Amplifiers - The Why and How of Good Amplification

Amplifiers - The Why and How of Good Amplification
Inc.
Web: http: // www. pearl - hifi . com
E-mail: custserv @ pearl - hifi . com
❦
Perkins Electro-Acoustic Research Lab, Inc.
86008, 2106 33 Ave. SW, Calgary, AB; CAN T2T 1Z6
Ph: + .1.403.244.4434 Fx: + .1.403.245.4456
Engineering and Intuition Serving the Soul of Music
Please note that the links in the PEARL logotype above are “live”
and can be used to direct your web browser to our site or to
open an e-mail message window addressed to ourselves.
To view our item listings on eBay, click here.
To see the feedback we have left for our customers, click here.
This document has been prepared as a public service . Any and all trademarks
and logotypes used herein are the property of their owners.
It is our intent to provide this document in accordance with the stipulations with
respect to “fair use” as delineated in Copyrights - Chapter 1: Subject Matter and
Scope of Copyright; Sec. 107. Limitations on exclusive rights: Fair Use.
Public access to copy of this document is provided on the website of Cornell Law School
at http://www4.law.cornell.edu/uscode/17/107.html and is here reproduced below:
Sec. 107. - Limitations on exclusive rights: Fair Use
Notwithstanding the provisions of sections 106 and 106A, the fair use of a copyrighted work, including such use by reproduction in copies or phono records or by any other means specified by that section,
for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright. In determining whether the use
made of a work in any particular case is a fair use the factors to be considered shall include:
1 - the purpose and character of the use, including whether such use is of a
commercial nature or is for nonprofit educational purposes;
2 - the nature of the copyrighted work;
3 - the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
4 - the effect of the use upon the potential market for or value of the copyrighted work.
The fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made
upon consideration of all the above factors
♦
PDF Cover Page
♦
♦
Verso Filler Page
♦
AMPLIFIERS
THE WHY AND HOW OF
GOOD AMPLIFICATION
By
G. A. BRIGGS
Author of" LOUDSPEAKERS; THE WHY AND
How OF GOOD REPRODUCTION"
"SOUND REPRODUCTION"
"PIANOS, PIANISTS AND SONICS"
and
H.
H.
GARNER
UNTIL RECENTLY TECHNICAL OFFICER TO THE ESSEX EDUCATION COMMITTEE
........but hearing ofcencimes
The still, sad music of humanity,
Nor harsh nor grating, though of ample power
To chasten and subdue.
Wordsworth.
Published by W H A R F E D ALE W I R E L E S S W 0 R K S
BRADFORD ROAD · IDLE ' BRADFORD 'YO:RKS
FIRST EDITION
MARCH
1 952
Copyright
Registered at Stationers' Hall
/vbde and Printed in England by Tapp & Tuothill Ltd. L ee ds , London and Johannc:iburg
CONTENT S
Page
Acknowledgme nts, Abbreviatio ns and Symbols
6
I ntroduction . .
7
Foreword
8
Chapter
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
9
I
Amplifier Quality
2
Valve Theory
18
3
Va lves, Voltage and Amplification ..
26
4
The Valve as a Power Amplifier
41
5
Deco upling and Instability
48
6
Push-pull Amplificatio n
57
7
Negative Feedback . .
63
8
The Cathode Follower
85
9
Phase Splitters
91
10
To ne Compensatio n .
.
106
II
Pick-up Input Circuits
123
12
Whistle and Scratch Fi lters
134
13
M icrophones and Mixing Circuits .
14 Power Supplies
.
138
152
15
Hum and Noise in Amplifiers
171
16
Measurement of Distortio n .
178
17
Ga mer Amplifier
180
.
18 A Few Questio ns Answered
Co nclusio n
193
197
Supplement: Useful Formula:, Db tables , Reactance tables ,
Transformer ratios, Lo udspeaker Watts
198
I ndex . .
203
Gamer Amplifier Circuits . .
5
ACKNOWLEDGMENTS
Our thanks are due to the Wireless World and other publications for
permission to reproduce diagrams as acknowledged under various
Figures. If as a result of two-fold authorship an acknowledgment has
been inadvertently overlooked, we offer our apologies here and now
to the authors and publishers concerned.
ABBREVIATIONS AND SYMBOLS
�Capacitance or Capacity in Farads (Units).
Cycles per second.
Delta Small change.
il
=Electro-motive force-Units Volts.
EMF
E or V
Voltage-Units Volts.
�Current-Units Amperes.
I or i
Kilohms, i.e. ohms X 1,000, or ohms x IOj•
K or Kn
� Kilocycles per second, i.e. cycles X 1,000.
Kc!s
= Inductance-Units Henries.
L
=Mil1i (one-thousandth).
m
Micro (one-millionth).
mu or fl.
C
c/s
�
c_
=
=
�
=
of an ampere or I amp
1
__
1,000
mA
=
Milliamps, i.e.
mH
=
Millihenries (Henries
mV
1
x
x
10-3.
)
10-' .
=Millivolts, i.e. __ of a volt or
1,000
I
volt
x
10-3•
M or M n
Megohms, i.e. ohms x 1,000,000 or ohms x ro6•
Mc!s
=Megacycles per second, i.e. cycles x 1,000,000.
�
I
.
°f an amp or
M·lcroamps, I.e.
fLA
1,000,000
fLF or Mfd�Microfarads (Farads x 10-6) .
=Microhenries (Henries x 10-6) .
lLH
,
=
fLY
n
pF
R or r
RMS
Z
w
=
Microvolts, i.e.
I
1,000,000
of a volt or
�Ohm.
Picofarads (Farads x 10-12) .
=Resistance (Units Ohms).
=Root mean square.
= Impedance.
21tf.
h
RL
gm
amp
volt
x
x
10-6 .
10-6•
�
=
AB B REVIATIONS applied
fL
a
k
f
g
I
I
to
VALVE C IRCUITS
�Amplification factor of a valve.
�anode, e.g. ra=Internal resistance, Ra
= cathode, e.g. Rk=Cathode resistance.
�filament, e.g. Vf=Filament volts.
grid, e.g. Vg=Grid voltage.
g l Control Grid.
g, Screen Grid.
g3 Suppressor Grid.
heater, e.g. Vb Heater volts.
= Load resistance.
=Mutual conductance of a valve.
=
=
6
=
Anode resistance.
INTRODUCTION
When the second edition of Sound Reproduction was published, I
sent a copy to Major Garner with the usual gesture of generosity, but
in reality fishing for the odd word of praise. He replied in appropriate
vein, but added a word of regret that the field of Amplification had not
been covered and suggested that there was a gap which might with
advantage be filled. I replied that in my opinion he was just the man
for the job, and we finally came to the conclusion that we should have
to put our heads together and produce a joint effort. The result of our
work now makes its appearance.
I naturally do not mind admitting that my colleague has done all the
donkey work, but I have also had to work very hard in order to preserve
some continuity of style-whether good or bad-throughout the book.
As usual, the main problem has been what to leave out. I have been
appalled at the vastness of the subject; an enormous amount of th::
original copy and scores of diagrams have been j ettisoned, much to the
chagrin of their hard-worked producers, in order to keep the volume
down to a reasonable size and price.
I must acknowledge help from many directions. This is the fourth
time that Mr. F. Beaumont, Chief Engineer of Ambassador Radio,
Brighouse, has acted as my technical sub-editor; as he has succeeded
in keeping me out of serious trouble, he must be good. Major Garner
joins me in acknowledging the immense value of the corrections and
suggestions which he has contributed to the present work. It is also
the fourth occasion on which my old friend F. Keir Dawson has
designed the cover and done the drawings. In this case, some of the
diagrams have involved the burning of much midnight oil. I think all
readers will agree on their general excellence-particularly for an
amateur.
Mr. E. M. Price, M.Sc.Tech., has again helped in technical and
oscillographic tests, and for final commendation there is my Secretary,
Miss E. Isles, who not only transcribes illegible manuscripts but
corrects mistakes in grammar at the same time.
G. A. BRIGGS
7
FOREWORD
This book is intended to supply the missing link between the previous
volumes, Loudspeakers and Sound Reproduction, which described in
considerable detail the equipment used before and after the process of
amplification. It should furnish ideas for the experimenter so that he
may obtain the best result from the amplifier and associated circuits,
with his available resources. For those who are blessed with the
necessary cash, there are several well-known designs for home con­
struction of amplifiers which give superb results, and commercial
amplifiers are now available which are a delight to use and hear. Such
readers may well be able to assess and appreciate their equipment all
the better after a perusal of this book. Many of us (the writer included)
have to think twice before we buy what we want, and if we achieve
good results by judicious and economic buying, we are all the more
pleased, and value our masterpiece so much more highly. It is hoped,
therefore, that the man who "makes one of his own", who is not
content to put in 40K and 0'5 mfd just because it says so in the book
but wants to know why he does it, will find most of the answers in
the following pages.
No doubt this book will be considered by some to insult their
intelligence by explaining a circuit in too great detail in words with too
few syllables, whereas others may complain that there are too many
technicalities; yet it is felt that a good foundation and appreciation of
basic principles is necessary to build the finished article. Professional
engineers may boggle at some of our short cuts to arrive at a general
conclusion, but we plead that the means justify the end-that of giving
some guidance and pleasure to the many amateurs striving to build
or buy that elusive thing, the perfect reproducer, and to obtain there­
from the maximum performance.
H. H.
8
GARNER
CHAPTER I
AMPLIFIER
QUALITY
The ideal audio amplifying system is, by definition, a device which
will give an increased output relative to the input supplied to its
terminals, without any additions to or subtractions from the nature of
that input.
The first consideration is the avoidance of distottion. Distottion can
be classified under a number of headings, and these will be discussed
separately as follows:
I.
HARMONIC DISTORTION.
in the original.)
(The production
. of harmonics not present
2. FREQUENCY DISTORTION. (Unequal amplification of a frequency or
band of frequencies.)
3.
4.
PHAS E
(Phase angle changes with frequency.)
DISTORTION.
(Acoustic unbalance due to reproduction at a
sound intensity differing from that of the original.)
SCALE DISTORTION.
5. DISTORTION OF TRANSIENTS.
(Sudden, sharp sounds becoming
muddy and blurred.)
6. INTERMODULATlON.
(One frequency interacting with another.)
7. SPURIOUS combination tones.
8.
MAINS HUM, CIRCUIT
9. INSTABILITY.
and
VALVE NOISE.
(The production of spurious oscillations, continu­
ously or intermittently.)
1.
HARMONIC DI STORTION
A harmonic, sometimes called an ovettone, is a tone at a frequency
which is twice, three times, four or more times that of the original
frequency or fundamental. Whilst all musical instruments produce
harmonics, an amplifier is capable of introducing harmonics on its own
account, and these naturally modify the original tone quality, and may
make a Strad violin sound like a one-string fiddle. The critical faculty
of the human ear to distinguish harmonic distottion varies, and the only
test of high fidelity reproduction is to compare the reproduced with
the original sound at the same time, as the " memory" of the ear is
most capricious. This is rarely possible, but if a reproducer is to please
it must not cause the listener to flinch when a violin is bowed in the
9
AMPLIFIER QUALITY
upper register. The "wiry" type of reproduction in the �o-\;alled
wide range reproducer can be attributed to :
Ca) Undue amplification of the high tones.
Cb) The presence of spurious harmonics.
Harmonic distortion is generally worse when a reproducer is working
at a level approaching its full output, and whilst the average output
power required for ordinary domestic listening is not more than 0 · 5
watt, this usually sounds better from a so-called 10 watt amplifier than
from a 3 watt version of the same standard.
Triodes show a preference for producing even harmonics, pentodes,
odd harmonics, but any non-linearity in operation, using any valve,
can produce combination tones.
As commercial amplifiers are now produced with a total harmonic
content of not more than 0·1 per cent., the bogey of distortion from this
cause may be said to have been frightened off, and I think we should
pay tribute to H. J. Leak for his pioneer work in this particular form
of ghost-laying. Pre-war figures accepting limits of 5 per cent. 2nd
harmonic and 2! per cent. 3rd harmonic in high quality amplifiers
now appear to be absurd.
The idea that spurious even harmonics are more acceptable than odd
ones seems to have arisen because the 7th produces a strong discord
with the fundamental, and the 9th is even harsher. This can easily be
tested on the nearest piano by playing the overtones of the second C
from the bass end, as follows:
6
I
7
2
3
8
4
5
Harmonic
._- ..
Note
True Harmonic Cc/s)
Tempered Scale Cc/s)
Harmonic
Note
True Harmonic Cc/s)
Tempered Scale Cc/s)
C
C
C
E
G
G
65· 4 13°·8 196 261 327 392
65"4 13°·8 196 261 33° 392
9
10
11
E
F�
D
588 654 719
587 659 740
12
13
14
A
G
B�
785 850 916
784 880 932
B�
458
466
C
523
523
15
16
B
C
981 1,046
987 1,046
It will be observed that the first really discordant even harmonic is
the 14th, but above the 16th the harmonics crowd more and more
closely together, and there is little to choose between them. We can
conclude, therefore, that all spurious harmonics are poisonous and
affect the timbre of sound, but the even harmonics are a rather slower
poison than the odd ones, and the higher they go the worse are the
results.
While on the subject, it is interesting to note that the oscillogram
of bottom C of a good piano shows traces of 30th harmonic, as in
Fig. 1/1.
10
AMPLIFIER QUALITY
FIG. 1/1.
Oscillogram of
piano tone at C 32'7 eis,
s h o w i n g t r a c e s of 3 0th
harmonic.
-
It is obvious that any harmonic distortion in valves would affect the
tone colour of such a note, even if actual harshness were not produced.
2.
FREQUENCY DISTORTION
This needs little explanation beyond saying that all frequencies should
be amplified so that their relative intensities are as near the original
as possible. The production of a continuous boom of bass, so beloved
of receiver manufacturers in the days of early moving-coil loudspeakers,
is not to be tolerated, whilst undue exaggeration of the sibilants of
speech due to high frequency over-emphasis is equally distressing.
3.
PHASE DISTORTION
Figure 1/2 shows two voltages in phase, the same two voltages
displaced through a phase angle of 180° and also out of phase by 90°.
Phase displacement can of course be of any angle, but over 180° the
phase displacement is becoming less, until at 360° you are back exactly
where you started.
FIG. I /2.-Diagrams to show
voltages in and out of phase,
with combined result in each
case.
Combin�d re5utt
�Rz:;,
c· Two volr"S'Cl$
))h<lSQ
by
O�: r
90°
of
�
Co .... bin'd
l"<lJi ..
1t
The audible effect of phase shift does not appear to be objectionable,
but, contrary to general belief, it is of great importance up to the speaker
diaphragm, from a transient point of view. It is now well known that
II
AMPLIFIER QUALITY
the tone quality of sounds is determined by attack and decay times, as
well as by harmonic content. To illustrate this point, we cannot do
better than examine the wave envelope of a typical piano note played
staccato, if the reader will forgive another reference to this instrument
in the first chapter.
FIG. 1/3.-Wave envelope of
piano tone, C 523 c/s as pro­
duced under normal domestic
room conditions.
From " Pianos, Pianists and
Sonics ", G. A. Briggs
The steep wavefronts at the start of the note are clearly shown;
the fairly exponential decay pattern lasting more than half a second is
partly due to the reverberation characteristics of the room. It is an
axiom that the steeper the pulse wavefront, the less the phase shift
must be. This gives great point to securing good transients, which in
turn demands both wide frequency response and a small and linear
phase shift over the given frequency bandwidth.
Incidentally, there is an explanation here of why conventional bass
and treble boost circuits often sound so muddy; it may also explain
to some extent why excessive pre-emphasis of high frequencies in
recording has such an unfortunate effect on the final tone quality.
In order to remove any impression that the amplifier is the only
black sheep in the phase family, let us admit here that a form of phase
distortion may be discerned when two loudspeakers in phase are used
without a crossover network, and the listener is not positioned equi­
distant from the speakers. Sound waves having to travel different
distances may arrive out of phase and cause varying degrees of can­
cellation of certain frequencies. The analogy could of course be
applied with equal force to many concert halls, where the acoustics
are so bad that cancellations from phase effects in certain positions
result in the unfortunate concert-goer hearing only about half of the
performance. (Such seats should be sold at half-price.)
4. SCALE DISTORTION
The ear is a most deceptive organ, and if the volume level as heard
from the loudspeaker is greater or less than the original (as normally
heard), then the ear "hears" certain frequencies predominating.
.
For example, at low level the double bass In an orchestra may be practi­
cally inaudible, yet at high level it could appear to be over-emphasised
12
AMPLIFIER QUALITY
due to no fault of the reproducing system. This is readily overcome by
the use of suitable compensation, and for domestic listening at the low
level which is nearly always necessary, it is desirable to incorporate bass
lift, and possibly some increase in the extreme "top" above 5,000 c/s.
5.
DISTORTION OF TRANSIENTS
Sudden, sharp noises of a percussion type may suffer a hangover
effect that reduces their clarity and realism. Figure 1/4 shows the
idealised waveform of a transient.
F I G . I / 4 .-S quare W a v e .
Note that the changes are
extremely rapid, as shown by
the vertical edges of the pulse.
u
A square wave is actually a combination of a large number of sine
waves of different frequencies, including the highest.
The next diagram, Fig. 1/5, shows an analysis of a square wave up
to the 15th harmonic; but actually the odd and even harmonics
extend up to infinite frequency. A first-class amplifier should deal with
a square wave without distortion; restriction of the HF response will
h
.,
I
I
I
I
I
I
+
o
FIG. 1/5.
Analysis
of square
wave
into its
harmonic
compon­
ents.
Sum of Harmonics
1 to 15
From ((Principles
of Television
Reception,••
A. W. Keen
S£r lsaac Pirman
& Sons, Ltd.
t
3
T
4
13
I
2
AMPLIFIER QUALITY
show up, but this would hardly be classed as distortion where it
operates above 20 Kc/s.
As a perfect square wave consists of an infinite number of pure sine
waves in the harmonic series, it will be clear that any distortion of
these pure sine waves will affect the shape of the square wave. Figure
1/6 shows two forms of distortion which may be produced.
FIG. I/6.-Distorted Tran­
sient Waveforms. The round­
ing of corners denotes poor
H F response.
