emitter
voltage. The result is that, in the first half-cycle, when the input signal is positive,
 Semiconductors 139

Fig. 7-10 DC +
Common base
amplifier

the bias voltage is reduced (

VE being closer to VB instead of farther from it), and in
the second half-cycle, the bias voltage is increased. This phase difference pro-
duces an output signal that is not phase inverted. It is in phase with the incoming
signal. Also, because now the input current to the transistor is not the leakage
current from the base, but the emitter current, there can be no current gain. In
fact, there is aslight current loss, because all current that enters at the emitter (
IE

=l c +1 B), and, since l

c = 1E —I B,then the output current, lc, must be less than the
input current. However, due to the fact that R2 is less in value than RI,there is a
large voltage gain. The characteristics of this configuration are as follows:

Common Base Amplifier Characteristics

Input impedance Low (30 fl to 150 9)
Output impedance High (750 11 to 1M(l)
Current gain (= V I
E) Less than unity
Voltage gain High
Power gain Moderate
No phase reversal

Because of its extremely high voltage gain, this type of amplifier can be used
in the first stage of apre-amplifier, where very low voltage signals require large
voltage amplification. The low input impedance can accept asignal from alow
impedance line or microphone, but some voltage loss occurs. The high output
impedance can be coupled to a high input impedance following stage; however,
because of the disadvantage of such ahigh output impedance and the difficulty of
correct impedance matching, this type of amplifier is not often used.
140 Chapter 7

Common Collector Amplifiers

This configuration is often called an emitter follower. The circuit configuration of
an NPN common collector is shown in Figure 7-11.

Fig. 7-11 DC+

Common collector
(emitter follower)
amplifier

v,

In this circuit, R 3 and R,, as well as C1 and C2,fulfill the same functions as
they did in the two previous circuits. Notice that in the common emitter circuit,
the emitter voltage remained constant. In the common base circuit, the base
voltage remained constant. So also in the common collector circuit, the collector
voltage remains constant, acting as a fixed reference that is common to the
incoming and outgoing signal. As before, the fluctuating voltage at the base
fluctuates the bias, and in so doing, fluctuates the current through the transistor.
The most striking difference between this circuit and the previous circuits is
that there is no load resistor, RI,here. The reason is that the output is taken from
the emitter. It is not the output voltage, but the output current that is important.
For this reason Ihave included the load, RL;the dotted lines show the current
through the transistor, which flows partly through RL and partly through R2.This
fluctuating current, flowing in the voice coil of aloudspeaker, forms the powerful
output from this circuit, which drives the speakers and produces the sound.

Common Collector Amplifier Characteristics

Input impedance High ( 10 k9 to 500 k12)
Output impedance Low (0.5 f to 100 9)
Current gain (= I
E/I B) High
Voltage gain Less than unity
Power gain High
No phase reversal
 Semiconductors 141

Notice that the voltage gain is less than unity. This is because the output
voltage at the emitter is in phase with the input voltage at the base. However,
change of bias depends on the difference between these two, and unity gain
requires that they be equal. This is impossible, because the change in VE must be
greater than that in VE to produce bias change and activate the amplifier.
Therefore, voltage gain is always just less than unity in this circuit configuration.
In the absence of voltage gain, all gain occurs as current gain. For this
reason, system designers use this circuit as the output stage of all power ampli-
fiers. At this end of the audio chain, the driver amplifier has already raised the
signal voltage to about 8or more volts, so voltage gain is no longer needed. What
is needed is power gain. As you know, power can be calculated from the product
of current and voltage. Because the signal voltage is substantially the same at the
input and output of this amplifier, the power gain is roughly proportional to the
current gain. This is what is needed to drive alow impedance loudspeaker, so this
amplifier configuration is used to drive loudspeakers or other equipment where
powerful electrical signals are needed.
One output transistor might not have sufficient current gain to provide the
power needed by aloudspeaker system. However, there is amethod of combining
two transistors in the common collector configuration that provides extremely high
current gain. This arrangement is called aDarlington pair. The two transistors can
be separate or incorporated in asingle package called aDarlington transistor. The
connections and current paths are shown in Figure 7-12.

Fig. 7-12
Darlington
transistor

Ihave labeled the two transistors Q1and Q2.Assuming that these are silicon
transistors, the required bias between the E and B terminals is 1.2 V (0.6 V for
each transistor). It is the change in this bias voltage that controls the current, just
as in asingle transistor.
The output current, li,is amultiple of asmall leakage current, 1 2,from the Q1
base. But 12 acts as the main current through Q2. This, in turn, is associated with
an even smaller leakage current, 1
3,through the Q2 base terminal.
Assume that the current gain of Q2 is 40, and that the gain of Q1is 30. Then
the total current gain is 40 x30, which equals 1,200. This is such a useful
arrangement that a Darlington pair is almost invariably used at the output of a
power amplifier.
I42 Chapter 7

Class A Amplifiers
The class of an amplifier depends only on the bias point of its active component.
As can be seen from the current/voltage graph in Figure 7-13, the current
conducted by a transistor is linearly proportional to the bias voltage above the
cutoff point. This straight line proportionality stretches from cutoff to saturation.
Notice that class B amplifiers are biased exactly at cutoff point. But more about
class Bamplifiers later.

Fig. 7-13
Transistor
characteristic
curve !max SATURATION

BIAS
CLASS B CLASS A VOLTAGE
BIAS POINT BIAS POINT

A class A amplifier should be biased half-way along the straight portion of its
slope. The transistor then conducts half of its maximum current. The effect of an
input signal is to fluctuate the bias voltage; it can be seen from Figure 7-14A that
this causes similar fluctuations in the current output. The disadvantages of wrong
bias voltage can be seen from Figure 7-14B and C.
Because half of the maximum current flows through the transistor at all
times, this class of amplifier is energy inefficient. In amplifier stages where very
little power is used, such as in pre-amplifier and line amplifier stages, this offers
no problem. In the output stage of apower amplifier, however, large amounts of
current are used, so this energy inefficiency causes power supply and cooling
problems. Consequently, it is not often that aclass A output stage is found in a
power amplifier. Designers look for a more energy efficient amplifier classifica-
tion. This can be found in the class Bamplifier.
 Semiconductors 143

Fig. 7-14 Effects

of bias voltage in OUTPUT CURRENT UNBALANCED
FLUCTUATIONS OVERLOAD CLIPPING
aclass A amplifier

i
r— 1
I
I
I
I
i
i
I /
\
OPTIMUM
BIAS POINT A
/
I
i , I
BIAS FLUCTUATIONS I i MAXIMUM UNDISTORTED
DUE TO INPUT SIGNAL I r— SIGNAL NOW REDUCED
I I%, I I
I I % 1 I

I 1 I

(A) Optimum class A bias—gives maximum (8) Bias voltage too high—half-wave overload
undistorted output by using the full current clipping due to saturation
range of the transistor from zero to saturation.

UNBALANCED
OVERLOAD CLIPPING

I\
I \
I l
1 /

// I MAXIMUM UNDISTORTED
I I SIGNAL NOW REDUCED
N
\
1
I

(C) Bias voltage too low—half-wave overload clipping due to cutoff

144 Chapter 7

Class B Amplifiers
Figure 7-13 shows that this class of amplifier is biased exactly at cutoff point, so it
appears to conduct only the positive half of an audio signal. During the negative
half-cycle, the bias voltage falls below cutoff, so there is no second half-cycle
response.
This problem can be solved by feeding the signal into two complementary
polarized transistors, namely an NPN and aPNP. The result is that voltage moving
in the forward bias direction for one transistor moves in the reverse bias direction
for the other. Let us call the transistors A and B. This arrangement ensures that
while A is conducting, B is cut off. And while B is conducting, A is cut off. The
transistors are then said to be in push-pull. However, special arrangements have
to be made to bias each transistor at cutoff point, because A ( if a silicon NPN)
requires + 0.6 V, and B ( asilicon PNP) requires — 0.6 V of bias. The problem is
how to maintain this differential voltage of 1.2 V between the inputs of the two
transistors.
One simple way to achieve this is to use adirect coupled class A driver stage.
This feeds the signal to the bases of the two complementary output transistors. By
including two silicon diodes between the complementary inputs, the two voltages
are held 1.2 V apart (see Figure 7-15).
The 1.2 V differential voltage results from the barrier potential (which numer-
ically equals the cutoff voltage) of the two diodes. Because the class A driver, Q1,
ensures that current flows through these diodes at all times, there is always this
differential of 1.2 V across the diodes. Also, because this voltage remains virtually
constant, whatever current is flowing (unlike the voltage across aresistor), there
remains only a DC component between the signal inputs, as can be seen from
Figure 7-16.
The shaded area illustrates which part of the signal each transistor is conduct-
ing. The rest of the time, its bias has fallen below cutoff. The complete signal is
reconstituted in the loudspeaker. Here the current alternates as it flows, first in
one direction, then in the other. You could say that while the signal is positive, Q2
conducts, and the current (electron flow) flows through the loudspeaker and Q2 to
DC positive ( + ); while the signal is negative, current flows from the DC negative
( — ) through Q3 and the loudspeaker to ground.
Returning to the schematic of Figure 7-15, the three branch currents (shown
by dotted lines) are labeled 1 0,l, and 12. 1
0 is avery small current that enables R2

and R3 to act as avoltage divider and produce the required quiescent voltage at
the base of Qi.l is a larger current. Its fluctuations produce the output signal
across RI,which forms the input signal to the next stage. This, in turn, produces a
much larger fluctuating current, 1 2,through Q2 and Q3, which powers the

loudspeaker.
With an understanding of the principles behind semiconductors, construction
of simple amplifiers can be undertaken, from which useful additional experience
 Semiconductors 145

Fig. 7-15
Complementary
class B output
stage fed by a
direct coupled 12

class A driver R1

R21
+0.6 V

Vg H OV

-
— 0.6 V
R3

— 15 V

DC—

Fig. 7-16
Differentially
biased audio
signal inputs
feeding a +0.6 V TO BASE OF 02
complementary
class Boutput
transistor

- 06 V TO BASE OF 03

can be gained. But more importantly, basic troubleshooting and maintenance can
be undertaken with an understanding of the working principles behind ampli-
fiers. This is far more satisfactory than relying on other people to maintain
equipment in good condition and make basic repairs.
146 Chapter 7

Field Effect Transistors

A completely different type of transistor is becoming commonly used to meet the
increasing demand for improved signal-to-noise ratio brought about by the advent
of digital audio. This is a unipolar device called a field effect transistor ( FET).
Unipolar refers to the fact that the current flows entirely through asingle doped
medium, either an N-doped channel or aP-doped channed. In this case, asingle
polarity (unipolar) conductor is used, whereas in aconventional transistor, cur-
rent has to negotiate two PN junctions and flow through both N- and P-type
material. For this reason, aconventional transistor is referred to as bipolar.
The FET has two main advantages over abipolar transistor. It produces less
background noise. This is because the electron flow through aPN junction causes
much of the noise in a bipolar transistor. Also, the FET has far greater input
impedance. In the form of an insulated gate FET, the input impedance is as high
as 15 Ma—almost infinitely high—so there is no leakage current, as is the case
with abipolar transistor.
It has been mentioned that the bias voltage controls the current through a
conventional transistor. Because about 2percent of the collector current is always
leaking out through the base, the main current flow becomes proportional to the
leakage current. The bipolar transistor is, therefore, called a current controlled
device. Because there is virtually no current leakage from a FET, this unipolar
transistor can be considered avoltage controlled device.
The disadvantage of a FET compared to a conventional transistor is that it
can handle less current than a power transistor. However, in a pre-amplifier,
where current flow is small anyway and low background noise is at apremium,
FETs are ideally suitable, and a FET pre-amplifier or op-amp is preferable to a
bipolar transistorized product.
The theory behind the operation of aFET is as follows. The structure consists
basically of asingle doped channel surrounded by asleeve of oppositely doped
material. The ends of the channel are called the source ( where the current enters)
and the drain ( where the current leaves). The surrounding sleeve is called the
gate. This controls the amount of current that can flow through the channel. The
basic structure is illustrated in Figure 7-17A and B, and the corresponding sche-
matic symbols in Figure 7-17C and D.
Notice that the direction of the arrow in the schematic symbol indicates
whether the channel is N or R As ever, the arrow indicates the direction of
conventional current flow (holes flow) across ajunction. Between the gate and
channel in Figure 7-17C, forward bias would occur if the gate (P material) were
more positive than the N channel. So, the inward pointing arrow signifies that the
channel is N. Similarly, the outward pointing arrow in Figure 7-17D indicates aP
channel.
If no voltage were applied to the gate, the channel would conduct current
with practically no resistance, so maximum current flows when there is no bias
voltage. (Bias voltage is the voltage between the source and gate, VsG .) Voltage in
 Semiconductors 147

Fig. 7-17 J-FETs

SOURCE U DRAIN SOURCE DRAIN

+ GATE

(A) N-channel construction (8) P-channel construction

(C) N-channel symbol (D) P-channel symbol

the same polarity as from Sto D is considered forward bias; therefore, avoltage at
the gate that is more negative than the source has to be applied in an N-channel
FET to limit the current flow. The polarities required for correct functioning are
shown in Figure 7-17.
Let us consider the action of an N-channel FET to see how the current can be
controlled by areverse bias voltage at the gate. As you know, adepletion zone,
where majority current carriers have been cancelled, occurs spontaneously at a
PN junction. (See the first section in this chapter.) By applying anegative voltage
to the gate of an N-channel FET, the negative field extends farther into the
channel than the extent of the normal depletion zone. This negative field repels
any free N current carriers and has the effect of increasing the depletion zone by
expelling negative carriers. This reduces the effective size of the channel through
which current can flow.
The greater the negative voltage applied to the gate, the farther the negative
field extends into the channel, reducing the channel size even more. If enough
negative voltage is applied, the N channel can be pinched off by the expanding
negative field effect. This completely cuts off all current flow from source to drain.
The gate voltage that entirely cuts off current is called the pinch-off voltage, as
opposed to the cutoff voltage in abipolar transistor (see Figure 7-18).
These FETs, in which the gate is in contact with the channel, are called
junction FETs ( J-FETs) to distinguish them from insulated gate FETs, which will
be discussed later. The characteristics of J-FETs are

• The J-FET conducts maximum current when no bias is applied.

• Reverse bias applied to the gate controls the current by reducing the
channel size, due to an electrical field that extends from the gate into the
channel.
148 Chapter 7

• If sufficient reverse bias is applied to the gate, the current channel can be
completely pinched off by the extent of the electrical field.
• J-FETs are atype of depletion FET, so called because the effect of bias is to
deplete the current carriers and reduce channel conduction accordingly.

Fig. 7-18 Effects

of negative
voltage applied to
the gate
rA
G

(A) Field produced by reverse bias reduces (8) Field pinches off all current when reverse
channel width and restricts current. bias reaches pinch-off level.

J-FETs work extremely well under reverse bias conditions. As long as their
circuitry never requires them to withstand forward bias, there is no problem. But
there are occasions in which an overload or some other situation could forward
bias a J-FET. If this should happen, it would cause serious trouble, because no
measurable current should flow from the gate into the channel. When the J-FET is
reverse biased, this cannot happen. But if the FET should become forward biased,
the gate/channel junction would also be forward biased and it would conduct a
large amount of current.
To deal with this problem, an insulated gate FET was developed. The insula-
tion between the gate and channel must be very thin, so that the electrical field
from the gate can penetrate the channel to asufficient depth. This insulation layer
is, in practice, created by plating athin film of nonconducting metal oxide onto
the inside of the channel. This is called ametal-oxide semiconductor field effect
transistor (MOSFET).
This layer of insulation has to be so thin that even the static charge on a
person's finger or the tip of asoldering iron can break it down. This would destroy
the transistor. To avoid this, the operator is advised not to touch the gate. A
soldering iron should have aflexible jumper connected by an alligator clip cable
between the tip and aconvenient ground during soldering.
There are several advantages to using MOSFETs instead of J-FETs. The first,
already mentioned, is that atemporary reversal of bias will not cause current to
flow from the channel through the gate. The other advantage is that acompletely
new family of FETs can now be created, because it becomes possible to use
forward bias without unfortunate consequences. This new family of FETS, called
enhancement MOSFETs, is shown in Figure 7-19.
Let us consider an N-channel enhancement MOSFET (aPchannel works the
same way, but with reversed polarities), as in Figure 7-19. The structure of an
enhancement MOSFET can be viewed as having the gate on one side of the
channel and asubstrate on the other. This substrate is heavily doped with P-type
 Semiconductors 149

Fig. 7-19
Enhancement
MOSFETs UNDOPED GATE

HIGHLY DOPED SUBSTRATE

(A) Structure

S S
N CHANNEL P CHANNEL

(B) Schematic symbols

Fig. 7-20
Summary of field
effect transistors

N CHANNEL S P CHANNEL S

(A) J-FETs (Depletion type—reverse bias reduces current.)

o
N CHANNEL S
P CHANNEL S

(B) MOSFETS (Depletion type—reverse bias reduces current.)

S
N CHANNEL S P CHANNEL

(C) MOSFETs (Enhancement type— forward bias increases current.)

150 Chapter 7

material. Its junction with the channel (aPN junction) produces alarge depletion
zone, which extends right across the channel to the gate. So the channel is
completely pinched off, even when no bias is applied.
The gate is not doped at all. When forward biased with apositive voltage, the
gate emits apositive electrical field that extends into the channel. This reverses
some of the depletion effect caused by the substrate, and effectively repels some
of the positive current carriers from the substrate that have neutralized N carriers
in the N channel. In this way, the positive field from the gate enhances the
number of free current carriers. This enables current to be conducted. The
greater the forward bias, the greater the amount of current that flows. The
substrate is usually internally connected to the source. It is not the voltage of the
substrate, but its doping that enables it to produce the required depletion effect.
To summarize, an enhancement MOSFET conducts no current when unbi-
ased. When forward bias is applied to the gate, current flows in proportion to the
bias, up to saturation point. A summary of the different types of FETs and their
modes of action is given in Figure 7-20.
Tape Recording Concepts, 153

Theory of Operation, 154

Mechanical and Electrical Requirements, 155

Internal Equalization, 161

Electronic Circuit Requirements, 162

151
 8 The Tape Recorder

Tape Recording Concepts

Surely one of the greatest achievements of mankind has been the ability to record
and reproduce sound with clarity and fidelity. Perhaps the most useful and
versatile mechanism for doing this is the tape recorder. Practically everybody
uses a tape recorder, but not many have the ability to use one to obtain top
quality sound recording and reproduction. Detailed technical knowledge and an
understanding of the principles underlying magnetic recording are necessary to
achieve this quality.
It is not necessary to be aprofessional studio engineer, nor is it necessary to
use the most expensive studio equipment. What is necessary is to use a well
manufactured consumer product and, above all, to understand thoroughly how it
functions and how best to utilize its abilities and limitations.
All equipment has limitations. The competent operator knows what those
limitations are and adjusts his/her activities accordingly. For example, it is
possible to overload a recording. Too high a recording level produces overload
distortion of the amplitude peaks. Too low arecording level definitely avoids this
problem, but then the quiet passages become lost in background noise. Adjusting
the sound level within the parameters of the dynamic range available, as
indicated by the recording level meter, can avoid both causes of sound
degradation.
Manufacturers are well aware of the comparative incompetence of many tape
recorder users and they cater to this market, so they have to try to make their
equipment as "idiot proof" as possible. To do this, they often incorporate abuilt-in
automatic level control. This acts as alimiter and avoids overload distortion by

153
154 Chapter 8

automatically attenuating peak signals. But it reduces the dynamic range

available, so it produces, at best, a second-rate recording. The knowledgeable
recording operator avoids using this built-in device whenever possible and relies
on his/her regulation of the recording level to obtain the best possible recording
results. The requirement, then, is to understand the theory and practice of tape
recording. And the first requirement is to understand the theoretical principles
behind magnetic recording.

Theory of Operation
Before going into adetailed description of how sound is magnetically recorded
onto tape, Ineed to mention the domain theory of magnetism. This theory
postulates that the molecular structure of iron (or other permeable material)
contains alarge number of domains. Each domain behaves as avery small bar
magnet, with its own north and south pole. Normally, the domains are arranged
in a haphazard manner, so that the individual polarities cancel out. Figure 8-1
shows these domains in ( A) an unmagnetized, (B) apartially magnetized, and (C)
afully magnetized state.

Fig. 8-1 Magnetic ••••111..

domains r „.11e

—11." •••••1111. 111».

(A) Unmagnetized (B) Partially magnetized (

C) Fully magnetized
(magnetically saturated)

When astrong magnetizing force, in the form of amagnetic field, is applied to

this material, it reorients the domains so that they cease to be haphazard and
start to align themselves in the direction of the magnetic field. This causes the
material to become magnetized. The degree of magnetization is determined by
the proportion of domains that become aligned to those that remain haphazard.
The magnetic tape used in tape recording consists of alayer of permeable
material (basically iron oxide) deposited evenly onto aflexible tape base. The iron
oxide contains multitudes of domains. Those that come under the influence of a
magnetic field become magnetized. That is, they become polarized in one
direction or another depending on the direction of the applied field. If the N pole
is to their right and the Spole is to their left, the domains become oriented with
their S poles to the right and their N poles to the left. If the applied field is
reversed, the domains also reverse their polarity. This is basically what happens
when sound is recorded onto magnetic tape.
The tape head consists of a small horseshoe piece of permeable material.
Think of it as soft iron that can easily be magnetized in one direction or the other.
A coil of thin insulated wire is wound around it, and as a current is passed
 The Tape Recorder 155

through this coil, the iron turns into an electromagnet, creating astrong magnetic
field in the air gap between its poles. If the current is reversed, the polarity of the
field is reversed. In fact, the field fluctuates in strength and polarity in exact
proportion to the current in the coil.
In atape recorder, the audio signal runs through the coil that energizes the
record head and produces in its air gap a fluctuating magnetic field that is
precisely proportional to the audio signal. Meanwhile, the tape is passing over the
head at constant speed, so the magnetic imprint on the tape exactly corresponds
to the instantaneous value of the audio signal at all times. This can best be seen
by an illustration of one full sound wave cycle, as shown in Figure 8-2.
Notice that as the polarity of the current flowing through the head reverses,
so the polarity of the magnetization imprinted on the tape also reverses. In this
way, the signal voltage is magnetically recorded onto the tape so that its intensity
and polarity exactly parallel the audio signal current.
When it is necessary to reproduce this recording, the tape is rewound and the
connections from the head are reassigned to the pre-amplifier of the reproduce
system. During playback, the tape is again transported past the head at constant
speed. But now the magnetized domains on the tape produce the fluctuating
magnetic field in the air gap between the poles of the tape head. It is interesting
to note that the process of electromagnetism is reversible. Just as a fluctuating
current in the head produces afluctuating magnetic field in the air gap, so, during
reproduction, a fluctuating magnetic field in the air gap produces a fluctuating
current in the head coil. True, this is avery small current. But it is applied through
ahead pre-amplifier, amplified again in the reproduce amplifier, and then passed
to apower amplifier, which drives the speakers.
To summarize, during the recording process, the audio signal is applied to the
coil of the tape head. This magnetizes the tape as it passes over the head, with a
polarity and intensity that exactly parallel the value of the audio signal voltage,
both in amplitude and frequency. During reproduction, the strength and polarity
of the magnetic recording on the tape produce afluctuating magnetic field across
the tape head, so that the induced current in the head coil exactly parallels the
magnetic field in amplitude, polarity, and frequency.
This sounds very simple. In essence it is. And, as you know, it works. But
when this new invention of magnetic recording was first being developed, some
very serious problems had to be solved. These problems and the solutions that
were eventually found to overcome them, are discussed in the next section.