It is interesting to note at this point that loudspeakers are usually
classed as the weakest link in the chain of distortionless reproduction
(especially by amplifier makers !). It must be admitted that the strange
effects which many loudspeakers produce on square wave input lend
some support to this sad criticism. The following diagram, Fig. 1/7,
c
B
A
D
F
E
G
FIG. 1/7.-Transient effects produced on square waves by various types
of loudspeaker.
A. Output from amplifier.
B. Output from 8-in. speaker 8,000 lines magnet, corrugated cone.
C. Improved result from 8-in. speaker with 13,000 lines magnet and
cloth suspension.
D. Io-in. speaker, 13,000 lines magnet and heavy one-piece corrugated
cone.
E. Improved result from Io-in. speaker 14,000 lines, heavy one-piece
cone with cloth suspension.
F. Io-in. speaker with rubber compliance in speech coil.
G. Io-in. speaker with jointed two-piece cone.
shows the effect of stiff suspension, cone resonances, inadequate
air-loading, etc., as seen via microphone and oscilloscope.
It is probably very fortunate that the loudspeaker is at the end of the
chain, and its output is projected into the air, where its performance
is camouflaged by a multitude of room effects and then accepted by
the wide tolerance of the average ear.
14
AMPLIFIER QUALITY
To return to the amplifier, satisfactory transient reproduction
obviously calls for a very wide frequency range. The subject is further
dealt with under the discussion of loudspeaker damping.
6.
INTERMODULATION
This effect may be observed whilst listening to a violin with an organ
accompaniment, the amplitude of the higher frequency sound varying
in accordance with the powerful low frequencies of the organ. The
effect may be overcome by separating these frequencies in a dual or
even treble loudspeaker system, if the intermodulation arises in the
speaker.
It is, however, important to remember that in the case in question
the distortion would only arise-even in a single speaker-when the
powerful low notes were actually producing non-linearity. This is
illustrated in Fig. 1 !7A, where we have oscillograms of the output of
loudspeakers fed from two separate tone sources, as follows:
a.
c.
b.
FIG. I /7A.-Oscillograms to illustrate intermodulation effects produced in
loudspeakers.
(a) Frequencies 82 and 900 cjs produced in one I s -in. speaker without
intermodulation.
(b) Non-linear 62 c/s note from IO-in. unit, with pure tone at 2,500 c/s
from 8-in. unit. Reasonably free from intermodulation.
(c) Treble note of Cb) transferred to Io-in. speaker now suffers severe
intermodulation from the non-linear 62 c/s note. This is indicated by the
difference in outline between top and bottom of trace, and also by general
shading effects.
The important point is that the intermodulation in a single speaker
is caused by non-linearity or overloading.
The same sort of thing can happen in pick-ups and amplifiers.
It is quite common practice, especially in America, to measure the
intermodulation product of amplifiers, divide by 4, and take the
answer as the distortion factor. The following tests with the EMI
Intermodul ation Test record JH.138, with 60 c!s, and 2,000 c!s at
1 2 db lower level, and the Garner amplifier described at the end of
this book, may throw some light on the problem. Fig. I!7B shows
IS
AMPLIFIER QUALITY
output from pre-amp and main amplifier with 15 ohms resistive load,
and the effect of deliberately overl oading the output and i ncreasing
distortion by removing N FB.
a.
b.
c.
FIG. I /7B.-Intermodulation test using EMI Record JH.I38.
(a) Output from Garner amplifier at 5 watts with full NFB, linearity is quite
good.
(b) Output increased to IS watts by reducing NFB. Note onset of non­
linearity.
(c) Output increased to 20 watts with excessive top lift. Severe inter­
modulation.
The performance at Ca) is quite good. Overloading at Cc) produces
distortion and intermodulation with a shape that looks like the work
of one of our modern sculptors, but testing for distortion by inter­
modulation always strikes the writer as equivalent to testing a man's
strength by seeing how much force is required to knock him down.
As intermodulation is a product of distortion or non-linearity,
surface noise from a record is intensified by intermodulation products
which may be produced by non-linearity in a pick-up, amplifier or
loudspeaker. It is claimed that surface noise is lower when a crystal
pick-up with absolutely linear movement is used, compared with a
type of pick-up which has some degree of non-linear output. The
reason is, of course, that the sound waves of music are linear with
overtones in the harmonic series, whereas surface noise from a record
is absolutely non-linear, which in the musical sense would be severe
distortion.
7. SPURIOUS COMBINATION TONES
This effect may be observed when two frequencies are fed into a
non-linear system, with the result that in the output there appear,
besides the original frequencies, various sum and difference combina­
tions of the originals, their harmonics and their combinations. The
number and strength of these combination tones increase as the har­
monic distortion percentage increases, and also increase as the order
of the harmonic increases. It is largely for this reason that insistence
16
AMPLIFIER QUALITY
is made on the necessity of avoiding any harmonic distortion
like the plague, particularly the higher order odd harmonics, 3rd, 5th
and 7th, the 4th and 6th usually being negligibly small.
8.
HUM, CIRCUIT AND VALVE N OISE
No amplifier can be considered as high fidelity if there is a constant
buzz or hiss present in quiet intervals. In the designs discussed in
this volume every hint will be given that may assist in a reduction of
self-generated noise.
9. INSTABILITY
This may take the familiar form of "motor-boating", when a plop­
ping noise is heard in the loudspeaker. The more insidious types
are of:
(a) Oscillations that are continuous, but above audibility or even of
radio frequency; and
(b) Oscillations that only take place when triggered off by a sudden
transient.
Figure 1/8 illustrates the effect of this type of oscillation on a square
wave.
FIG. r/8.-Effect of oscilla­
tions set-up in amplifier by a
transient.
These insidious faults are to be suspected in many home-built ampli­
fiers, particularly in those using negative feedback, and as such will
be discussed later. It is very necessary to check for condition (a)
with an oscilloscope, and for condition (b) with the oscilloscope and
a square wave generator.
*
*
*
*
It is not suggested that there is anything new in the foregoing
summary of the qualities affecting the design and construction of
good amplifiers. The list outlines most of the problems which confront
the home constructor; but it is comforting to know that it is now
possible to buy first-class amplifiers from reputable makers, which
have been competently designed, built and tested.
17
CHAPTER 2
VALVE
THEORY
As this chapter is inserted simply for the benefit of the amateur who
is new to the subject, or for the older hand who may have forgotten
all the theory he ever knew, let us begin by including a list of the
symbols which will be encountered in its perusal.
E or V
Volts.
I
Current, in amps.
R
Resistance.
mA
Milli-amps.
Ea or Va
Anode voltage.
Eg or Vg
Grid voltage.
la
Anode current.
Grid current.
Ig
Anode resistance.
ra
i)
Small change (not necessarily in copper!)
(Pronounced delta).
Vg
Grid/Cathode voltage.
gm
Mutual conductance.
mu, meaning amplication factor.
fL
The inclusion of this list makes it quite clear that the chapter is not
intended for the expert, unless of course he wishes to read it in order
to exercise his critical faculties.
=
=
=
=
=
=
=
=
=
=
=
=
=
*
*
*
*
An artist can make a good picture without studying the chemistry
of his paints; in the same way, a knowledge of the physics of the
Thermionic valve is not altogether essential to the investigator primarily
concerned with the uses to which it may be put. He can examine,
mathematically as well as electrically, the behaviour of a host of circuits,
knowing not why the valve works, but only how it works. In this book
our more practical concern is with the outward applications of valves
rather than their inward design; nevertheless, as an appropriate
introduction to the study of electronic circuitry, a brief survey of the
theory of valve action is thought to be desirable.
1.
THERMIONIC EMI S SION
It is generally accepted that, if matter is pulverised, broken up and
sub-divided into tiny pieces, a point will be reached at which any
18
VALVE THEORY
further sub-division will destroy the very nature and properties of
the substance itself. When this point is reached, we are left with a
molecule of the substance. Further investigation of a molecule would
reveal that it is comprised of atoms, a water molecule being made up
of two atoms of hydrogen and one atom of oxygen, or common salt
comprising one atom of sodium (a metal) and one atom of chlorine
(a gas). The molecules or atoms in a body are not tightly packed
together, even in the hardest and most dense substances, and it is
perfectly true to say that" iron has pores "-it has, in its inter-atomic
spaces. The atom is the smallest part of an element capable of taking
part in a chemical reaction. Investigation into the nature of an atom
would reveal that it comprises a solar system in miniature. A central
core of a fixed quantity of negative and positive charges, together with
neutrons and probably other particles, is surrounded by orbital electrons
revolving around the nucleus. In certain substances which are known
as conductors, some of the electrons in the outer orbits are but loosely
attached to their particular solar system, and there is a more or less
constant flitting of these electrons across the wide inter-atomic spaces
at normal temperatures.
In the absence of an external electric field the velocities of the
free electrons have random distribution. No electric displacement
occurs in any fixed direction; but if a difference of potential is main­
tained between two points in a conductor or a semi-conductor, a
steady drift of electrons results, flowing from the region of lower
towards the region of higher potential. It should be noted here that
a battery, in spite of having a plus sign on its positive terminal, is
actually deficient in electrons at the positive pole or plate. Therefore,
in the electron sense, the flow is from the negative to the positive
pole. This is very confusing when studying the old and classical
text-books on electricity and magnetism which insist that current
flows from plus to minus. One could note at this point that an
insulator is a substance that has no free electrons which can be
readily displaced.
From the foregoing summary it will be seen that the flow of electric
current is but the movement or scurrying of electrons within the con­
ductor. If the rate of flow is greatly accelerated, there is such intense
activity within the body of the conductor and so many collisions occur
between electrons, that heat is generated to the point of incandescence
and ultimately volatilisation. At the point of incandescence electrons
actually leave the surface of the conductor. If a thin filament of a
refractory metal like tungsten is enclosed in an evacuated glass envelope,
together with a plate or anode, as in Fig. 2/1, and the plate is given a
positive charge relative to the filament, this plate will collect some of the
electrons escaping from the surface of the filament and the current will
flow in the external circuit back to the filament as indicated by the
current measuring instrument.
19
VALVE THEORY
METER
: H.T
: BATTERY
F I G . 2 / I .-Simple Diode
Valve Circuit.
+- D irection of Electron flow.
.
FILAMENT
HEATING
BATTERY
This emission of electrons is known as Thermionic Emission, and
is the basic principle underlying the construction of practically all
modem valves, the invention being due to an Englishman b y the name
of Fleming. The word " valve" is a mechanical term which implies
that gases or liquids can only flow through it in one direction, just as
the valve of a cycle tyre will allow air to pass into the tube but not out
again. The simple circuit just described can be justly termed " valve"
because it is only when the anode is positively charged by the external
battery that electrons will be attracted to it. If the anode is charged
negatively with respect to the filament, electrons will be repelled and
thus no current will flow in the external circuit. The device outlined
is termed a " diode". The filament, or more accurately the cathode,
could well be, and was in early types of valves, a thin tungsten wire,
but tungsten is quite a poor emitter, and the fact that it will only emit
when its temperature is brought up to white heat is a dis advantage in
two ways. It is wasteful of battery power, and the heat rapidly causes
the metal to become crystalline and so extremely fragile.
FILAMENT COATING
Methods were sought to overcome these disadvantages and it was
found that the addition of a very small quantity of the metal thorium
greatly improved the thermionic emission efficiency and such filaments
only required to be heated to yellow heat. These valves were then
styled dull emitters as distinct from those with filaments operating at
incandescence and styled bright emitters.
Further investigation to improve the efficiency revealed that coating
a filament with a mixture of strontium and barium oxides produced
higher emission at an even lower temperature, and these are the types
commonly used in the modem valve.
MAINS VALVES
The demand for " all mains" sets rapidly called for a valve whose
cathode could be heated by using a low voltage AC source readily
20
vALVE THEORY
obtained from a step-down transformer. However, the 50 cycle
current caused the emission to be varied at 100 times a second when the
AC current was at its minimum, due to the low thermal capacity of
the fli ament. This defect was overcome by making the cathode a nickel
cylinder coated with emitting material. The whole system was heated
by a reasonably high wattage heater wire, the economy of current not
being so important as when storage batteries were used. This type of
cathode is known as the indirectly heated cathode and serves exactly
the same function as the thin filament. Fig. 2/2 shows the schematic
for an indirectly heated valve.
.-ANODE
/
FIG. 2/2.-Schematic of in­
directly heated Diode Valve.
CATHODE
HEATER
Typical directly heated valves consume 50 mA at 1'4 volts, 50 mA
at 2 volts and a few legacies I amp at 4 volts. Typical indirectly heated
types operate with 4 volts '65 to I amp, 6'3 volts '3 to '9 amps, 12 volts
. 15 amps, etc.
THE DIODE
Referring again to Fig. 2/1, if the voltage applied between anode
and cathode is steadily increased and the current readings are plotted
on a graph, a curve will be produced as shown in Fig. 2/3. It will
la
I:z:
w
O!
cL
::J
V
w
Cl
0
:z
<I:
FIG. 2/3.-Typical voltage!
current curve of Diode Valve.
ANODE VOLTAGE
Ea
be seen that at low anode voltages the current increases but slowly.
Then, as the anode is made more positive, the electron current grows
rapidly, then less rapidly, and finally remains constant. Any further
21
VALVE THEORY
increase in anode voltage produces practically no corresponding
increase in current. At this point the electron current is said to be
saturated. The actual value of current at which saturation takes place
depends on the temperature of the cathode (for a given emitting
material) and the electron capacity of the cathode surface. An y attempt
at increasing the value of anode current beyond saturation by applying
an excessively high anode voltage will most likely result in the destruc­
tion of the cathode. The initial knee on the curve is due to the electric
field which surrounds the cathode, and contains a large quantity of
negative electrons called a " space charge". Sufficient force must be
applied to overcome this tendency to form a traffic jam at and near
the cathode surface.
The resistance offered to the flow of the anode current is commonly
known as the anode impedance, or internal resistance of the valve, and
this depends upon the anode voltage and the relative geometrical
dispositions of the anode and cathode. From the curve in Fig. 2/3,
it will be seen that the relationship is never quite linear, although for
the higher anode voltages it becomes very nearly so. We cannot,
therefore, apply Ohm's law to arrive at the internal resistance of a
valve; this can only be determined by taking the ratio of a very small
change in anode voltage to the resulting very small change in anode
current. This is expressed mathematically as cma, a (pronounced delta)
ala
being the mathematical term for a very small change. Ea of course
anode current.
means anode voltage and la
=
I.
(mAl
..
OA90
•
FIG. 2/4.-Anode Characteris­
.
2
o
5
10
15
20
25
Vg (V)
22
,
tic of Mullard DA90 Diode
Valve.
VALVE THEORY
Reference to Fig. 2/4 for a Mullard DA90 valve having a 1'4 volt
heater consuming 0 '15 ampere will show that for a change from 10 to
15 volts on the anode a current change from 2 to 3'2 mA takes place.
5
Therefore, the anode impedance of this valve would be
3'2 - 2mA
5 x 1,000
ohms (expressed in basic units of volts and amperes) = 4,166
1'2
ohms.
3. THE TRIODE
If a wire mesh grid is interposed between cathode and anode as
shown schematically in Fig. 2/5, we have what is called a triode valve.
F I G. 2 / s .- S che m a t i c o f
GRID
Triode Valve.
CATHODE
This fine mesh grid is in the electron path between cathode and
anode, and if it is given an increasingly negative charge the electrons
will be progressively repelled until no more electrons can reach the
anode, this point being known as cut-off. Fig. 2/6 shows the action
so
EC52
I.
40
30
20
-10
-5
o
23
FIG. 2/6.-Grid Characteris­
tic of Mullard ECS2 Triode.
VALVE THEORY
graphically in the characteristic curve of a Mullard type EC52 valve.
It will be seen that with a constant anode voltage of 300 volts, the
application of 7 volts negative to the grid will completely cut off the
flow of anode current. Even with the anode voltage increased to 400
volts, the grid is capable of cutting off anode current when made 10
volts negative. Fig. 2/7 shows a family of anode characteristics for
various values of grid volts, with a shape very similar to that of the
diode, except that these curves are not taken up to the point of satura­
tion, otherwise the valve would be damaged.
H
0
'3
�
0
w
,.
f-'
f-
ECC33
-t
TIfL
Vh = 6'3V
..:..
,p
..:..C;; �
.I ,
it ,�
)
..,
o�
I\)
,j'�
"'.
0
t,J
103
.-i [-.
� I-
,
f-'�
Iif';O..:."j.. _
::"
f- f\
�
,
0
0
,
0
1::>0
400
300
200
FIG. 2/7.-Anode Characteristic of Mullard ECC33.
�
,0,
f-
va(vT
A dotted line indicates where the wattage dissipated at the anode as
heat imposes a limit to the anode current.
The anode resistance of this triode valve with zero �rid volts
(Vg
0), calculated exactly as in the case of the diode, is as follows:
8Ea
150 - 100
50 X 1,000 (ohms)
-'
- 9,000 0hms.
ala
14'5 - 9
5'5
From Fig. 2/7, which is known as the anode charag:eristic of the
valve, it can be seen that with the anode voltage at 200 volts and the
grid at minus 2 volts, 1 1 mA of anode current will flow. Increasing
the grid voltage to - 3 volts reduces the anode current to 7'25 mA.
This relationship shows that for I volt change on the grid there would
be 3'75 mA change in anode current. This is known as the mutual
=
_
_
_
24
VALVE THEORY
conductance and is expressed in mA/V. In this case the valve has a
mutual conductance of 3'75 mA/V.
One further relationship can be established from the family of curves
shown in Fig. 2/7. With the grid at -2 volts and anode volts at 190,
the anode current is 1 0 mA. Increasing the grid voltage (Eg) to -3
volts with the anode still 190, the anode current falls to 6'3 mA. To
restore the anode current to IQ mA the anode volts would have to be
increased to 225 volts . Thus a I volt change on the grid is equal to
changing the anode volts by 35. This measure of the effectiveness of
the grid compared with the anode in controlling anode current is the
SEa 35 giving an amamplification factor of the valve, in this. case
SEg
plification factor of 35.
=
1
SUMMARY
(l
= fJ. = Ea
SEg
SEa
Anode Resistance
=
(lIa
SI a
Mutual Conductance = gm =
SEg
Amplification
ra =
There is a fixed relationship expressed as
.
(lEa
SEa SIa
SEg = aEg X SIa
In the valve quoted (Fig. 2/7), fJ. = 35, gm 3'75 mA/V and r. = 9,000
fL =
gm
X
ra I.e.
ohms. Checking, 35 should equal
3' 75
1 ,000
X
9,000, but actually equals
33'75 which is within reasonable limits of error in reading the graphs.
CHAPTER 3
VALVES,
VOLTAGE
AND
AMPLIFICATION
In the following chapter, an attempt is made to give an outline of the
various methods of coupling and biassing different types of valve,
without introducing technicalities which would be beyond the scope of
the average amateur. Such chapters are always extremely difficult to
write, as protecting readers from mental strain is liable to give the
writers a few headaches. The subject is a dry one and covers a wide
field; to make it entertaining would be almost as difficult as producing
music from a slide rule. It is hoped that the result will enable the
amateur to improve his general conception of valve practice, with
consequent better understanding of amplifier " why and how".
HT+
R
FIG. 3/1.-The Triode as an
OUTPUT
V
amplifier.
VOl.TS
k-----��----�--HT-
THE TRIODE AS AN AMP LIFIER
Fig. 3/1 shows a triode with a resistance interposed between anode
and the anode battery (usually referred to as the high tension battery).
Between grid and cathode is shown an AC generator Eg• From the
grid characteristic in the previous chapter, it will be realised that, as
the grid is made increasingly negative, anode current will fall. However,
when the grid is made increasingly positive anode current will increase
because the positive grid, instead of limiting the flow of electrons from
cathode to anode, will now assist and accelerate the process. These
variations in anode current have to flow through the resistance R,
and by Ohm's law a voltage will be developed across this resistance in
sympathy with the grid fluctuations. If a voltmeter is placed between
anode and cathode it will show a rise in voltage when the grid is going
increasingly negative (less anode current flowing, less volts dropped
across R), but when the grid is going increasingly positive, the volt26
VALVES, VOLTAGE AND AMPLIFICATION
meter reading will fall due to the increase of anode current causing a
bigger drop along resistance R. From this emerges the fact that when
the grid is going increasingly negative the anode is going increasingly
positive, thus showing that the grid and anode are in effect 1800 out
of phase. Given a suitable value of R, the voltage fluctuations as read
between anode and cathode will be greater than the voltage fluctuations
between grid and cathode. Thus our valve is acting as an amplifier.
D ISTORTION
If the grid is made positive with respect to cathode, it will actually
behave like a second anode and will collect electrons unto itself ; not
all of them, of course, due to the wide spacing of the grid mesh. The
source of AC voltage applied to the grid has internal resistance. The
FLOW OF GRID
CURRENT ELECTRONS
\
.--:-oE-�.- +
..----- HT +
FIG.
3/2.-Effect of Grid
Current. Opposition voltage
developed across internal
resistance of input voltage
source.
- - - -
Eg
GENERA OR
L-------�--HT­
OPPOSITION
INTERNAL
VOLTAGE
ACROSS
RESISTANCE
electrons collected by the grid will flow through this resistance back
to the cathode, but by Ohm's law this will produce a voltage across the
internal resistance of the generator and this voltage will be in opposition
to that of the generator itself. The net result will be that, on the positive
la
EXPECTED
AMPLITUDE
FIG. 3/3.-Distortion due to
Grid Current.
AMPLITUDE
OBTAINED
Vs
27
INPUT VOLTS
VALVES, VOLTAGE AND AMPLIFICATION
half cycle of input the actual input waveform will be distorted and there­
fore the output at the anode will not be a faithful replica of the input.
This state of affairs arises from the grid being allowed to go positive.
If, however, a small battery is introduced as in Fig. 3/4, and this
battery has such a voltage that the positive swing of the generator can
never drive the grid positive, then grid current cannot flow and dis­
tortionless amplification should result.
la.
.------ H T t
LOAD RL
_ AN O D E
PO INT
I:g
Vg
I
I
I
INPUT
FlG.
BIAS BATTERY
�I��----- H T -
3/4.-Grid Bias, to avoid grid current.
Another factor that must be considered is that to obtain distortionless
output, the grid voltages must only be swung over the straight or
sensibly straight part of the grid characteristic. This means that the
battery voltage should be so chosen that the valve works either side of
the mid point of the straight part of the grid characteristic. This
battery is known as the bias battery;to give grid bias. The resistance R
is called the anode load, and is referred to as " RL ".
CHOICE OF VALUE FOR ANODE LOAD
From the foregoing description of the valve as an amplifier, it would
seem that the bigger the value of load resistance employed, the better,
so that very large voltage changes occur across it. However, even when
the valve is not amplifying, a certain standing value of anode current
is flowing and this will cause a voltage drop across the load. It follows
that, with a large resistance, a very high value of high tension is required
if the valve is to be worked at a high tension voltage which will give a
sensible length of grid characteristic. It can be shown that the gain
obtained from a valve equals the amplification factor :
X -
load resistance
--------
load resistance + anode resistance.
Formula A
=
fL
_
RL _
RL +
ra
VALVES, VOLTAGE AND AMPLIFICATION
It will be seen that the stage gain can never quite equal the amplification
factor of the valve, although it can approach it for high values of load
resistance. This circuit arrangement is not as perfect as it would
appear because a hidden factor creeps in which is the shunt capacitance
across the load resistance. This shunt capacitance will have an
increasing " admittance " with increase of frequency. Therefore the
total load will be less as frequency rises. This change of " admittance "
in ohms in a capacitor, in relation to frequency, will often be met with ;
for example, it is the basis of conventional crossover networks in
loudspeakers.
LIM ITATIO N OF ANODE LOAD
The serious reduction of high tension voltage due to the voltage
drop across the load resistance suggests the use of a load with a high
impedance to alternating current and a low DC resistance. A choke
may be used at both audio and radio frequencies providing it has a
suitable inductance for the frequency or band of frequencies it is
desired to amplify. But as the inductive reactance of a choke varies
with frequency, the load will be less effective as frequency goes down.
This means that the lower the frequency we want to amplify, the larger
must be the inductance of the choke. Agai n, shunt capacitances are
fairly high in this type of load so that a response tailing off at the higher
frequencies would result.
RC INTERST AGE COUPLING
Having developed an ample voltage, it is necessary to apply this
voltage to a succeeding valve for further amplification. Referring to
Fig. 3/4, the output voltage appears across the load resistor, and we
wish to apply this voltage between grid and cathode of the succeeding
FIG. 3is .-INTERSTAGE
Resistance Capacity Coupling.
Z=
ANY S U ITABLE FORM OF LOAD
valve. If direct connection were made between the load resistor and
the second valve, the grid of this valve would have the high tension
voltage applied to it. As we have already seen, a positive grid is to be
29
VALVES, VOLTAGE AND AMPLIFICATION
abhorred, quite apan from the fact that the valve would probably be
destroyed by the excessive space current which would flow. So the
use of a coupling capacit.:Jr is suggested, this capacitor having infinite
impedance to DC and, if a suitably large value is chosen, a negligible
reactance to even the lowest audio frequency voltages likely to be
encountered. Reference to Fig. 3/5 shows the actual arrangement
employed.
The bottom or anode end of the load resistor RL is connected to
the grid of V2 via the coupling condensor Cc. The top end of the
load resistor is connected to the cathode of V2 via the high tension
battery which is assumed to have negligible internal resistance. The
presence of the resistance Rg is necessary in order that the valve V2
may be supplied with the necessary grid bias voltage. The load resis­
tance is actually shunted by the coupling condenser and grid resistance
in series. It is important to appreciate this point because the effective
load of the valve is influenced by the presence of the coupling condenser
and the grid resistance. If the coupling condenser is of too small a
value, its reactance will increase with decrease of frequency, giving a
decrease of voltage across the resistance Rg and the input to the valve
V2 will not be a faithful replica of the input voltage as applied to the
first stage V I . It will also be seen that the value of Rg should be kept
as high as possible to avoid giving a reduced total value of load for the
valve VI. The same method of interstage couplings could well be
adopted when a choke or a tuned circuit is used as the anode load for
VI.
TRANSFORMER COUPLING
Another method of interstage coupling is shown in Fig. 3/6. This
is the inter-valve transformer .
AUDIO �
TRANSFORMER
.-------1r-HT+
11
F I G . 3 / 6 .- T r a n s fo r m e r
Coupled Amplifier.
Es
f--I_-�----4-- HT-
The primary of the transformer constitutes the anode load for V I, the
secondary applies the voltage between the grid and cathode of V2 and
also supplies the DC path for the applica�on of the bias voltage: .This
transformer can be of a step-up ratlo pnmary to secondary, glvmg a
30
VALVES, VOLTAGE AND AMPLIFICATION
step-up of voltage, and would appear at first sight to be an ideal method
of interstage coupling. However, this is not altogether the case,
because to obtain a sensibly constant value of anode load at all fre­
quencies the primary must have a high value of inductance. This
demands many turns of wire with considerable self capacity giving a
shunting effect at the higher frequencies. The step up ratio in practice
cannot exceed I to 7 and in most cases I to 3 or 4 is employed. The
loss of inductance due to the flow of DC in the primary, in addition
to the alternating component of anode current, can be avoided if the
circuit of Fig. 3 /7 is employed.
ANODE
LOAD
FIG. 3/7.-Parallel-fed trans­
former. Removes DC from
primary and improves bass
response.
f"'...J
E9
��----����--- HTThis is known as shunt or parallel feeding and enables a smaller
transformer with lower self capacity but the same inductance to be
used, but either method of using an interstage LF transformer is not
regarded with much favour where the highest fidelity of reproduction
is demanded. The transformer method of interstage coupling is
frequently employed in radio circuits where the HF transformer will
have an air core or dust iron core to avoid eddy current losses.
DIRECT COUPLING
In order to avoid the use of a condenser in the interstage coupling
with troubles due to increasing reactance at low frequency and phase
ANODE
-----�--p__ IH Tt
LOAD"'"
+
100
FIG· 3 /8 . -D IRECT­
COUPLED AMPLIFIER.
By eliminating coupling con­
denser bass response is main­
t a i n e d , a n d p h a s e s hift
avoided.
VI
L-�---� + 102
31
VALVES, VOLTAGE AND AMPLIFICATION
:!hift, the direct coupling as shown in Fig. 3/8 may well be employed.
It will be appreciated that the cathode of V2 must be raised to a
positive potential slightly higher than that at the anode of VI in order
to avoid grid current. This method is employed in the well-known
Williamson amplifier.
DETERMINATION OF THE BIAS POINT
The value of grid bias must be chosen with a view to avoiding two
sources of trouble :
(a) Too Iow a bias may permit of the grid being driven positive with
grid current flowing.
(b) If bias is excessive then large excursions on the grid in the negative
direction may produce cut-off of anode current and distortion of
the negative half cycle of input after amplification.
THE TRIODE AS AN O SCI L LATOR
;----- H T+
FIG. 3/9.-The valve as an
oscillator, with tuned grid
circuit and tuned anode load.
Cga is internal capacity gridl
anode of the valve.
Fig. 3/9 shows a triode valve with a tuned circuit in the grid and
another tuned circuit as the anode load. Assuming that inductive
coupling cannot take place between the two tuned circuits, it would
most probably be found that the amplifier was unstable and tended to
behave as an oscillator. Now a valve can only behave as an oscillator
when sufficient energy is fed back from the anode into the grid circuit
to overcome the losses in the grid circuit due to resistance. Such feed­
back takes place through the internal capacity in the valve between the
anode and the grid Cg" the two electrodes behaving as the two plates
of a condenser. The capacity may amount to several pico-farads and
is quite sufficient to cause instability. This feedback is referred to as
positive feedback and is sometimes deliberately introduced into a tuned
circuit to reduce the resistive losses, as in a set with reaction. This
increases the gain and improves the Q of the circuit, thereby narrowing
the band of frequencies which it will accept. Reference is made to
32
VALVES, VOLTAGE AND AMPLIFICATION
this phenomenon because when the construction of a radio feeder
unit is considered, it is imponant to realise that every step must be taken
to avoid feedback in the radio frequency amplifier ; otherwise the
tuned circuits will not accept the full bandwidth of the station, with
consequent loss of high notes contained in the outer fringes of the side
bands. Whilst the capacity anode to grid is normally the one with which
we are most concerned, it should be appreciated that capacity also
exists between anode and cathode and grid and cathode.
MILLER EFFECT
.------ HT+
Cga
�- - .,.
- - - - .,
:
= Cak
' -+
.---+
Cgk ..L
"'T'"
I
L
_ _ _
_ _ _ _ _
I
i
,
I
.J
FIG. 3/ Io.-Inter-electrode
capacities in a valve.
Eg
When an alternating voltage is applied between grid and cathode of
the triode amplifier shown in Fig. 3/10, an alternating current flows
in the small condenser formed by the grid and cathode electrodes, Cgk>
just as in any other condenser. This means that the grid input impe­
dance is by no means infinity, and as frequency increases the input
damping will become more marked.
In a similar way an alternating current flows in the condenser
formed by the grid and anode, Cga, but since the instantaneous voltage
between these electrodes is considerably larger than the signal voltage
on the grid, the current in the grid/anode capacity Cga is larger than
it would be were no amplification taking place. Looked at from the
grid circuit the increased current is equivalent to an increase in input
capacity of the valve and the effective input capacity is (I + A) times
(where A is the amplification of the stage) the capacity which would be
expected from an examination of the actual inter-electrode capacity
Cga alone.
Readers desiring to study the Miller effect more fully are recom­
mended to an article by " Cathode Ray " in the August 1949 issue of
Wireless World, which gives a well reasoned and lucid survey. The
third hidden capacity is capacity anode/cathode, Cak> which will
shunt the load of the valve.
33
VALVES, VOLTAGE AND AMPLIFICATION
THE SHIELDED OR SCREENED GRID VALVE
In early days of radio communication, the feedback due to inter­
electrode capacitance already described, made it exceedingly difficult
to obtain stable radio frequency amplification. Attempts were made to
balance out the positive feedback by applying an equal amount of
feedback in the opposite sense, i.e. negative feedback. This system
was known as neutralisation, but was never a particularly practical
solution. At the same period, the idea of putting an electro-static
shield between the grid and anode was evolved. A positive voltage
was applied to the screen to draw the electrons through to the anode.
SC R E E N
GRID
F I G . 3 / 1 I .-The t e t r o d e
valve.
,---- + 1 20
+--
t
80
This greatly reduced the anode to grid capacity but did not completely
eliminate it. Examination of the grid characteristics of typical screen
grid or tetrode valves shows them to be not dissimilar to those of an
ordinary triode.
la
(, U R R E N T
ANODE
FIG. 3 / I 2.-Anode charac­
teristic of a Tetrode Valve.
('HARACT E R I S T I C
E... A N O D E
VOL.TS
Examination of the anode characteristic shows that it is a very
different shape from that of the triode. At low anode voltages, the
anode current rises linearly, but after a certain critical value of anode
voltage, the current actually falls and continues to do so until it again
34
VALVES, VOLTAGE AND AMPLIFICATION
takes on its normal upward trend. This kink in the characteristic is
due to the fact that when electrons impinge on the anode sufficient
energy is released to produce new electrons from the anode material,
called secondary emission electrons. These are then attracted by the
positively charged screen until the anode voltage is raised sufficiently
to pull them back to itself. This kink in the anode characteristic is a
serious drawback to this type of tetrode valve in that the voltage swings
at the anode must be limited so that the valve works over the straight
part of the characteristic.
THE PENTODE
This valve is a development of the tetrode and a further wide mesh
grid is interposed between the anode and the screen. This grid is
normally connected to cathode and its presence repels the electrons
produced by secondary emission from the anode and these return to
the anode. The anode characteristic of a typical pentode shows that
the tetrode kink has been removed but that the characteristic is still
very unlike that of a triode. The grid characteristic of a pentode is,
however, very similar to that of a triode. This means that gm is similar
to that of the triode. Referring to Fig. 3 /12, and remembering that
the anode resistance of a valve is obtained from the relationship E�
la
it will be seen that on the flat part of the characteristic, the anode
resistance will be very high and in practice a value of one megohm is
by no means unusual. From the relationship [J. equals gm X ra, it will
be seen that ra being very large fJ. will become very large. In other
words, the valve will have a high amplification factor, and this implies
that a high stage gain can be obtained from one of these valves used as a
resistance capacity coupled amplifier. The screen is maintained at a
potential of from half to two-thirds that of the anode and is coupled
to the cathode by means of a bypass condenser, thus ensuring that
AC fluctuations on the screen, remembering that the screen takes some
little part of the space current, will be obviated.
OPTIMUM VALUES FOR RC COUPLED AMPLIFIERS
By now the reader should have a fair appreciation of the basic
principles underlying the choice of coupling values, but it is quite
outside the scope of this book to tell him how to determine them for
each and every valve. (Valve makers publish data for all their valves.)
As the value of grid leak in the succeeding stage affects the stage gain
and maximum voltage output, the obvious inference is to use as large
a value of Rg as possible. There is usually a maximum value of grid /
cathode resistance that a valve can tolerate, particularly in the case of
a power valve.
35
VALVES, VOLTAGE AND AMPLIFICATION
METHODS OF BIAS SING
It is necessary to bias a valve in order to prevent grid current, and
up to the present a battery has been used for the sake of simplicity.
However, the separate bias battery is a thing of the past as much
simpler ways of biassing can be employed. If it is borne in mind that
grid bias is fundamentally a matter of making the grid some few volts
more negative than the cathode, then biassing problems do not exist.
r---IH
-- T+
+
IN PUT
RS
+
FIG. 3 / 1 3.-Cathode Bias­
sing. Method of obtaining
grid bias by resistance in
cathode.
RK
Fig. 3/13 shows an arrangement that is commonly adopted to
produce so-called " Cathode Bias ", " Self Bias " or " Automatic
Bias ". (The last two terms are not liked as they can so easily be
confused with another method used largely in RF technique.)
It will be seen that a resistance RK has been introduced between
cathode and HT- of an amplifying system. The space current of the
valve will flow through RK and in accordance with Ohm's law, IR
E,
a voltage will be produced across it, so that the cathode is more positive
than the HT- line. The grid is connected to the HT- line through its
grid leak, Rg, and therefore the cathode is more positive than the grid,
which is another way of saying that the grid is more negative than the
cathode-the condition we seek for grid biassing.