Mechanical and Electrical Requirements

A tape recorder (and its more sophisticated version, the video recorder) is a
combined product of mechanical and electronic engineering. The mechanical
parts of the system have to produce atape transport mechanism that can move
the tape past the head at precisely constant speed while holding it in vertical
156 Chapter 8

Fig. 8-2 Magnetic

recording of afull
sound wave cycle

(A) Plan view of head

HEAD
TAPE TRAVEL

..I- •••,- - ,... "', '\ - - - tt ..- -I- -

- , .1 / ,e — -.-. .- ,e le
TAPE
— ...- — ... \ -.- / / / / / — — /
— •- \ \ -... -.... -..- — \ \ -...- — — \ \ \

1__I ILie
N-- S S- N N-LS

PEAKS PRODUCE
S- N POLARITY

AUDIO
SIGNAL
CURRENT

TROUGHS PRODUCE
N- S POLARITY

(B) Magnetic recording process

alignment with the head. It must also be able to either fast forward or fast rewind
the tape when required. It must be able to maintain tape tension within limits that
do not stretch or break the tape. In addition, the mechanical construction of the
head must fulfill many exacting demands that will be discussed later.
 The Tape Recorder 157

Electromagnetic Requirements
Perhaps the most difficult problem that first had to be overcome was
electromagnetic. The previous section described how the magnetic imprint on the
tape had to parallel the signal current passing through the record head exactly.
But in fact, when this method of magnetic recording was first tried, this linear
relationship between magnetization and head current did not take place. There
was extreme distortion, which, if uncorrected, would have made magnetic
recording impractical.
The problem was related to what could be thought of as internal static
resistance to the movement of domains when a reversal of magnetization was
initiated. It was a form of magnetic hysteresis. Hysteresis is described in the
appendix at the end of this book, but briefly, it appears that once the process of
magnetization is under way, it continues in proportion to the magnetizing force.
However, as soon as there is areversal of the magnetizing force, aresponse lag
occurs. The result is that aconsiderable change of magnetizing force has to take
place before any corresponding demagnetization occurs. This is what causes
distortion at the peaks of the waveform. And this is what, at first, seemed to make
the dream of magnetic sound recording unobtainable.
This static internal friction, which opposes the initial process of reversal, can
best be illustrated by an analogy. If iron filings are sprinkled onto apiece of paper
and a magnet is placed beneath the paper, the filings move into a pattern,
aligning themselves with the magnetic lines of force. If the magnet is slowly
moved, the filings do not move at first. Eventually, however, they start to move
and from then on continue to follow the motion of the magnet. If the magnet
reverses direction, there is another lag in the movement of the filings. This occurs
at every reversal of motion. It is due to static friction between the filings and the
paper.
The problem in magnetic sound recording is how to make the domains
respond without atime lag at the points of magnetic reversal, which is at the wave
peaks. To continue with our analogy, the problem boils down to making the iron
filings respond instantly, without any delay due to static friction, when the
magnet starts to reverse its direction.
If, as apossible solution, the paper is jiggled rapidly up and down, so that the
filings are always in motion, they continue to align themselves in accordance with
the magnetic field lines as they fall. However, in this situation, there is no delay in
their response to areversal of motion as the magnet starts to reverse its direction.
This is because the filings are freed from static friction by the up and down
motion produced by the jiggling.
What could be used in magnetic recording that corresponds to this jiggling of
the iron filings? The superimposing of ahigh-frequency, ultrasonic signal onto the
audio signal was found to be the answer. This superimposed frequency, applied
during the recording process, is called bias voltage. It is always AC and ultrasonic;
in fact, it is usually between 80 and 200 kHz. It can be filtered out during
reproduction by a bias trap (a sharp cutoff stopband filter). What is left is a
158 Chapter 8

perfect, undistorted magnetic recording, which can be reproduced with perfect

fidelity. The optimum bias voltage to be used depends on the characteristics of the
recording tape, and to a slight extent, on the preference of the recording
engineer. A method for determining asuitable bias voltage is described later.

Record/Reproduce Head Requirements

The construction of the record/reproduce head was also alimiting factor in the
performance of early tape recorders. The problem was in achieving a narrow
enough head gap so that the full range of audible frequencies could be recorded.
It was easy to record the lower and mid frequencies, but the highest frequencies
were a problem. That is because there is a definite relationship between the
highest recordable frequency and the width of the head gap. The higher the
frequency, the narrower the air gap has to be, or alternatively, the faster the tape
speed has to be, or both.
In the early days, high tape speeds of 71
2 or 15 ips were necessary. Even then,
/
frequencies could be recorded only up to about 6kHz. Manufacturing techniques
have since improved. Cassette recorders, running at FA ips, can now record up to
16 kHz, an octave higher than could formerly be recorded at 71 2 ips.
/
The reason why head gap, tape speed, and maximum frequency are
interrelated results from the wavelength occupied by each sound cycle.
Remember that tape runs at aconstant speed; therefore, low frequencies of, say,
20 Hz produce 20 cycles (or waves) in the length of tape that runs in 1second. At a
speed of 71/ ips, a length of 71
2 / /20 inches is occupied by each cycle. This is a
2

0.375-inch wavelength. The head gap has to be not more than half this
wavelength, so that the positive half-cycle and negative half-cycle are not
included in the head gap at the same time. If they were, they would cancel each
other out.
At this tape speed of 71
2 ips, the maximum size of head gap can be calculated
/
for the most difficult situation, which occurs at the highest frequency. For good
results, arecorder should record up to 15 kHz. At atape speed of 71 2 ips, 15,000
/
cycles occupy atape length of 71 2 inches. So 1cycle occupies a wavelength on
/
tape of 712 /15000, which is 0.5 mil—that is, 0.5 of a thousandth of an inch.
/
Because the head gap must be not more than half this length, ahead gap of 0.25
mil is needed. In fact, most audio heads for this tape speed are made with a0.2-
mil gap. However, at 15 ips, the wavelength occupied is twice as long (twice as
much tape having passed during this time period), so this same head can record
up to 30 kHz (twice the frequency), other factors being equal.
Now consider how narrow the gap must be on acassette recorder in which
the tape speed is only 1 / ips. This must record 15,000 cycles in alength of 1
4
1 /
4
1

inches. So each cycle, each wavelength, is 1.875/15000, which is 0.125 mil. But
the head gap must be less than half this amount, so the head has to be ground to a
gap width of 62.5 microinches.
Other factors that have to be accommodated in tape head construction
include such things as use of highly permeable material that can respond readily
 The Tape Recorder 159

to the magnetizing field. Also, hardness is required to resist premature wear

resulting from the friction of the tape. Unfortunately, these two requirements tend
to be mutually opposed. But special ferrite materials are being used for tape head
construction, which give long hours of use with acceptable permeability. Head
wear is afactor for which the recording engineer must be alert. Once atape head
wears past acertain point, the high-frequency response falls off sharply. Regular
frequency response tests with atest tape are needed to ensure that the recorder
continues to operate satisfactorily. Loss of HF could indicate head failure due to
wear.
The reason why HF loss occurs due to wear is that wear tends to increase the
width of the head gap. One of the telltale signs to watch out for (you cannot see
the head gap; it is too small) is on the face plate surrounding the head. If the tape
has worn this down so that it has channeled out arecess for itself, with shoulders
above and below the tape path, then you can be sure that the tape has eaten into
the head by the same amount.
It is also necessary to keep the head clean. Any accumulation of dirt on the
surface of the head(s) has adisastrous effect on the high-frequency response. The
reason is that dirt has the effect of lifting the tape away from the head surface.
This enables the magnetic field to spread out on each side of the head gap before
it reaches the tape. Thus, it effectively widens the head gap. It has already been
shown that any widening of the head gap reduces the high-frequency response.

Tape Transport Requirements

Another requirement of the mechanical tape transport system is that the tape
pass the head at precisely constant speed. Because no electromechanical system
is perfect, there are bound to be slight fluctuations in the friction and motor
torque, so special precautions have to be taken to even out any resulting speed
variations. The main precaution is the provision of a heavy, smooth-running
flywheel attached to the capstan motor shaft. The rapid rotation of this flywheel
evens out most of the inevitable speed fluctuations by virtue of its rotational
inertia.
It should be understood that the tape speed is entirely determined by the
capstan motor, which drives the capstan. During the record and reproduce modes
of operation, the tape is held tightly against this capstan by arubber pinch roller.
This prevents any tape slip between the capstan and the tape. So the tape speed
is the same as the speed at which the circumference of the capstan is moving.
Through the use of an accurately speed-controlled capstan motor and heavy
flywheel, speed is held as constant as possible. Remember that the take-up motor
and rewind motor should not affect the tape speed in any way during the record
or reproduce modes. They affect the tape speed only during fast forward and fast
rewind, when the tape is out of contact with the capstan and the heads.
During the record and reproduce modes, there is avery slight torque from
both the rewind motor and the take-up motor. This slight torque is merely to
prevent any slack between the supply reel and the heads, and to take up the slack
160 Chapter 8

as the tape leaves the capstan (so that it doesn't accumulate on the floor). Test this
out. During playback, stop the take-up reel by hand. You will see that the tape
continues to pass the heads at the same speed due to the action of the capstan and
pinch roller. Some unwound tape will then start to accumulate. Stop the recorder
and wind up the slack by hand. Don't just let go, or the rapid winding of the take-
up reel might stretch or break the tape when the end of the slack has been
reached.
Any fault in this constant speed system usually manifests itself as what is
called wow or flutter. By playing aconstant mid-frequency note from atest tape,
it is easy to hear the effect of any speed fluctuations. A fairly slow, cyclical rise
and fall in the speed sounds like a repeated "wow, wow" superimposed on the
sound. This can be caused by afault in the capstan motor, producing acyclical
fluctuation in motor speed. It can also be caused by excessive take-up torque
produced by the take-up motor; or by friction, as bent and pinched flanges on the
take-up reel catch on the tape at each revolution and cause intermittent tape slip
past the capstan.
Try holding the take-up reel stationary for afew seconds and see if this cures
the problem. If it does, you have the cause. The trouble might also come from the
supply reel. To find out, try unwinding afew feet of tape; then, holding the supply
reel stationary, play back the slack tape. If that cures the problem, the fault was
connected with the supply reel. If neither of these are at fault, examine the
capstan, pinch roller, and capstan motor for problems.
The other audible effect of unsteady speed is called flutter. This is a rapid
cyclical fluctuation of speed, which sounds like a flutter superimposed on the
sound. It is often caused by an accumulation of dirt (usually magnetic coating
from the tape) adhering to the capstan. This lump of dirt produces a slight
eccentricity in the radius of the capstan. Because this is rotating at constant
speed, the intermittent excess radius causes the tape to speed up momentarily
every time that piece of dirt inserts itself between the capstan and tape (once per
revolution). This problem is easily cured by cleaning the capstan and pinch roller
with denatured alcohol on the end of a cotton swab. This should be done
regularly whenever the heads are cleaned.
In portable recorders, the speed regulation of the capstan motor is often
achieved by an electromechanical governor. In line-powered machines, it may be
achieved by using a synchronous motor. Synchronous motor speeds are
automatically linked to the AC line frequency that powers them. Variable speeds
can be obtained by using apower amplified signal-generator output to drive the
capstan synchronous motor. The oscillator frequency can then be used to control
the tape speed (within limits). However, digital logic is taking over where
electromechanical systems used to be employed.
Digital control of tape speed can be achieved quite simply. A multitoothed
wheel is included on the capstan drive shaft. A narrow beam of light is directed
through this toothed wheel onto a photosensitive cell, whose output goes to a
digital counter. At each sample period (which may be every half-second or so),
the number of teeth interrupting the light beam is digitally counted. This count is
 The Tape Recorder 161

then compared, using agreater than/less than comparator, to abinary number

stored in ROM (read only memory), which is incorporated by the manufacturers.
The ROM has stored in its memory the exact required number of teeth that should
pass the light beam during one sample period, assuming correct tape speed. The
output of the comparator indicates if the recorder is running too fast or too slow,
and aspeed adjustment is automatically made until the two numbers coincide.
This test is carried out every half-second or so, depending on the sampling
frequency. It ensures extremely accurate speed control, and it can also be linked
to other compatible timing systems to synchronize two or more recorders.

Internal Equalization
The equalization and alignment routine needed for good quality recording is
described in Chapter 11. Before we are ready for that, however, it is necessary to
say something about the record and reproduce equalization already built into the
tape machine. It is generally accepted that bass notes tend to become lost in
background noise, and also that, during recording, the high frequencies become
attenuated due to the bias voltage. Bias, as you know, is necessary to avoid
distortion. It also helps to reduce background noise. But the fact is that it
somewhat attenuates the high frequencies. The greater the applied bias, the more
attenuation of high frequencies there is.
To counteract these losses, the record amplifier contains an equalization filter
that slightly boosts the low frequencies and considerably boosts the high
frequencies. This boosting of the signal before recording is called pre-emphasis.
The bass boosting is intended to be cancelled by an equal bass cut during
reproduction. Its only purpose is to reduce noise. The high-frequency boosting is
necessary to counteract losses.
During playback, an equalization filter is incorporated in the reproduce
amplifier. This reproduce filtering is called post-emphasis. It is needed to
counteract the bass boost supplied during recording, and again to boost the high
frequencies to counteract new losses. In order for recording to be compatible on
all machines, the post-emphasis filter has been standardized. In this country, it is
called the NAB Standard Reproducing Characteristic. (NAB stands for the
National Association of Broadcasters.) In Europe, there is aslightly different post.
emphasis standard. It is called the CCIR Reproducing Characteristic. (CCIR stands
for Consultative Committee for International Radio.) Its main difference is that
there is no bass cut of 10 dB at 30 Hz.
There is one other important filter added to the post-emphasis network. Called
an integration filter, it acts as apole that boosts the low frequencies by 6dB/octave
at each lower octave below 15 kHz. This is needed because the output from the
reproduce head falls off by 6dB at each lower octave as the frequency falls.
Remember that the head gap is designed to equal half the wavelength of the
signal at 15 kHz. At each lower octave (when the frequency is halved), only half of
162 Chapter 8

the correspondingly longer wave is contained within the head gap, so the signal
amplitude falls by half (6dB). You can also look at it this way: In the inductor that
forms the head, it is the rate of change of magnetic flux that induces the voltage.
Lower frequencies become flatter as their wavelengths extend, so the slope (rate
of change) reduces in proportion to the reduction of frequency.
To summarize, during recording, there is apre-emphasis of the signal, which
boosts the low and high frequencies. During reproduce, there is apost-emphasis
consisting of two filters. One is the NAB filter, which cuts the low and again boosts
the high frequencies. The other acts as apole and boosts the low frequencies to
counteract the reduced response of the head as the frequency falls. The result of
all this filtering should be zero, in that the reproduced signal should be precisely
the same as the recorded signal.

Electronic Circuit Requirements

Every tape recorder requires both arecord and areproduce head pre-amplifier,
each with its own filtering network. It also requires an ultrasonic oscillator. The
oscillator serves two purposes. In the record mode, it provides the AC erase
voltage to the erase head. Also, at an attenuated level, it supplies the AC bias
voltage to the record head. During reproduction, this oscillator is deactivated, so
that no erase or bias voltage is applied.
The oscillator has to comply with two important constraints. Both its output
frequency and voltage must be temperature stable and not drift as the equipment
warms up. Also, the waveform must be symmetrical. Any asymmetry in its output
waveform increases background noise. The oscillator circuit contains an
amplified output, so that it can supply sufficient voltage to magnetically saturate
the tape as it passes over the erase head. This ensures that the tape is completely
erased. In Figure 8-3, it can be seen that there is a pad ( resistive network)
between the oscillator and the record head. Its function is to reduce the oscillator
output to asuitable voltage level to supply bias to the record head. This must be
low enough that even maximum signal voltage, combined with the bias voltage,
does not quite saturate the tape magnetically. Of course, various options are
possible. Instead of having separate potentiometers for controlling record and
replay gain, asingle potentiometer can be used. It is assigned to its appropriate
position by operation of the record/replay switch.
The bias trap at the output of the record amplifier prevents any bias voltage
from feeding back and overloading the record amplifier output. Similarly, during
replay, no ultrasonic bias signal should be amplified by the reproduce amplifier.
This explains the positioning of the bias trap.
A more sophisticated professional tape machine might be a modest 4-track
portable using 1 /-
4 inch tape. Or it might be a 24-track, 1-inch studio recorder
designed to operate in conjunction with aconsole. A typical circuit block diagram
for one track of such amachine is shown in Figure 8-4.
 The Tape Recorder 163

LOUDSPEAKER

PLAYBACK
LEVEL

RECORD/REPLAY
HEAD

RECORD
LEVEL

Fig. 8-3 Block diagram of consumer-type tape recorder circuit

The main difference between this and the consumer model is that here there
are three heads; aseparate head is used for record and reproduce. The advantage
of this is that the recording can be monitored directly from the tape by the
reproduce head while it is actually being recorded. Also, in this machine, the record
and reproduce equalization can be user aligned and the bias can be set by the user
to produce optimum results. (See Tape Recorder Alignment in Chapter 11.)
Notice that there is no power amplifier or loudspeaker in astudio recorder.
The output at line level is intended to be fed to a console or mixer. An
independent power amplifier and monitor speakers are installed separately from
the recording machine. (However, aprofessional portable for location recording
would probably contain apower amplifier and monitoring speaker.)
Between the bias adjust and record head is abuffer. This is an amplifier with
avoltage gain of less than unity. It has the effect of reducing the erase voltage to
the level needed for biasing the record head. An amplifier is used in this circuit,
164 Chapter 8

REPLAY
HEAD
A
OUT
INPUT? REPRO

RECORD
CALIBRATE

RECORD
HEAD

IN
RECORD C
REPRO
LEVEL

REC RD

ERASE
BIAS
HEAD
ADJUST

OSCILLATOR
C

Fig. 8-4 Typical circuit block diagram of aprofessional tape machine

rather than the passive resistors shown in Figure 8-3, because in this circuit, the
bias voltage is adjustable. Such adjustments would cause achange in the current
drawn from the oscillator, if a pad were used. This, in turn, would reduce the
oscillator stability. By using a buffer instead of a pad, any alteration of power
needed to supply bias to the record head is provided by the buffer amplifier, while
its input impedance remains unchanged. In this way, the input is buffered from
any variation in load at the output. This ensures improved stability in the
oscillator's output voltage to the erase head.
You can see from the switch labeled A in the repro path of Figure 8-4 that
there are three switch positions available. In the Repro position, the tape is
reproduced, as would occur during playback or during monitoring of arecording.
During arecording (or while setting up prior to arecording session), it is useful to
be able to compare the incoming signal with the signal actually being recorded
onto the tape. This comparison can be made by switching from Repro to Input
The Tape Recorder 165

and back to Repro again. This does not affect the continuity of the recording. In
the Input position, monitoring of the input signal can be obtained before it
reaches the recording amplifier. Other than the half-second or so time interval
(by which the reproduced signal lags the input, due to the head positioning), there
should be no difference in tone or quality between the input and reproduced
signal.
Yet another useful facility is the ability to synchronize the recording of one
track with a previously made recording on another track. This is useful for
overdubbing additional music lines or instruments onto aprevious recording. In
this way, asingle musician can accompany him/herself on anumber of different
instruments, all to be reproduced simultaneously. For this purpose, the track
being reproduced requires that switches A and B be in the Sync position. You can
see from Figure 8-4 that the record head then reproduces the recording from that
track, through the sync amplifier to the output. It only remains to patch this
output to the monitoring headphones worn by the musician(s), and they can then
record onto another track whatever additional music (or speech, or sound effects)
are required, while listening to the previous recording.
If the reproduce head were used for synchronizing, any newly recorded
signal would be misplaced on the tape by the distance separating the record head
from the reproduce head. It would then be about a half-second late. By
reproducing from the record head (which is vertically aligned with all other
record heads), any new recording exactly synchronizes with the original track.
On amultitrack machine, anumber of tracks can be mixed down onto asingle
track using this synchronizing method; however, it is often preferable to mix
down onto aseparate recorder.
Digital Systems, 169
Analog/Digital Interfacing, 171

Digital Audio Applications, 178

Advantages of Digital Audio, 181

167
 9 Digital Audio

Digital Systems
A system consists of three parts. They are (a) the input sensors, (b) the central
processing unit (CPU), and (c) the output devices. Practically everything can be
considered asystem. A car is asystem, atree is asystem, and ahuman being is a
perfect example of a system. In the case of a human, the senses are the input
sensors. These are connected by the sensory nervous network to the brain, which
is the central processing unit. As aresult of certain inputs from the senses, the
brain may decide that some action is needed. In this case, command impulses are
sent from the brain through the motor nervous network to the appropriate
muscles, which are the output devices. This does not mean that ahuman being is
nothing more than a machine. It signifies that a human is capable of making
appropriate reactions to his or her environment.
The difference between an analog and a digital system is that in a digital
system, processing is done in digital form, using digital logic. Digital logic
characteristically recognizes only two values: aone or azero, sometimes called a
high or alow, or alternatively, true or false. From the viewpoint of digital logic, an
analog value that is increasing in asmooth curve is seen as rising in discrete steps.
This difference can be illustrated by comparing an analog and a digital
voltmeter. Irrespective of accuracy, the least significant digit displayed by a
digital meter must move up in unit steps. If 7is being displayed, the next higher
value must be 8. However, the pointer of an analog meter, which might also be
indicating 7, moves over an infinite number of values as it traverses to 8. These
might be so close together that they are not readable. But theoretically, the
transition is smooth and continuous, whereas in a digital instrument, the
movement is in steps.

169
 170 Chapter 9

Because digital values are limited to is and Os, the binary number system is
used to store and process numerical quantities. Just as the decimal system is based
on 10 (each more significant column being 10 times the value of the previous
column), so the binary system is based on 2. Each more significant binary column
is twice the value of the previous column. In fact, the binary column values can be
thought of as whole number powers of 2. Binary column values are shown below.

Exponential 25 24 23 22 21 2o • 2- ' 2-2 2 -3

Designation
Column Values 32 16 8 4 2 1 • 1/2 1
/4

The binary system lends itself ideally to digital logic, because it contains only
two values, either 0or 1. The value 2is written as a1in the next higher column.
For instance, binary 2is written 10—a Iin the 2s column and a0in the Is column.
Binary 5is 101—a 1in the 4s column, 0in the 2s column, and a1in the is column.
The subscript 2is added to abinary number to indicate that it is in base 2form.
Subscript 10 is added to adecimal number. Thus, the value of 10 2 is 2, while the
value of 10 10 is 10.
To convert from binary to decimal, simply add the values of all binary
columns in which a1is present, as shown in Problem 9-1.

Problem 9-1: Convert binary 1001 2 to decimal.

Answer: Add the column values:

(1 x8)+(0 x4)+(0 x2)+(1 x1) = 8+0 + 0+ 1 = 9 io

Because a digital audio system cannot represent a changing value as a

smooth curve, it must sample the analog signal at frequent intervals and
represent each sample in the form of a binary number. Consequently, an
approximation to the analog curve can be built up in the form of discrete steps,
each with its own binary value (see Figure 9-1).

Fig. 9-1 Analog

vs. digital
representation of
asignal

(A) Analog wave (

B) Digital approximation

In the language of digital logic, agroup of digits that are combined to form a
numerical value (or asingle piece of information) is called aword. Each piece of
data within that word is called abit. The expression "bit" is derived from the two
words binary digit. The bi in bit comes from the first two letters of binary. The
 Digital Audio 171

final tcomes from the final letter in digit. So a bit is one piece of digital data,
either ahigh or alow (a1or a0).
A waveform can be more accurately approximated by alarge number of small
steps, rather than by afew large ones. For instance, a2-bit word can produce only
four binary values: 00, 01, 10, 11. These correspond to 0, 1, 2, 3in decimal. If a2-bit
quantization system were used, the step size would have to be large and an analog
wave would have to be broken down into three increments. If a3-bit word could be
used, eight different values would be possible: namely, 000, 001, 010, 011, 100, 101, 110,
111, corresponding to decimal 0, 1, 2, 3, 4, 5, 6, 7. In this case, the wave could be
broken down into seven smaller increments. The greater the number of increments,
the more accurate the representation of the wave. This is illustrated in Figure 9-2.

Fig. 9-2
Representation of a
wave

(A)2-
bit word size (B) 3-bit word size

In fact, the number of possible values for agiven word can be calculated from
2, where n is the number of bits. Thus, the 3-bit word we have used gives 23
different values. This is 8, the values being 0 through 7. Although there is an
advantage in using large binary words to represent a smooth waveform
accurately, there are also disadvantages. The chief disadvantage is the amount of
hardware and the cost of large-word processing. A balance has to be reached, and
16 bits is now almost universally accepted as the standard word size for digital
audio processing. Many computers are built to handle 16-bit words, and
interaction between computers and digital audio is an already developing trend. It
is interesting to note the large number of different values that can be obtained
using a16-bit word. The number of values is given by 216 ,which is 65,536.
In general then, adigital audio system has to convert an analog waveform
into a continuous stream of 16-bit binary numbers. This is called quantization.
These numbers can then be digitally processed, equalized, and mixed, have
digitally delayed echo added, and be possibly stored in digital form. During
reproduction, the signal must again be returned to analog form. So, between the
digital processing stage and the power amplifier, conversion is again required—
this time from a sequence of binary numbers back to a corresponding analog
waveform that represents the air pressure fluctuations needed to generate the
final sound. A complete sequence of steps is illustrated in Figure 9-3.