The current flowing through RK will be the direct current component
of anode current plus the alternating current component, therefore
both a steady mean potential and an alternating voltage will appear
across RK. This condition is undesirable for the following reason­
when the grid of the valve is going less negative the anode current will
increase and the cathode will go more positive with respect to the
HT- line. Now the grid input of a valve is applied between grid and
cathode, and therefore the potential across RK is in effect in series with
the input, and a study of Fig. 3/13 shows that the two potentials are in
anti-phase, or out of step, and the net input voltage is the difference .
between the two.
The effect may be overcome by shunting RK with a condenser of
such a capacity that its reactance at any frequency within the range it
=
VALVES, VOLTAGE AND AMPLIFICATION
is desired to amplify is negligibly small relative to RK• It is usual to
employ a low voltage electrolytic condenser for this purpose, having
a capacity of from 25 to 100 microfarads. This ensures that the voltage
across RK is a steady DC potential.
It might suggest itself that if it is desired to attenuate the bass
frequencies, this could be accomplished by deliberately using a shunting
condenser of a small value, thus achieving a form of " tone control".
In practice, however, it is more difficult to produce the bass notes than
lose them !
The partial cancellation of the input voltage by the voltage fed back
to the input across the cathode resistor is often referred to as degenera­
tion, the opposite of reaction or regeneration. As reaction is referred
to as positive, in phase, feedback, it is logical to refer to degeneration,
out of phase feedback as " Negative Feedback" and as it is due to the
current through the valve the example outlined above is tertned
" Current Negative Feedback". The subject of Negative Feedback
is treated in some detail in a later chapter.
D IRECTLY HEATED VALVE BIAS ARRANGEMENT
r------ H T+
H E AT E R
Rg
"------./. WINDI N u
INPUT
RK
.-----43
11
FIG. 3/I4.-Cathode biassing
applied to a directly heated
valve.
CK
Cathode biassing may be applied to a directly heated valve in the
manner shown in Fig. 3/14. It is necessary to use a centre tap on the
heater winding to establish a mean cathode potential, otherwise there
will be superimposed on the bias voltage a 50-cycle ripple voltage
equal to the heater voltage.
BAC K BIASSING
An examination of Fig. 3/15 shows that a resistance Rb has been
inserted in the HT- lead, and the total space current of the valve or
valves flowing through this resistance will produce a voltage across it.
If the grid of the valve is connected to HT- via its grid leak then the
grid will be more negative than the cathode, which is the result desired.
37
VALVES, VOLTAGE AND AMPLIFICATION
A bypass condenser must be in shunt with the resistance Rb to avoid
the production of alternating voltages across it, just as was used in
Cathode biassing. This method of obtaining bias is well adapted for
,----- H T+
'---� LT +
L-
FIG. 3/Is.-Back biassing,
HT- lead biassing.
or
Rb
�,-���_+- HT-
_
_
_
'----- LT-
use with directly heated battery valves where the cathode biassing
system is not convenient to use, as the resistance Rb may be tapped
to obtain other values of bias.
CALCULATION OF THE B IAS RESISTANCE
The calculation of the bias resistance is a simple application of Ohm's
E
law, R = , where E is the desired bias voltage and I is the current
I
flowing through the bias resistance.
To take a practical example of cathode biassing, a Mullard output
pentode type EL37 with 250 volts on both anode and screen requires
a bias voltage of 13'5. The current flowing through the bias resistor
will be the sum of the anode and screen currents, I OO and 13'5 mA
respectively; therefore
(Remember the 1,000 because Ohm's
13'5
X 1,000
RK=
law
is in the basic units of amps)
100+ 13'5
�
= 120 ohms approx.
=
1 1 3'5
Take another example, this time HT- lead biassing; a small battery
set has as its output pentode a Mullard DL93, preceded by three other
valves taking a total HT current of 5"5 mA. The output valve takes a
total of 17'5 mA and the bias voltage required is - 7'5. The total HT
current is 11'5+5'5 mA; therefore the HT - lead resistance
7 '5
X 1,000 ohms
17'5+5'5
= 326 ohms approx.
VALVES, VOLTAGE AND AMPLIFICATION
Note carefully that in this system the total HT current of the whole
set or amplifier must be ascertained to make the calculation.
When using any particular valve as an RC coupled amplifier, a
direct calculation of the bias resistance is not feasible because, due to
the high value of load resistance, the actual HT voltage at the anode
bears no relation to the figures of "Operating Conditions" as shown in
a valve manufacturer's catalogue. In consequence the space current
will be reduced, and for the correct bias to be obtained a higher value
of bias resistor will be needed.
The calculations involved are complex, and the reader is recom­
mended to refer to tables produced by the valve maker for each valve
suitable for use as an RC coupled amplifier, in which the values of
RL and RK and Rg of following valve are listed, together with stage
gain and maximum output voltage for a given distortion figure.
For American type valves a comprehensive table is given in the
(American) Amateur Radio Relay League's Handbook, easily obtained
in this country.
LEAKY GRID BIAS
A system sometimes used for early stage audio amplifiers is leaky
grid or condenser bias. This method depends on the flow of grid
current and for that reason is not to be recommended.
GENERAL
•
A point to note is that in both Cathode biassing and Back biassing,
the bias voltage is obtained at the expense of the actual value of HT
voltage applied between anode and cathode of the valve, but this is
rarely a serious problem, especially in mains operated sets.
VARIABLE
mu
la
FIG. 3 / I 6 .-Grid characteris­
tic of variable-mu valve. The
slope or gm decreases with
increase of bias.
-Vs
Reference to Fig. 3/16 shows the grid characteristic of a variable-mu
valve. Comparison w th t e grid characteristics shown in Fig. 2/6
shows that the essenual dIfferences are that the point of cut-off is
much more remote and that the characteristic is not truly straight over
�
�
39
VALVES, VOLTAGE AND AMPLIFICATION
any portion. This is achieved by incorporating a grid varied with a
non-constant pitch, i.e. the wires are close together, say, at the ends,
and wide apart in the centre. The net effect is that very considerable
negative voltage must be applied to the grid before it can completely
prevent the flow of electrons to the anode through the widely spaced
parts of the mesh, resulting in the long gradual curve as shown in the
diagram. Mutual conductance is defined as
SI.
SEg
.
It will be appreci-
ated that this value will be greater when the grid is slightly negative
and the slope steeper, than at a higher value of grid volts when the
slope is less steep ; hence the name variable-mu. The gain afforded
by an amplifier incorporating such a valve will depend on the working
bias point selected, and the arrangements of Fig. 3/ 17 would enable the
stage gain to be varied by adjustment of the value of VRK, where
VRK forms part of a potentiometer across the HT supply.
....-
....o--..
- HT +
-
FIG. 3/17.-Circuit for con­
trol of gain by variable bias.
Rs
RK
VRK
CK
The inclusion of RK is desirable, so that when the value of VRK is
turned to zero the valve cannot be worked with zero bias but will have
a minimum bias value dependent on the value of RK.
It will be appreciated that, due to the constant curvature of the
characteristic, distortion is inevitable, but this will not be serious if
the input is kept to a very low value. The chief application of these
valves being in RF circuits, the latter condition will normally be
satisfied ; the type is mentioned here because it has a restricted
application In giving an automatic gain control effect in audio
amplifei rs.
40
CHAPTER 4
THE
VALVE
AS
A
POWER
AMPLIFIER
The problem so far has been to produce an exact facsimile of the
input voltage, but amplified to give greater voltage excursions at the
output. The problem in the final stage is to make a valve produce a
distortionless power output to do work : the work of actuating a
headphone diaphragm or causing the mass of a loudspeaker cone and
coil assembly to move and displace a column of air. Electrical work or
energy is measured in watts, the product of volts and amperes. Thus,
besides large voltage changes, we require comparatively large current
changes. The factors that govern the current that can pass through the
valve are :
(a) the extent of the emissive surface on the cathode linked with heater
wattage which together govern the number of electrons available,
and
(b) the permissible heat dissipation at the anode without causing
overheating and the possibility of softening the vacuum due to
driving out occluded gases from the electrode structure. Study of
these points shows that a power valve will generally speaking have
a larger cathode and larger electrodes than a valve used only for
voltage amplification.
CHOICE OF OUTPUT VALVE
TRIODE v . PENTODE
For the triode there is a low inherent distortion due to straightness
and parallelism of grid characteristics coupled with a low anode
resistance. Against the triode is the low stage gain due to its low mutual
conductance ; a large voltage input is needed to drive it fully and it is
greedy of high tension current for the output obtained. The triode
also produces a fair measure of 2nd harmonic distortion although
steps can be taken to eliminate this fault as will be seen in later chapters.
For the pentode is its high gain for a comparatively small input
voltage and its good return for the expenditure of anode current.
Against it is its higher inherent distortion due to lack of straightness
and parallelism of grid characteristic, although these defects can very
largely be overcome, as will be seen in Chapter 1 5 . The pentode has
a very much higher anode resistance ; this question of internal resis­
tance of the output valve and its effect on the quality of reproduction
is discussed later in this chapter and also in Chapter 8.
41
THE VALVE AS A POWER AMPLIFIER
THE TRIODE OUTPUT STAGE
The valve can be considered as an alternator and as such there will
be a value of load into which it can deliver its energy to the best effect.
It is helpful to consider the case of a simple battery possessing a certain
internal resistance. When current is taken from the battery it has to
flow through this internal resistance, made up of the resistance of the
elements of the battery. A battery may be said to have an EMF of
1 ' 5 volts, but on load it would not show a potential difference of 1 ' 5
volts across its terminals . When current is drawn, a certain voltage
drop will occur across the internal resistance and the potential difference
at the terminals is the difference between the EMF and the voltage
across the internal resistance.
Fig. 4/r.-Battery EMF = I ' 5 volts.
Internal resistance 0'5 ohm.
External
Current in
PD across
Power in
Resistance
Circuit
external R
external R
7'0 ohms
0'2 amp
1 '4
2'5
0'5
1 '25
1 '0
"
"
1 '0
"
1'5
0'25 "
2'0
0'1
2'5
0'5
"
0'05 "
2'73
"
"
"
"
"
"
volts
1 '0
0"75
0'5
0'25
0 ' 1 36
"
"
"
"
"
"
0'28 watt
0'625
"
1 '000 "
1 ' 125
1 '000
0'625
0'372
"
"
"
"
Columns 1, 2 and 3 in the table shown in Fig. 4/1 should make this
clear. Reference to column 4 of the table shows that the watts dissipated
in the external load steadily increase until a maximum is reached when
the external load resistance equals the internal resistance of the battery.
After this, reduction of the load causes more current to flow, but
only produces a lower dissipation of energy external to the battery,
where, after all, it is of most use.
From the foregoing example it can be seen that the greatest voltage
appears across the external load when this load is so very large that
practically no current flows, and in consequence there is little voltage
drop across the internal resistance of the battery. This agrees with our
conclusion that when the greatest voltage output is wanted from a valve
connected as a voltage amplifier, the load should be as high as possible :
many times the anode resistance of the valve.
Let us take the case of a valve as a generator of an alternating poten­
tial as shown in Fig. 4/2 (a). This represents a valve of anode resistance
ra feeding into a load RL• The alternating input to the grid is Eg volts.
Fig. 4/2 (b) shows the electrical equivalent of the circuit, in which the
THE VALVE AS A POWER AMPLIFIER
valve is shown as a generator of internal resistance ra delivering [LEg
volts into the load RL when [L is the amplification factor of the triode.
Our argument shows that for maximum power in the load, RL must
equal ra, and the practical significance is that when a triode valve is
( a)
(b)
r------ H T +
RL
ra
ra
E9
EQU IVALENT
CIRCUIT
OF
(a)
FIG. 4/2.-The valve as a generator.
working into a loudspeaker approximating to a pure resistance, we get
maximum power into the speaker when its resistance is equal to the
anode resistance of the valve.
Most readers already know that for various mechanical and electrical
considerations, it is usually undesirable to produce a moving coil
loudspeaker having a coil resistance much in excess of 1 5 ohms, which
is nowhere approaching the anode resistance of a valve and cannot be
placed directly in the anode circuit for an efficient development of
power across it.
To overcome this difficulty, it is necessary to introduce a step-down
transformer into the circuit, the primary being placed in the anode
circuit of the valve and the secondary connected to the coil of the
loudspeaker. The turns ratio of this transformer can be determined
approximately from the following formula :
ILoad required
V
by valve
:
I
Resistance of speech coil
The load usually required by a triode valve is approximately twice its
anode resistance because the load offered by a loudspeaker is not
resistive. When the load RL is twice ra the most efficient condition is
obtained at the expense of an increased driving voltage on the grid.
THE BEAM TETRODE
beam tetrode employs the principle of focusing the stream of
electrons flowing from the cathode to the anode. Furthermore, the
control grid and screen grid have the same winding pitch and are
A
43
THE VALVE AS A POWER AMPLIFIER
assembled in the valve so that they are in optical alignment. The effect
of the latter is to reduce the value of screen current as compared with
a pentode of similar power. This is due to the screen wires being in the
" shadow " of the grid wires which, being negatively charged, cause
the electrons to diverge. This reduction in screen current represents
a saving of power, giving a higher overall power efficiency.
Beam forming or confining plates, connected to cathode, are employed
to shield the anode from receiving any electrons coming towards it
from the direction of the grid support wires, where the focusing of
the electrons is imperfect. The presence of these plates, bent round
as they are, prevents the slow velocity secondary emission from the
anode reaching the screen, and the anode characteristic exhibits no
kink as in the normal tetrode. This means that greater swings of anode
current over the approximately linear part of the characteristic can
produce greater power output. Furthermore, due to their general
shape, the harmonic distortion produced is principally 2nd, with
but little 3rd, which is the opposite to a pentode and more like a
triode. In push-pull the 2nd harmonic is cancelled and a pair of
such valves gives a better return in terms of undistorted power output
than a pair of equivalent pentodes for the same HT requirements.
Well known examples of this type of valve are the Osram KT66
and American types 6V6, 6L6 and 807.
PENTODE OR TETRODE LOAD
When using pentode or tetrode output valves the method of deter­
mining the value of the load previously outlined is not applicable, as
such valves have very high values of anode resistance, 50-70,000 ohms,
and work most satisfactorily into loads generally between one-third and
one-sixth of their anode resistance. These output valves can give an
intolerable degree of distortion (without NFB) unless matched most
carefully to the loudspeaker.
A pentode or tetrode will give more inherent distortion than a triode
in any case, due to lack of parallelism in its characteristics. It is apparent
that a value of load should be chosen that will give the least possible
total harmonic distortion, but this is pretty certain not to be the value
of load that will give greatest power output.
ACTUAL VALUE OF LOAD FOR A POWER VALVE
The calculations needed to determine the actual value of load
required for any particular type of power valve are quite complex.
The requirements of maximum power output and minimum harmonic
distortion are conflicting as already indicated, and it is quite beyond the
scope of this book to treat fully the methods adopted. The interested
reader might well refer to Radio Designers' Handbook, edited by F.
Langford Smith and published by Iliffe & Sons Ltd.
44
THE VALVE AS A POWER AMPLIFIER
The experimenter would do well to obtain the recommended values
of load resistance for any particular valve from the makers and abide
closely to their data.
THE LOAD LINE AND POWER AMPLIFIERS
A power amplifier is certain to be working into an inductive load
and it is assumed that the inductance of the transformer primary is
sufficiently high to offer a sensibly constant load with no appreciable
shunting of the AC load at all frequencies it is called upon to handle.
However, if it does not fulfil these conditions, or even if a capacitance
is placed across the primary to reduce top response, then the load will
be reactive and not a constant value .
The general effect is either to increase distortion for the power
output the valve is capable of giving, or to produce less output for the
quoted distortion figure.
LOUDSPEAKER DAMPING
This is a term used indiscriminatively to describe two separate and
yet inter-related phenomena.
First the ability of a loudspeaker moving system to come to rest
immediately the input applied to its terminals ceases, and secondly
the damping out of resonances, particularly the bass resonance of the
moving system, by the action of the output stage. Treating the second
one first, it will be seen from the impedance curve of a typical loud­
speaker shown in Fig. 4/3 that the impedance of the speaker rises to a
high value at 52 c/s.
lOO
18
"
OD
,.
I zl :!
(Il) ;
"
'0
-
---
"
10
o
N
�
� 51 S g
2��
�
� § U��§�
� � � fi U��
C Y C L ES PER S ECOND
IMPEDANCE CU�VE O F W I7/CS. UNIT
FIG. 4/3.-Typical loudspeaker impedanc(curve.
The effective resistance shunted across the primary of the output
transformer is the resultant of the anode resistance and the load
resistance in parallel. If the load resistance were a fixed resistance load
say of 2,500 ohms for an Osram P)4 of anode resistance 860 ohms
4S
THE VALVE AS A POWER AMPLIFIER
then the resultant would be 2, 5 00 and 860 in parallel = 6o(j ohms.
However, with a speaker load, if the impedance of the speech coil
rises six times at the bass resonant frequency, the shunt impedance
on the primary will be 1 5,000 and 860 ohms in parallel = 8 1 4 ohms.
Taking the case of a pentode output valve, Mullard EL32, working
into a load of 8,000 ohms with an anode resistance of 20,000 ohms, a
rise of speech coil impedance by a factor of 6 will give a shunt impedance
of 20,000 and 48,000 ohms in parallel = 29,000 ohms approximately
as compared with a normal value of about 7,180 ohms.
By comparing the figures quoted, it can be seen that the voltage
output will rise very considerably in the case of the pentode, and a
pronounced audible effect will take place when a note is fed to the loud­
speaker in the region of its bass resonant frequency. The triode
however, by virtue of its low anode resistance, tends to swamp the
rise in load resistance and gives a much less pronounced effect.
This must not be taken as a condemnation of the pentode or tetrode,
as methods can be employed to give them in effect a very low anode
resistance and these are described in the chapter on Negative Feedback.
The foregoing arguments have assumed that a good output trans­
former has been used, i.e. one with adequate primary inductance to
reproduce down to the lowest audible frequencies. It should, however,
be realised that a transformer having a low primary inductance will
be helpful in reducing bass resonance with a pentode output stage,
due to low value of inductance acting as an appreciable shunt across
the load. The power output will, in this case, be reduced to a value
very much less than the rated output of the valve.
Reverting to the question of damping of a loudspeaker for the repro­
duction of transients, the moving system should come to rest im­
mediately the transient ceases and should not come to rest like a
pendulum. Loudspeakers vary enormously in this quality of transient
response.