Analog/Digital Interfacing
The analog signal is in the form of afluctuating voltage. In order to be suitable for
digital processing, this signal has to be converted to aseries of binary numbers,
172 Chapter 9

Fig. 9-3 Block

diagram of digital mic LOUDSPEAKER
audio chain ND PRE- POWER
AMP AMP

DIGITAL
A/D PROCESSING DIA
CONVERTER (including possible CONVERTER
(ADC) digital recording (DAC)
and reproduction)

whose values are proportional to the instantaneous voltages of the wave at each
sample moment. The analog-to-digital converter is dedicated to performing this
function. A symbolic representation of this conversion is shown in Figure 9-4. It
uses 4 binary digits, although in reality, 16 binary digits would be used to
quantify each sample.

Fig. 9-4 Analog

waveform
sampling

ANALOG WAVEFORM

6 8 7 3 -2 -5 - -2 DECIMAL SAMPLE VALUES

0000 0110 1000 0111 0011 - 0010 - 0101 - 0101 - 0010 4- BI - BINARY SAMPLE VALUES

Sampling Requirements
Certain constraints are needed in order to complete the sampling successfully.
Sample times must be precisely spaced, and the sampling frequency must be high
enough so that the highest audible frequency can be adequately represented. A
good quality audio system must reproduce frequencies up to 20 kHz. Because at
least two samples are needed to represent each wave (one in the positive half-
cycle and one in the negative half-cycle), the sampling frequency must be at least
twice the highest audio frequency. Therefore, sampling has to be at or above 40
kHz. In fact, the standard sampling frequency for professional digital magnetic
recording is 48 kHz; for compact disc recording it is 44.1 kHz.
Accurate timing of this sampling process is important. It is controlled by an
electronic clock, which produces asquare wave at accurate intervals. To achieve
sampling at 48 kHz, for instance, aclock consisting of asquare wave generator is
used to produce a48 kHz waveform. The sampling system can be triggered by the
leading edge of this square wave. Other parts of the digital system need to work
considerably faster, so that processing of each sample can take place in the
 Digital Audio 173

interval between samples. Usually the faster clocks oscillate at whole number
multiples of the sampling frequency. Frequencies up to ahundred times this clock
frequency may be used.
Not only must the voltage from the fluctuating audio signal be extracted
precisely at sampling time, but this must be held steady while its value is being
converted. The problem is that any electronic component takes a little time to
respond to an input signal. This is called the component's acquisition time. So the
sample value has to be held steady during this period. To achieve this, asample
and hold circuit is used. This uses a capacitor and two semiconductor switches
that charge the capacitor up to sample voltage at precisely the right moment,
where it is held during the required time.
The basic sample and hold circuit is illustrated in Figure 9-5. The two field
effect transistors, J1and J2, act as electronic switches. There is anegative voltage
at S1and S2 that normally keeps them switched off. The two op-amps B1and B2
are called buffers. They output the same voltage as they receive at their inputs, so
there is no voltage gain. They can, however, produce high current gain. This
prevents current from being drawn from the circuit connected to their inputs. In
effect, they buffer this circuit from current drain—hence their name, buffers.

Fig. 9-5 Sample

and hold circuit

TO A/D
CONVERTER

51

The action of the sample and hold circuit is as follows. Just before sample
time, a positive voltage pulse is applied by the clock system at SI.This
momentarily switches on J1 and discharges the capacitor to ground, so that the
previous value is cleared. Then, precisely at sample moment, apositive voltage
pulse is applied to S2, momentarily switching on J2. This allows asurge of current
from op-amp B1 to charge the top plate of the capacitor to the instantaneous
voltage of the analog waveform, VA .This value is held by the capacitor while the
conversion to the corresponding binary value takes place in the A/D converter.
Op-amp B2 allows the required current to be drawn by the A/D converter without
discharging the capacitor, so that its voltage remains constant. From this brief
description of the sample and hold circuit, it can be seen that conventional
electronic components are often used within the framework of adigital system.
174 Chapter 9

Secondary Sampling Requirements

Having examined the main activity of sampling, we have covered the
fundamental principles underlying digital audio. However, we must still deal with
all sorts of secondary problems that arise. A major problem resulting from
sampling is the generation of an audible alias frequency. The word "alias" means a
false name. Similarly, an alias frequency is a false frequency. It is not true and
should not be present. Unfortunately, an alias frequency can be created by the
process of sampling. The sampling frequency and the audio signal frequency
inevitably come into and out of phase alternately. This produces athird frequency
called abeat, whose frequency is numerically equal to the difference between the
two interacting frequencies.
An example of this occurs in the cinema, when you see the spoked wheel of a
car apparently rotating backward, standing still, or rotating slowly forward. This
is sometimes called a stroboscopic effect. It is produced by the interaction
between the frequency with which a wheel spoke passes a point, and the
sampling frequency of the camera. (A movie camera normally takes 24 frames/
second.) In this case, a given position of a spoke is in phase with the camera
picture at one moment. But, if the camera speed is slightly greater than the spoke
frequency (the frequency at which spokes pass a given point), the next picture
occurs when the following spoke has not quite reached this position. This
difference in phase continues giving the illusion that the wheel is rotating
backward. The number of times per second that the spokes come into and out of
phase with the camera is the difference between their frequencies. This is why a
beat frequency of this nature is also called adifference frequency.
Because, in an audio system, the audio frequency is always less than hall the
sampling frequency, the difference between them must be greater than the
highest audio frequency, so it will be ultrasonic and not heard. (For instance,
40K — 15K = 25K, which is ultrasonic.) This would be fine, but there's asnag. Most
music instruments produce ultrasonic harmonics. These, being much higher than
half the sampling frequency, produce audible alias tones. For instance, a high
harmonic of 30 kHz subtracted from asampling frequency of 40 kHz produces an
alias frequency of 10 kHz, which is within the audible range. Here lies the
problem; but it can be solved.
Only the high harmonics above 20 kHz (ultrasonic) can cause audible alias
frequencies. Therefore, alow-pass filter is installed just before the sampling stage.
This is designed to cut off sharply all high harmonics above 20 kHz, while leaving
the audible range of frequencies intact. By eliminating these high harmonics, the
resulting alias frequencies are also eliminated. This low-pass filter is called an
anti-alias filter.
One other worrisome little situation exists. Both during quiet passages and
during pauses, background noise in adigital audio system is different from white
noise. White noise is the natural noise in our surroundings. It consists of equal
levels of all frequencies combined in arandom manner. It is also the type of noise
produced by an analog system, so it can be accepted by the listener.
 Digital Audio 175

In digital audio, the stepped nature of low-level sound, such as background

noise, is not "white." It sounds like low-level clicking and is called granulation
noise. This is not acceptable; in fact, it is quite objectionable to the ear. Adding a
low-level audio signal made up of white noise just before the anti-alias filter adds a
random element that eliminates this problem. This added white noise is called
dither. Its amplitude needs to be only about one-third step size, so it barely
increases the noise level. A dither generator, therefore, eliminates this
granulation noise problem entirely.
Let us now summarize the process involved in digital audio systems, as it has
been described up to now. The fundamental requirement is the sampling of the
analog waveform at just over 40,000 times/second. At each sampling moment,
the instantaneous voltage of the wave is measured and converted to a binary
numerical value. This conversion process is performed by the analog-to-digital
converter (ADC), which acts as a digital voltmeter whose readout is binary. A
number of secondary processes are needed to ensure that the sampling process is
completed successfully. These are

1. Addition of low-level white noise called dither. This is needed to convert

background noise from unacceptable granulation to white noise.
2. Filtering of the analog signal by an anti-alias filter. This prevents generation
of audible alias frequencies.
3. Use of asample and hold circuit prior to A/D conversion. This ensures
accurate quantization at precise sampling moments.

D/A and A/D Conversion

After digital processing (and during reproduction of a digital recording), the
sampling process has to be reversed, to convert the sampled binary values back
to asequence of voltages. This automatically reconstructs the analog waveform,
which is then amplified by the power amplifier and fed to loudspeakers. A block
diagram of the overall sequence is illustrated earlier in Figure 9-3. Just as
secondary problems associated with sampling must be overcome during A/D
conversion, there are also two secondary problems during D/A conversion. The
first requires the use of an output sample and hold circuit. The second requires
another low-pass filter just before the power amplifier. This output LP filter also
has acutoff frequency of 20 kHz, but it is called an anti-imaging filter. The precise
need for and function of these two secondary components will be described
shortly. First it is necessary to say something about the two main components in
the digitization process: the A/D converter and the D/A converter. Their
positions in the audio chain are shown earlier in Figure 9-3.
Both A/D and D/A converters perform the fundamental requirements of the
digital processing and recording of an audio signal. The A/D converter converts
an instantaneous voltage to abinary value. The D/A converter converts abinary
176 Chapter 9

value to an instantaneous voltage. Although the A/D converter occurs first in the
audio chain, Iwill describe the D/A converter first, because aD/A converter is
included in the architecture of the A/D converter. Therefore, it is necessary to
have some idea of how it works before dealing with the more complex circuit.
The function of the D/A converter is to convert a binary value to a
corresponding voltage, which is proportional to the value. There are several ways
to do this. Perhaps the simplest method uses a summing op-amp (described in
some detail in Chapter 3).
Instead of summing all of the input channels equally, however, we must give
each input avoltage gain that corresponds to its column position in the binary
number system. The circuit of a summing amplifier is illustrated in Chapter 3
(Figure 3-7). The expression for its output voltage is

Rf Rf Rf
vo = Vi
, + V2
, ..... + V3
, — + . . .
Ri R2 a3

This formula can be thought of as summing each input voltage times its
channel gain. To convert abinary number to acorresponding voltage, all that is
needed is to amplify each more significant column with twice the gain of the
previous column. A simple 5-bit digital-to-analog converter can be made in this
way (see Figure 9-6).

Fig. 9-6 5-bit

digital-to-analog
converter

Rf = 16 kfl
1ka 2 kil 4 IQ 8 ka 16 kfl

(= Rd

If the inputs consist of a 5-bit binary number, which we will call X, then its
digits are X4, X3, X2, X1, X0. Rf is the multiplier. By varying Rf,the maximum
output voltage can be adjusted to any required value, while the proportionality of
the inputs remains unchanged. (In adigital system, 16 input bits are needed, but
conversion is achieved in the same way.)
Consider a binary input of 11001. For the sake of this example, we will
consider a logic high to be 1V. Of course, a logic low is 0 V. Then the output
voltage is given by
 Digital Audio 177

v. _ (1 x 1
16
11
C) + ( i
( x 1
26
11
{) + ( 0
( x 1
46
11
()
( .
4. (0 x 1
86
11
{) + ( 1
( 16(
x 161
1()

= 16 + 8 + 0 + 0 + 1

= 25 V

The decimal value of 11001 2 is 25. So it is clear that the output voltage of this
converter exactly corresponds to the value of the binary input. However, there is
a small problem associated with this conversion process. At the moment of
transition from one voltage to the next, there is ashort glitch. This is asudden,
false, fluctuating output voltage that occurs just before the new output voltage has
stabilized. These glitches have to be eliminated or the output waveform might
look something like that in Figure 9-7.

Fig. 9-7
Uncorrected
output glitches

r
To solve this problem, the output sample and hold circuit is installed just after
the DIA converter. This samples the true value of each step after the fluctuations
have ceased, and holds that value until the next sampling moment. That is after
the next output value has stabilized.
We are now left with a waveform that corresponds to the required analog
output except that it is made up of numerous, small voltage steps. By passing this
stepped waveform through a low-pass filter, called the anti-imaging filter, the
vertical transients (the leading and trailing edges of the square wave) can be
shaped sufficiently, so that aperfectly smooth analog waveform is obtained.
The cutoff point of this final output filter is 20 kHz, so it passes all audio
frequencies and still effectively smoothes out the steps into acontinuous curve. It
is interesting to note that the entire digital process starts and ends with low-pass
filters that have acutoff frequency of 20 kHz. These are the anti-alias filter and
the anti-imaging filter, respectively.
Now we can describe how the AID converter works. Its purpose is to convert
the instantaneous sample voltage to abinary value. The method most often used
in digital audio systems is called the successive approximation register ( SAR), a
block diagram of which is shown in Figure 9-8.
Each bit in the 16-bit register, from the most significant to the least, is rapidly
tested by the converter's control unit, first with a1, then with a0, until the binary
value is found that most closely corresponds to the analog voltage.
The control unit contains, in addition to other circuitry, a slow and a fast
clock. The slow clock speed is that of sampling frequency. At each sampling
moment, it clears the 16-bit register by resetting it to 0. Then the fast clock
178 Chapter 9

Fig. 9-8
Successive
approximation
register (A/D
converter)

BINARY
OUTPUT

sequentially tests each bit in the register, starting at the most significant bit (on
the left) with a 1; then, if the binary value is too high, changing it to a O.
Meanwhile, the binary value is fed to the D/A converter, where it is converted to
a corresponding analog voltage, VI,which is compared by the analog voltage
comparator with the input voltage, VA .If it is too high, the comparator outputs a
high, telling the control unit to change that bit to a O. If it is too low, the
comparator output goes low, telling the control unit to keep it as 1. In this way, the
A/D converter rapidly converts an instantaneous voltage to its binary value,
where it is held until the next sampling moment. The cycle is then repeated. A
block diagram of the stages before and after the central processing unit is shown
in Figure 9-9.

Digital Audio Applications

Digital audio data can be recorded onto any digital storage medium. Each
different medium requires its own processing and modulation method, both for
recording and reproduction. As aresult, considerably more than 16 bits of binary
data per sample have to be made available for successful storage (of course, stereo
recording doubles the audio data). Additional data containing processing
information is also needed. These additional data bits are called redundant bits.
Part of this redundant data consists of asynchronizing code that identifies the
start of each binary word. Another part contains an error detection and
correction code, which enables a comparison to be made of reproduced and
recorded data. This error detection system allows the reproduction processing
stage to determine if an error has occurred. In addition, it usually permits the
error to be corrected. If this is not possible, an automatic error concealment
system is activated, so that the error is not noticed.
Stereo audio samples require 32 bits of data (two 16-bit channels) to be
recorded at a rate of just over 40 kHz. This requires a recording frequency of
 Digital Audio 179

Fig. 9-9 Block

diagram of digital ANTI - INPUT
A/D
audio system VA too ALIAS SAMPLE CPU
CONVERTER
FILTER AND HOLD

DITHER
GENERATOR

(A) Input stages

OUTPUT ANTI -
SAMPLE IMAGING
CONVERTER
AND HOLD FILTER

-DI LOUDSPEAKER

(B) Output stages

about 1 4 million bits/second. The additional redundant data that also has to be
/
1

included brings the total digital data to be recorded up to about 41 4 million bits/
/
second. This high-frequency requirement presents limitations and challenges
peculiar to digital audio recording. Because of this, the average computer
memory is able only to hold comparatively short sequences of music. The most
useful memory storage medium for digital audio is magnetic tape or compact disc.
A compact disc can store an hour of digitally recorded music. Its reproduction
quality is superior to an LP disc in frequency range, reduced background noise
and consequent increased dynamic range, reduced harmonic distortion, longevity
(being sensed by a laser rather than a stylus), improved wow and flutter
characteristics, and in general, amore life-like quality of sound.
The recordings are produced by imprinting microscopic pits onto the flat
surface of a substrate. During reproduction, the sensing laser is processed to
interpret apit edge (an upward or downward transition at the start or end of apit)
as alogic high. A flat surface is interpreted as alogic low. Currently, most compact
discs are recorded by the manufacturer and can be used only for reproduction.
180 Chapter 9

However, alternative CD formats are being developed that can be both recorded
and reproduced by the user.
Magnetic tape is the other most useful medium for digital recording. On this
medium, data can be both recorded and reproduced; however, an analog audio
recorder cannot be used for digital recording. This is because the high recording
frequency needed to accept digital data at over 4million bits/second is far above
the top frequency threshold of an analog recorder. There are two types of
specialized recorders that can be used to magnetically record digital audio.
One is the digital audio stationary head ( DASH) recording machine. This
really consists of aprofessional multitrack recorder with the necessary processing
stage added to adapt the digital data to the recording format. This machine can
successfully record very high frequencies by using a high tape speed. Or it can
synchronously record on multiple heads to compensate for lower tape speeds.
The relationship between the number of heads per channel and the tape speed is
shown in Table 9-1.

Table 9-1
Relationship Tape Speed Heads Per Channel
between tape 7V2 ips 4
speed and heads
15 ips 2
per channel
30 ips 1

At 30 ips, the high frequencies can be achieved by asingle head. At 15 ips, the
digital data is demultiplexed (divided into two parallel streams) and recorded
simultaneously on two heads. Each head records half of the data. On playback,
these same two heads reproduce the two recorded tracks, which are then
multiplexed (combined) onto a single line, to achieve the original sequence of
digital data. Similarly, at half this speed, four heads are used synchronously, each
head recording aquarter of the data onto four separate tracks.
The other type of magnetic recorder that can be used comes under the
category of rotary head machines. In fact, it is avideo recorder with aspecialized
digital processing stage that adapts the digital data to a format suitable for
recording on a video recorder. Video machines are designed for very-high-
frequency recording, so there is no problem with frequency limitations. The
recording heads rotate at high speed inside ahelical tape guide drum, so that the
heads write thin, diagonal tracks onto the slowly moving tape as it passes.
Both professional and consumer recorders of this type are available. Most
CDs are mastered in this way from arotary head recording machine. From the
consumer's viewpoint, digital recording equipment—even the video recorder
combined with the necessary processing stages—is costly. However, a new
recording format has recently been introduced. This is known as R-DAT, which
stands for rotary digital audio tape, or simply DAT. Its development is already
bringing digital audio recording within the reach of audio enthusiasts as well as
serious musicians.
 Digital Audio 181

Advantages of Digital Audio

It has been mentioned that compact disc specifications are superior to LP discs in
many respects, most notably in lower background noise and increased dynamic
range. Background noise is quite different in digital systems than in analog
systems.
In the analog system, it is largely the current flow through transistors (or
tubes), amplified in stage after stage, that causes background noise. There is also
added tape hiss in tape recording. One result is that each time an analog
recording is re-recorded, the background noise and any equalization errors are
increased. This tends to cause degradation of quality after a number of re-
recordings.
In digital systems, noise is of adifferent nature. It is not noise, as such, but
quantization errors that cause the problem. Although an analog waveform is
broken down into an extremely large number of very small steps, there is still
some error probability in each step. This, of course, becomes more noticeable in
quiet passages, when the step size becomes large compared to the wave
amplitude. The maximum possible quantization error amounts to half the step
size (see Figure 9-10).

Fig. 9-10
Maximum
quantization error

TRUE ANALOG
VOLTAGE
MAXIMUM QUANTIZATION ERROR
OF + HALF-STEP SIZE

The ratio of signal to noise in an analog system is given by the maximum

signal voltage divided by the noise voltage. In adigital system, this ratio is strictly
speaking asignal-to-error ratio. This is avalue equal to the maximum number of
steps (corresponding to maximum signal) divided by the maximum error, which is
half astep. As noted earlier, to overcome the problem of granulation noise, white
noise in the form of dither at one-third step size is added to the signal. This
converts the signal-to-error ratio into what can justifiably now be called signal-to-
noise ratio, the granulation noise having been converted to white noise.
182 Chapter 9

In a16-bit system, the digital signal-to-noise level is about 98 dB, taking into
account quantization errors and added dither. In fact, most digital components
are specified as having a signal-to-noise level of "over 90 dB." This compares
favorably with analog tape recorders, in which the figure is about 65 dB.
Digital systems have another advantage over analog, in that when re-
recording, the noise level does not increase. Consequently, digital recordings can
be re-recorded many times. The reason for this is that the quantization error
occurs only during sampling. Provided the digital signal is reproduced and re-
recorded in digital form, sampling does not have to occur again; therefore, there
is no increase in noise. Tape hiss and transistor noise have no effect, because they
are always at such alow voltage that they are interpreted as alogic 0. From the
point of view of subsequent logical analysis, they don't exist.
Another important advantage of digital audio lies in the field of signal
processing. The processing that has been described so far has been anecessary
part of analog/digital interfacing in digital recording and reproduction. Once the
analog signal has been successfully converted to digital and fed to the central
processing unit, any amount of additional processing can be applied. All ot the
acceptable state-of-the-art processes that amodern analog console can achieve
can also be achieved digitally. For instance, a digital console can digitally mix
numerous input channels; it can apply digital delay and reverb; and it can expand
or compress, filter, and equalize any or all of the channels. Then, of course, it can
digitally record these onto amultitrack master. At alater time, these tracks can
be reproduced through the console and mixed down to astereo master, just as
with an analog console and multitrack recording system.
However, digital processing can also do more. Many digital consoles are now
being made with adigital memory that can remember all of the operations made
during the mixing process, and reproduce these operations without human
intervention. For instance, if the first half of a mixdown is perfect, but ashort
passage needs remixing, the first half can be automatically repeated and anew
mix opportunity can then be punched in or edited as needed.
Other, sometimes strange and often superbly imaginative, effects can be
created on digitally recorded sounds. This can be achieved by writing the
relevant processing in the form of a software program instruction set and
interfacing it through a computer. In this way, audio processing is becoming
available to an extent far beyond what was previously available to the analog
audio engineer. It can fairly be said that what can be done by interfacing
computers with digital audio extends future processing possibilities to the limits of
human imagination. This newly developing situation offers great problems, great
challenges, and vast possibilities for the future of one of the most important of
human activities—making music.
Power Supplies, 185

Power Supply Construction, 191

Signal Tracer Construction, 193

Mixer Construction, 198

Technical Considerations, 204

183
 10 Practical Audio Circuits

Power Supplies
All active electronic circuits need to be energized by a DC source of suitable
voltage. For most applications, it is convenient or absolutely necessary to use the
AC line voltage as an energy source. In these cases, the AC line leads directly to
the power supply inside the component cabinet. It is the function of this power
supply to convert the 115 V AC current to aDC source of suitable voltage to drive
the electronic circuits. Sometimes anumber of different voltages are needed. The
power supply can be adapted to supply these.
It is useful and necessary to know how a power supply works, because all
line-powered circuits contain one, and it is the first stage to be tested when
troubleshooting or repairing afaulty component.
There are three main parts to apower supply. They are

1. The transformer. This transforms the 115 V line voltage to just above the
final required voltage. For transistor circuits, this can be anything from 5V
to 35 V. For vacuum tube circuits, this can be in the region of 250 V or
more.
2. The rectifier. This converts the AC voltage from the transformer secondary
to aDC voltage. However, it is apulsating DC, not asmooth DC, and a
smooth DC is needed to energize the electronic circuits.
3. The smoothing and regulating stage. Here the pulsating DC is changed to a
smooth DC, largely by use of asmoothing capacitor. Due to internal
resistances of the secondary coil and rectifiers, the output voltage of this
smoothed DC is not entirely stable. When current is drawn, the voltage falls

185
186 Chapter 10

in proportion to the amount of current. So this stage usually includes a

regulating device that maintains virtually constant output voltage, however
much current is drawn (within the limits for which the power supply was
designed to operate).

Let us now examine each stage in detail. In the next section, we will see how
to construct a power supply that is useful for powering commonly used signal
processing components.

The Transformer
A transformer consists of aprimary and asecondary coil. The more powerful the
transformer, the larger and more costly it is, because it requires more iron in its
core and larger coils. Don't overload atransformer or it will overheat and it might
burn out.
The secondary consists of acoil with or without taps; it can also consist of
several coils, depending on how many different output voltages are needed, and
also on the type of rectifier that is used.

The Rectifier
Rectification is the process of converting an AC source to DC. There are two main
types of rectifier. One is the half-wave rectifier, which uses only one diode. The
other is the full-wave rectifier. This latter can be divided into two categories: the
full-wave rectifier that uses two diodes, and the bridge rectifier that uses four
diodes. Figure 10-1 illustrates the transformer with the secondary connected to
the different types of rectifier.
The diode is the active component in all of these rectifiers. Notice that the
terminals of the transformer secondary oscillate in reverse polarity to each other.
When the top terminal is positive, the bottom terminal is negative. In the next
half-cycle, both polarities reverse, so you can trace the current flow during each
half-cycle.
Look first at Figure 10-1A. In the first half-cycle, point X is positive and point
Y is negative, so electron flow is from Y, up through the load resistor, through the
diode (with no opposition, because diodes conduct electrons in the opposite
direction to the arrow), and back to X, the most positive point in the circuit. The
electromagnetic action then forces the current down through the secondary and
out again at point Y. This is the complete circuit during the first half-cycle. Note
that electron flow is upward through the load; therefore, the top terminal of the
load must be positive (electrons flow only to amore positive voltage).
In the second half-cycle, electrons try to flow out of point X, which is now
negative. They are attracted to point Y; however, they cannot pass through the
diode in the direction of its arrow, because the diode is now reverse biased.
Therefore, no current flows in the circuit in the second half-cycle. All of the
voltage develops across the diode, none across the load, and no current flows
Practical Audio Circuits 187

through the load. Because current flows only during one-half of the cycle, this is
called a half-wave rectifier, and the output consists of fluctuating positive half-
cycles only. This is illustrated in the waveform drawn in the top right corner of the
diagram in Figure 10-1A. Because there are half-wave gaps between each DC
pulse, alarge, comparatively costly smoothing capacitor is needed to smooth out
these pulsations. But in all other respects, it is asimple and effective circuit.
We can now examine the two-diode full-wave rectifier in Figure 10-1B. Notice
that this requires acenter tapped secondary. The center tap is held at ground
potential.
In the first half-cycle, the current flow is shown by the heavier dashed line. It
is true that current (electron flow) tends to originate from the most negative point,
namely Y, but it can't pass the diode in its reverse biased polarity, so no current
flows through the bottom half of the coil. However, current can flow from the
center tap (which is at ground potential), up through the load, and back through
the top diode to point X, the positive terminal. Therefore, only the top diode and
top half of the secondary conducts during the first half-cycle.
In the second half-cycle, all polarities are reversed, and the current flow is
shown by the lighter dashed line. Again, current can leave the secondary only by
the center tap. It flows again upward through the load and back through the
bottom diode (which is now forward biased) to point Y. Thus, only the bottom half
of the coil conducts during the second half-cycle. This type of rectifier has the
disadvantage of requiring a double length of secondary coil with a center tap.
This adds to the cost of the transformer. You will notice from the waveform shown
that a DC voltage occurs during both half-cycles; hence its name, full-wave
rectifier. The fact that there are no half-wave intervals between pulses means that
it is easier to smooth this waveform, so a smaller and less costly smoothing
capacitor is adequate. However, this benefit is countered by the need for amore
costly center tapped transformer. There is not much choice between the one-
diode and two-diode circuits, in cost or effectiveness.
Figure 10-1C illustrates abridge rectifier. The advantage of this is that the whole
transformer secondary is used in both half-cycles, and there is no center tap. This
keeps down the transformer cost. Also, the full wave is rectified, which means that
the same comparatively small smoothing capacitor can be used as was suitable for
the two-diode circuit. For the added cost of two extra diodes (which is negligible),
this circuit incorporates the advantages of both previous circuits with none of the
disadvantages. The first half-cycle current flow is shown by the lighter dashed line,
the second half-cycle by the heavier dashed line. Notice that two diodes conduct in
the first half-cycle. The other two conduct during the second half-cycle. The
discriminating factor is the direction of bias that is applied to the diodes.
In the first half-cycle, electron flow is from point Y. At the bottom of the
bridge, only DI can conduct. (Electrons cannot flow in the same direction as the
arrow.) From the DC negative point, current cannot pass D2,so it goes through
the ground bus, up through the load to the DC positive, and then through D3.
Therefore, only D1 and D3 conduct in the first half-cycle, and D2 and D4 conduct
in the second half-cycle.
188 Chapter 10

Fig. 10-1
Transformer and
rectifier stages
_n___n_n___
x

Y
(A) Transformer feeding ahalf-wave rectifier

.-
(8) Transformer feeding atwo-diode full-wave rectifier

(C) Transformer feeding abridge rectifier

The Smoothing and Regulating Stage

In Figure 10-1, the rectifier is shown connected directly to the load. This is done so
that the current paths through the rectifier can be clearly seen. But in fact, there
is always asmoothing and regulating stage between the rectifier and the load.
 Practical Audio Circuits 189

A capacitor is adevice for storing electrical energy in the form of acharge. It

is this ability to store electrical energy that enables it to smooth out most of the
pulsations and fluctuations that emerge from the rectifier stage (see Figure 10-2B).
In fact, it is usually possible to locate the power supply in any electronic circuit
simply by looking for the largest electrolytic capacitor.

Fig. 10-2
Smoothing
capacitor action
in apower supply

(A) Smoothing capacitor following half-wave rectifier

i
0V I__ L _ I__ _

(B) Pulsating DC from rectifier (C) Effect of smoothing capacitor

AC RIPPLE

DC COMPONENT 7
0V

(D) AC ripple riding on aDC component

The capacitor acts as an electronic shock absorber. It absorbs energy in the

form of a charge when the rectified voltage is at its peaks, and it gives back
energy when the voltage falls to its troughs. This is the basic smoothing action.
Let us look at the simple half-wave rectified power supply circuit in Figure 10-2A,
to obtain aclearer picture of the smoothing capacitor's action. Assume that the
bottom terminal of the secondary is grounded and that all AC voltage fluctuates
at the top secondary terminal. We need only consider the first half-cycle, because
no current can flow in the second half-cycle, due to the action of the diode.
When the power supply is first switched on, the rectified voltage rises to its
first peak and charges the top plate of the capacitor to peak value. From the
schematic in Figure 10-2A, you can see that current (electron flow) passes not
only through the load to point X (the peak positive polarity), but also from the top
plate of the capacitor to point X. It is this flow of electrons from the capacitor's top
plate that causes it to become charged to the positive peak. (Absence of
negatively charged electrons is what causes the plate to become positively
charged.)
190 Chapter 10

Shortly, the rectified voltage falls to zero (see Figure 10-2B). This is where the
capacitor's value comes in. The top plate of the capacitor still remains at positive
peak voltage. The result is that current can continue to flow upward through the
load. When this current reaches point Y, it turns downward onto the top plate of
the capacitor, which is now the most positive point in the circuit. In this way,
current can continue to flow through the load, although the rectifier output has
fallen to zero.
As current (electrons) accumulates on the capacitor's top plate, it gradually
becomes more and more negative; in other words, it loses some of its positive
charge and so its voltage drops. But it is such a large capacitor that its voltage
hasn't time to fall very far before the next positive peak from the rectifier occurs
and charges the capacitor up to its peak voltage again. The resulting output from
this rectifying and smoothing stage is shown in Figure 10-2C. After the initial
surge, caused when the power is first switched on, the voltage settles down to a
rippling effect averaging just below peak value. This waveform can be described
as an AC ripple riding on the shoulders of aDC component, as shown in Figure
10-2D. (AC ripple results in hum.) For good quality power supply action, the AC
ripple should be very small compared with the DC component. In fact, the ripple
from the smoothing capacitor can be almost completely eliminated by the
regulator that follows.
Without aregulator, the output voltage of the power supply would be affected
by the current flow. A small current flow would not discharge the smoothing
capacitor much, nor would it produce much drop in the resistances of the diodes
and secondary. But at maximum current flow, the resistances and capacitor
discharge level would cause aloss of about 20 percent of the output voltage. Such
voltage fluctuations would cause serious problems in electronic circuits. What the
regulator does is to reduce the voltage to about 70 percent of its maximum value
at all times. This is just below the level to which it would fall under maxinwm
current conditions. In this way, the output voltage remains substantially constant,
whatever current might be drawn.
The voltage regulator consists of aconstant voltage circuit made up of one or
more transistors; but these days, nobody bothers to construct his own voltage
regulator. These are mass produced for the electronics industry in the form of a
chip, or inexpensive circuit. A person simply purchases either a positive or
negative regulator designed to output whatever voltage is required. Usually, for
transistorized circuits, a 12 V or 15 V regulator is used. Thus, the final power
supply circuit might be like that shown in Figure 10-3A. This is abridge rectified
positive output power supply.
An advantage of the regulator is that it serves two important purposes. By
using asuitable value of about 70 percent of the peak secondary voltage, it not
only regulates the output voltage, keeping it constant at all normal loads, it also
cuts out the AC ripple that results from the effect of the smoothing capacitor. The
smoothing capacitor is chosen to be of sufficient capacitance so that the voltage
never falls below 70 percent of peak value, even when maximum current is
drawn. Because the RMS (Root Mean Square) value of an AC voltage is just over 70
 Practical Audio Circuits 191

Fig. 10-3 Bridge

o
rectified positive o
10, VOLTAGE
output power o DC —
IC>
REGULATOR
supply ler•

(
A)Schematic diagram

VP

OV

(B) Smoothed but unregulated output (C) Smoothed and regulated to 70 percent of peak voltage

percent of its peak value, it works out that the rated secondary voltage of the
transformer should be just over the required output value of the voltage
regulator—about 15 percent over to allow for losses due to resistance in the
rectifier stage. For aregulated output of 12 V, atransformer secondary voltage of
about 12.6 V to 15 V is suitable. For an output of 15 V, asecondary of about 16 V to
18 V is suitable.

Power Supply Construction

The power supply circuit described here corresponds to that illustrated in Figure
10-3A. The theory of its action has already been covered in detail. The first step in
construction consists of deciding on a suitable layout for the components. A
suggested layout is shown in Figure 10-4.

FILTER 10 KS2
TRANSFORMER RECTIFIER CAPACITOR REGULATOR RESISTOR

— DC +
DC +

115 V
LINE DC
INPUT OUTPUT

DC —
—
—N \DC
FUSE TERMINAL STRIPS

Fig. 10 -4Suggested layout for power supply

192 Chapter 10

List of Components
1 Cabinet
1 Transformer (Primary 115 V, Secondary 12.6 V to 16 V, 250 mA
approximately)
4 Silicon diode rectifiers ( 1A)
1 LF, 30 V, axial)
Electrolytic capacitor (500 i
1 Voltage regulator (7812)
1 Heat sink to fit voltage regulator
1 Tube, heat sink compound
3 Terminal strips (4or 5tag)
3 Assorted colors of hook-up wire (20 gauge, solid)
2 Terminal posts (one red and one black)
1 Line cable with plug
2 Fuse holders
1 Slow-blow fuse (V2 A)
1 Quick-blow fuse ( 1A)
1 Resistor ( 1/
2 W, 10 kg approximately)

Construction Procedure
Notice in Figure 10-4 that the stages have been laid out in sequence from the line
input on the left to the DC output on the right. Drill a few equally spaced
ventilation holes at the bottom of one end and at the top of the other end. This
will allow convection air flow to cool the components. Pass the line cable through
its access hole on the left. With plenty of slack inside, tie aknot inside the access
hole to anchor the cable. Drill fixing holes for all of the components.
Keep the wiring tidy. Use exact lengths of wire, and make all bends right
angles by bending with long nose pliers. The three terminal strips consist of four
or five solder tags fixed to an insulated strip. The tag connected to the fixing
bracket should be used for aground connection only. No voltage must be affixed
to it, because the fixing bolt extends through the cabinet. For professional results,
color code all hook-up wiring: AC should be green or blue, DC + should be red,
and DC— and ground should be black.
The cathode end of a diode is the one indicated by the silver band. When
more than one wire has to be connected to any one soldering tag, anchor each
wire separately, but do not solder until each has been put into place.
Spread alittle heat sink compound on the back plate of the voltage regulator
before bolting it to the heat sink, and fix the regulator to the heat sink before
soldering into place. This helps to keep the regulator cool during soldering.
Test each stage as it is completed. When the transformer has been installed,
insert the line voltage fuse ( 1
2 A slow-blow) and line plug, and test to obtain 115
/
 Practical Audio Circuits 193

AC volts at the primary and approximately 16 AC volts at the secondary. Don't

forget to remove the line plug before continuing with the construction.
The rectifier output should be approximately 20 DC volts when the smoothing
capacitor has been installed. When installing this capacitor, make sure it is
polarized correctly. If incorrectly polarized, the capacitor will work for a few
seconds, then overheat and blow up.
The unregulated voltage at the regulator input should be the same as across
the capacitor—about 20 V. The regulated output voltage should be 12 DC volts.
The 10 kg resistor across the output terminals is merely to ensure slight current
flow to maintain correct regulator voltages.
For agiven power, far less current flows at the high-voltage input than at the
low-voltage output. For this reason, the 1
2 A fuse is used for the line voltage and
/
the 1A fuse for the DC output. When the power is first switched on, asudden
surge of current flows through the rectifier to charge the capacitor; therefore, a
slow-blow fuse is assigned to the input. It permits this momentary surge to exceed
its rated value for ashort period, while still ensuring that not more than 1/2 A will
flow long enough to cause damage. The 1A output fuse protects the regulator
from burnout, should the output be short circuited.

Signal Tracer Construction

A signal tracer consists of a low-power audio amplifier connected to a small
loudspeaker. It is useful during troubleshooting because it can audibly detect an
audio signal between stages of an amplification chain. A signal tracer schematic
is shown in Figure 10-5A. It is similar to the inverting op-amp discussed in Chapter
3, and the amplification theory is the same.
The integrated circuit pin connections are identified in Figure 10-5B. Note that
pins 2and 3are the inverting and non-inverting signal inputs. They have nothing
to do with battery + and — . The battery is connected between pin 6and ground.
As was mentioned in Chapter 3, the gain of an inverting op-amp is given by
Rf/R i,which in this schematic is R2/R 1 or 680 k0/100 kit, which is 6.8. This
should be adequate for our purposes.
It is not possible to use aregular op-amp, because it has insufficient power to
drive aloudspeaker. The op-amp used in this circuit is the 386. It includes apower
amplifier output stage that can produce up to 0.4 W of audio power. This is quite
sufficient to drive the small loudspeaker in our signal tracer, with more than
enough volume. A power supply consisting of a 9 V battery makes the unit
completely portable.
Volume control is provided by the 10 1(1/ potentiometer, R4 in the schematic.
This potentiometer configuration is astandard circuit for gain or volume control
in almost any audio circuit.
A potentiometer consists of afixed resistor between the outer terminals (pins
3 and 1), and a moving wiper, which is the center terminal (pin 2) in the
194 Chapter 10

Fig. 10-5 Signal

tracer PHONE R2
JACK C1 •vvy
680 ki)
—0--• 1p.F
3
R4
10kÇ 2 100 kt≥
IC

4 1000 ¡ IF

680 k12 02
JAF

9V

(A) Schematic

1 •

— 2 --- — 7

Vin + 3 — — 6 DC+

GND 4 —
-- 5 Vow

(B) 386 pin-out

schematic. The wiper can move up and down the fixed resistor, as indicated by
the arrow in the schematic symbol. The fixed resistor acts as the load to the
incoming signal. The wiper taps off progressively lower signal voltages as it
moves down from pin 3 (maximum signal) to pin 1, which is connected to ground
where the signal is zero.
In some situations, such as the shelving equalizer described in Chapter 5, a
linear potentiometer is used. Linear taper means that the resistive element is of
constant resistivity throughout its length, so the resistance between points 1and 2
is linearly proportional to the distance moved by the wiper (slider in a slide
potentiometer, or knob rotation in arotary potentiometer). However, this type is
not suitable as avolume control.
Because the subjective experience of an increase in loudness is logarithmic, a
large signal increase at high listening levels gives the same apparent loudness
increase as a small signal increase at low listening levels. For this reason, the
resistive element in avolume control is tapered in such away that the resistivity
rises at an ever increasing rate as the wiper is moved up. This type of
Practical Audio Circuits 195

potentiometer is said to have alogarithmic or audio taper. The loudness appears

to increase linearly in proportion to knob rotation.
If alinear potentiometer were used, all of the loudness increase would appear
to occur at the start of the knob rotation, and the remaining rotation would
appear to do practically nothing. It is important, therefore, not to connect the
outer terminals of the potentiometer (pins 3and 1) in reverse. This would make
the apparent lack of linearity worse instead of better.
To ensure that the response of the potentiometer remains logarithmic, it is
necessary that the input resistance of the following circuit (R1in Figure 10-5A) be
high compared to the resistance of the potentiometer. If it were equal to or lower
than that of the potentiometer, it would tend to short the lower section of the
potentiometer to ground, thus distorting its logarithmic taper. A suitable value for
11 1,therefore, is ten times higher than the resistance of the potentiometer; its
effect is then negligible. That's why it is specified as 100 ka. In consequence, R 2 is
necessarily set at 680 kft to obtain the required 6.8 gain.
Because aunipolar positive power supply is used (the 9V battery), the output
from pin 5consists of an AC signal riding on the shoulders of aDC component. A
coupling capacitor, C2, is therefore necessary to block the DC while coupling the
AC signal to the loudspeaker. Note that the positive terminal of the capacitor must
go to pin 5. C2 in series with the effective resistance of the loudspeaker forms a
high-pass filter. It is necessary to ensure that its cutoff frequency is low enough to
pass all audio frequencies without significant loss. Assuming a40 speaker, the
cutoff frequency occurs when the reactance of the capacitor rises to 4 O.
However, this small speaker cannot, and need not, output very low frequencies,
so we can permit cutoff at 40 Hz. The calculations for finding a suitable
capacitance for C2 are, therefore,

1
XC2 = 271C2

So,

C2 2.2-fX c2

Cutoff occurs when Xc2 = R; therefore, substituting R for Xc2 ,we get

1
C2 =
27rfR

With f = 40 Hz and R = 49,

C2 2rx40x4
196 Chapter 10

So,

C2 = 995 ¡LP 1000 IX

Thus, we can use a 1,000 g capacitor. It would be a good idea to have the
maximum working voltage of this capacitor high enough to exceed the 9 V
battery potential. Ihave, therefore, specified 12 V as its voltage.
It is also advisable to isolate any DC component that may accompany the signal
input, so acoupling capacitor, C1,is required. This capacitor, in series with the 10 ka
potentiometer, forms ahigh-pass filter; therefore, its reactance at the lowest audible
frequency should not be more than 10 ka. Because of this high input resistance, a
much smaller capacitor can be used than was needed to couple the 4a loudspeaker.
This input capacitor need be only about 2g, with aworking voltage of 100 V
The 0.2 p.F capacitor, C3, is described as adecoupling capacitor. It decouples
the signal from the 9V battery feeding the amplifier between pins 6and 4. To
understand the need for this, remember what was said in Chapter 1about the
generator part of acircuit.
Every voltage source must be assumed to be in series with an output
resistance. In this case, it is the internal resistance of the 9V battery with which
we are concerned. As the amplifier's current drain from the battery fluctuates,
increasing at signal peaks, so the voltage available at pin 6 drops, due to the
internal resistance of the battery. This causes a signal to become, in effect,
superimposed on the 9V battery supply. The decoupling capacitor, C3, acts as an
electronic shock absorber. Due to its energy storage capacity, it smooths out these
voltage fluctuations, thus decoupling the signal from the power supply voltage. If
this were not done, the battery source would feed signal back into the amplifier.
This would cause all sorts of internal resonance, which would cause distortion and
increase background noise, and which could even make the amplifier unstable.
Because C3 is not an electrolytic capacitor, it is not polarized. Connections
can, therefore, be made in any polarity.

List of Components
1 Cabinet
1 8-pin dual-in-line package (DIP) IC socket
1 386 audio power amplifier IC
1 10 ka potentiometer (log or audio taper)
2 680 Ica resistors ( 1
2 W)
/
1 100 1(0 resistor ( 1/
2 W)

1 Nonpolarized capacitor ( 1AF, 100 V)

1 Electrolytic capacitor ( 1,000 µF, 12 V)
1 Nonpolarized capacitor (0.2 p.F approximately, 16 V)
1 Low-voltage mini switch (SPST or SPDT)
 Practical Audio Circuits 197

1 General purpose printed circuit board (PCB)

1 Small loudspeaker
1 Phone jack ('/4-inch, mono)
1 Roll hook-up wire (20 gauge, solid)
1 9V battery (MN 1604 or equivalent)
1 Battery holder
1 Battery connector

Construction Procedure
The first step is to lay out the position of the loudspeaker, potentiometer, and
switch on top of the cabinet, and mark the fixing holes. Drill sound vent holes in
the area that will be covering the speaker cone.
Lay out and mark fixing holes for the printed circuit board (PCB), battery
holder, and input jack in the cabinet. The construction of the PCB can then begin.
There is always at least one copper bus running the entire length of the
board. Assign one of these as the ground and connect all grounds to this bus.
The heart of the circuit is the amplifier IC. All PCBs are designed to accept
the pin spacing of this chip. The IC holder must be soldered in first, and the chip
plugged in after the pins have cooled. Start with the IC holder, because all other
components are related to its pin layout.
There is achannel in the center of the board, with conductors on each side
designed to accept this IC holder. Notice that there is asemicircular notch and/or
aspot at one end of the chip and holder. This identifies pin 1so that the pin-out
can be correctly identified and the chip is not inserted back to front.
All components and wiring must be situated on top of the PCB (that is, on the
plastic surface). Wire connections should extend through the holes, and are then
soldered onto the copper conductors underneath. Excess wire is cut off.
The holes along each edge of the PCB are designed as access points. All
input/output flexible wires should terminate at these edges, never in the center of
the board. Solid hook-up wire can then connect to the appropriate component, if
no copper conductor is available.
A spacer should be threaded on the fixing screws under each corner of the
PCB. This avoids bending and breaking the corners of the board while the fixing
screws are being tightened.
Once the wiring is completed, it is a good idea to test for correct voltage
polarity at pins 6and 4of the IC holder. Remove the chip first, because wrong
polarity here will destroy the chip.

Mixer Construction
It is not difficult to construct a4-channel mixer based on an op-amp used as a
198 Chapter 10

summing amplifier. The theory behind a summing op-amp is described in

Chapter 3. The circuit to be employed is shown in Figure 10-6.
It is suggested that a741 op-amp be used. This is one of the most common and
best op-amps. It is designed for line-level signal voltages of about 0dBv, which is
suitable for signal processing stages. It is also designed to be energized by a
bipolar power supply, which is one that supplies two oppositely polarized
voltages, equally balanced on each side of ground. The three power supply
terminals are then at voltages such as + 12 V, ground, and — 12 V. This simplifies
the amplifier circuit, because the signal can then oscillate in both directions about
this zero potential, without having to pass internal capacitors. However, this op-
amp can operate equally well at somewhat lower or higher voltages, provided
they are bipolar.
With alittle ingenuity, we can use the 12 V unipolar power supply illustrated
in Figure 10-3 to power this circuit. Our unipolar power supply can easily be
adapted to supply + 6V and — 6V balanced on either side of ground, and it will
then power the mixer satisfactorily.
All we have to do is change the zero ground reference point from the minus
terminal of the power supply to the new ground potential half way between the two
output terminals. This is done by means of avoltage divider formed by R1and R2 in
the schematic of Figure 10-6A. Note that the newly formed ground between the two
resistors must be connected to all other grounds in the schematic by means of a
ground bus. It must not be connected to the black terminal of the power supply,
which is now the — 6V source. The two decoupling capacitors, Cs and C6, also help to
maintain the stability of this new ground. They act as energy reservoirs, countering
any tendency of the newly formed ground to shift polarity in either direction.
The potentiometers in the mixer act as input faders or gain controls. As
described in the Signal Tracer Construction section, all audio volume controls must
be in the form of logarithmic potentiometers. Therefore, connect the outer
terminals correctly, according to the terminal numbers: terminal 1to ground,
terminal 2 to the wiper, and terminal 3 to the signal input from the input
capacitors. The use of coupling capacitors C1 through C4 and C7 is advisable to
ensure that aDC component from any source doesn't interfere with the operation
of the mixer.
The maximum gain of each channel is determined by the ratio of the
feedback resistor, R7, to the input resistor, R3 through R6. Signals entering and
leaving should both be at the same level, so the overall gain should be unity.
However, for best results, the faders should operate normally at about two-thirds
of their maximum height. Then there is headroom to allow increase of gain by a
few dB, as well as to permit fading to any required extent. It is recommended that
an extra 6dB of boost be available to any channel, so the maximum gain has to
be 2. (A gain of 2corresponds to alevel gain of + 6dB.) This determines the value
of the feedback resistor at 200 kft then the gain of any channel can rise to a
maximum value given by 200,000/100,000 = 2.
As was the case with the signal tracer, the coupling capacitors form an HP
filter, because they are in series with the input resistor of the following stage. This
Practical Audio Circuits 199

+6 V

(A) Schematic

• 8

2— 7 DC+

V,„+ 3 —• — 6 Vout

DC— 4— ,-- 5

(B) IC 741 pin-out

Fig. 10-6 4-channel mixer

200 Chapter 10

mixer is designed to feed an amplifier or other signal processing stage whose

input resistance must be at least 600 9, but is probably far more. Taking the worst
possible case of 600 at the lowest audible frequency of 15 Hz, the output
capacitor, C 7 ,can be evaluated by

1
Xc =
2rfC

At break frequency, below which all signals are cut off,

Xc = R

SO,

1
R=
2rfC

and

1
C=
27rfR

Substituting 15 Hz for fand 600 1

2for R,
-

1
c =
27r x15 x600

= 18 x10 -6 farads 20 id

This gives aloss of 3dB. Because we wish to lose only about 1dB, we will use a
100 ,
uF capacitor for C 7 .
Notice that the input capacitors feed high 10 Id/ loads, so their reactance can
be much higher. This permits 1id for the input capacitors.
R 8 acts as aground resistor. It should equal Rf to balance any DC component
that might develop, but its value is not critical.