Results are improved by high flux density, free suspension and
efficient systems of mounting, all of which help to reduce " ringing ".
The coil of a loudspeaker moving in a powerful magnetic field will
behave as a dynamo and when falling back to rest will generate an
EMF in opposition to the one originally causing it to move. This
EMF will cause a current to flow in the coil and the field produced
will interact with the flux of the field and tend to cause a movement in
the opposite direction to the original one, taking the system past its
dead centre position instead of coming to rest immediately. Now the
anode resistance of the output valve as reflected into the loudspeaker
via the output transformer acts as a shunt damping resistance across
the speech coil, and it can be seen that the lower the anode resistance
of the valve the better the damping on the speech coil. It would appear
that the triode valve would score here over the pentode or tetrode,
THE VALVE AS A POWER AMPLIFIER
but the same measures can be applied to these as for improving the
resonance damping. The " damping factor " of an output system can
be expressed in general terms as the ratio of the load resistance to the
anode resistance (RL to ra).
As this is the case, the damping at frequencies at which the impedance
of the loudspeaker rises will improve, since RL is greater and a greater
effective damping factor is obtained. It will be seen therefore that
whilst the increase of load resistance, at bass resonance of the loud­
speaker in particular, tends to increase the output fed to the loudspeaker,
the improved damping factor tends to improve the reproduction as
regards transients .
In the case of the P]4, the damping factor would be :
2,500 : 860, 3 : I .
The EL32 . . . . . . 8,000 : 70,000, I : 8 ' 5 .
47
CHAPTER 5
DECOUPLING
AND
INSTAB ILITY
In view of the fact that the popularity of negative feedback has
resulted in more and more cases of instability in amplifiers, often unsus­
pected in home-built equipment, it is considered worth while to devote
an entire chapter to the study of the problem.
For the benefit of readers who are not accustomed to inspecting the
antics of amplifiers with the aid of an oscilloscope, we will begin by
the study of a few oscillograms of typical effects of instability, before
proceeding to a technical, albeit less pictures que, diagnosis.
EXAMPLES
For the purpose of the first test (and others described elsewhere in
the book) an amplifier with uncorrected push-pull tetrode output was
used and was known to be unstable with any more than 20 db NFB.
A variable resistance provided a means of adjusting feedback to any
value between 26 db and zero, giving an effect equivalent to source
impedances varying between I ohm and about 100 ohms. A rough-and­
ready test for ascertaining this important impedance value in an
amplifier (which incidentally may vary with frequency) is to feed in
a steady tone between 500 and 1 ,000 cjs from AF oscillator or constant
frequency record, and measure the voltage across the secondary of
the output transformer without loudspeaker load : then to connect a
calibrated variable resistance across the secondary and adjust the value
until there is a voltage drop of 50 per cent. The resistance reading
then approximates the amplifier impedance in question. With the
tetrode amplifier, variation of feedback produced the following impe­
dance readings, with corresponding effect on the voltage rise due to
bass resonance of a loudspeaker. The transformer ratio was for load
matching to a 1 5 ohms speaker.
TABLE 1
Tetrodes and Voltage NFB.
Feedback
o db
6 "
8
"
20
"
26
"
14 "
*
Volts
Input
I
10
12
32
75
*
Volts
Output
7
7
7
7
7
*
No load.
Impedance
measured
1 00 ohms
13
10
4
2
Serious instability intensified by absence of speech coil load.
*
"
"
"
"
DECOUPLING AND INSTABILITY
Table I shows the impedances achieved by varying the feedback up
to 20 db. The gain control of amplifier was set very low and was not
moved during the test. At 26 db the oscillations made readings
impossible. (They blew the fuse of the voltmeter.)
Fig. 5/1 illustrates the presence of oscillation at a frequency as high
as 150,000 c/s in the tetrode amplifier. It was found that reducing
feedback (in this case) from 26 db to 20 db completely cleared the
oscillation, and improved the quality of reproduction-despite the
very high frequency of the disturbance.
26 db
22 db
20 db Feedback
FIG. Si l o-Oscillation at I S 0 Kc/s in tetrode amplifier.
micro-secs.
Time base 5 0
It is fairly easy to check the frequency. The oscillograph was set
I
with the time base at 50 micro-sec., '000050, which is - --- sec. There
20,000
are 7'5 waves in the complete trace, so the frequency is 7'5 X 20,000
150,000. It is hardly necessary to say that an ordinary AC voltmeter
will not give a reading at such frequencies, but it is necessary to add
that a dummy speech coil, fed by this input, started to go up in smoke.
(The output transformer is still standing up to grave abuse, and is a
credit to its makers, Excel Radio, of Shipley, Yorkshire, who threw
together the " guinea pig " amplifier for the purpose of these tests.)
An amplifier which gives poor results, with excessive "top " and
possibly excessive heat in the output valve (or valves) should be suspect.
An AC voltmeter connected across the speech coil of the loudspeaker
should obviously give no reading with the absence of input to amplifier.
An oscilloscope similarly connected should show no amplitude on the
"Y " axis. The discovery of unwanted voltage by either of these
simple tests would clearly indicate that the "patient " required medical
attention. Unfortunately, the oscillation might only occur at a parti­
cular part/amplitude of a low frequency cycle. This would not show
statically, i.e. with zero input, so diagnosis of the fault would require
the use of an oscilloscope and a signal source. The writer recently
tested an amplifier in which this fault produced a picture rather like
a string of sausages, as shown in Fig. 5/2, which has been drawn from
memory.
=
49
DECOUPLING AND INSTABILITY
The amplifi�r received some attention, with slight improvement, and
th e sausages dIsappeared from the menu, but there is still a tendency
FIG. 5 /2 .-0sciIlogram of shock-excited parasitic oscillation of
extremely high frequency, carried on 50 cycle mains hum.
towards motor-boating, which is easily provoked by switching on
without a reasonable load on the output stage. This is coupled with
instability which is provoked by turning the treble control fully on.
These combined operations produce some rather fascinating pictures
on the oscilloscope, and incidentally burnt out quite a robust variable
resistance which was tried as a dummy load. (See Figs. 5/3, 5/4 and
515·) In the first of these oscillograms, Fig. 5/3, a steady tone at 1 ,500
cls was fed into the amplifier, to be mixed with the motor-boating and
HF oscillation.
FIG. 5 /3.-Picture of combined effect of m�tor-boati�g and para �itic
oscillation, with steady tone at 1,500 cls, showmg severe mtermodulatlon.
Three traces are included in Fig. 5/3 because the period of the motor­
boating was too slow for the complete cycle to be reproduc� d in one
shot. It is interesting to observe how the 1,500 cycle note IS pushed
around by the low frequency of the motor-boating ; also (in the third
picture) how the high frequency oscillations beat in sympathy with the
1,500 cls " carrier ".
The next oscillogram, Fig. 5/4, shows at A the combination of motor­
boating and two bursts of HF oscillation, taken with motor-speed of
1 '2 ins. second (which establishes the frequency of the LF disturbance
at about 5 c/s), and at B a single trace to show the effect of connecting
a loudspeaker to the output of the amplifier.
When comparing these oscillograms, it is necessary to bear in mind
50
DECOUPLING AND INSTABILITY
that A is a continuous trace with a moving film, whereas B is a snapshot.
The final oscillogram in this set, Fig. 515, shows the effect of the
HF osc;illation on a pure tone of 800 c/s.
The audible effect of condition B-C was to introduce a hissing effect
A
FIG. 5/4.
B
A. Motor-boating and HF oscillation. No loudspeaker. Film speed 1 ' 2
ins. sec.
B. Motor-boating stopped by connection of speaker. Two shock­
excited bursts of oscillation remain. Time base 50 m/so
to the pure tone of A which was easily heard and could not be over­
looked by the normal ear. Unfortunately, a listening test on music
resulted in condition B-C being mistaken by experienced loudspeaker-
A
B
c
FIG. 5/5. -Effect of oscillation on a pure tone (Loudspeaker in circuit).
A.
B.
C.
800 cycle note, with treble control at normal position. Time base
IS m/so
Oscillation caused by turning treble control to maximum. Time
base IS m/so
Same as B, but time base on oscilloscope altered to 5 milli-secs.
builders as an increase in HF response, instead of being classified as
distortion, which it undoubtedly is. This only goes to prove how
difficult it is to trace this type of trouble-which may be dangerous
to valves and coils-by simple listening tests. In fact, an oscilloscope
51
DECOUPLING AND INSTABILITY
is an invaluable aid in checking any amplifier for a wide range of
suspected and unsuspected faults.
TREATMENT
The cure for oscillation produced by in-phase feedback due to
coupling through the common HT power supply is to reduce the
internal resistance of this supply at the frequencies involved. The
connection of a large condenser across the HT supply will do this,
and a value of 16-100 mfd is suggested, the larger the better.
A 16 mfd condenser has a reactance of 332 ohms at 30 c/s whilst
100 mfd has a reactance of but 53 ohms, thus too small a value will not
effect a cure at low frequencies. It should be noted that this feedback
effect can only take place over three or more stages of amplification,
but serious difficulties can arise in an amplifier of high gain, and a more
elaborate system of decoupling becomes necessary.
,...-----r---;--.Ni! N·;-H T +
RHT
+
�
���
__
�
__
__
__
��_______�__+-�
__
__
__
__
H T-
FIG. 5/6.-Three stage amplifier. The resistance of the HT supply forms
a common coupling resistance to all stages.
In Fig. 5/6 is shown the outline of a typical amplifier. VI is RC
coupled into V2, and V2 into V3 the output stage, which being a power
valve will make large excursions of anode current. Ideally there will
be no internal resistance in the HT battery or other source of HT
supply but some will exist in practice (RHT Fig. 5/6), and the large
current variations in the anode circuit of V3 will produce alternating
voltages across RHT. The valves VI and V2 also deriving anode volts
from the same source, will receive a DC potential with an AC potential
superimposed. Taking the instant when the grid of VI is becoming
more negative, its anode volts will be increasing due to reduced voltage
drop across its anode load. The grid of V2 will therefore be going
positive and its anode negative, and the grid V3 will be going negative
and its anode positive. It will be seen that the variations at the anodes
52
DECOUPLING AND INSTABILITY
of V3 and VI are in step, and it is quite possible for the variation in
HT voltage at the anode of VI to be quite as large as that caused by the
input voltage on the grid. If this is the case the input could be removed
and the system would continue to operate as an oscillator, supplying
its own input at some frequency determined by the circuit values, and
the value of the common coupling resistance. If in the audio range,
this would give rise to the familiar noise of " motor-boating ". The
exact frequency of oscillation is difficult to forecast and may quite
easily be above audibility, at 20 Kc/s or more, and its presence is
difficult to detect, as already shown in Fig. 5/1 .
Large
Co n de n se r
� �-�
Cd r
Rd Re s i s t o r
Decoup1 in9
VI
of F i g . 5/6
FIG. sI7.-Decoupling the
HT Supply.
Figure 5/7 shows the modification that can be applied to VI of Fig.
5/6. An additional resistance Rd is inserted between the HT supply
and the load resistance RL with a condenser Cd to the ground line from
the junction of Rd and RL•
The time constant in seconds of the system Rd and Cd is obtained
by multiplying Rd X Cd where Rd is expressed in ohms
and Cd in farads. To be completely effective the time constant
should be such that it is longer than the time internal of
any frequency to which the amplifier will respond. For example,
25 c/s takes 1/25th of a second for one complete cycle ('04 second).
Suitable decoupling values would be 40,000 and 2 nf, having a time
2
constant of 40,000 X
sec. = � sec.
'08 sec. which gives
100
1,000,000
an ample margin. If an increase of Rd is possible, bearing in mind
the reduction of available HT (not usually serious in an early voltage
amplifying stage where the grid voltage swing is limited), then Cd
could be decreased in proportion ; 80,000 and I mfd would give the
same result as 40,000 and 2 mfd. A rule of thumb is to make Rd =
� of RL with an appropriate value for Cd. The addition of the decoupl=
53
DECOUPLING AND INSTABILITY
ing Rd and Cd will also assist in the reduction of hum when a mains
HT supply unit is used, but this aspect is treated in the chapter devoted
to Hum.
2.
INSTABILITY
Whilst motor-boating is one particular form of instability, it has been
indicated that supersonic oscillations may be taking place and these are
unlikely to be caused by coupling due to the common supply, but to
hidden factors such as valve inter-electrode capacitance. Fig. 5/8 shows
an output stage with valve inter-electrode capacitances indicated.
Rb
�-r------�r H T +
Cb
FIG. 5/8.-Feedback through
inter-electrode
capaciti e s
causing oscillation i n output
stage.
R b and Cb act as
tone control by reducing
higher frequencies, thus preventing oscillation.
Ca9 :
mn�1l11
unum..
TypiCll va!ues :
Rb 10K n
Cb 0'0 1 mfd
�
�
Due to phase shift in the transformer at high audio frequencies the
capacity Cag, small but nevertheless present even in a pentode
(greater in an output pentode due to larger electrodes), will feed
back enough energy into the grid, thus further shifting the phase, so
that oscillation will take place. Pentodes and tetrodes are particularly
prone to this trouble because they tend to produce more high order
harmonics, and with increase of frequency the reactance of Cag will
decrease and so help to produce oscillation. An output pentode will
normally have a resistance (Rb) and capacitance (Cb) in shunt with
the transformer primary as Fig. 5/8, to reduce the high frequency
response which is harsh and objectionable due to the presence of
harmonics, and is intensified by the normal increase of loudspeaker
coil reactance at high frequencies. The component Rb and Cb will
normally remove the possibility of feedback and oscillation.
Further steps to reduce the possibility of oscillation are the inclusion
of " Stoppers " in the grid, screen and anode circuits.
The presence of resistance in an oscillatory circuit damps the oscilla­
tions and may even preclude oscillation from starting. The grid
stopper will not reduce the input voltage to the grid as normally no
current flows in the grid circuit, and may be of a large value up to
54
DECOUPLING AND INSTABILITY
0'1 megohm for example, bearing in mind limitations of resistance
between grid and cathode. The presence of the capacity between grid
and cathode also limits the size of grid stopper that one may introduce,
10
HT+
ST O P P E R S
FIG. 5 /9.-" Stoppers ". Re­
sis:ances in anode, screen and
grid leads reduce tendency to
spurious oscillation.
as the grid stopper and the capacity (R and Cgk, Fig. 5 /10) form a low
pass filter and attenuate the higher frequencies. Ri and Cgk behave
as a potentiometer in which the reactance of Cgk becomes less with
increase of frequency.
Cgk '..
,
I N PUT
Cgk
-'"
FIG. 5/I o.-Diagram to illustrate formation of low-pass filter by grid
stopper and capacity grid/cathode ; higher frequencies are attenuated.
However, as considerable current flows in the anode circuit, both DC
and AC, a large resistance cannot be tolerated due to the power lost
ther ein, and a value from 5-50 ohms is suitable. When using a pentode
or tetrode 47-470 ohms would be suitable values in the screen ; any­
thing greater would unduly reduce the screen potential.
To be completely effective, stopper resistors should be wired in
circuit directly on to the pin of the valve base concerned.
Whilst continuous parasitic oscillation is fairly easy to detect, the
triggered variety previously mentioned may require a Sherlock Holmes
55
DECOUPLING AND INSTABILITY
technique. The trouble may only arise durin g op eration with a signal
input, usually started off by a transient and rapidly dying away. It
will cause muddiness of reproduction. A transient is a rapid build-up
and dying away of signal and for checking purposes may be produced
by a square wave generator. Comparison of the input and the output
waveform on oscilloscope will reveal any differences. Distortion of
the shape of the waveform will reveal amplifier deficiencies usually of
the OP stage as a whole, whereas the addition of a " tail " of oscillations
of high decrement will reveal parasitic oscillation.
The reason why a transient will produce oscillations is that the shock
excites the resonant circuit of stray capacity and inductance, usually
leakage, the resonance dying away rapidly due to the resistive losses
of the circuit.
A square wave obviously starts with vertical transients, and contains
harmonics up to and above audio frequency limits ; it is therefore the
ideal " yardstick " for the characteristics in question.
CHAPTER 6
PUSH - PULL
AMPLIFICATION
When the need arises for obtaining a great power output there are
several ways of approaching the problem. First one thinks of using a
larger valve with a greater anode dissipation. This is perfectly feasible
as single output valves are capable of producing 20 watts of audio ;
the same result could be achieved by using two valves in parallel as
shown in Fig. 6/ 1, where double the power output range of one valve
would be obtained for the same grid swing.
.----- HT1"
FIG. 6/I.-Valves connected
in parallel to increase power
output capacity.
Both systems would demand an output transformer capable of carrying
the large anode current without saturation of the core. This requires
a generous core which is expensive. Push-pull output probably has
its greatest merits in that saturation of the core by the DC component
is obviated. There are other important advantages ; in fact, the writer
considers the expense and complication of push-pull to be worth
while even for an output of 4 watts because of the low distortion
obtainable. Figure 6/2 shows two valves arranged in a push-pull
circuit.
In the absence of signal input to the grid of V1> the quiescent anode
current of the valves V2 V3 will be equal, or very nearly so, depending
on how closely they are matched. Therefore the anode current in the
two halves of the output transformer primary T2, flowing in opposite
directions, will tend to cancel out the magnetising effect on the core
of the transformer. This means that the transformer core will only need
to provide sufficient primary inductance to give adequate loading at
the lowest frequencies.
At any instant, the ends of the secondary winding of the input
transformer TI will be at opposite potentials with respect to the mid­
point which is connected to the common cathode, so that the grid of
57
PUSH-PULL AMPLIFICATION
one valve is swung positive at the same instant that the grid of the other
is swung negative, hence the anode current of one valve is out of phase
with that of the other. The net effect is that of two valves both deliver­
ing power to the load. A common analogy is that of two men sawing a
HT+
TI
STA G E
DRIVER
T2
LOAD
OUTPUT
:HAGE
FIG. 6/2. Push-pull output stage.
tree trunk with a double-handled saw-one man pushes as the other
pulls, and vice versa, with equal force. (provided they are both working
equally hard !).
In push-pull operation the even harmonics, 2nd, 4th, etc., are
cancelled in the symmetrical anode circuit so that for the same output
the distortion will be less than with parallel operation of the same pair
of valves. This feature is of special importance when triodes are used
as the distortion produced by a triode is very largely 2nd harmonic.
Fig. 6/3 illustrates the grid characteristics of valves V2 and V3 showing
la
l�
la
I
I
I
I
I
�:
I
I
APPLIED
i � IN
la. DECREAS I N G
INCREASING
ANTI
-
I
I
I
SIGNALS
L-!
PHASE ...----T i
FIG. 6/3 .-Grid characteristics of Valves in Push� pull.
PUSH-PULL AMPLIFICATION
that the grids are operated in anti-phase on the straight part of their
characteristic, with mid-point biassing known as the class A condition.
The exciting voltage measured between the two grids will be twice
that required for one valve, or parallel valves, but this presents little
difficulty in practice, and the transformer T I in Fig. 6/2 readily enables
the grids to be fed in anti-phase. Other methods of obtaining the
" phase split " are described in Chapter 9.
The value of anode load for each valve is not the same as that required
for single output working. The precise value is always quoted in the
maker's published characteristics and is a figure arrived at by a careful
computation of conditions
that will give the minimum odd harmonic
.
distortion.
The dominant harmonic with Power Pentodes is the 3rd, and there
is very little reduction of distortion with push-pull operation. If,
however, the load resistance per valve is decreased the effect is to
increase the 2nd harmonic distortion which is cancelled out, and
to decrease the 3rd harmonic, thus improving the overall performance.
Beam Power Tetrodes have considerable 2nd harmonic, but less
3rd and higher order harmonics than pentodes and are thus very
suitable for push-pull operation.
la.
PO I N T OF PROJECTED
CUT
- Eg
FIG. 6!4.-Classes of Bias.
OFF
n
r
l>
(f>
'"
()
So far the description of a push-pull amplifier has been taken to
mean a pair of valves biassed to the mid-point of their grid characteris­
tic. This is referred to as Class A amplification. However the valves
may be iassed slightly in excess of the Class A conditio , normally
to the pomt where the bottom curve is commencing (Fig. 6/4). This
is re erred to as Class AB biassing and the general effect is to step up
consIderabl� the power output available from a pair of valves as
compared WIth the Class A condition, because each valve is contributing
much greater anode current swings in the primary of the output
.
transformer. Reference to FIg. 615 shows that each valve is driven
�
�
�
59
PUSH-PULL AMPLIFICATION
well beyond cut-off on the negative half cycle input, but that the
current-swings to the positive half cycle of input " marry " in the
primary of the output transformer to give a result which approximates
the waveform of the input. The same pair of valves biassed to the
Class AB position may be driven even harder on their grids so that
grid current is produced. This is referred to as the Class AB2 condition.
When no grid current flows it is referred to as the Class AB I condition.
In both the Class AB I and AB2 conditions greater output is obtained
than in the Class A condition, but at the expense of fidelity. A pair
of valves may be worked still harder by biassing them to the point of
projected cut-off shown in Fig. 6/4. This is in effect biassing the valves
almost to the point of cut-off and is known as Class B amplification.
la
C O M B I � E D OUTPUT
O F V1 & V 3
POINT O F
PROJECTED
( UT - O F F
-E
g
B I AS
POINT
V3-,.\
·
OUTPUT OF
·
·
.
.
/
:
.
:
.
.
.
.
� . . ..
U "4 DtSTOA.T E D
I N PUT
FIG. 6/5.-0perating condition for Valves in Class B Push-pull.
Class B operation would never be considered for high fidelity results ;
its use is limited to public address work where high power at low cost
is required. This statement is not intended to imply that PA equipment
is usually designed on these lines. Very high standards of quality are
nowadays provided and, indeed, expected.
CLASS C
A further method of biassing known as Class C involves biassing the
valve to � X cut-off voltage and exciting the grids with a high input
voltage. The " pips " of anode current of high value that will flow can
be used to excite a tuned circuit which will " fill in " due to the fly­
wheel action of such a circuit, but this is of course only applicable to
radio frequency technique.
BALANCE
A problem that often worries the experimenter when dealing with
push-pull circuits is the question of " balance ". We have assumed
60
PUSH-PULL AMPLIFICATION
in this chapter that the pair of push-pull valves were identical in every
way, that the grids were fed with exactly equal voltages, exactly 1 800
out of phase, that the loads for the anodes had exactly the same induc­
tance and DC resistance-now let us come down to the hard facts of
reality.
It is desirable that the pair of valves shall be reasonably alike and
it is always possible to purchase a pair " matched " by the manufacturer.
Matched is put in quotes because it is pretty certain that the valves
even then will not have identical anode currentS. It is impossible to do
better than I mA in 50 mA, but 2 per cent. is nothing to worry about.
The need for matched pairs arises to avoid possible differences of
20 per cent. which may occur when valves are selected at random.
The question of equal input voltages and opposite phase is dealt
with in the chapter on Phase Splitters.
The value of load for the two halves of the transformer should be
fairly closely matched. This is attained by having the same turns!
ratio in the two halves of the transformer. In other words, the number
of turns in the two halves of the primary winding must be identical,
regardless of the effect on the DC resistance of each half. As a coil is
wound, the size increases and the length of wire per layer increases
at a proportionate rate. This may result in a difference of 10 to 25 per
cent. between the resistance of the inner and outer sections. Many
amateurs are unduly worried by such differences.
In a large, expensive transformer, wound with 4, 6 or even more
primary sections (and 3, 5 or more secondaries neatly sandwiched to
reduce leakage inductance) the different resistances are balanced by
suitably " marrying " long and short primary sections. It should
nevertheless be remembered that the main object of section-winding
is to improve the coupling between primary and secondary ; the
equalising of resistance is of secondary importance (in spite of being
in the primary).
In a small or medium size transformer, considerations of space and
cost usually make section-winding an impracticable proposition. It so
happens that the need is less because the leakage inductance is ipso
facto lower in a small winding.
The actual effect of a difference of 10 per cent. in resistance values
works out as follows ;
The load is actually made up of two parts-the DC resistance and
the inductance. The reactance of the inductance XL is equal to
wL = 2 ",fL. Therefore the impedance Z
yR2 + X2L, where R
is the DC resistance. It will be instructive to take a likely value for L
and see what part of the total load R actually is. A triode valve of
ra
1,000 ohms requires a load of about 3,000 ohms, 16H would be a
very generous primary inductance, and say 100 ohms of DC resistance.
XL at 30 cycles would equal 3,014 ohms approx. Thus the DC
=
=
61
PUSH-PULL AMPLIFICATION
resistance is approx. 3l per cent. of the load at 30 cycles, and but
0'33 per cent. at 300 cycles. This should show that small differences
in resistance will have but little effect on the effective load. Due to the
direct component of anode current there will be a voltage drop across
the primary. Taking likely figures, 50 mA for a certain Tetrode's
anode current and 10 per cent. difference in primary resistances,
100 ohms one half and I I O ohms the other, we arrive at the following
result ;
50 X 100
The voltage drop across one
VoItS = 5V ·
1,000
50 X I IO
other
Volts = 5·5V.
"
"
"
"
=
=
1,000
The difference of 0' 5 volt in the anode voltage of one valve will not
affect its anode characteristics and load line in any way and will not
upset the push-pull working.
In all mass produced output transformers as used in the average
radio set, the primary winding is continuous, with tappings brought
out as required. The secondary winding is put on either before or
after the primary. In rather more expensive types, one half of the
primary is wound on, the coil is removed to another machine to receive
the secondary winding, and then returned to the first machine for
completion. These antics have the effect of reducing the leakage
inductance, increasing the cost of production, and intensifying the
difference in resistance between the two primary windings-which
may now amount to 25 per cent. The voltage drop in a typical case
with 50 mA, 200 ohms one half and 250 ohms the other, would be
I CV and 12'5V respectively. The difference in anode volts is still
only 2 · 5V.
From a quality point of view, the reduction of leakage inductance
is more important than the increase of disparity in the resistance
readings.
62
CHAPTER
7
NEGATIVE FEEDBACK
There are two types of negative feedback-current and voltage­
both of which are investigated and explained in this chapter.
Voltage feedback is the form usually employed for the improvement
of amplifier performance, at the expense of available gain. It would
remove ambiguity if the use of the initials " NFB " could be discon­
tinued in favour of " CFB " for current feedback and " VFB " for
voltage feedback.
For the benefit of those readers who are new to the subject, or
possess only a vague idea of its function, a few illustrations of the effect
of NFB (voltage type) on amplifiers and loudspeakers now follow.
So far as the loudspeaker is concerned, it will be observed that 14 db
feedback with uncorrected tetrode output was adequate in levelling
the response of the amplifier reasonably well, and also in removing
the bass resonance of the speaker. This 14 db application of feedback
A
B
c
Cc) 26 db
CA) NFB Zero
CB) 14 db
FIG. 7 !I .-Oscillograrns showing effect of NFB on tetrode response.
reduces the source impedance from 100 to 4 ohms, where the optimum
load is matched to 1 5 ohms. Increasing NFB to 20 db lowers the source
impedance to 2 ohms, and 26 db to I ohm. It is evident that the reduc­
tion of source impedance is becoming more and more difficult. It is
therefore absurd, so far as the loudspeaker is specifically concerned, to
increase the NFB beyond a reasonable point of safety and stability.
The use of a variable resistance for the control of feedback in home­
built amplifiers is an attractive proposition, which would in some cases
NEGATIVE FEEDBACK
lead to the elimination of HF squirting and to prevention of cruelty
to loudspeakers, transformers and valves.
FREQUENCY RESPONSE
Figure 7/1 shows the effect of NFB on the response of uncorrected
tetrodes, as seen by oscilloscope. The glandular swelling is neatly
removed by 14 db feedback.
The 26 db curve is identical with the response of the AF oscillator
as fed into amplifier.
SPEAKER RE SONANCE
No apology need be made for including details of loudspeaker
performance in a book on amplifiers . After all, the improvements
which are achieved in amplifier quality must culminate in improved
speaker performance, otherwise they become abortive ; research
should be co-ordinated as much as possible.
A very simple test for the effect of NFB on speaker resonance is
to measure the volts developed in the voice coil. The following
Table 2 gives results with the tetrode amplifier of Fig. 7/1, already
used in Chapter S, and a typical 8-in. speaker mounted on a small
baffle, with voice coil resistance of 10 ohms .
TAB LE 2
Approx.
Source
Impedance
100 ohms
13 "
10 "
4 "
2 "
I ohm
9'S ohms
Effect of voltage NFB on cone resonance.
8-in. unit on baffle.
Voice coil
Voice coil
Volts at
Volts at
Feedback
cone
SOO cls into
Tetrodes
resonance
speaker
o db
6 "
8 "
14 "
20 "
26 "
Oscillator direct
to speaker
3'S
3'S
3'S
3'S
3'S
3'S
10'0
6·8
6'2
4'S
4'0
3'7S
3'S
6'0
Frequency
of cone
resonance
8S
82
80
77
74
71
cls
"
"
"
"
"
80 "
The addition of the voice coil load to the amplifier reduced the
instability at 26 db mentioned in Chapter S, Table I, but the ISO Kc/s
oscillation pictured in Fig. slI must have been present as an invisible
guest.
For the tests of Table 2 the power was in every case set at 3' S volts
at SOO cls into the loudspeaker. The source impedance of the AF
N.B.
The oscillator output stage was triodes in push-pull without
NFB.
NEGATIVE FEEDBACK
oscillator at this frequency was about 10 ohms. It is interesting to
note that the voltage rise at cone resonance was in this case about the
same as the rise with tetrodes with 8 db feedback, where the internal
impedance is also 10 ohms.
Another interesting disclosure, which the writer had not previously
observed, is the fact that the frequency of the cone resonance goes
down as the intensity of the resonance is reduced by feedback. This
is an obvious advantage ; but whether it is always desirable to absorb
all the bass resonance of a loudspeaker by feedback is another question.
It is conceivable that where the frequency of the resonance is very low
it may be an advantage to retain some of it in order to make up for
losses in other parts of the reproducing system, and for the inefficiency
of small listening rooms at very low frequencies.
Not infrequently, keen listeners complain of lack of bass after
installing an amplifier with a high damping factor, which tends to
make the cone move with constant velocity at low frequencies. A
similar effect may be produced by using a magnet with very high flux
density, but results are always affected by the mass of the cone and
coil. For example, flux density of 1 3,000 lines with a I-in. centre
pole-total flux 54,000 lines-completely damps the cone resonance
of an average 8-in. speaker ; but the same flux density with a I i-in.
centre pole-total flux 145,000 lines-does not succeed in damping
the cone resonance of the average 1 2-in. speaker. A really expensive
17,000 line magnet does the trick here, but some users then complain
oflack of bass and write the makers insinuating they have been swindled !
It should always be remembered that " perfection " in a single link
in a reproducing chain is still impossible, and usually undesirable.
The following oscillograms show the effect of NFB on resonance,
linearity and response in a more interesting and vivid way. The
oscilloscope is far more sensitive in recording loudspeaker performance
than the mechanical stylus used in conventional pressure response curves.
The free-field readings were taken with the loudspeaker mounted
in the wall of the research room, facing into a large field at a height
of some 1 6 ft. above ground level ; they are, therefore, free from any
special characteristics which may be associated with anechoic rooms.
Figure 7/2 illustrates the response of a typical 12-in. speaker used
with the tetrode amplifier, with snapshots of the waveform produced
by the speaker at 60 and 50 cycles per second, under different conditions
of feedback.
The cone resonance at 60-65 cls with zero NFB is very pronounced.
Frequency doubling and trebling are shown as shadows in the trace,
and disappear at about 75 c/s. With 14 db of feedback the bass
resonance disappears, but there is still some distortion at 60 and 50
c/s. The LF waveform is again improved by increasing the NFB
to 26 db, but the shape is still non-sinusoidal at 50 c/s.
It is very important to note here that any steps which are taken to
NEGATIVE FEEDBACK
reduce the amplitude of cone movement at resonance (e.g. reflex
loading) automatically improve the waveform by reducing frequency
doubling. It follows therefore that the difference between the wave-
A
B
c
(A)
Zero NFB
(B) 1 4 db NFB
(C) 26 db NFB
(100 ohms source)
(4 ohms source)
(I ohm source)
FIG. 7/2.-Free-field response of 12-in. unit with corrugate d cone
suspension, plus waveform at 60 and 50 c/s. Input 3'5 volts at 500 c/s.
Mic. 12 ins. on axis.
form at A and B is due to reduced cone movement as well as to improved
quality from NFB, whereas the difference between B and C is due
entirely to the improved amplifier quality from the extra feedback.
(The possibility of distortion from HF instability is being ignored here.)
Before leaving Fig. 7/2, a word must be said about the response in
the region of 1 ,000 c/s upwards. The HF output is reduced with
increase of feedback-as one would expect after seeing Fig. 7/I-but
the rise in the region of 1 ,500 to 3,000 c/s cannot be blamed entirely
on tetrodes, nor can it be entirely removed by NFB. Unfortunately,
it is in the nature of loudspeaker cones to display a range of maximum
efficiency usually covering between one and two octaves in extent :
the smaller the cone, the higher the frequency range of this rise in
output. With a small cone and very light voice coil, it is possible to
cover the range of 5 to 1 5 Kc/s with remarkable efficiency. Incidentally,
these oscillograms, by reproducing the full positive and negative
half-cycle of the sound waves, make the rise in output appear twice
as bad as it really is, compared with the normal sound pressure curves,
so there is no need for readers to consign their 1 2-in. speakers to the
lumber-room as a result of shock from seeing these illustrations.
The next illustration, Fig. 7/3, shows results with the same speaker
as the previous one, but fitted with cloth suspension. It will be observed
that the cone resonance has gone down from 65 to 45 c/s. There is
66
NEGATIVE FEEDBACK
still frequency doubling at 50 cls with zero NFB, but it has largely
disappeared at 60 c/s.
As regards the HF performance, the tendency for soft surrounds
A
B
CA) Zero NFB
CB) 26 db NFB
FIG. 7/3 .-Free-field response of 12-in. unit with cloth suspension.
Conditions as in Fig. 7/2.
to smooth out the sharper peaks of resonance is discernible by com­
parison with Fig. 7/2.
REFLEX LOADING
The third picture in this series, Fig. 7/4, is intended to show the
effect of NFB in reducing resonances associated with reflex loading .
A typical 12-in. speaker with corrugated cone suspension was mounted
A
B
CA) Feedback Zero
CB) 14 db
FIG. 7/4.-Live-room response of 12-in. unit in reflex cabinet, showing
effect of NFB on resonances at 40, 75 and 120 c/s. Tetrode source. Input
3 ' 5 volts into 12 ohms at 500 c/s. Mic. 12 ins. on axis.
NEGATIVE FEEDBACK
in a reflex cabinet with inside dimensions of32 ins. x 15 ins. x 16 ins. and
a port area of 9 ins . X 3 ins . The readings were taken in a room 16 ft. X
15 ft., with the cabinet standing in a corner. The start of room reflections
is clearly shown. (In fact, the use of oscillograms may greatly enhance
the value of live-room readings in assessing speaker performance.)
There is a fundamental cone resonance at 40 cls which is (fortunately?)
hardly affected by NFB, but the main resonance at 75 c(s is virtually
removed, and the next one at 120 cls is rounded off.
Although outside the purpose of the present investigations, it is
interesting to note that the first outbreak of room reflections begins
at about 220 cls, with still stronger effects at 500 c/s. These are related
to the corner position. Moving the cabinet away from the corner
produces strong reflection effects as low as 1 20 c/s.
PRACTICAL APPLICATION
Two references have already been made to actual examples of negative
feedback ; one when describing cathode biassing where it was seen
that the omission of the bypass condenser from across the cathode
resistor reduced the effective input voltage to the amplifier, and the
other when describing the split load type of phase splitter which was
seen to be an exaggerated version of case one. Sundry hints have been
made that NFB would prove to be a panacea for all ills, but whilst
this is not strictly true, it will be seen from this chapter that its dis­
criminating application can be most beneficial in the reduction of
harmonic distortion and in generally improving amplifier performance.
NFB is a vast subject ; it is hoped that the survey given will enable
readers to acquire a fair appreciation of its commoner uses.