List of Components
1 Cabinet
1 8-pin DIP socket
1 741 op-amp IC
4 10 kl potentiometers (log or audio taper)
2 51(9 resistors ('/2W)
4 100 1W resistors eh W)
 Practical Audio Circuits 201

2 200 kû resistors ( 1/
2 W)

1 22 resistor ( 1
2 W)
/
4 Nonpolarized capacitors ( 1g, 12 V)
2 Nonpolarized capacitors (0.2 id approximately, 16 V)
1 Electrolytic capacitor ( 100 id, 16 V)
1 General purpose printed circuit board
5 Phone jacks ( 1/
4-inch, mono)

Construction Procedure
After passing the power supply cable through the access hole, the cable should be
anchored inside the cabinet, so that an external pull does not apply force to its
junction with the PCB.
The next step is the PCB construction. The rules for PCB use that were
applicable during the signal tracer construction should always be followed.
Position the IC holder across the center channel of the PCB, so that each of its pins
contacts aseparate conductor. The PCB is designed for compatibility with an IC
spanning the center channel.
First, consider the connections to pins 2and 6. There are five connections to
pin 2and three to pin 6. Because ageneral purpose PCB normally accommodates
only two connection holes for each pin, an alternate connection procedure is
needed. It is suggested that, of the two full-length bus conductors on the PCB, one
should be assigned as the ground bus. The other should be cut in half, using a
razor or sharp knife. It is agood idea to make two cuts close together and remove
the piece of copper between them. Use the half of this bus on the pin 1side as the
input bus. Use the other half as connection to pin 6. Connect the input bus to pin
2, using ashort length of hook-up wire. Use some of its remaining holes to accept
the five resistors assigned to pin 2—namely, R3,R4,Rs,R6,and R7.
Imentioned that ashort wire is needed from the input bus to pin 2. This is
because pin 2 is sensitive to interference, so only short wire lengths should be
assigned to it.
Once these two priority connections have been made, other connections can
be assigned to any remaining conductors as seems appropriate. As was suggested
during construction of the signal tracer, test for correct power supply voltage and
polarity at pins 4and 7before plugging in the IC chip. Incorrectly applied polarity
here will destroy the chip.
In rotary potentiometers, there is ametal lug on the body casing that extends
forward, parallel to the shaft. Do not leave this lug in position without special
arrangements. It is intended that avery small hole be drilled to accommodate this
lug when the potentiometer is being installed. The purpose is to prevent rotation
of the body of the potentiometer by excessive torque applied to the knob. If you
do not want to drill this additional hole, do not leave the lug in position, because it
will tilt the potentiometer out of true when its fixing nut is tightened. Bend the lug
202 Chapter 10

back or break it off with pliers; then the potentiometer will sit flush with the side
of the cabinet. In this case, however, it is agood idea to insert atoothed washer
between the body of the potentiometer and the cabinet. The teeth will prevent
rotation of the potentiometer during use and will serve the same purpose as the
lug was intended to serve.

Additional Pre-amplifier Stages

This mixer is designed to accept line-level signals of about 0dBv, such as would
be available from the line output of atape machine or equalizer. If you want to
feed microphone signals to the mixer, additional pre-amplifier stages are needed.
It is comparatively easy to construct pre-amplifier stages for this purpose, but a
larger cabinet is needed to accommodate the extra printed circuit board. It is a
good idea to use dual 741 op-amps. These consist of two op-amps on asingle chip.
One set of power supply pins is used to power both amplifiers. Two dual 741 op.
amps can supply microphone input facilities to all four channels.
By use of anon-inverting amplifier circuit (described in Chapter 3) and alow-
voltage slider switch, each pre-amp can be controlled to give either again of 50
or again of 1 (suitable for line-level inputs). This enables each channel to accept
either microphone or line inputs at the flick of aswitch. The circuit schematic for
amic pre-amplifier stage is shown in Figure 10-7.
Additional components needed for two pre-amp stages are listed below. Note
that the two op-amps on the chip are labeled A and B. Ihave indicated the pin
numbers that should be used for the B amplifier in parentheses on the single
schematic. Pins 8and 4are common to both op-amps on one chip.

Additional Components Needed for Two Pre-amps

1 8-pin DIP socket
1 Dual 741 op-amp IC
1 General purpose printed circuit board
2 10 kfl resistors eh W)
4 500 ka resistors ( 1
2 W)
/
2 x, 16 V)
Nonpolarized capacitors ( 1t
4 Nonpolarized capacitors (0.2 pf, 16 V)
2 Low-voltage switches (SPDT or SPST)

If you prefer to use balanced line microphone cables, use the circuit in Figure
10-8 instead of that in Figure 10-7. In this case, a1
4 -
/ inch stereo input jack must be
used. In addition, two 10 kg resistors, a500 kit resistor, and a1i tF.capacitor are
needed for each unit. Also, aDPDT (double pole, double throw) switch is required
instead of the SPDT switch specified for the circuit in Figure 10-7.
This balanced input circuit has several advantages over that in Figure 10-7.
Both the Mic input and Line input can accommodate balanced lines. This notice-
 Practical Audio Circuits 203

Fig. 10-7 LINE MIC

Microphone
pre-amplifier with
Mic/Line switch 500 k.S-
2

+6 V DC

10 kS1

(6) TO
(
7) C1 C2
1/2 Dual 741 O
C3 C4

(
5)

1 0.2 pF
00 kO,

— 6 V DC
Y5

(A) Schematic

Vnnt A 1 — • — 8 DC+

V— A 2 — — 7 Vout 6

\f in + A 3 — — 6 V1n —B

DC— 4 — — 5 V, n + B

(B) Dual 741 op-amp pin-out

ably reduces noise interference. If you want to use an unbalanced line, it is

completely compatible. By plugging in a mono phone plug instead of a stereo
plug, the mono sleeve automatically shorts the non-inverting input to ground. The
amplifier then functions as an inverting amplifier with unbalanced input.

Technical Considerations
204 Chapter 10

Fig. 10-8 •VvN.,

Balanced input 10 k≤≥
mic pre-amplifier
with Mic/Line 500 kl2
OPTIONAL
switch
3- PIN XLR +6 V DC
SOCKET

j 1 µF
STEREO
JACK 1 0-o-re 0
TO
10 kfl 2 Mic Line
\•• (
7) Cl C2

C3 C4
_ Mic Line
0-&-•
1 ¡.IF 10 kJ-2

_
r 0.2 i.tF

DC 500 kJ)

equipment. The best way to acquire these practical skills is to build and test simple
audio components. Istrongly recommend that you build one or more of the
circuits described in this chapter. Not only will the component be useful, but more
importantly, the hands-on practice will develop skills and understanding that
cannot be achieved in any other way.
One of the most necessary skills is good soldering technique. Bad soldering
causes more problems than anything else when constructing or repairing
electronic circuits. The following points should be noted.
When soldering onto aprinted circuit board, asmall, precision 12 or 14 W
soldering iron should be used, never a 25 W iron. Use light-duty rosin-cored
solder, never normal-duty solder. Keep the tip of the soldering iron sharply
pointed and completely coated with solder at all times. When the iron is used for
the first time, coat the top with solder as soon as it is hot enough. This prevents
corrosion. If any part of the tip becomes brown due to corrosion, unplug the iron
and use asmall, fine-toothed file to clean the surface. Reheat the iron and coat
the tip with alayer of solder before it can become corroded again.
Have aroll of desoldering braid handy. If aconductor becomes shorted by an
oversize lump of solder, use desoldering braid to suck up the liquid solder and
remake the joint. The technique for using desoldering braid is to lay the end of the
braid over the solder to be removed. Press the soldering iron tip down onto the
braid, so that the heat goes through the braid and melts the solder underneath.
Practical Audio Circuits 205

The braid will then soak up the solder like asponge. Remove the braid and cut off
the solder-covered end. Repeat the process until all excess solder is removed.
Beware of making dry joints. This is the most important point. The newcomer
to soldering naturally imagines that the soldering iron and solder should be
applied to the wire that is to be soldered, and that the solder can then be spread
to the soldering tag on each side of the wire. This is wrong and an almost sure way
to make adry joint.
A dry joint looks like agood joint, but the connection is intermittent. This is
because, when the solder was liquid (melted), it didn't wet one of the two pieces of
metal that had to be joined. The solder probably appears to be covering the
metal, but it is not fused onto the metal, so the slightest vibration moves the two
pieces of metal apart, and an intermittent connection is formed. As you know,
intermittent connections can be extremely difficult to find.
When soldering awire to atag (or to the copper conductor of aprinted circuit
board), do not let solder get onto the wire, to start with. Place the soldering iron
on the metal tag (or copper conductor) on one side of the wire and heat this for 2
or 3 seconds. Then apply some solder to the tip of the iron and move the tip
backward and forward along the metal tag until the solder positively wets the tag.
Do the same to the tag on the other side of the wire, heating it first, then applying
solder and spreading it with the tip. Finally and only when solder is adhering to
the tag on both sides of the wire, move the soldering iron across and in contact
with the wire, applying alittle more solder, so that the whole joint is heated and
liquid together. Then, remove the soldering iron and make sure the wire doesn't
move until the solder has solidified. Push and pull the wire to test the joint. If the
wire moves relative to the tag, it is abad joint and needs resoldering.
The reason why the metal tag must be soldered before the wire is that the tag
contains more metal than the wire and takes far longer to heat up. If solder is
applied first to the wire, it becomes fused onto the wire; at this point, however,
the tag is comparatively cold and the solder won't fuse onto it, so adry joint is
produced.
Semiconductors can easily be destroyed by overheating, so special
precautions have to be taken. General purpose transistors used for low-power
signal processing—such as op-amps and diodes—do not get heated while they are
operating. However, transistors and integrated circuits that conduct sizable
current flow are constructed with a metal back plate, which is designed to be
bolted onto a heat sink. Any semiconductor with such a back plate should be
installed with aheat sink to dissipate its heat during operation. Also, the cabinet
should contain ventilation slots.
In the power supply circuit illustrated earlier in Figure 10-3A, the voltage
regulators are of this nature. Heat sink compound should be applied to the back
plate, and the back plate should then be bolted onto a heat sink before
installation. The heat sink compound assists thermal conduction. During layout of
these components, make sure that the heat sinks cannot touch each other. Each
heat sink is electrically connected to its transistor. A short circuit and burnout of
the component can result if two heat sinks at different voltages touch each other.
206 Chapter 10

While soldering any semiconductor, precautions should be taken to avoid

overheating. A temporary heat sink, consisting of any metal object, should be
placed so that it touches the electrode between the semiconductor body and the
point where the soldering is taking place. This drains off some of the heat
produced by soldering before it reaches the body of the semiconductor. An
alligator clip can be temporarily attached for this purpose. Or asmall screwdriver
might be wedged between the electrodes of a transistor while they are being
soldered. As an additional precaution, limit the time spent in soldering any one
transistor connection to not more than 4 or 5 seconds. If soldering is not
completed by then, leave the joint to cool while other connections are being
made. Then spend another 4or 5seconds finishing off the soldering. In this way,
semiconductors will not be damaged during installation.
Audio Chain Troubleshooting, 209

Power Supply Troubleshooting, 212

Tape Recorder Alignment, 214

207
 11 Troubleshooting and
Maintenance

Audio Chain Troubleshooting

An important ability of an audio engineer is to be able to locate afault when it
occurs, and if possible, to be able to repair it. Excessive time, trouble, and
expense result from having to take each piece of faulty equipment to arepairer. In
addition, faults often occur in a cable or patch cord, which can easily be
misinterpreted as occurring in acomponent that is, in fact, working perfectly.
Troubleshooting is like solving a detective mystery. The same logical
deductive process needs to be used as would be appropriate in a criminal
investigation. The method is simple. Narrow down the possible list of suspects
until there is only one left—the guilty party. And don't forget to include in your list
all cables and patch cords used in the interfaces between components.
Let's take a typical audio chain as an example. We will start with a
microphone followed by a mic pre-amp, a line amp, an equalizer, a mixer, a
power amp, and ending with aloudspeaker (see Figure 11-1).
Suppose this system is switched on. (On many occasions, components have
been said to be "not working" when they were either not plugged in or not
switched on. So watch out for this one!) When sound is fed into the microphone,
nothing comes out of the loudspeaker. The audio chain can be tested to find the
position of the fault, starting from one end and working to the other. It is asimple,
step-by-step process.

Troubleshooting Left to Right

If we decide to work from left to right, we need asignal generator feeding the mic
pre-amplifier in place of the microphone. This is set to 1kHz (audio test frequency)

209
210 Chapter 11

Fig. 11-1 A typical

audio chain mic LOUDSPEAKER

MIC E POWER F
PRE- AMP AMP

LINE D
EQUALIZER MIXER
AMP

at alow output voltage level of about — 35 dBv, equivalent to what amicrophone

outputs. Also, a signal tracer is needed. This consists of a small, low-power
amplifier with abuilt-in loudspeaker. A signal tracer is easy to make, as described
in Chapter 10. It is useful to connect alead terminated by an alligator clip to the
ground input of the signal tracer, and alead terminated by aprobe to the hot lead
input. Then the ground can be kept permanently connected to the ground bus of
the audio chain, while probing the signal at various interfaces along its path.
Now connect the tracer input to points B, C, D, E, and F in sequence (see
Figure 11-1). The stage before the point where the signal disappears is the faulty
stage. For instance, suppose asignal is detected at points Band C, but not at point
D; this means that the signal is passing through the pre-amp and through the line
amp, but failing to pass through the equalizer. It is obvious that the EQ is where
the fault lies. Don't assume yet that the fault must be in the EQ itself. Remember
the cables. At least 50 percent of all faults occur in cables, not in audio
components. So test the interface cable between the line amp and the EQ. Also,
test the interface cable from the EQ to the mixer. If both of these cables are good,
you can strongly suspect that the EQ is the culprit.
During aprofessional recording session, it is not possible to make repairs. The
procedure then is to find and pull out the faulty component and replace it with agood
one. Speed and accuracy in finding the fault and getting the system working again is
what is needed. Then, at aconvenient time, the faulty component can be repaired.

Troubleshooting Right to Left

Imentioned that the audio chain can be tested either from left to right or from
right to left. If we decide to start the troubleshooting from the right, a signal
tracer is not needed. Only asignal generator is used. Again, it is useful to have a
cable with the ground output from the signal generator terminated by an alligator
clip and the hot lead terminated by a probe. The idea is to inject a signal (of
suitable voltage) at points E, D, C, B, and A in sequence, until the point is found
 Troubleshooting and Maintenance 211

where no signal output is heard from the loudspeaker. The component to the
right of this point must be the faulty one.
But there is a catch. The output from the signal generator has insufficient
power to drive the loudspeaker audibly. Therefore, the sequence of events has to
start with atest of the loudspeaker. To do this, disconnect the speaker from the
power amplifier. Use the multimeter or FET meter to measure resistance on the
Rx1range, and connect the meter probes to the loudspeaker terminals. If the
speaker makes an audible click when the probes are connected, and another click
when the probes are disconnected, then the speaker is good. If there is no click,
and the meter reads infinitely high resistance, the voice coil is open circuit and
the speaker is not working. (The same method can be used to test a dynamic
microphone, but in this case, it is advisable to set the meter to the Rx1ka range.)
If the speaker is working, reconnect it to the audio chain and proceed with
the test by injecting the audio signal at points E, D, C, B, and A in sequence. Note
that the signal voltage level to be injected anywhere along the audio chain should
be at about 0dBv—that is, about 0.8 V. However, at the input of the pre-amplifier,
the injected signal should be about — 35 dBv, which is about 15 mV.
There is one other improvement that can speed up this method of
troubleshooting. Start by injecting the signal at the mid-point of the audio chain,
instead of at one end. If the signal is injected at point D as afirst step, and if it is
heard coming out of the loudspeaker, it means that all stages following point D
are working, and the fault must lie to the left. Then proceed to points C, B, and A
in sequence. If no signal is heard from injection at point D, then the fault follows
this point; proceed to points Eand Fin sequence. This is the quickest possible way
to locate afaulty stage.

Cable Testing
Imentioned that often asuspected cable or patch cord has to be tested. The method
is this. Set the multimeter to measure resistance on the Rx1range. With both ends
of the cable held close to each other, connect the probes of the meter to the two
ground connections, one at each end of the cable. The meter should read less than 1
ohm. Then connect the meter probes to the two hot lead terminations. Again, the
meter should read less than 1ohm. If there are three conductors, acontinuity test
like this should be applied to all three conductors. If, in any case, the meter reads
infinitely high resistance, that particular conductor is open circuit.
Next, test for short circuits on any of the cable terminations between ground
and each of the hot leads in turn. The meter should read infinitely high resistance.
If it reads 1ohm or less, there is ashort circuit.

Electronic Circuit Troubleshooting

A method similar to audio chain troubleshooting can be used when
troubleshooting an electronic circuit, such as an amplifier or mixer. First, check
to see that the DC power supply voltages are present. Then, to find the faulty
212 Chapter 11

stage, scan the signal path either from input to output or from output to input, as
was done in the audio chain troubleshooting procedure.
The test equipment needed is the same—namely, an audio signal generator, a
signal tracer (optional but useful), and an AC/DC VTVM or FET meter. When
scanning from input to output, inject asignal at the input and use the signal tracer
to probe for this signal at various places along the signal path. The point where
the signal is lost identifies the faulty stage. Alternatively, with an amplifier or
signal tracer connected to the output, inject asignal at various stages from output
to input. It is useful to have a manufacturer's schematic diagram available. This
helps to identify the signal path. It also provides information on the correct
quiescent voltages at the transistor terminals.
When the faulty stage has been found, use the FET meter or VTVM to
measure the DC voltages (under quiescent conditions) at the terminals of the
transistor. This should give aclue as to the fault. If the power supply voltage is
missing, find the break in the supply line. If power is present but other voltages
are faulty, it may be that the transistor needs replacing. By measuring the voltage
drop across aresistor, the current flow can be found. Thus, it is always possible to
find out if a transistor is conducting; just find the current through its emitter
resistor. Theoretical understanding is necessary, but as in learning to ride a
bicycle, there is no substitute for practice.

Intermittent Faults
The most difficult faults to find are the intermittent faults. You know Murphy's Law on
intermittent faults: "When the equipment is being tested, it works perfectly. Only
when it is replaced in its cabinet or put back into service does it immediately stop
working."
In the case of cable testing, it is agood idea to wiggle the ends of the cable
while testing for continuity or short circuits. Bending the cable near its connectors
might disconnect aloose connection, while bending in the other direction might
give an illusion of continuity.
The only other advice Ican give about intermittent faults is gently to kick,
joggle, or tap on components or wired connections. If aclick or crackle is heard,
that is aclue. When working inside an electrical circuit that is powered, use an
insulated rod, such as aballpoint pen sleeve with the refill removed. With this,
gently tap all wires and components until one clicks or crackles when tapped.
This may lead to a dry joint or some other loose connection that might be the
cause of the intermittent fault.

Power Supply Troubleshooting

The action of apower supply was described in Chapter 10; from this it is easy to
apply the overall troubleshooting method if power supply failure should occur.
Troubleshooting and Maintenance 213

However, before talking about dealing with apower supply that has completely
died, Ineed to mention acommon problem that is more subtle in its effects.
It sometimes happens that a 60 Hz or 120 Hz line frequency hum is heard
superimposed on the audio signal. There are several possible causes. One lies in
the power supply—especially if the equipment is old—and results from partial
failure of asmoothing capacitor.
When an electrolytic capacitor gets old, some of its electrolyte might dry out.
Or the capacitor might start to leak its charge internally. Either of these faults is
capable of reducing capacitance. The drying out of electrolyte means that a
smaller area of plate is charged. A leaking of charge means that the time during
which acharge is held is reduced, so the capacitor reacts just as asmaller one
would.
If this happens to the smoothing capacitor, the AC ripple increases in
amplitude, and the loss of voltage between peaks causes the troughs of the ripple
to descend to avalue below the output level of the regulator. Then asevere ripple
is audible in the audio signal.
If the smoothing capacitor is suspect, there is an easy way to check it. Obtain
another large electrolytic capacitor. It needs to be of about the same capacitance
as the one in the power supply, but its value is not critical. It also needs to have at
least the same working voltage. With the line plug removed from the wall socket
and the equipment switched off, discharge the existing smoothing capacitor by
holding the blade of ascrewdriver across its terminals. This will probably cause a
flash and abang. (There is amore civilized way of discharging acapacitor: hold a
low value resistor across the terminals. This discharges it in about 2 seconds
without abang.)
Connect your spare capacitor temporarily in parallel with the one in the
power supply. An easy way to make atemporary connection is to use two flexible
insulated wires terminated by alligator clips. Watch out for two things.
Electrolytic capacitors are polarized. The positive terminal of the capacitor must
be connected to the more positive terminal of the circuit; so make sure that the
temporary capacitor is connected with the same polarity as the one in the circuit.
The other precaution, already mentioned, is that the DC working voltage of your
temporary capacitor must be equal to or greater than the one in the circuit. With
connections complete, switch on and see if the hum is reduced or eliminated. If it
is, the problem can be solved by replacing the old capacitor with anew one. After
the test is completed, switch off and discharge the capacitors as before. Then
remove the temporary capacitor.
If complete failure of the power supply should occur, it is easy to find the fault
and make the necessary repairs. As a first step in all troubleshooting, make a
visual inspection of the circuitry. Look for any wires that have come adrift, or any
blackened or burnt components. Also, test all fuses for continuity. If ablown fuse
is found, don't just replace it. The cause of the failure also needs to be corrected.
Replace any fuse with the correct current rating or it will be worthless.
The next step is to work through the circuit from left to right, just as was
suggested in the previous section. However, now we are looking for electrical
214 Chapter 11

power, not an audio signal, so avoltmeter is all that we need. The discovery of a
faulty voltage instantly indicates where the problem lies. Knowledge of the theory
behind apower supply circuit will tell you what voltages should be present at the
various stages from source to output.
Starting from the left, there should be 115 AC volts across the transformer
primary. There should be a few more AC volts than the power supply's rated
output across the secondary. (For instance, a 15 V power supply should have
about 18 AC volts across its secondary.) At the output of the rectifier, which is
most readily accessed across the terminals of the smoothing capacitor, there
should be roughly the same voltage, but now it should be DC rather than AC. At
the output terminals, there should be the rated voltage. If the final voltage is too
high, the regulator needs replacing. If, at any point, the voltage is zero, the
component to the left is the faulty one.

Tape Recorder Alignment

For tape recorder alignment the following equipment is needed:

• Denatured alcohol and cotton swabs

• Head demagnetizer
• VTVM or FET meter or good-quality multimeter
• NAB test tape
• Signal generator

Cleaning and Demagnetizing

Before arecording is made and before any alignment is carried out, the heads and
tape guides should be cleaned and demagnetized. Demagnetization is
particularly important before the test tape is threaded, because a residual
magnetism on the heads will partially erase the high-frequency test tones. If that
should happen, the test tape will no longer be useable as areference guide.
Cleaning is best carried out with denatured alcohol on acotton swab. Wipe
over the surface of the heads, tape guides, capstan, and pinch roller. Clean up any
surplus alcohol with the dry end of the swab.
Next, use a head demagnetizer. This consists of a 115 AC line voltage
transformer with an extended core. Some have a switch button that has to be
depressed while in use. Others are energized as soon as they are plugged in.
These should not be left plugged in when not in use. The demagnetizing process is
as follows:

1. Make sure the tape recorder is switched off; otherwise, the high magnetic
flux applied to the heads could burn out the head pre-amplifier.
 Troubleshooting and Maintenance 215

2. With the demagnetizer held in the on position, bring the metal end close to
each head surface and tape guide in turn.
3. Move the demagnetizer slowly at all times. The demagnetizing process
requires slow movement away from the object in order to demagnetize it.
4. Never switch off the demagnetizer until it is at least 2feet away from the
parts being demagnetized. If this should happen, repeat the
demagnetization over again.
5. Keep the demagnetizer away from all magnetic tapes.

The need for slow movement away from the object being demagnetized
results from the fact that the alternating magnetic field produced by the
demagnetizer is strong. It magnetically saturates the object, first with one
polarity, then with the other. As it moves slowly away, its field becomes
progressively weaker. Consequently, each reversal of field not only cancels the
previously induced magnetic polarity, but produces a slightly weaker opposite
polarity. It is this gradual reduction in strength of each oppositely produced
polarity that eventually produces complete demagnetization. A sudden
movement away from the object, or aswitching off of the demagnetizer close to
the object, is likely to leave it more strongly magnetized than it was before.
It might be well to note at this point that the erase head on the tape deck
works in asimilar way. An amplified value of the high-frequency bias voltage is
used to energize the erase head. This head is placed before the record head. In
the record mode, the strongly amplified bias voltage is applied to the erase head
and it magnetically saturates the tape as it passes over the erase head. As the
tape moves farther and farther from the erase head, the opposing polarities
become progressively weaker, so that when the tape arrives at the record head, it
is completely demagnetized and ready for its new recording.

Basic Alignment Procedure

Depending on the sophistication of the tape recorder, alignment might occupy
only afew steps or many. In the case of asimple machine, with no facilities for
reproduce or record equalization, the recommended steps are these.
After cleaning and demagnetizing, check the playback level of the test tones on
the NAB test tape. Connect an AC voltmeter to the line output (or if there is none, to
the loudspeaker terminals) with a suitable loudspeaker load. A loudspeaker is
necessary in any case, in order to hear the specified frequency of each test tone.
With the machine in reproduce mode, play the first tone. This is normally a
reference level tone at the standard test frequency of either 1kHz or 700 Hz.
Adjust the gain to register 0dB on the VU meter, or 0dBv at the line output, or 5
or 10 dBv at the loudspeaker output terminals. In this case, subtract 5or 10 dB
from all output readings.
The next tone on the test tape is 15 kHz. It is for azimuth alignment. Azimuth
is the angle between the line formed by the head gap and the direction of tape
216 Chapter 11

travel. The head gap should be perpendicular to the edge of the tape. Because
high frequencies are so closely packed, their response falls off rapidly if there is
any azimuth error. Therefore, azimuth is always set and checked at the highest
frequency of 15 kHz. (The position of the azimuth setting screw for each head
should be indicated in the manufacturer's manual.)
Play the 15 kHz test tone and note its level on the output voltmeter or VU
meter. Then turn the azimuth setting screw very slowly, first one way, then the
other. The azimuth should be set to give maximum output at 15 kHz. There are
sometimes a number of false peaks at various azimuth settings. The correct
setting is the peak that gives the highest output of the various peaks that may
occur.
Once the azimuth has been accurately set, it should not be necessary to reset
or check it frequently. (Equalization checks should be done more often, however,)
Some operators like to seal the azimuth setting screw with adab of paint between
the side of the screw and its surrounding surface. This prevents the screw from
working loose as aresult of vibrations.
Next on the test tape, follow asequence of test tones from 12 kHz down to
about 50 Hz, each about 15 seconds. Tabulate each frequency and output level. If
the reproduce equalizer is functioning correctly, the levels should all be within
about 2dB of the 1kHz reference level. If they are alittle less accurate than this, it
is not important, as long as you know what they are. It is agood idea to plot a
graph of the reproduce frequency response level plotted vertically on afrequency
base scaled logarithmically. Keep this for reference. Then, if there is any sudden
change, such as afall-off of high frequencies, you will know that some corrective
action is needed.
Although asimply made consumer tape recorder does not allow you to alter
the equalization of the record system, the record characteristics need to be
known. In this way, recordings can be made as compatible with standard NAB
specifications as possible.
One of the most important things to understand about tape recording is the
significance of bias voltage. An increase in bias tends to reduce distortion and
background noise, but it also reduces the high-frequency recording response. So
the operator has to make atrade-off between improved signal-to-noise ratio and
loss of high frequencies.
A good guide with which to start is 3dB over bias at 10 kHz. This is described
in detail later. Many consumer machines have no facilities for adjusting bias. It is
then necessary to choose a tape whose bias requirements allow satisfactory
recording performance. By always using the same tape, consistent results can be
ensured. These can be established in tlie overall record/reproduce test described
next. Some machines have abias level switch that can give either high bias or low
bias. Use the low bias setting for tapes specified as requiring normal bias. Tapes
that require high bias should only be used on machines that have a high bias
facility or adjustable bias.
To carry out the overall record/reproduce test, start by removing the test tape
from the machine and replacing it with ablank tape. Feed to the line input a 1
 Troubleshooting and Maintenance 217

kHz signal from the signal generator set at + 4dBv. (This is specified as + 4dBm,
but assuming astandard load, equals + 4dBv.) Adjust the record level to give — 10
dB on the VU meter. (It is wise to test record 71
2 ips tapes at — 10 dB. This lets you
/
add the required degree of pre-emphasis to the HF tones without over-saturation
of the tape.) Keeping the record gain unchanged and maintaining exactly the
same input voltage at all frequencies, record about 10 seconds, first of the 1kHz
reference frequency, then of all other frequencies that are included on the test
tape.
After recording is complete, rewind and play back your test recording. During
the 1kHz reference tone, adjust the output gain to obtain 0dBv at the line output,
or the same 5or 10 dBv that was measured at the loudspeaker terminals. Then
tabulate the output level at each frequency and plot the overall response graph.
This gives acomprehensive view of the tape recorder's characteristics.
For best compatibility of recording performance, the overall record/
reproduce graph should not necessarily be flat. It should coincide as far as
possible with the playback characteristics of the NAB test tape. To the extent that
it does so, your recordings will be perfect when reproduced according to the NAB
standards on any high-quality machine. Once it is known where any discrepancy
lies, it is often possible to counteract this during a recording session. But it is
necessary first to have good information about your equipment.

Professional Alignment Procedure

The alignment procedure for professional tape machines involves a somewhat
different approach. The difference results from the fact that these machines have
separate record and replay heads, with user-adjustable equalization for both
record and replay amplifiers. They also incorporate user-adjustable bias voltage.
The method of cleaning and demagnetizing is exactly the same as previously
described. This must always be carried out as afirst step.
Before continuing to describe the alignment procedure, Imust mention that
there is adifference between using a71 / ips test tape and a15 ips test tape. With a
2

15 ips tape, all frequencies are recorded at operational level, which is normally
250 nanowebers/meter. This should be calibrated at 0dB on the VU meter. In the
71
/ ips test tape, the reference level and frequency test tones are all recorded at
2

—10 dB. This avoids the risk of overload distortion at high frequencies, during
which increased pre-emphasis is applied. At the end of these test tapes there is
normally areference frequency recording at operational level, 0dB. This can be
used as a final verification of operational level, if required. At each step in the
following alignment procedure, the correct level settings for 15 ips machines are
indicated, followed by the corresponding values for 71 / ips machines in
2

parentheses.
First, align the reproduce system by replaying the test tape. At the reference
tone of 1kHz, adjust the reproduce level to obtain 0dB ( — 10 dB) on the VU meter.
Next comes the azimuth setting tone. Set the playback head azimuth as
previously described.
218 Chapter 11

The various frequency tones follow, each usually about 15 seconds long. Use a
high-frequency tone of about 12 kHz to adjust the HF reproduce equalizer and set
the output level to that of the reference tone. Then replay the reference tone and
reset the output to 0dB ( — 10 dB) once again, in case the adjustment has caused
any change in the reference value. Use the reproduce level control for this
purpose.
Next, wind on until atone of about 100 Hz occurs. Use this to adjust the LF
reproduce equalizer to reproduce the correct reference level of 0dB (— 10 dB) for
this LF recording. Again, reset the original 1kHz reference tone to 0dB ( — 10 dB)
by use of the reproduce level control. (Only on the 71
2 ips test tape, there follows
/
a final reference tone recorded at operational level, 0 dB. This enables you to
make any final adjustment to the reproduce level to obtain exactly 0dB on the
VU meter.)
Now that the reproduce equalizer has been aligned for both HF and LF
reproduction, it is time to replay all of the frequency tones in sequence without
any further adjustment. Tabulate each frequency against its output level and plot
these on semi-log graph paper—the level vertically against a frequency base
scaled in octaves. This completes the reproduce equalization. It should be flat
with 2dB.
The next step is to align the record characteristics. Before doing anything
else, remove the test tape and exchange it for ablank tape ready for recording.
This type of sophisticated recorder has two great advantages over the
consumer model. First, the bias voltage is adjustable. Because every type and
make of tape requires adifferent bias, any tape can be used to its best advantage
by using the optimum bias for that particular tape. Remember that once the bias
has been set, it is not necessarily valid for any other type of tape.
The other advantage is that the recording can be monitored direct from the
tape while it is being recorded. The sequence of heads is erase, record, replay; so
what has been recorded can be monitored through headphones a fraction of a
second later, as the tape passes over the replay head.
The alignment of the record stage has to follow these steps. First, set the
record head azimuth. With blank tape threaded, connect asignal generator to the
line input. Adjust it to supply + 4dBv at 15 kHz. Run the machine in the record
mode, with the record level set to produce 0dB (— 10 dB) on the VU meter.
Adjust the azimuth setting screw of the record head (not the replay head). As
you turn this azimuth screw first one way and then the other, the record head
azimuth is altered. It is only when it is identical with the replay azimuth that
maximum signal output occurs. This can be read from the VU meter. The correct
setting occurs when the output VU meter reads maximum value.
Now find the optimum bias value for this particular tape. Adjust the signal
generator to output 10 kHz at the same level. Run the machine in the record mode
and turn the bias control fully counterclockwise, so that the bias voltage falls to
zero. Gradually turn up the bias voltage, watching the replay level on the output
VU meter. As the bias increases, the level of the recorded signal will increase
until a point is reached where it stops increasing and starts to fall as the bias
Troubleshooting and Maintenance 219

continues to rise. Reduce the bias until the maximum output is again found and
note this value. Continue to increase the bias until the output has fallen by 3dB
below its maximum value. This is one method of finding the optimum bias for this
tape. It is called 3dB over bias at 10 kHz. Other bias setting methods can be used.
It is useful to consult the tape manufacturer's literature to find its recommended
bias setting procedure.
Now that the record head azimuth and bias voltage have been correctly
aligned, it is possible to align the equalization of the record amplifier. Refer back
to the circuit configuration in Figure 8-4; it will help you understand the
significance of the following steps. Set the signal generator to the reference
frequency of 1kHz. Ensure that its output level remains constant at + 4dBv at all
frequencies. Run the tape recorder in the record mode and adjust the record level
to produce 0dB (- 10 dB) on the VU meter. (It is assumed that the reproduce level
control remains at the operating level established during reproduce alignment.)
Now set the signal generator to ahigh frequency of about 12 kHz (keeping its
output at + 4dBv), and adjust the HF record equalizer until the replay level is 0
dB (— 10 dB) on the VU meter. Return the signal generator to the reference
frequency of 1kHz and recheck that the recording produces 0dB (— 10 dB) on the
VU meter. Readjust the record level control slightly, if necessary, to obtain this.
Next, set the signal generator to alow frequency of about 100 Hz and adjust
the LF record equalizer to obtain 0dB (- 10 dB) on the VU meter. Recheck the
reference frequency of 1kHz for 0dB (— 10 dB) replay, adjusting the record level
slightly, if necessary. The record equalization is now complete.
Step the signal generator through all of the frequencies listed in the test tape
and note the resulting level of each recorded frequency, as indicated on the
reproduce VU meter. Tabulate and plot the results of the overall record/
reproduce response. This graph should closely coincide with the replay responses
from the NAB test tape. You will then know that the recording characteristics will
give excellent results when reproduced on any good-quality NAB machine.
In the case of 712 ips machines, an additional step is necessary. With the
/
signal generator at reference frequency and + 4 dBv, make a recording,
increasing the recording level until 0dB is obtained on the VU meter. This aligns
the record level control to the correct operating level.
If there is a record calibrate control, a further step is needed. This record
calibrate alignment is the same for both 15 and 71/ ips speeds. Make an additional
2

recording of 1kHz reference frequency with the signal generator at the usual + 4
dBv. Now switch the VU meter to Input ( switch A in Figure 8-4) and adjust the
record calibrate level to obtain 0dB on the VU meter. Switch back to Reproduce.
The meter should read the same.
If there is a sync facility, next align the sync level. Switch off the signal
generator and wind back some of the tape that you have just recorded at
reference level. Play back this tape. With switches A and B (both operated by a
single switch—refer back to Figure 8-4) in the Sync position, adjust the sync level
to obtain 0dB on the VU meter. The meter should now read the same in both the
Sync and Reproduce positions. This completes the alignment of the tape machine.
Output: Voltage of Differential Amplifiers, 223

Output Level of High-Pass Shelving Filters, 225

Output Level of Bandpass Filters, 230

Transfer Function of LF Equivalent Circuits, 237

Hysteresis, 239

221
 A Formulas and Derivations

Output Voltage of Differential Amplifiers

(See Differential Amplifiers in Chapter 3.)
The formula for finding the output voltage of a differential op-amp can be
derived as follows. First, find the outputs for each separate input voltage. Let us
call the output from VI,Vu ,and that from input V 2 ,V L 2. Superimposing them
gives the total output voltage (refer to Figure 3-8).

1. Output due to V 1

V IL1 = V 1 Gy

But,

Gy =
Rf (for an inverting op-amp).

So,

(A-1)
vLi = - V,

(The negative sign is due to the inverting character of this input.)

2. Output clue to V2

VL2 - V2 'Gy

223
224 Appendix A

But,

Gy = 1 + ---f
- (for anon-inverting op-amp),
R,

or

Ri + Rf
GV = Ri

So,

(A -
2)
VL2 = V2'
(R i
+i Rf
)
R

However, V2'is the voltage at the op-amp itself. Its value in terms of the
applied voltage, V2 has to be found. Between V2 and ground there are two
resistors— R, and Rg—forming avoltage divider. From the Voltage Proportionality
Law, we know that

V2 V2
Rg = Rf+R g

So,

V2' = V2 (R, + g Rg )
R

But,

Rg = Rf

so this equation can be written

Rf

V 2' = V 2 (R i Rf

Substituting this expression for V2'in Equation A-2 gives

Rf Ri + Rf)
VL2 = V2 (Ri Rf X Ri

Canceling the common term (Il i+ Rf)leaves us with

R (A-3)
VL2 = V2 1
Ri
 Formulas and Derivations 225

The final output voltage can be found by superimposing the two output voltages,
VL,and VL2 ,from Equations A-1 and A-3.

Rf Rf

VL = V 2 — v

Factoring, we get

VL = -J (V2 — V,)
R;

This demonstrates the derivation of Equation 3-2, in Chapter 3.

Output Level of High-Pass Shelving Filter

(See The Pole/Zero Approach in Chapter 4.)
Draw the Bode plot and graph the output level of the high-pass shelving filter
in Figure A-1.

Fig. A-1 HP
shelving filter

Zo
Transfer function, H(S) = =

but

Zo = R2

and

R,(1/ SC)
Zi = + R2
+ ( 1/ SC)
226 Appendix A

Multiplying the numerator and denominator of the fraction by SC, we get

Zi = R2
SRC + 1

So,

H(
S) = = RI

SR 1C + 1 +

Multiplying the numerator and denominator by (SR IC + 1) produces

R2 (SR IC -I- 1)
H(
S)
= SR 1R2C + R1 + R2

Divide out K, the DC gain. (By inspection of the schematic, it can be seen that at
zero frequency, the capacitor becomes an open circuit, so the gain is given by the
voltage divider formed by the two resistors only.) So, the DC gain is

K — R2
R 1 + R2

The transfer function thus becomes

1)
H(
S = R2 SR 1C + 1 (A-4)
) R I 4. „s2
K s R 1R 2 C

RI -I- R2

The equivalent standard form would be

(A-5)
H (S) = K (ST ' + 1)
ST2 1

By comparing Equations A-4 and A-5, it is clear that the coefficients of Sare the
time constants. So the constants are

R112 2C
K = R2 T1 = RIC
R 1 + R2 7 2— R 1 + R2
Formulas and Derivations 227

Substituting the actual component values,

20
K = 1200 + 20 = 0'
0164
= 1.2 x10 3 x1.5 x10 -6

1.2 x10 3 x20 x1.5x 10 -6

• - 1200 + 20

and

Level K = 20 log 0.0164 T1 = 1.8 x 10 -3 s

72 = 2.951 x 10 -5 s

So level K = — 35.7 dB.

From these time constants, we see that

f
l
1 1 = 88.4 Hz (azero)
- 21-T 1 27(x 1.8 x10 -3

1 1
f2 = 2rT2 - 2r x2.951 x10 -5 — 5.39 kHz (apole)

To construct the Bode plot, start by drawing a vertical dotted line at each
break frequency (see Figure A-2). Note that at the lowest limit of frequency (when
S = 0), the level is — 35.7 dB. So draw a horizontal line at — 35.7 dB from the
Y-axis until the 88.4 Hz break frequency is reached. This is azero, so the graph
turns up there by 6dB/octave. A 6dB/octave slope is simply astraight line that
rises (or falls) by 6dB, for a horizontal displacement of one octave (double the
frequency). Continue this slope until the next break frequency is reached at 5.39
kHz. This is apole, so the graph turns down here by — 6dB/octave, bringing the
Bode back to ahorizontal straight line. This completes the Bode plot.
For calculation purposes, any (STn+ 1) term can be replaced by

So the standard form of

(ST 1 + 1)
H(S) = K ,
228 Appendix A

becomes

H(f) = K

Thus, the level change is shown as

(f +

LVG = 20 log K

Substituting the values for K and the break frequencies,

(A-6)
+ 1
e\ 2

Lv0 = 20 log 0.0164

¡ ( f_ \ 2

+1
5.390)

Fig. A-2 HP 4
shelving filter I
- I
BODE PLOT
response o

8
ACTUAL 'FILTER
— 12

m — 16
-o
o
> 20

— 24

—28

— 32
• •
—36
f(Hz)
— 40
15.625 31 25 62.5 125 250 500 1kHz 2 kHz 4 kHz I 8 kHz 16 kHz

f2
Formulas and Derivations 229

The filter response can now be plotted from Equation A-6 to show the output
level at all frequencies. It is suggested that the value be found at each octave
within the audible range from 15.625 Hz to 16 kHz. The graphs can be
conveniently plotted on a frequency base scaled in octaves, forming a base 2
logarithmic scaled horizontal axis. An HP 11 calculator program can be used to
facilitate calculations. After programming, return to the run mode and proceed as
follows. For the first run only, initialize with keystrokes g RTN, 15.625, STO O.
Then key fA. The frequency will be displayed. To find the output level at this
frequency, press R/S. It is necessary only to key fA and R/S for each subsequent
reading. The frequencies will be incremented in octaves automatically. To repeat
the entire sequence, key in 15.625, STO 0, then fA and R/S. ( The program for
this calculation is shown in the box below, along with the response in Table A-1.)

HP 11 Program for Shelving Filter Analysis

Program Initialization
fLbl A gRTN
RCL 0 15.625
R/S STO 0
88.4 fA
R/S
g x 2

RCL 0
5390

g x 2

V--)
C
.0164

glog
20

STO 1
RCL
2

STO 0
RCL 1
gRTN
230 Appendix A

Table A-1
Shelving filter Frequency (Hz) L vG dB
response
25.625 — 35.6
31.25 — 35.2
62.5 — 33.9
125 — 30.9
250 — 26.2
500 — 20.6
1,000 — 14.7
2,000 — 9.2
4,000 — 4.5
8,000 — 1.6
16,000 — 0.5

Output Level of Bandpass Filter

(See Bandpass Filters in Chapter 4.)
Draw the Bode plot and graph the output level of the bandpass filter shown in
Figure A-3.

Fig. A-3
Bandpass filter v; o

To simplify the process, call the impedance of the network between A and
ground, ZA .This consists of the series connected R2 and C2 shunted across R,
(that is, in parallel with R1). So,

R,[R 2 + ( 1 / SC 2)]
zA =
+ R, + ( 1/SC 2)

Multiplying the numerator and denominator by SC 2,we get

Z RI(SR 2 C 2 + 1) (A-7)
A S(R1 R2)C2 + 1
 Formulas and Derivations 231

The simplified circuit that we are considering now looks like that in Figure A-4.

Fig. A-4 v, o
Simplified
cl
bandpass filter
circuit
Vo

Between Viand ground there is asimple series circuit, so VA can be found

from

V. Z •
VA = A

where Z, = total input impedance.

Vi ZA (
A-8)
VA = ( 1/SC 1) + ZA

Substituting in Equation A-8 for ZA from Equation A-7, produces

ViRi(SR2C2 + 1)
S(R i + R2)C 2 + 1
VA = 1 Ri(SR2C2 + 1)
SC 1 S(R i + R2)C2 + I

Adding the two denominator terms by use of acommon denominator of

SC I[S(R i + R2)C 2 + 1]

Vift i(SR2C2 + 1)
+ R2)C 2 + 1
VA S(R i + R2)C 2 + 1 + SC IRI(SR 2C2 + 1)
SC I[S(R i + R2)C 2 + 1]

and canceling like terms,

\T ilt i(SR2C2 + 1)SCi (

A-9)
VA S(R i R2)C 2 + 1 + SC IR I (SR 2C2 + 1)
232 Appendix A

With this value for VA ,we can now find Vo from the series circuit consisting of
R2 and C 2 in Figure A-3.

V = V A(1/ SC 2)
()
R 2 + ( 1/SC2)

Multiplying numerator and denominator by SC 2,we get

VA
V. — Qn„
+

Substituting for VA from Equation A-9 we see that

V,RI (
SR 2C2 + 1)SC I
= S(R 1 + R2 )C 2 + 1 + SC IRI(SR 2C2 + 1)
Vo
SR2C2 + 1

Cancelling like terms and multiplying out the denominator gives

V,SRICI
V0
SR 1C2 + SR 2C2 + 1 + S2R1R2C,C 2 + SR ICI

Rearranging in standard form,

v.
S2RIR2C1C2 + S(R ICI + R2C2 +RI C2) + 1

Divide both sides by V,. This gives the transfer function, Vo/V,. (Note that dividing
the right-hand side by V, simply cancels V, from the numerator.)