1
FiG. 7/s.-Equivalent circuit
of an amplifier with an out­
put of fl times the input
flEg (Ohm's
voltage. 1 =
law) .
r. + R L
CURRENT FEEDBACK
Reference to Fig. 715 shows the now familiar equivalent circuit of
an amplifier. Consideration of this circuit will show that the curre�t
in the circuit will depend on the application of Ohm's law, and that it
E
will be equal to fl g amps.
r. + RL
-
68
NEGATIVE FEEI: BACK
Referring now to Fig. 7/6 with the cathode resistance RK un­
bypassed, if the value of the anode load is decreased, say by connection
of another loudspeaker, then the signal current through the circuit
HT+
I I�
FIG. 7 j6.-Bias resistance
RK un-bypassed giving
CURRENT NFB in an
output stage.
HT-
will increase and therefore the voltage developed across RK will be
greater. This will reduce the effective input volts still further, causing
a decrease in anode current which tends to offset the original rise due
to the reduced value of load RL, i.e. the feedback is tending to make
the valve into a constant current generator.
Now if the current has been maintained constant it would appear
that the total value of resistance in the circuit of Fig. 7/5 has remained
constant. But the load has been reduced by parallel connection of an
additional speaker, therefore the anode resistance (ra) must have been
increased. This is undesirable in an output stage. The advantages of
low anode resistance giving a high value of damping factor have already
been explained.
We can conclude that current NFB is not desirable in the output
stage, and the omission of the cathode bypass condenser is to be
deprecated, although this is often done in the mistaken belief that
some NFB is better than none-regardless of type.
On the other hand, current feedback can often be usefully employed
to raise the input impedance of a valve. An analysis of the arrangement
is made later in this chapter.
VOLTAGE FEEDBACK
An output circuit of this type is shown in Fig. 7/7.
A proportion of the output voltage has been fed back into the
grid/cathode circuit by means of the potentiometer chain C, RI and R2.
The anode being in anti-phase with the grid, will develop a voltage
across R2 opposing the grid voltage Eg, and effectively reducing the
voltage input between grid and cathode. Assuming that ra (the anode
resistance) is high compared with RL (the load resistance), and RL is
increased, say due to a speaker resonance in the case of an output stage,
then the voltage across the load will tend to rise. This, however, will
NEGATIVE FEEDBACK
�
increase the e�dbac V?lts and so tend to reduce the input volts and
cancel the ong1Oal rISe 10 output volts. This means that voltage NFB
tends to make the valve into a constant voltage generator. If the voltage
�
HT+
C
FIG. 7/7.-Amplifier with a
fraction of the output voltage
fed back in anti-phase to the
input.
F E E DBACK
across the circuit of Fig. 715 tends to remain steady, it is as if the total
resistance load on the generator is constant. As RL the load resistance
is known to have increased, the effect of voltage NFB is an apparent
reduction of the anode resistance, and so is a desirable state of affairs
in the output stage, as seen in Chapter 4 on the output stage.
Figure 7/7 is not a practical circuit as it is nearly always necessary
to have one side of the input to an amplifier earthed, and this arrange­
ment would make earthing impracticable. The presence of C is neces­
sary to act as a blocker to the HT voltage, and its value in the feedback
network would have to be chosen with due care to ensure that feedback
did not take place at different values for different frequencies (unless
so desired). If C is small, then more feedback will take place at high
frequencies than at low, allowing bass notes to receive less attenuation,
thus giving a form of tone control which has decidedly useful applica­
tions.
PHASE SHIFT EFFECTS
When thinking of phase relationships in connection with NFB it
should be remembered that by definition NFB is feedback which has
a component out of phase with the input voltages. Ideally the feedback
should be 180° out of phase at all frequencies but the presence of a
reactive component will mean that the phase shift can be something
other than 180°.
Figure 7/8 shows a vectorial diagram. If e is the input voltage, the
output voltage E and thus the feedback voltage BE (where B is the
fraction fed back) will ideally be 1 800 out of phase. If there is a change
of phase due to a reactive component the output voltage may be as
shown at Ell and the feedback voltage BE will be at the angle e with
OE. The effective feedback voltage is therefore reduced to OA)
70
NEGATIVE FEEDBACK
(BEl cos 6). When the angle 6 is 90°, there is no feedback vol!a�e,
and if 6 exceeds 90° there will be an in phase feedback voltage glVlng
positive feedback and instability.
.
A single RCC stage can never cause an undeSirable phase angle
FIG. 7/8.-Vectorial diagram of feedback
voltage less than 1800 out of phase, show­
ing how magnitude of effective feedback
is reduced from OBE to OA. Method of
arriving at actual feedback voltage where a
phase shift other than 1 80° is irlVolved.
rotation of more than 90°, even at the extreme limits of frequency
where there is the greatest possibility of phase shift. Therefore
feedback over a single stage is safe from the risk of instability.
If feedback is applied over two or more RCC stages it is possible for
regeneration to occur at the extreme ends of the frequency scale,
although by careful design stability can be achieved up to three or even
four stages if the degree of feedback is not too high.
e
INPUT
VOLTAGE
:9
E FFECTI V E
INPUT
l
FIG.
t
AM PLI F I E R
W I TH G A I N
l RI
OF
r--
A T I M ES
FEEDBACK 'I/�LTAGE
BE
R2
E
OUTPUT
VOLTAGE
I
7/9.-Block schematic of voltage feedback circuit.
Figure 7/9 shows a block schematic for NFB of the voltage type. A
potential divider, RI and R2, across the output of the amplifier feeds
back a proportion of the output voltage in opposition to the input
voltage e, leaving a net input of eg to the grid of the first valve.
The fraction of voltage fed back, called B, is equal to
RI
RI + R2
. If no
voltage is fed back, that is e=eg, then the basic gain of the amplifier,
called A, will be the ratio of the input and output voltages, that is
A=�
.
eg
If feedback is now applied, the input volts e will have to
be raised to obtain the same output voltage E, and the new gain of the
E
amplifier, AI> will equal ; this value for Al will be less than A.
e
7I
NEGATIVE FEEDBACK
Now e is equal to the sum of the input volts to the grid circuit, eg,
and the feedback volts, BE, but as the feedback factor B is negative
for negative feedback, e = eg - BE. Thus Al may also be expressed
E
as
eg - BE.
The Gain Reduction Factor due to feedback can be expressed as the
ratio of the old gain without feedback to the new gain with feedback,
E
B
_
E
A
E /eg
g
I_
that is : �, but A= and AI=
.'.
Al
eg
Eg - BE
Al e /eg - BE
eg
= I - BA.
The block schematic of Fig. 7 /9 forms the basis for a practical circuit
shown in Fig. 7 /10. This leaves room for experiment on an existing
amplifier.
---
HT+
FIG. 7/ 10.-Typical NFB circuit (with potentiometer control)
over three stages.
The feedback is taken from the secondary of the output transformer.
One side of this is earthed and the feedback network comprises
RI and R2. R2 is a potentiometer of, say, 200 ohms, and RI is 800 ohms.
With the slider of R2 at the end nearest the cathode of VI' 20 per cent.
feedback would be obtainable, variable down to zero at will.
If, on switching on, the amplifier gives a violent moan, one can
conclude that positive feedback has been inadvertently applied. It
should be switched off hurriedly, and the connections to the secondary
of the output transformer reversed.
It is always advisable to apply a minimum of feedback when first
checking for sense of feedback.
One should not be surprised if an amplifier of three stages as shown,
or even two stages, to which the experiment can equally well be applied,
oscillates when a large measure of feedback is applied. To check if
72
NEGATIVE FEEDBACK
the feedback circuit is working, the connection to the slider of R2
can be broken and a rise in gain should be observed.
EXAMPLE : Taking a practical example, if 20 per cent. feedback
E x 20
or
is employed, the fractional feedback voltage B will be
100
0·2 E. Assuming the amplifier without feedback had a gain A of 20
20
A
times, then the new gain Al will be
I - BA I - ( - 0·2 X 20)
20
= - = 4·
1+4
Another way of expressing the same thing, which is often employed,
is to say that the gain with feedback is equal to the gain without feed20
back divided by I + AB. Using the figures above : Al
I + (20 X ·2)
20
= - = 4.
This latter expression is more convenient if not strictly
5
accurate, as the fact that B is negative in the first expression can, and
often does, lead to errors in calculation !
CONCLUSIONS TO DATE
(I) Voltage feedback lowers the output impedance of a valve.
(2) As a result of (I), better damping of the loudspeaker resonances
is obtained.
+
(3) Voltage feedback re
ces the gain of an amplifier.
(4) As the gain is reduced, so is the available output ; this can be
counterbalanced to some extent by providing a bigger signal
input to the amplifier, but this may necessitate a further stage of
amplification, with more distortion.
(5) Voltage feedback can cause instability in an amplifier due to
phase shifts at extremes of frequency producing positive feed­
back. These phase shifts can arise from two main sources:
(a) An insufficiently sectionalised output transformer with a high
leakage inductance producing phase shift at high frequencies.
This applies most particularly when the feedback is taken
from the secondary winding, which is usually desirable to
correct for distortion arising in the transformer.
(b) Inadequate size of interstage coupling condenser causing phase
angle rotation at the lower frequencies.
73
NEGATIVE FEEDBACK
OTHER EFFECTS OF NFB
The next sections deal with other effects of NFB, and may well be
skipped on a first reading, as they are summarised at the end of the
chapter.
THE EFFECT OF VOLTAGE FEEDBACK
ON FREQUENCY RESPONSE
If an amplifier has different gains at various frequencies due to
deficiencies of one kind or another, the application of feedback will
tend to make the gains at these frequencies more nearly equal.
Let A= gain of amplifier at one frequency.
Let X= gain of amplifier at another frequency.
The ratio of gains without feedback will be
A
With feedback the new gains will be : A1 - I + BA
j
�.
X
X1 - I + BX
and the ratio of gains with feedback will be :
Al
A
_X
I + BX
I + BA
Xl
A
I + BA
X
_
_
I + BX
I + BX
A
=
X
'
X
I + BA
A
--
This equals the ratio of gains without feedback multiplied by the
1 + BX
'factor
,
I + BA
Assuming the gain A was 20 and the gain X was 25, the new ratio
of gains will be with 20 per cent. feedback :
20
1 + ('2
X
25
1 + ('2
20
- _
25
X
1 +5
1+4
X
X
20
25
25)
20)
X
6
5
-
120
125
= -.
The ratio of 120 : 125=96 per cent, is considerably better than
1 20 : 1 50=80 per cent., thus demonstrating that voltage NFB tends
to even out the gain of an amplifier at all frequencies .
74
NEGATIVE FEEDBACK
THE EFFECT OF VOLTAGE FEEDBACK
ON HARMONIC DI STORTION
The application of voltage NFB will assist in reducing harmonic
distortion generated within the amplifying stage itself, and a proof,
not entirely rigorous, but giving a close approximation for practical
purroses is appended.
Let D= distortion voltage in the output without feedback.
Let Y
=
distortion voltage in the output with feedback.
Now the distortion voltage fed back to the input will be BY, but at
the input there is no component at this frequency, so no cancellation
can take place, and the fed back voltage is amplified, giving an output
of ABY. This amplified distortion is out of phase with the original
distortion voltage and the resultant voltage will be :
Y = D - ABY
. . . Y + ABY
=
. . . Y (1 + BA)
D
=
D .. . Y=
D
.
I + BA
From this it can be seen that the harmonic distortion is reduced by
a factor approximating to the gain reduction factor.
THE EFFECT OF V OLTAGE FEEDBACK
ON AMPLIFIER NOISE
The inevitable circuit noise generated in an amplifier will tend to
be reduced in the same way that harmonic and frequency distortion
is " ironed " out by feedback, but what one gains on this particular
swing one may lose on the roundabout of an extra stage to make up
for the gain lost by feedback. There is little advantage in using feed­
back on the early stages of amplifiers handling very small voltages
where distortion is not likely to creep in.
THE EFFECT OF CURRENT FEEDBACK
ON INPUT IM PEDANCE
Figure 7/I I shows a valve amplifier in which a resistance RI has been
inserted in the cathode circuit. The alternating anode current will
have to flow through RI and a voltage will be developed across it in
opposition to the input voltage, i.e. current NFB.
Now the input impedance, Z, will equal
� and in the absence of Rl
I
would equal Rg, the grid leak, neglecting inter-electrode capacitances
and Miller effect for the moment. When feedback is applied the effec75
NEGATIVE FEEDBACK
tive voltage appearing across Rg will be lessened and I will fall. If I
is reduced then the value of Z in the expression Z =
LOAD
H T+
� will increase.
FIG. 7/1 I .-Useful current
feedback circuit ; the pres­
ence of Rl will give rise to
current feedback and one
effect is to increase the input
impedance.
E
HT-
Practical figures may serve to clinch the argument.
Assume E, the input = 1 volt, Rg, the grid leak = 0'25 meghom,
I X 1 06
micro-amps= 4 [LA. With feedthen without feedback 1=
0'25 X 106
back the effective input voltage across Rg (eg), might well be reduced to
0'25 X 106
I [LA, and now the input impedance=
0'25 volt, then I
0'25 X 106
1
- --6 = I megohm.
I X 10
It will be seen that with an increase of feedback, the voltage across
the grid leak will be made smaller, and the input impedance will rise
in proportion, with, of course, increasingly reduced stage gain.
When the Miller Effect was discussed, it was seen that the input
capacitance to a valve was much greater than might be expected,
thereby reducing the input impedance so that the effect of current
feedback will largely wipe out Miller effect to great advantage, particu­
larly on the input circuits for crystal pick-ups and microphones.
In the circuit of Fig. 7/ I I , the omission of the cathode resistor bypass
condenser will give the same effect as the inclusion of Rl> but if the
value of resistance needed to get adequate feedback is greater than is
required for biassing purposes, then RI can be included and CK
omitted in addition.
=
FINAL SUMMARY OF EFFECTS OF NFB
(1) A reduction of harmonic distortion.
(2) Improved linearity of frequency response.
NEGATIVE FEEDBACK
(3)
(4)
(5)
(6)
A reduction of noise.
A reduction of gain.
A modification of internal resistance of an amplifier.
Greater stability with changing supply voltages, ageing valves,
and difference between individual valves.
(7) A modification of input resistance to an amplifier.
Some not so obvious practical implications of these effects are to
be noted as being of considerable importance.
Possibly one of the biggest problems in the mind of the constructor
is concerned with speaker matching ; how far can one mis-match
before trouble starts ?
Referring back to the section on voltage feedback it was found that
the feedback tended to offset any change in the load resistance of the
valve. The load resistance in the case of an output valve is the loud­
speaker, and changes on the secondary of the transformer are reflected
back to the valve's anode circuit. It is not suggested that the reader
would be so casual as to connect a 3 ohms speaker to a transformer
wound to match a 1 5 ohms speaker*, but if this experiment is carried
out, using an amplifier with a heavy degree of feedback, it is very
surprising how tolerant the amplifier has become. The main effect
will be a restriction of power output available, and a loss of bass, due
to the valve being underloaded. If a 1 5 ohms speaker is connected to
a 3 ohms secondary, there will be loss of power, without the attenuation
of bass . Another aspect of the same problem would be the use of a
transformer ratio of, say, 26 : I with valves and loudspeaker calling
for a ratio of 20 : I . With NFB, such liberties may be taken with
impunity.
Heavy NFB is used in commercial amplifiers, say in a school, where
the load may vary from one speaker, demanding 3 watts, to eight
demanding 24 watts, and almost perfect regulation is achieved. Listen­
ing at the first speaker to be switched on, no change in level is percept­
ible when the other seven are switched in circuit, and the quality
remains remarkably uniform.
METHODS TO COUNTERACT INSTABILITY
In-phase feedback may occur in audio amplifiers over which a large
amount ofNFB has been applied. Phase shifts at extremes offrequency
in various parts of the circuit, mainly the output transformer, produce
instability either continuously or on peaks and transients only, but
always disagreeably.
Instability usually occurs in a feedback amplifier at a high audio or
supersonic frequency. The response of a quite usual amplifier is
*1 often do this.-G.A.B.
77
NEGATIVE FEEDBACK
fairly fiat up to some high frequency and then drops off sharply. This
rapid fall ciff means that the phase shift runs rapidly up to 1 800 before
the amplitude has dropped appreciably, and instability results. The
old dodge of applying condensers across anode loads of RC coupled
stages may work by reducing the amplification rapidly but is not
nee:essarily a cure. What is really required is a stabilising device that
will provide attenuation without phase shift. The arrangement in
Fig. 7/12 will practically meet the requirement because the " tailing
off " of the high frequency response is gradual and once the phase
H T -t
33K
FIG. 7/12.-Modification of
anode load giving a gradual
fall of amplification with rise
of frequency, thus avoiding
large phase shift. Values are
those of Williamson amplifier.
20 0 p F
471<.
47 K
VI
. 6J 5
shift peak is passed the phase displacement remains small up to several
hundred Kc/s and so the danger of high frequency parasitics is lessened.
This circuit is used in the Williamson amplifier.
FIG. 7 / 1 3 .-NFB circuit in
which feedback increases with
rise of frequency due to
shunting of R2 by C.
- - - - -+----r--
-
-
-
-----'
The arrangement of Fig. 7/ 13 is particularly valuable for obtaining a
high stability margin under conditions of wide changes in load, and
for applying to awkward cases. The resistor RK, the cathode resistor
of an early valve, has feedback applied across it from the secondary
of the output transformer by means of the network Ri> R2 and C.
At low frequencies the reactance of C is made great and the feedback
is proportional to RK/RK + RI + R2• At the higher frequencies the
presence of C corrects for the phase shift introduced by the leakage
inductance and self capacity of the output transformer.
Yet another method of cancelling phase shift effects consists of
shunting the two halves of the primary of the output transformer by
condensers as in Fig. 7/14.
NEGATIVE FEEDBACK
The capacity tends to reduce gain with increase of frequency so
that as the critical frequency is approached where leakage reactance
of the output transformer is liable to cause instability, the gain has
fallen to such a degree that there is not enough positive feedback to
H T ..../"
FIG. 7 / 1 4.-Method of stabilising amplifier with heavy NFB, as em­
ployed in Garner amplifier described at end of book.
cause trouble anyway. The inclusion of resistors is to damp out
possibility of forced oscillations in the primary circuit. To translate
into the sordid terms of £ s. d. this " economy " operation of extracting
bad teeth enables the user to work with an 8-section OP transformer ;
the RC network used on the primary costs but a fraction of the higher
price involved in winding a transformer with 16 sections.