SR iCI (A-10)
H(S) =
S2R,R 2C,C 2 + S(R ICI + R2C2 + RiC2) + 1

The standard form of acomparable filter would be

Sr i
H(S) =
(ST2 + 1)(ST3 + 1)

When the denominator terms are multiplied out, this becomes

ST i (
A-11)
H(S) = s2727-3 + S(7-2 + 73) + 1
Formulas and Derivations 233

By comparing Equation A-10 with the standard form of afilter in Equation

A-11, it is clear that the time constants, which are the coefficients of S, are
given by

=C i

7 2 7 3 = R1R2C1C2
(Call this "P") (A-12)

T2 -I- T3 = Ric, + R2c2 + Ric2 (Call this " Q") (

A-13)

Substituting the component values,

= 6.5 x103 x .05 x10 -6 = 3.25 x10 - as

T2 T3 = 6.5 x10 3 x3.3 x10 3 x .05 x10 -6 x .02 x10 -6 = 2.145 x10 -8s("P")

7-2 + 73 = 6.5x 10 3 x .05 x10 -6 + 3.3x 103 x .02 x10 -6

+ 6.5x 103 x .02 x10 -6 = 5.21 x10 -4s("Q")

Using Pand Q for the identity, for simplicity, and solving simultaneously for T,

T2'r3 = P

7 2 + r3 = Q

7
-2 = T3

Substituting for T2,

P
/3 — Q
3

Multiplying by T3,

P + 7
-32 = QT3

In standard form,

T3 2 — QT3 + P =

Because this equation was derived from both the T2 and T3 relationships, one
of the solutions gives T2; the other gives T3. A quadratic equation can be solved
by the formula

—b ± Vb 2 — 4ac
xi and x2 —
2a
234 Appendix A

This requires the quadratic to be in the standard form of

ax 2 + bx + e = 0

In our equation,

732 — Q73 + P = 0

By comparison of forms,

a = 1
b = —Q
c = P

Therefore,

,, Q ± VQ 2 — 4 P
72 and , 3 =
2

Substituting the values of P and Q, we get

5.21 x10 -4 ± V(5.21 x10 -4 )

2 — 4x2.145 x10-8
72 and ,,,,..3 =
2

So,

72 = 4.7593 x10 -4 s

and

73 = 4.507 x10 -5 s

Recalling the value of 71,which was found directly from RICI,

71 = 3.25 x 10 -4 s

The break frequencies can now be obtained from

1 1
— 490 Hz
fi = 27rT I = 27r x 3.25 x 10 -4

1
— 334 Hz
f2 = 27r1T2 = 27r x 4.7593 x 10 -4

1 1
f3 = 27r73 _ — 3530 Hz
27r x 4.507 x 10 -5
Formulas and Derivations 235

Recall the equation for the standard form of this filter:

ST 1
H(S) —
(ST2 + 1)(ST3 + 1)

For calculation purposes, we can rewrite the numerator as follows:

S = HI and Ti = i
col

so,

co f
ST i = — —
co 1 — II

The denominator terms can be written

,
/[(k)2+11 [g + i]
‘

Therefore, for calculation purposes, the transfer function becomes

H(f) = ri

4 2+ 1] [(k) 2+ il
and the level is given by

LVG = 20 log H(f)

Thus,

f
ri
LVG = 20 log

,
iKky +11 [( 3)
\ 2+ 11
236 Appendix A

Entering the actual break frequencies, we get

(A-14)
490
Lv G = 20 log

1
-\1[(33f4) 2 + 1] [(35f30) 2

We can now construct the Bode plot. We can also draw the calculated
response curve from Equation A-14. For these graphs, it is best to use semi-
logarithmic paper, in which the horizontal axis is incremented in powers of 10, or
in octaves (base 2). Either gives a constant horizontal increment per octave. To
draw the Bode plot, start by drawing vertical dotted lines at each break
frequency. Because the transfer function is

H(
S) = ST1
(ST2 + 1)(ST3 + 1)

the numerator, ST1,causes an upward sloping line of 6 dB/octave (see Figure

A-5). This line strikes the 0 dB level when the value of ST 1 equals 1. A gain of 1
corresponds to 0 dB, so the Bode construction can start with a + 6 dB/octave
slope, starting at the lowest level and reaching the 0dB level at frequency f
l. (In

order to draw a 6 dB/octave slope on base 10 logarithmic paper, it might be

useful to realize that 6dB/octave forms aslope of 20 dB/decade. Decade equals
ten times the frequency.)

Fig. A-5 BANDPASS FILTER

Bandpass filter 3
response o BODE PLOT

—3
I/

6 3 dB
ACTUAL FILTER
9

12
BANDWIDTH

21

24

27

30

—33 ((Hz)
10 2 3 4 5 6 8 100 I 1kHz 10 kHz 100 kHz

f2 fil f,
 Formulas and Derivations 237

The two terms (ST2 +1) and (ST3 + 1) are both poles, being denominator
terms. Therefore, at break frequencies f
2 and f
3,the slope is reduced by 6 dB/

octave. Because the original slope was + 6dB/octave at break frequency f, draw
a horizontal line starting from the upward sloping line at the point where it cuts
frequency f2. (This appears to be at about — 3dB.)

The next pole turns the Bode down to a — 6dB/octave slope at frequency f 3.

Notice that the Bode never quite reaches the 0dB level. The horizontal part of the
plot is at about — 3dB.
In order to plot the filtering effect, plot the calculated level gain at each
octave, starting from 15.625 Hz. An HP 11 program for this calculation is shown in
the box on page 238, along with the response in Table A-2. For the first run only,
initialize with keystrokes g RTN, 15.625, STO O. Then key fB. The frequency
will be displayed. To find the output level at this frequency, press R/S. It is only
necessary to key fB and R/S for each subsequent reading. The frequencies will
be incremented in octaves automatically. To repeat the entire sequence, key in
15.625, STO 0, then fB and R/S.

Table A-2 Level

gain response Frequency (Hz) L vG dB
15.625 — 29.9
31.25 — 23.9
62.5 — 18
125 — 12.4
250 — 7.8
500 —5
4000 — 4.1
2,000 — 4.7
4,000 — 6.9
8,000 — 11.2
16,000 — 16.7

Transfer Function of LF Equivalent Circuits

(See Bandpass Characteristic in Chapter 6.)
From the LF equivalent transformer circuit in Figure 6-11A, the transfer
function is defined as

VT ZT
H(S) = = (by the Voltage Proportionality Law).
V; Z,
238 Appendix A

Program for Calculating Level Gain

Program Initialization

fLbl B gRTN
RCL 0 15.625 First run only
R/S STO 0 ]
334 fB
+ RIS
gx 2

RCL
3530

gx 2

'Ix
1/x
RCL
490

g log
20

STO 1
RCL
2

STO 0
RCL 1
gRTN

But,

Z — SL mR L (
L and RL being in parallel).
L SL m + RL

And,

SL m RL
Z, = Ro
SL m + RL
 239
Formulas and Derivations

Using acommon denominator,

SL Mo
R + RoLR + SL ML
R
SL m + R,

But,

SL m RL
H(S) ZL SL m + RL
Zi SL m R. + SL m RL + RoRL
SL m + RL

The two fractional denominators (SL m + RL)cancel; therefore,

SLmRi.
H(S) —
SL m (R. + RL) + R.R L

Divide out k, the DC gain, which occurs when reactance is zero, and,

K L
— R. + RL

Then,

RL SL m
H(S) —
+ RL sLro R. R.+RLRL

Multiply the numerator and denominator by R. + RL , to obtain the standard

RR
form, which was given in Chapter 6.
o L

/ R. + RLI
RL SL m R. RI,
H(S) — R. 4, SL m R. + 11 1.1
\ R. RL

Hysteresis
(See Bandpass Characteristic in Chapter 6 and Mechanical and Electrical
Requirements in Chapter 8.)
Hysteresis is ascientific term that describes acertain type of energy loss that
occurs when an action is initiated or reversed. It is due to akind of friction.
240 Appendix A

To illustrate this, consider an extreme case in which arusty old spring has to
be compressed and then released a number of times. A certain amount of
compression force has to be applied before the spring even starts to move. This is
due to friction. From then on, the spring compression increases in proportion to
the increasing force.
When it is time for the spring to be allowed to expand to its original state, it is
found that as the force starts to be reduced, the spring does not immediately
respond. It remains stationary due to friction. When the force is reduced enough,
however, the spring starts to expand, and this expansion continues in proportion
to the reduction of the applied force.
At each reversal of the applied force, there is alag due to friction, before the
movement of the spring responds. This is atypical hysteresis phenomenon. The
distance moved, d, as the spring is compressed could be plotted vertically against
the applied force, F, plotted horizontally. This is shown in Figure A-6A. Notice that
aloop is formed. This is called ahysteresis loop. The area within the loop gives a
measure of the loss of energy that occurs at each cycle of compression and
expansion.
A similar phenomenon occurs when a magnetizing force from a magnetic
field, H, is applied to a piece of permeable material, such as iron. There is a
similar lag in the strength of magnetization, B, which has been induced in the
iron. Similarly, at each reversal of magnetizing force, there is a lag in the
magnetic response of the iron. This is due to the friction that the domains
experience inside the iron as they try to reverse their orientation.
A graph is shown in Figure A-6B of induced magnetization in the iron, B,
plotted vertically against the magnetizing force, H, plotted horizontally.

Fig. A-6
Hysteresis loops

(A) Hysteresis in arusty spring (8) Hysteresis of soft iron

Formulas and Derivations 241

The amount of magnetization remaining in the material after the

magnetizing force has been withdrawn (has reduced to zero) is indicated by the
height, Br.This is called the retentivity of the magnetic material.
The slope of the straight portion of the graph represents the ease with which
the material can be magnetized by the magnetic field. This is called its
permeability. It is the rate of increase of magnetization (magnetic flux) with
respect to an increase of magnetizing force (field strength). The permeability is
often designated by the Greek letter IL. Thus, ti = dB/dH.
In cgs (centimeter, gram, second) units, the permeability of air or avacuum is
taken as unity. If the material is soft iron, the permeability is much greater,
because iron has the ability to concentrate amagnetic field into asmall area. The
relative permeability of iron to air is about 600 to 1.
Note that the flux density is the intensity of magnetization within the
material. The magnetic field strength is the magnetizing force that is producing
flux in the material. This is often specified in ampere turns per meter, because an
electromagnet produces afield that is proportional to the current and the number
of turns, distributed along its length, which is measured in meters. Permeability is
the ease with which a material can be magnetized. Retentivity is the amount of
flux remaining after a material has been fully magnetized and the magnetizing
force has been removed.
 eá!
l

J-.
4
•
e s
4 ‘
ki
t.!,.
k
_
e
w•-•
!-rt 4!1
.,

eiet.,
.›..:*-É
1A--
-
,* 4i; 4. .
1
54,
•tik »1' 14, - -n- t.-t• e-. rksr #.4º ,s4Ê,
r
e
'4; 5. 911.iii`; 4,2,* + 4 14‘• 4 e
-
41 r;--
, ,';•. 4.

41 4* i.gr ,
t- %4I 1,4 4 *-4 , fie 4 e qp,
twe ;riiç:e*T'es
7- 4 '114
e..tlit, 1

I** lk * r
sv * • --"*- _
r. . * jie:eitr

e 44. ,T

-4 , ...„. ,
+ Y - • • , , .,. . e
.., 1-.,:...1.-. 41 , itt r
. -*".' :* : .
4, 4 - 4 1.i
. '.
tr.,-..f.. 4 -'im •, 4- 4:: *:•-., -
i e .t,---4 ,
, i
-W..iii .siiir* ii ., ¡,: , e..-,
;. 4 ,e.-
44-.4 4 ':4 .,
'
4 '1 :-: .
1 '
,W--* ,
-.Ae'rli..-* *:.
.Pre ..:F:ii4i "' ie- -, . ilià .44r+ bi.,l‘. :4: -:#...47'.:,4- lé, , ï;. -.,.
-. _
e
.
. + +, le * il..., 4f,
',Am.- " re .j. 4
.1.. ;. , .. ...,it.»It'-- ir ei 'le k* --:.: - ',
-ite'. ‘i; 1 e 4-.`,1A • ' '',*
e,'4e ' à.'. Le Ls. I , —
- - .,.4. ' z"
re ,
_.,. -
4».
, ,,,, „ •,. , , -* lk.,-..4% ...., 4«.
: +., 2.,...#t If .-4ü ii.
, : w .., 4, .-.4:>1.-nts,' di:Y.* .4- •
. , ,.• _.. , àh.. ..,
ïj.. -+...-•ef...-..,'14 i' 76 Lt. .
,,
- r
1,4 ••
e, It! , • „
-xt. e
.litir,*-. ''' tb•."-iii- ti, '1 -* :# e'.. -- • -e-:...-
, 4.#. , rffi '..¡Veli, 4) 4 .,, ; 1,-t- ,
-41-*•#* ,i1., .
1
,-.
. -
71'2 -.4r- • •
 4 .4 m. .-* * ..* L* .
;

-,.,.:., ldir ,
ir 1.A .....-..1
1.-- .
, itk, 4..4 4. 4
•ii?-.* :4'4 -$:,, iii*,-t-
- *1- ' +. $
- • e- q' ,
0,
,, * 4 4 .40- *: -e
. 41

e
, . 4
, 4,- 1.; t :is- . ..

:* * °!r
/lr.4
,

,,*
' ir, ,
e ..,,,,„.4.-s.-1 - ,...-,
s ml..,-...si. #I,. .
411 : --
- F. ..'"i•-, 14. k .,

*k"- !,.i2 4 1.'• 4t.// --:

: 4-. *..,, -,?¡., ,.4, -gt. e :. 4r- 4.,-e ,.-
-e- -7 ''''1
s
7,- 5
.',i-e- .*:.* it- ' it -'
ê -.1
i_.-' i
* • .

..'' ..4'` .'

r.'
*-••'' r''' 'I
t' 4 4ii ... i
.7 '-
ed .z.
5
1 s
,.,-,F
,

es- , ee

*le

e.,44.. -4

41 `-d
e e ^e

1*/. ,* :tft • ./4/,, Ifé : 414- lt(

Je • ir

.4et jtee •
eft *, *.
.,,

?. is.• ,
.p.. ,se. sj;:„
. i
, ,.. ,
É 1-41:''* ';*•- 'it' 1,1
!!e.
. e
: ... "
- * .
fl. ,xf.`4î( ,,P.;?-;,* 1
.4::: -
4 4;• ,
* 4 * --•-• -
i. - ,
. ii
.. ls.i.*..à.
p.
4/... ,,/* e .g.../ es. ' k- 4,..-:"A"-4/eb: tf• ‘"ki. * lit ''‘'.>"0 * ''' le''' '
4!
. 4;-,..:• *

.4..,. : 4:....e , 4:i--.4...4

. 1.
- r: .*.::?iir. i*:•tl.,,.4- -#--« 4.i. . 1. :,,...,, - r

•Ig-
Glossary

Active Component A semiconductor or edge, thus forming aright angle with the
vacuum tube. Any electronic component edge of the tape.
whose action is altered in response to an
Balanced Line A line that carries asignal
applied voltage other than the input, or
or voltage on two conductors in which
whose action depends on input polarity.
the voltage is equally balanced on either
Alias False. In digital systems, an invalid side of ground potential. If the voltage is
audible frequency created by the AC, the two line voltages are always 180°
interaction of the sampling frequency out of phase with each other. In other
with the true audio frequency. words, they are always of equal and
opposite polarity.
Amplitude The height of the peak value of
Bandwidth The extent of the range of
awave above the zero reference value.
frequencies within certain limits.
Analog A continuously variable waveform. Frequencies at which the signal has fallen
A smooth curve, forming acontinuum of by more than 3dB from its maximum are
values between any two points. said to have been cut off. These are
called the cutoff frequencies, and they
Analog-to-Digital Converter A device form the limits of the band. The
that converts aseries of sample voltages bandwidth is the frequency difference
to their corresponding numerical binary between the high cutoff frequency and
values. the low cutoff frequency.
Asymptote A straight line to which acurve Bar A measure of air pressure. One bar
approaches ever more closely, but never equals 10 5 pascals. One pascal is 1
quite reaches. newton per square meter. The value of
10 5 is chosen because it is a convenient
Audio Signal A fluctuating voltage, whose approximation to standard atmospheric
value fluctuates in proportion to the pressure, which is 1.03 x10 5 pascals.
instantaneous pressure of asound wave.
Bel The exponent of the ratio of avalue to
Azimuth In general, a vertical line from areference value. In bel units, the value
the zenith to the horizon. In magnetic is specified as alevel. Thus, apower level
tape, a line perpendicular to the tape is the log (exponent) of P/P ref ,where Pis

245
246 Audio Technology Fundamentals

the power in watts and ' r areference

ef is zero phase angle) and a j term (the
power, usually taken as 1mW. Similarly, reactive effect, at 90° phase angle).
avoltage level is the log of V/V ref ,where However, resistance is frequency
V is the voltage and \ ireis the reference independent; therefore, in filter analysis,
voltage, usually 0.775 V. In audio the number term is zero. Consequently,
technology, levels are evaluated in the complex frequency variable,
decibels, rather than in bels. A decibel is designated by the letter S, contains only
one-tenth of abel. ajterm. It is given by S = jw, where w is
the radial frequency— that is, the
Binary A type of number system, based on
frequency measured in radians/second.
2. In a binary system, each more
One cycle (360°) equals 271-radians, so jw
significant column is twice the value of
= j2/rf, where fis the cyclical frequency,
the previous column. This means that a
measured in cycles/second.
binary number contains only two digits,
0and I. The value 2is interpreted as a 1 Compressor A component that reduces
carried into the next more significant the dynamic range of an audio signal.
column. This is achieved by a voltage controlled
amplifier that reduces the gain of high-
Bode Plot A graph of output level in dBs,
level signals.
plotted on a frequency base scaled in
octaves (that is, logarithmically). The Constant Voltage Coupling A coupling
graph is made up entirely of straight between two stages, in which the voltage
lines. Each line represents the resistive or reaching the second stage is at least 90
reactive effect of one or more filtering percent of the open circuit voltage of the
elements. Each angle, where the lines first stage.
break into different slopes, occurs at a
Corner Frequency See Break Frequency.
break frequency.
Break Frequency The frequency at which Current The rate of flow of electrons. As
a reactive filtering effect of ± 6 dB/ electrons flow in one direction, the
spaces left behind, called holes, appear
octave, starting from zero frequency,
to flow in the opposite direction. Thus,
strikes the Bode plot. At this frequency,
the new reactive effect causes abreak in current can be visualized as electron
flow (negative current flow) or, in the
the direction of the Bode plot, either
increasing or reducing its slope by ± 6 opposite direction, holes flow (positive
current flow, sometimes called
dB/octave. The break frequency is
sometimes called the corner frequency or conventional current flow).
knee. Cutoff Frequency The frequency at which
Capacitance The ability of a capacitor to the signal falls off by 3 dB from its
store an electrical charge. maximum value.

Capacitor A component that can store Darlington Pair Two transistors

electrical energy in the form of an connected as emitter followers, so that
electrical charge. the current gain of the first is amplified
by the current gain of the second. Often
Charge The electrical state of a particle. used as a high-gain audio amplifier
An excess of electrons creates anegative
output stage.
charge. A reduction of electrons
(sometimes called an excess of holes) Decibel One-tenth of abel. See Bel.
creates a positive charge. The unit of Digital A type of system in which the
charge, the coulomb, consists of processing is done by digital logic.
6.25 x10' 8 electrons. Digital logic recognizes only two values,
Complex Frequency Variable An AC called 1and 0, high and low, or true and
frequency in complex number form. A false. There can be no continuous
complex number contains a number variation of values in digital systems. All
term (the resistive effect of a filter, at value changes must occur in discrete
Glossary 247

steps, the minimum step being of unit of actual ground (the earth) is so huge
size. that no amount of current flow into or
Digital-to-Analog Converter A device out of it will measurably change its
voltage.
that converts instantaneous binary
values to their corresponding voltages. Harmonics Whole number multiples of the
Domain The smallest known magnetic fundamental frequency. Twice the
field within a magnetized material. A fundamental is the second harmonic,
domain is presumed to act as asmall bar three times the fundamental is the third
magnet with anorth pole at one end and harmonic, etc.
asouth pole at the other. In the Domain Henry The unit of inductance.
Theory of Magnetism, asolid permeable
material is assumed to contain multitudes Hertz A unit of frequency. One hertz
of domains. equals one cycle/second.

Doping The process of diffusing a small Holes See Current.

quantity of non-semiconductor material Hysteresis Loss of energy due to an
into the pure material of which alternating physical activity, such as
semiconductors are made. Doping magnetizing and demagnetizing an iron
material is of two kinds: trivalent doping core (see Appendix A).
contains three valent electrons, and
pentavalent doping contains five valent IC Integrated circuit. A miniaturized active
circuit constructed on a single
electrons.
semiconductor chip.
Dynamic Range The ratio of the greatest
to least sound intensity, or greatest to Impedance The total opposition to current
least signal voltage. flow resulting from the combined effects
of resistance and reactance.
Equalizer A component that can boost or
cut the signal level at a predetermined Inductance The ability of acoil, whenever
frequency. A shelving equalizer can it conducts current, to extract electrical
supply boost/cut at the high or low energy and store this in the form of a
frequencies, its level of gain flattening magnetic field.
out at the frequency extremes. A Inductor A coil.
parametric equalizer can boost or cut the
signal at any selected frequency within Ion An electrically charged atom.
the audible range. Kirchofrs Current Law The total current
Expander A component that expands the entering any point in acircuit equals the
dynamic range. This is achieved by a total current leaving that point.
voltage controlled amplifier that reduces Kirchoff's Voltage Law In aseries circuit,
the gain of low-level signals. the sum of the voltage rises equals the
FET Field effect transistor. A unipolar sum of the voltage drops.
semiconductor through which the Level The magnitude of aphysical quantity
current flow is controlled by an in bel or dB units.
electrostatic field
Level Gain The difference in level between
Filter A device for changing the frequency
two points.
response of a signal, so that its output
level becomes frequency dependent. Its Load That part of acircuit in which useful
gain (or loss) will then be different at work is done; the component that the
different frequencies. circuit was designed to energize.

Gain The ratio of output voltage (or power) Logarithm The exponent of a power to a
to input voltage (or power). given base.

Ground Zero reference point of voltage. Microphone A transducer that converts

Called ground because the capacitance acoustical energy into electrical energy.
248 Audio Technology Fundamentals

Mixer An audio component in which Pre-amplifier A component designed to

signals from various channels can be amplify very small signal voltages, such
mixed. Each channel input incorporates as those from a microphone or tape
a fader, which determines what head.
proportion of the channel signal is
Pre-emphasis Equalization applied to a
incorporated in the final output (or
signal immediately prior to recording by
mixdown).
amagnetic record head.
MOSFET Metal-oxide semiconductor field
Power Amplifier The final stage of
effect transistor. This is a field effect
amplification in an audio chain. Its
transistor in which the gate is insulated
function is to increase the power of the
from the channel by a thin layer of
processed signal, so that it can drive the
metal-oxide insulating material. It has
loudspeakers sufficiently hard to produce
certain advantages over the junction FET
the required sound volume.
(J-FET).
Newton A unit of force in the MKS (Meter, Quiescent Voltage The DC voltage at any
Kilogram, Second) system of units. point in acircuit, when power is switched
on but no signal is passing.
Noise A random fluctuating mix of all
audible frequencies. Low-level noise Radial Frequency The frequency of an AC
forms anatural background to all audible signal measured in radians/second, as
sounds. opposed to cycles/second. One cycle
equals 2/r radians.
Ohm's Law Current flow is proportional to
the voltage, and inversely proportional Reactance Opposition to current flow due
to the resistance. to the reaction of acapacitor or inductor
to the AC frequency.
Open Circuit A circuit in which the output
terminals are open—that is, with no load Rectifier A component, such as a diode,
connected. that passes current only in one direction
and can, therefore, be used to convert AC
Operational Amplifier (0p-Amp) A high.
to DC.
quality, low-power differential amplifier,
suitable for line-level signal voltages, Semiconductor Material made up of atoms
usually manufactured as an integrated having an outer sphere containing four
circuit. Its characteristics are high gain, valent electrons. Such material can be
high input impedance, and low output used to construct an active electronic
impedance. component comprising two or more
differently doped wafers of semiconduc-
Pascal A unit of pressure, defined as 1
tor material. Examples include diodes,
newton per square meter.
transistors, and integrated circuits.
Passive Filter A filter made up of passive
components, which cannot amplify. Its Shelving Equalizer See Equalizer.
effect is obtained by selectively Signal A series of voltage (or current)
attenuating certain frequencies. fluctuations that parallels the fluctuating
PCB Printed circuit board. air pressure as sound waves pass agiven
point.
Pole A single filtering element, which
causes an attenuation of signal level, as Transducer A component that converts
frequency rises, at arate of 6dB/octave. acoustical, mechanical, or magnetic
The break frequency at which such an energy into electrical energy, or vice
effect starts is often called apole of the versa.
function. Transfer Function The ratio of a filter's
Post-emphasis Equalization applied to a output voltage to input voltage. It is
signal immediately after being necessarily a function of frequency,
reproduced by amagnetic replay head. because the purpose of a filter is to
Glossary 249

produce an output that is frequency Watt The unit of electrical power, equal to
dependent. one joule of work per second. It can be
evaluated from the product of voltage
Transistor See Semiconductor. times current.
VTVM Vacuum tube voltmeter. It has an Zero A single filtering element, which
extremely high input impedance and can causes an increase in signal level of 6
be used to obtain accurate voltage dB/octave as frequency rises. The break
readings, even of very low voltages in frequency at which such an effect starts
highly resistive circuits. is often called azero of the function.
-
Index

Asymtotes, 74
A Audible frequencies, 4
Active filters Audio
characteristics of, 95 applications, digital, 178-180
high-pass, 98-99 chain, 8
inverting, 95-99 troubleshooting of, 209-212
low-pass, 96-97 circuit, 10-12
non-inverting, 99-100 input resistance, 12
Alias frequency, 174 output resistance, 12
Alignment signal voltage, 12
procedure, tape recorder standard, 32-33
basic, 215-217 components, interfacing, 12-16
professional, 217-219 digital, 169-182
of tape recorder, 214-219 advantages of, 181-182
Amplifiers signal, 6-10
Class A, 142-143 systems, 6
Class B, 144-145 Azimuth alignment, 215-216, 218
common base, 138-139
common collector, 140-141
common emitter, 136-138
differential, 63-66
output voltage formulas and Bandpass
derivations, 223-225 characteristics, transformer, 118-124
inverting, 57 filter
non-inverting, 57-59 output level, formulas and derivations
stepped gain, 59-61 for, 230-237
summing, 61-63 passive, 88-90
Analog to digital conversion, 175-178 Bandwidth, of filter, 89
Analog/digital interfacing, 171-178 Bar, 29
Anti-alias filter, 174 Barrier potential, semiconductor junction,
Anti-imaging filter, 175 131-132
Applications, dB, 34-39 Bel, 22-24

251
252 Audio Technology Fundamentals

Bias Coupling
effect on class A amplifier, 143 constant voltage, 15
tape recorder, 157-158 loss, transformer, 118-119
transistor, 134 power matched, 15-16
voltage, adjustment, 216, 218-219 transformer, 111-112
Bode plot Current carriers, semiconductor material,
construction of, 227-230 131
vs actual response, 85-86 Cutoff
Break frequency, 84 frequency, filter, 78, 89
filter, 75-76 point, filter, 77
Breakdown, zener, 132
Bridge rectifier, 190-191
Buffers, digital circuits, 173
Building out resistor, 117-118 D
Darlington transistor, 141
DAT, 180
dB
formulas, summary of, 44
C applications, 34-39
Decibel, 22-24
Cable testing, 211
Demagnetizing, and cleaning tape head,
Capacitive reactance, 69
214-215
Capacitor, smoothing, 189
Depletion FET, 148
Capstan motor, 159
Differential amplifiers, 63-66
Chain, audio, 8
output voltage, formulas and
Characteristic curve, diode, 132
derivations, 223-225
Characteristics
Digital
high-pass filter, summary of, 83-84
audio, 169-182
low-pass filter, summary of, 80-81
advantages of, 181-182
Circuit(s)
stationary head, (DASH), 180
audio, 10-12
audio applications, 178-180
standard, 32-33
control of tape speed, 160-161
clock, digital system, 172
systems, 169-171
electronic, troubleshooting of, 211-212
to analog conversion, 175-178
sample and hold, 173
Diode(s), 129-132
tape recorder, 162-165
characteristic curve, 132
Class A amplifiers, 142-143
forward biased, 130
Class B amplifiers, 144-145
reverse biased, 130
Cleaning, and demagnetizing tape head,
schematic symbol, 130
214-215
Distortion, transformer generated, 124-125
Clock, digital circuit, 172
Dither, 175
Common base amplifiers, 138-139
Domain, theory of magnetism, 154
characteristics, 139
Doping, semiconductor material, 129-130
Common collector amplifiers, 140-141
Dynamic range, 22-23
characteristics, 140
Common emitter amplifiers, 136-138
characteristics, 138
Components, audio, interfacing, 12-16 E
Constant voltage coupling, 15
Construction Electromagnetic requirements, tape
of amixer, 198-204 recorder, 157-158
of apower supply, 191-193 Electron flow, in transistor, 133-135
of asignal generator, 193-198 Electronic circuits, troubleshooting of,
Corner frequency, filter, 75-76, 78 211-212
Index 253

Electrons, valent, 129-130 Flutter, cause of, 160

Emphasis, post, 99 Formulas
Enhancement MOSFET, 148 dB, summary of, 44
Equalization and derivations
aligning of, 219 bandpass filter output level, 230-237
tape recorder, internal, 161-162 high-pass shelving filter, output level,
Equalizers, shelving, 100-106 225-230
output voltage, differential amplifiers,
223-225
transfer function, LF equiv circuits,
237-239
Free electrons, 129
Faults, intermittent, 212 Frequency
Field effect transistors (FETs), 146-150 alias, 174
construction, 147 audible, 4
types of, 147-148 break, 84
Filter(s) filter, 75-76
active, 95-106 corner
characteristics of, 95 filter, 75-76
inverting, 95-99 cutoff, filter, 78
anti-alias, 174
anti-imaging, 175
bandpass
output level, formulas and derivations G
for, 230-237
Gain
passive, 88-90 level, 24-28, 30-32
bandwidth of, 89
resistance, 39-44
break frequency, 75-76
Granulation noise, 175
corner frequency, 75-76, 78
Ground
cutoff
loop, 113
frequency, 78, 89
virtual, 54
point, 77
high-pass
active, 98-99
passive, 80-83 H
summary of characteristics, 83-84
integration, 161 Harmonics, 3
low-pass Head(s)
active, 96-97 tape, 154-155
passive, 72-79 cleaning and demagnetizing, 214-215
summary of characteristics, 79-80 gap requirements, effect of tape speed
non-inverting active, 99-100 on, 158
passive, 69-92 per channel, relationship with tape
high-pass, 80-84 speed, 180
pole, 84 Hearing, sound and, 3-6
problem, 78-79 High-pass filter
response, shelving, program for HP 11 active, 98-99
calculator, 229 characteristics, summary of, 83-84
stopband, 91-92 passive, 80-84
transfer function, 70-71 shelving, output level, formulas and
Filtering derivations, 225-230
concepts, 69-72 Hole flow, transistor, 133
effect, of transmission line, 123 Hysteresis, 239-241
reactive effect, 84-85 loss, transformer, 118
254 Audio léchnology Fundamentals

I Logic, digital, 169

Loop, hysteresis, 240
Impedance Loss
audio circuit, 12 coupling, transformer, 118-119
matching, 14-15 hysteresis, transformer, 118
loudspeaker, 121-122 Loudspeaker matching transformer,
microphone to cable, 122-123 121-122
with transformer, 113-118 Low-pass filter
Inductive reactance, 70 active, 96-97
Input resistance, audio circuit, 12 characteristics, summary of, 80-81
Insulated gate FET, 147 passive filter, 72-79
Integration filter, 161
Intensity
sound, 21-22
audible, 4 Ael
Intensity level Magnetic domains, 154
problems, 23-24 Majority carriers, semiconductor materials,
sound, 23-24 131
Interfacing Matching transformer, 113-118
analog/digital, 171-178 loudspeaker, 121-122
audio components, 12-16 microphone to cable, 122-123
Intermittent faults, 212 Mechanical requirements, tape recorder,
Inverse square law, 22 155-156
Inverting Minority carriers, semiconductor
active filters, 95-99 materials, 131
amplifiers, 57 Mixer
Isolation, with transformer, 111-113 construction of a, 198-204
adding pre-amplifier to, 202-204
four channel, 199-200
J MOSFETs, 148-150

Junction FET, 147

N
K Noise
granulation, 175
Kirchhoff's current law, 62 white, 174-175
Non-inverting
active filters, 99-100

L amplifiers, 57-59
N-type material, 130
Level
gain, 24-28, 30-32
HP 11 program for calculating, 238
power, 24-28
o
pressure, 28-30 Ohm's law, 14
resistance, 39-44 Op-amp(s), 53-66
gain, 39-44 characteristics, 53
voltage, 30-32 construction, problem, 56-57
LF equiv circuits, transfer function, Output
formulas and derivations for, level, high-pass shelving filter, formulas
237-239 and derivations, 225-230
Load power, 26 resistance, audio circuit, 12
Index 255

voltage, differential amplifiers, formulas Reactance

and derivations, 223-225 capacitive, 69
inductive, 70
Reactive effect, filtering, 84-85
Record/reproduce head requirements,
tape recorder, 158-159
Rectifier, power supply, 186-188
Passive filters
Reference level tone, 215
bandpass, 88-90
Resistance level, gain, 39-44
high-pass, 80-84
Resistor, building out, 117-118
low-pass, 72-79
Retentivity, magnetic material, 241
Permeability, magnetic material, 241
Reverse biased, diode, 130
Pinch roller, 159
Rotary
Plot, Bode, 85-86
digital audio tape, 180
Polarities, in transistors, 133-134
head, 180
Pole, filter, 84
Pole/zero approach, 84-88
Post emphasis, 99
Power
load, 26
level, 24-28
Sample and hold circuit, 173
matched coupling, 15-16
Sampling requirements
supply, 185-191
digital circuits, 172-174
construction of a, 191-193
secondary, 174-175
rectifier, 186-188
Schematic symbols
smoothing and regulating stage,
transistor, 133
188-191
FETs, 149
transformer, 186
Semiconductor material
troubleshooting of, 212-214
current carriers in, 131
Pre-amplifier stages, adding to mixer,
doping, 129-130
202-204
Shelving
Pressure
effect, 99
level, 28-30
equalizers, 100-106
sound, 28-30
filter
Problem(s)
high-pass output level, formulas and
filter, 78-79
derivations, 225 230
intensity level, 23-24
response, program for HP 11
op-amp construction, 56-57
calculator, 229
power level, 25, 35-37, 41, 43-44
Signal
supplementry, 44-49
analog vs digital representation, 170
voltage level, 33, 35-38, 41, 43
audio, 6-10
Program
generator, construction of a, 193-198
HP 11, calculator
tracer, use in troubleshooting, 210
for calculating level gain, 238
voltage, audio circuit, 12
shelving filter response, 229
Sine wave, 5
Project construction, technical
Smoothing capacitor, 189
considerations, 204-206
Solder joints, dry, 20-5
P-type material, 130
Sound
and hearing, 3-6
intensity, 21-22
audible, 4
intensity level, 23-24
Range, dynamic, 22-23 pressure, 28-30
R-DAT, 185 Sound pressure level, 28
256 Audio Technology Fundamentals

SPI.; See sound pressure level Transformer(s), 109-125

Standard audio circuit, 32-33 action, 109-110
Stepped gain amplifiers, 59-61 applications, summary of, 125
Stopband filter, 91-92 bandpass characteristics, 118-124
Successive approximation register (SAR), coupling, 111-112
177 distortion generated by, 124-125
Summing amplifiers, 61-63 hysteresis loss, 118
Systems, audio, 6 impedance matching function, 113-118
isolation function, 111-113
loudspeaker matching, 121-122
power supply, 186
voltage changing function, 109-125
Transistor(s), 133-135
T bias, 134
Darlington, 141
Tape
electron flow in, 133-135
head
field effect, 146-150
cleaning and demagnetizing, 214-215
construction of, 154-155 polarities in, 133-134
schematic symbol, 133
effect of wear on response, 159
Transmission line, filtering effect of, 123
recorder
Transport, tape, requirements of, 159-161
alignment of, 214-219
Troubleshooting
basic, 215-219
professional, 217-219 audio chain, 209-212
electronic circuits, 211-212
bias voltage, 157-158
power supply, 212-214
consumer type, 162-163
electromagnetic requirements, 157-158
electronic circuit requirements,
162-165 V
internal equalizaton, 161-162
mechanical requirements, 155-156 Valent electrons, 129-130
professional type, 162-165 Virtual ground, 54
record/reproduce head requirements, Voltage
158-159 bias, tape recorder, 157-158
tape speed, 158 changing with transformer, 109-111
theory of operation, 154-155 level, 30-32
recording problems, 33, 35-38, 41, 43
concepts, 153-154
speed, 158
digital control of, 160-161
effect on head gap, 158 W
relationship with heads per channel,
White noise, 174-175
180
Wow, cause of, 160
transport, requirements of, 159-161
Test, of cables, 211
Thermal runaway, 137
Timbre, 5
Transfer function
z
filter, 70-71 Zener
LF equiv circuits, formulas and breakdown, 132
derivations for, 237-239 diode, 132
/ -Y-1....1-- 3 ,==CC=.
 Sound Recording Handbook Audio Production Handbook for Sound Engineers: Recording Demo Tapes
John Worm Techniques fer Video The New Audio Cyclopedia at Home
Destined to become the audio in- David Miles Huber Glen Ballou, Editor Bnwe Bartlett

dustry's new "standard" reference, This reference book examines the The most authoritative audio refer- This easy-to-follow guide details
Sound Recording Handbook is important role that audio plays in ence on the market, this book how to create aprofessional-quality
written by one of the foremost video production. It is awell- offers the professional engineer or demo tape without the expense of
audio experts. It assumes abasic rounded assessment of the equip- technician aone-stop guide to the arecording studio. It describes
understanding of recording and au- ment, techniques, and technology complete field. what equipment is needed, how it
dio technology as it targets the required to understand and create Editor Glen Ballou and 13 other works, and how to use it, then ex-
intermediate-level professional audio in today's world of media audio authorities have written the plains how to use the newly creat-
engineer, student, or audiophile. production. 31 chapters that are combined into ed tapes to get engagements and
It's systematic and in-depth treat- Bridging the gap between the cur- 7parts as follows: recording contracts. Clearly or-
ment of the entire sound recording rently merging technologies of au- ganized and thorough, its extensive
• Acoustics— Fundamentals; coverage includes judging sound
chain contributes to its success. dio and video production, this book Psychoacoustics; Small Room
Everything from sound basics, outlines modem audio production quality, troubleshooting bad sound,
Acoustics; Common Factors; promoting the demo tape, on-
microphones, speakers and tape and post-production techniques for Acoustic Design; Recording Stu-
recorders to studio recording ses- video. It thoroughly covers the location recording, and sam-
dio Design; Rooms for Speech, pling/sequencing/MIDI.
sions and mixing techniques is often confusing and misunderstood Music, and Cinema; Open Plan
covered in this important hand- time code, e:ectronic editing, digital Rooms Topics covered include:
book. New topics such as time audio, multi-track audio, and live • Electronic Components for • The Recording and Repro-
code, Dolby sound recording, and broadcast stereo. Sound Engineering— Resistors, duction Chain
digital audio are also addressed, Unlike any other book on the mar- Capacitors, and Includes; • Equipping Your Home Record-
making this the most comprehen- ket, this book addresses the specific Transformers; Tubes, Discrete ing System
sive coverage of recording studio needs of the audio track in video Solid-State Devices. and Inte- • Setting Up the System
technology available on the market. tape production and the new audio- grated Circuits; Heat Sinks, • Recording aSoloist or Small
Topics covered include: for-video standards set for the in- Wire, and Relays Acoustic Group
• Basic Theory dustry. • Electroacoustic Devices— • Recorder/Mixer Features
III Music & Psychoacoustics Topics covered include: Microphones; Loudspeakers, En- • Signal Processors
• Microphones closures, and Headphones • Microphone Techniques
•I The Audio Tape Recorder/Video • Audio Electronic Circuits and • Tape Recording
• Stereo Microphones Tape Recorder
• Speakers Equipment—Amplifiers; Attenua- • Session Procedures
• Synchronization tors; Filters and Equalizers; De- in On-Location Recording of
• Delay & Reverberation • Audio Production for Video
II Equalization lay; Power Supplies; Constant- Popular Music
• Audio Post-Production for Video and Variable-Speed Devices; VU • Judging Sound Quality
• Dynamic Range • Introductory Electronic Editing
• Tape and Heads and Volume Indicator Devices • Sampling, Sequencing, and MIDI
Techniques B Recording and Playback— Disk, • Uses for Your Demo Tape
• Tape Transports III Appendices: Preventing and
• Noise Reduction Magnetic, and Digital • Appendices: Training Your
Troubleshooting Time Code • Design Applications—Sound Hearing, Basics of Sound, Refer-
• Consoles Problems; Definitions and Stan- System Design; Systems for the ence Sources
450 Pages, 712 x9
/ 4,
/
3 Hardbound dards of SMPTE, MC, and Pilo Hearing Impaired; the Broadcast
ISBN: 0-672-22583-2 System 256 Pages, 71
2 x9
/ 4,
/
3 Softbound
Chain; Image Projection ISBN: 0-672-22644-8
No. 225583, $49.95 352 Pages, 71 / x93/ 4, Softbound III Measurements— Audio
2 No. 22644, $ 19.95
ISBN: 0-672-22518-2 Measurements; Fundamentals
No. 22518, $29.95 and Units of Measurements
1,264 Pages, 8x10, Hardbound
ISBN: 0-672-21983-2
No. 21983, $ 79.95

Visit your local book retailer or call

800-428-SAMS.
 How to Build Speaker Introduction to Professional John D. Lenk's Modern Recording Techniques,
Enclosures Recording Techniques Troubleshooting & Repair Third Edition
Badmaieff and Davis Bruce Bartlett of Audio Equipment David Mlles Hube and Robert A. Runstein
John D. Lenk
A practical guide to the whys and This all-inclusive introduction to Recording engineers, technicians,
hows of constructing high quality, the equipment and techniques for This manual provides the most up- and audio engineering students will
top performance speaker en- state-of-the-art recording—whetherto-date data available and asimpli- appreciate this updated version of
closures. A wooden box alone is in residences or professional stu-fied approach to practical the best-selling Modern Recording
not aspeaker enclosure—size, baf- dios or on location—offers awealthtroubleshooting and repair of major Techniques. It has been completely
fling, sound insulation, speaker of valuable information on topics audio devices. It will enable both updated with new information on
characteristics, and crossover points not found in other books on audio the beginning and the intermediate state-of-the-art audio topics includ-
must all be carefully considered. recording. level technician or hobbyist to ap- ing digital audio, random access
The book contains many detailed ply tips and tricks to any specific audio, and the ase of digital tech-
Geared primarily for the audio hob-
drawings and instructions for build- byist or aspiring professional, this equipment. nologies in audio production.
ing the various basic types of en- book delivers acomprehensive dis- This book also includes such time- The book provides abasis for intel-
closures, including the cussion of recording engineering saving hints as circuit-by-circuit ligence and understanding of
infinite-baffle, the bass-reflex, and and production techniques, includ- troubleshooting based on failure or recording technology, allowing the
the horn-projector types, as well as ing special coverage of micro- trouble symptoms, universal step- reader to get afeel for the entire
different combinations of these. phones and microphone techniques, by-step procedures, and actual scope of procedures. The book's
This practical book covers both the sampling, sequencing, and MIDI. It procedures recommended by comprehensive coverage makes it
advantages and disadvantages of provides up-to-date coverage of manufacturers' service personnel. an ideal reference for the practicing
each enclosure type and includes a monitoring, special effects, hum Topics covered include: or aspiring recording engineer.
discussion of speaker drivers, cros- prevention, and spoken-word Topics covered include:
sover networks, and hints on the • Introduction to Modern Audio
recording, as well as special sec-
techniques of construction and Equipment Troubleshooting • Introduction
tions on recognizing good sound
• Troubleshooting and Repair • Sound and Hearing
testing. and troubleshooting bad sound.
of Amplifiers and Loudspeakers • Studio Acoustics
Topics covered include: Topics covered include: IM Troubleshooting and Repair of • Microphones: Design and
• Speaker Enclosures • The Recording and Reproduction Linear-Tracking Turntables Technique
• Drivers for Enclosures Chain Ill Troubleshooting and Repair of • The Analog Audio Tape
• Infinite Baffles III Simple Home Recording Audio Cassette Decks Recorder
II Bass-Reflex or Phase-Inversion • Setting Up the Studio II Troubleshooting and Repair of • Digital Technology
Enclosures II Microphones and Microphone AM/FM Stereo Tuners • MIDI and Electronic Musical
• Horn Enclosures Techniques II Troubleshooting and Repair of Instrument Technology
• Combination Enclosures • Control-Room Techniques CD Players • Synchronization
• Crossover Networks • On-Location Recording 208 Pages, 81 2 x11, Softbound
/ • The Amplifier
• Construction and Testing • Judging the Recording • The Audio Produétion Console
ISBN: 0-672-22517-4
Techniques • Appendices: dB or not dB, No. 22517, $ 21.95 IM Signal Processors
144 Pages, 51 2 x8
/ 2,
/
1 Softbound Introduction to SMPTE Time •I Noise Reduction Devices
Code, and Further Education B Monitor Speakers
ISBN: 0-672-20520-3
No. 20520, $6.95 • Product Manufacture
416 Pages. 71 / xg
2 3/
4,Softbound
• Studio Session Procedures
ISBN: 0-672-22574-3
• Tomorrow's Industry: Just
No. 22574, $ 24.95
Around the Corner
400 Pages, 71
2 x9
/ 3/
4,Softbound
ISBN: 0-672-22682-0
No. 22682, $ 26.95

Visit your local book retailer or call

800-428-SAMS.
 Musical Applications of Sound System Engineering, The Microphone Manual: Principles of Digital Audio
Microprocessors Second Edition Design and Application Second Edition
Second Edition Don and Carolyn Dams David Vies Huber Ken C. Pohlmann
Hal Chamberlin
Like the first edition, this compre- This excellent reference bridges the Beginning with the fundamentals
This expanded and revised edition hensive text will provide you with gap between the equipment of numbers, sampling, and quan-
covers analog, digital, and useful information for the day-to- manufacturer and the microphone tizing, this is acomprehensive
microprocessor sound and music day work of designing sound user by clearly introducing and ex- look at digital audio, complete
synthesis. Its nonmathematical lan- systems—with much more material plaining microphone design, charac- with the latest technologies such
guage makes the material accessible and in-depth coverage of subjects. teristics, and theory. The book is as CD-I, CD-V, and DAT.
to musicians and computer users, It is apractical manual that careful- written for intermediate to ad-
This second edition of apopular
as wed as engineers. ly examines methods of accurately vanced audio users, professional au-
text serves equally well as atech-
New synthesis techniques, non- predicting such variables as acous- dio and video technicians and
nical reference, auser's hand-
linear waveshaping, Vosim, and the tic gain, clarity of sound, and re- engineers, and students. The latest
book, or atextbook and is
Fourier transform are covered and quired electrical input power while microphone technology— including
written by one of the country's
supported with program listings in plans are still on the drawing wireless microphones, clip and
leading audio experts. It includes
BASIC and 68000 assembly board. boundary microphones, electrical
new information on digital signal
language. Topics covered include: characteristics of the microphone,
processing, CD technology, and
single and stereo microphone place-
An entirely new section examines II Audio Systems magnetic storage, as it seeks to
ment techniques— is fully detailed
the practical application of synthe- • Mathematics for Audio Systems provide an in-depth understanding
and illustrated.
sis theory in actual synthesis • Using the Decibel of this ever-changing technology.
products, including professional and • Impedance Matching Topics covered include
Topics covered include:
studio equipment, novelty products • Interfacing Electrical and • Basic Theory of Operation
MI Audio and Digital Basics
using modem synthesis techniques, Acoustics Systems • The Microphone Transducer
• Fundamentals of Digital Audio
and sound generation circuits. • Loudspeaker Directivity and • Microphone Characteristics
• Digital Audio Recording and
Topics covered include: Coverage II Electrical Imerface-Cabe/
Reproduction
• The Acoustic Environment Connector
• Music Synthesis Principles II Alternative Digitization Methods
• Large-Room Acoustics • Microphone Accessories
• Sound Modification Methods l• Coding, Interfacing, and
• Small-Room Acoustics • Fundamentals of Single-
MI Direct Computer Synthesis Transmission
II Designing for Speech Microphone Techniques
Methods • Error Correction
Intelligibility • Fundamentals of Stereo-
• Computer-Controlled Analog • Magnetic Storage
II Designing for Acoustic Gain Microphone Techniques
Synthesis • Digital Audio Tape (DAT)
• Microphones 111 Applied Microphone Tech-
• Digital-to-Analog and Analog-to- • Optical Storage and Transmis-
• Loudspeakers and Loudspeaker aiques in Music Production
Digital Converters sion
Arrays • Applied Microphone Tech-
II Control Sequence Display and • The Compact Disc
• Using Delay Devices niques in Video/Film Production
Editing ▪ Digital Signal Processing
II Installing the Sound System • Speech and Music Reinforcement • Digital Audio Workstations
III Digital Synthesis and Sound
• Equalizing the Sound System • Appendices: The Use of
Modification • Audio and Acoustic 474 Pages, 712 x9
/ 3/
4,Softbound
Omnidirectional Microphones for
• Source-Signal Analysis Instrumentation ISBN: 0-672-22634-0
Modern Recording, Microphone
Ill Product Applications and the • Putting it All Together No. 2263 4, $29.95
Techniques for Predictable
Future • Specifications Tonal Balance Control, the
816 Pages, 614 x9
/ 4 ,Hardbound
/
1
688 Pages, 712 x9
/ 3/
4,Hardbound PZMN Boundary Booklet, List-
ISBN: 0-672-45768-7 ISBN: 0-672-21857-7 ing of Popular Professional
No. 45768, $39.95 No. 21 85 7, $ 49.95 Microphones, Glossary
336 Pages, 71 / x93
2 /,Softbound
4
ISBN: 0-672-22598-0
No. 22598, $ 29.95

Visit your local book retailer or call

800-428-SAMS.