SE LECTIVE NEGATIVE FEEDBACK FOR
L OW LEVE L LISTENING
It is sometimes considered advisable for the degree of negative
feedback to vary the gain of an amplifier differentially (a) at different
volume control settings ; (b) according to the frequencies being
reproduced by the amplifier.
It is quite possible to arrange for either bass boost or cut and/or
treble boost or cut by including frequency discriminating networks
as part of the negative feedback loop. These may consist of either
inductances or capacitors, or sometimes a combination of the two.
Similarly it is easy enough to arrange for the volume control to act
additionally as a potentiometer across the injection points for negative
feedback and thereby to vary the amount of applied feedback.
We will consider both these conditions when applied to the simple
amplifier shown in Fig. 7/15.
The amplifier is designed to provide 3 · 3 watts maximum into the
15 ohms load of the speaker. Therefore, approximately 7 volts RMS
will be developed across this load. The output valve requires 5 volts
79
NEGATIVE FEEDBACK
RMS at its grid, and since Vi has a gain of 10 times it is obvious that
a signal of ·5 volt RMS is required between grid and cathode at the
Rs
II
·IM
�
INPUT
Gain
+ 250v
+
.'
z
0
....
:: 7",
R .., s
a (. f O � S
LS
Inp"-'�-JoMIMr-+
5v RMS r
)( 10
Gives
TREBLE r···········
BooST :
IOOIl
e T ::mm
(�j riooot
eT)
Appmx
:
o 8 uF :
15 n
LoS. ]'lw.
fOf 3-3..
O.Mpu�
F8R
2
T
1•••• _
Circuit by F. //. Beaumo1lC
FIG. 7/IS.-NFB circuit giving higher proportion of extreme bass and
treble as the volume level is reduced. Independent control of treble and
bass can be included. (See text.)
input. Taking the case of the volume control being adjusted for
maximum input and with the feedback disconnected, · 6 volt input
will be needed however, since ·1 volt is dropped across the series
resistance Rs. Introducing feedback so that the gain of the amplifier
is halved with the volume control at maximum, it will be seen from the
values given on the circuit that six times this amount of feedback
must be applied at the earthy end of the volume control. In other
words, if the output were fully maintained at such a control setting
3 volts RMS must appear as a result of feedback at the point Z. Since
we have 7 volts across the speaker, a divider consisting of FBR! and
FBR2 of the values given on the diagram would meet the require­
ments. At minimum volume control setting, a signal of 3·5 volts
RMS would now be needed at the grid of Vl so that negative feedback
has reduced the gain 7 times or 17 db. At the maximum volume setting,
however, gain reduction is only 2 times or 5 db, and intermediate
settings of the volume control will vary the amplifier gain proportion­
ately.
Now it is permissible to introduce counter distortion at lower volume
settings to correct for the non-linearity of sound perception of the
human ear when plotted against frequency. This means that we
want more bass and more treble and less middle at low volume settings.
Dealing with the bass boost required if we insert a capacitor in series
with FBRl of such a value that the impedance in this branch of the
feedback network totals 6 times FBRl alone at, say, 50 cjs we shall
obtain a bass boost at that frequency of I I db. Similarly, if we connect
a capacitor across FBR2 of such a value that the impedance between
Z and earth is only one-sixth of the value of FBR2 at, say, 10,000 cjs
we shall have a treble boost of I I db. Thus, with both capacitors
80
NEGATIVE FEEDBACK
connected, bass and treble boost result. The amount of boost becomes
relatively larger the nearer the volume control is set towards minimum.
A ready means of correcting for aural characteristics is therefore to
hand. The amount of boost permitted by this method is restricted by
considerations of stability, since the two capacitors naturally introduce
a phase shift. Care must be taken with the output transformer design,
and in general such selective feedback is only permissible over one or
two stages.
If it is desired to vary the amount of boost introduced, a variable
resistance of say 1,000 ohms connected across the bass boost capacitor
CB and another variable resistor of about 2,000 ohms in series with
the treble boost capacitor CT would allow adjustment of either.
On the other hand, bass cut could be accomplished by inserting a
capacitor in series with FBR2 and treble cut by connecting a capacitor
across FBR1 .
NOVICE'S CORNER
Our imaginary correspondent writes : " Can you elucidate the
following points? Please reply in simple language as I have forgotten
all my algebra and I can only multiply up to 10 times."
No. 1 .-1 have an amplifier with pentode output and a medium-price
output transformer with leakage inductance in excess of ·IH.
I should like to apply negative feedback to improve quality. I am
prepared to accept a reduction to I watt output. 15 ohms speaker.
How do I proceed ?
Answer.-As the output valve is a pentode, the distortion is probably
largely odd harmonic. The best method would be to apply
feedback from the secondary of the output transformer and feed
it into the cathode circuit of the previous valve. A 200 ohms
potentiometer would be inserted between cathode resistor and
earth, the slider being connected to a 1 ,000 ohms resistor-thence
to transformer secondary. If excessive feedback is attempted,
the high leakage inductance of the OP transformer will result in
instability, either audible or supersonic. If the loss of gain is too
severe for available input, the feedback loop could be taken to the
bottom end of the grid leak of the output valve, thus limiting its
operation to one stage, and incidentally reversing the sense of the
required connection to the output secondary winding.
No. 2.-1 have an amplifier similar to No. I but fitted with power
valve instead of pentode. What must I do for the same results ?
Answer.-The triode valve, providing it is reasonably well matched
to its load, will not produce much distortion inherently, and the
application of NFB will give little advantage. The low anode
resistance of the triode is already giving a measure of loud­
speaker damping, and whilst NFB will provide increased damping
81
NEGATIVE FEEDBACK
it is a moot point whether any real advantage will accrue. If
distortion is present it is probably in the driving stage, a triode
needing a large drive voltage. (The pentode and tetrode output
valves score here, as they need less driving, and with NFB can be
arranged to give plenty of damping to the loudspeaker.) To
deal with distortion, check that the penultimate stage is supplied
with adequate HT voltage and is correctly biassed.
No. 3.-1 have an amplifier with push-pull output of about 10 watts,
which I am prepared to drop to 2 watts by NFB if there is any
benefit. What do you advise ?
Answer.-If the push-pull amplifier is giving satisfactory results,
leave it alone. If distortion is present, find out where it comes in.
Suspect :
Gross mis-match between valves and load.
Unmatched pair of output valves.
Poor quality output transformer.
Phase splitter grossly out of balance.
Ageing valves or low HT voltage.
Wrong bias conditions.
Short on bias resistor due to electrolytic bypass failure.
Load resistor or screen feed resistor to early valves changed
in value.
NFB is not a patent purge to eliminate distortion ; it is asking
for trouble to apply it to a basically unsatisfactory amplifier. It
will however improve frequency response and L S damping,
and could be applied over penultimate stage as shown in Fig. 7/16,
or to previous valve as described in Answer No. I .
22K
SO l<
"
HT
22K
ll ll
OP
III_JJ
·2 5
1
M
*
x
25K �+-------'1�
VALUE DEPENDS ON VALVE
*
I NT E R- R E LATED
FIG. 7/16.-PP Amplifier with NFB applied over penultimate stage,
with less reduction of gain than if fed into cathode of V I .
82
NEGATIVE FEEDBACK
No. 4.-lf my output conditions are matched to a 3 ohms speaker
instead of 1 5 ohms, how would this affect the NFB arrangements ?
Answer.-The voltage appearing across the secondary of the output
transformer will be less with the 3 ohms unit. For 3 watts the
voltage is 3V across a 3 ohms voice coil, but 6'7 volts appear
across a 1 5 ohms coil. Therefore, the ratio of the feedback
potentiometer arms will be lower for the 3 ohms secondary to
obtain an equivalent feedback voltage.
No. 5 .-1 have an amplifier with NFB which is used with a 15 ohms
speaker. If I connect another speaker in parallel and reduce the
load to 7t ohms what effect does this have on the NFB voltage ?
Does it increase the NFB or reduce it ?
Answer.-When the second loudspeaker is connected, the voltage
across the two will be reduced, but this will reduce the feedback,
so the gain of the amplifier will rise and a compensating effect
will take place. The output therefore tends to remain level and
voltage feedback tends to produce a constant voltage generator.
This accounts for the fact that NFB helps to overcome a mis­
mateh in the output transformer or speaker load.
No. 6.-1 should like to know a simple way of estimating the amount
of NFB which is being applied in one or two typical cases.
Answer.-As the premise is that 10 X table is the highest one we know,
it is rather difficult to answer this question. An approach in
terms of input sensitivity for equal outputs seems to present the
fewest mathematical obstacles. If an input of 0'25 volt will give
an output X without feedback, and I ' 5 volts are needed to give
the same output X, then we have divided the gain by 6. This is
clearly a voltage ratio of 6 : I , which is approximately 15'5 db.
(The db/ratio can be checked by reference to the table in the
supplement.) Also, in the case in question, we must have fed
back 1 '25 volts .
Looked at from the other end, let us assume we produce 1 5 volts
acress a 1 5 ohms load. The power is now 1 5 watts. Feedback is
applied and the output volts fall to one-sixth of the former value,
i.e. 21 volts, which is equivalent to 0'416 watt. This is a power
ratio of 1 5 : 0'416, say 36 : I which is again 1 5' 5 db approx.
If it is desired to restore the output to 1 5 watts without re­
ducing NFB, this can be done by increasing the input volts by
6 times, and the grid/cathode circuit of the first valve still only
receives its original net input and all the valves in the amplifier
proper are only called upon to handle the same voltage excursions
as without feedback. Also, the power-handling capacity of the
output stage is not destroyed in any way. This is one advantage
of overall feedback to an early stage, which more than offsets
phase shift difficulties that may creep in. When feedback is
NEGATIVE FEEDBACK
applied to fewer stages or only one stage, there is always the
danger of overloading if the input level is increased to counter­
balance the feedback losses.
For reading voltages as outlined in this answer, a valve volt­
meter on the input and ordinary AC meter across a dummy
resistive load on the output are quite satisfactory.
CHAPTER 8
THE
CATHODE
FOLLOWER
Although the use of a valve in a cathode follower circuit actually
converts it into a de-amplifier, it has so many useful and interesting
aspects that the system cannot be ignored in a book about Amplifiers.
BASIC CIRCUIT
Inspection of the basic circuit shows that the load is connected in the
cathode circuit instead of the usual anode circuit, and the output is
taken from cathode and earth. The effect is to reduce the stage gain
,----- H T �
I N P UT
LOAD RESISTANCE
ou
puT
FIG. 8/L-Basic Cathode Fol­
lower circuit in which the
output is in opposition to the
input, giving a " gain " of
less than unity.
- H TL--...... --�-
to a value slightly less than unity, for the total output voltage appears
across the cathode resistor and opposes the input voltage, thus giving
100 per cent. feedback.
The salient features of such an arrangement are to give a high input
impedance and a low output impedance.
STAGE GAIN
The stage gain can be calculated from the usual formula
A
where A = gain of amplifier without feedback, and
Al =
I + BA
B = the fractional voltage feedback. Since B in this case is unity, the
A
expression becomes Al = __, but as A is always larger than unity,
I + A
20 _ 20 20 Al =
the gain is slightly less than unity : e.g. A
1 + 20
21
0'95 . This means that the input voltage must always be slightly higher
than the output voltage required ; in fact, obtaining sufficient grid
drive for the stage often constitutes a difficulty.
=
THEORY OF OPERATION
This particular circuit arrangement has caused more obscure
THE CATHODE FOLLOWER
" explanations " of how it works than any other circuit the writer has
ever encountered, but he considers that the best approach to the prob­
lem is to observe the facts and then explain them.
The amplifier at first glance appears to be very similar to that in
which the current feedback was obtained by omitting the cathode
bypass condenser and this can quite well be the case as the stage has a
high input impedance.
I VOLT
r----;-- HT+
I
__.L
,:
FIG. 8/2.-Voltages appearing
across inter-electrode capaci­
ties in a cathode follower.
INPUT
I VOLT
r.v
O·95V
I
This can also be argued by reference to Fig. 8/2 in which the actual
voltages applied across the inter-electrode capacitances are shown. If
the grid is made 1 volt less negative, i.e. 1 volt positive, the cathode
follows it to nearly 1 volt positive (hence the name cathode follower).
The voltage across Cgc (capacity grid-cathode) will now be o·oSV.
This will certainly cause less current to flow than the 1 volt that would
appear across it in an ordinary amplifier as shown in Fig. 8/3.
,...--- HT+
RL
FIG. 8/3 .-High value of
voltage appears across Cga in
normal amplifier, cf. Fig. 8/2.
I VOLT
�--�---- HT-
Now the voltage appearing across Cga (capacity grid-anode) is only
1 volt in the cathode follower, but in Fig. 8/3 it would be eg X (1 + A)
volts . In an amplifier with a stage gain of 20 times, this would amount
to 2 1 volts for 1 volt input. Furthermore, much more current would
flow. The cathode follower behaves like a Miller effect in reverse,
and has the highest input impedance obtainable in any circuit with a
particular valve.
Taking the cathode follower's second characteristic, that of a low
86
THE CATHODE FOLLOWER
output impedance, this is not a characteristic of a current NFB arrange­
ment, but rather that of a voltage feedback arrangement. It is argued
that the feedback is the voltage across the load, the whole of it, but it
cannot be both kinds of feedback at once, yet it certainly appears to
be so.
CATHODE FOLLOWER OUTPUT STAGE
Take a case as shown in Fig. 8/4 in which the cathode follower is
used as an output stage. The load in the cathode as presented by the
r------ HT+
FIG. 8/4.-Cathode Follower
as an output stage driving a
loudspeaker.
L---��---- HT-
loudspeaker via the transfonner is the optimum load quoted for the
particular valve employed.
If the load rises, as it will do at the frequency of the bass resonance
of the loudspeaker, then the voltage across the load will increase, giving
rise to greater feedback, thus tending to cancel the increase in output.
Conversely, connect another loudspeaker in parallel with the original
one and the load impedance will fall ; this gives less feedback and the
gain rises. These are all the characteristics of a low impedance generator
and are admirable for an output stage.
The argument of the output stage still goes on : whether to use the
cathode follower, a low impedance triode, or a pentode or tetrode with
NFB, but the general tendency seems to be to use the pentode or
tetrode with NFB giving adequate loudspeaker damping. The extra
damping factor conferred by cathode follower output is considered to
give no appreciable improvement in speaker perfonnance. It is possible
to have too much of a good thing and give a loudspeaker excessive
damping.
It is certainly much easier to provide the drive for an output stage
of a more normal type. For instance, a PX25 triode would require a
grid drive of approximately 230 volts to give 6 watts output as a cathode
follower, but with an output resistance of 120 ohms. The nonnal grid
drive is 33 volts. It is a big problem to provide seven times as much,
demanding a voltage amplifier with a very large value of anode load
and HT supply to match. An inter-valve transfonner might be used
to give a voltage step-up, but the shunt capacities would take their toll.
THE CATHODE FOLLOWER
PRACTICAL DESIGN
If the reader desires to experiment, the circuit of Fig. 8/5 could be
used. This has given very satisfactory results, such as one only expects
from a more ambitious output stage.
R - 450 n
I
C l " 0 · 2 5 "u F
R2- 1 0 0 K
C 2= 8 "u F
Rs-
C3- 25).! F IOYW
25 K
R4= 500 K
Vt lO J 7 O R
S I M I LAR
V2- &Y6
C4= O ' I,u F
R5- 250K
R �- I O O n
FIG. 8/ 5 .-Practical amplifier circuit for a cathode follower output stage.
A 6V6 Tetrode is used in the output stage, but will of necessity
become a triode with an expected output of 0·8 watt. This valve
works into a load of approximately 3,000 ohms, through a suitable
matching transformer, ratio 14 : I with 1 5 ohms speaker, or 32 : I
with 3 ohms LS. The matching is by no means critical due to the low
internal resistance of output. To develop 0·8 watt across 3,000 ohms
49
volts 54 volts
requires about 49 volts ; this means that an input of
0'9
is needed.
A 6J7 pentode as an RCC amplifier is capable of giving 81 volts
peak across a following grid leak of 0'25 megohm, the recommended
value for a 6V6, using a 300 volts HT supply. The anode load of the
6J7 should be lOoK, screen resistor o' 5 megohm, cathode resistor
450 ohms, screen bypass conveniently 0'25 mfd, cathode bypass 25
mfd. The coupling condensor due to the high input impedance of the
output stage need not be large, 0'1 mfd is adequate. The 6J7 will give
a stage gain of 82 times, thus a I volt input will fully load the output.
I volt is easily obtained from a radio feeder, but if a pick-up is used a
low gain triode might be employed ahead of VI' The bias voltage for
the 6V6 is in the order of 15 volts, thus with a space current of 50 mA
the DC resistance in the cathode should be H X 1,000 ohms
300
ohms. If the primary of the output transformer is less than 300 ohms
then it should be made up to 300 ohms as shown in Fig. 8/6A.
If the primary of the output transformer has a resistance greater than
300 ohms, then the grid is returned to a potential divider across the
=
=
88
THE CATHODE FOLLOWER
HT supply as shown in Fig. 8/6B, thus raising the grid to a positive
potential equal to the excessive bias produced in the cathode circuit.
.-----_r_--- H T +
-L:"------1-- H T+
INPUT Rg
g
R
INPUT
R
CB
B
A
A.
B.
FIG. 8/6.
Method of obtaining more bias for a cathode follower output stage.
Method of cancelling excessive bias voltage developed across trans­
former primary.
For example, transformer primary 400 ohms, therefore excess bias
(400 - 300) X _J�= 5 volts. A suitable divider across the HT would
1,000
be one in which the elements are in the ratio of 295 : 5 e.g. 295,000
ohms : 5,000 ohms. A low 300K would be suitable.
The bypass condenser of Fig. 8/6A makes no audible difference if
R is quite small, say less than 100 ohms, which it normally would be,
but the omission of a bypass condenser in Fig. 8J6B produces mains
hum.
COMPARABLE TETRODE OUTPUT
It is very instructive to compare the cathode follower amplifier with
the one shown in Fig. 8/7, where the 6V6 is used as a tetrode with
NFB. The loading is 5,000 ohms, bias
15V, input say 12 volts.
12V
The 6J7 stage gain is 82, therefore the input required is
0· 15V
82
without feedback.
Assuming 3 watts output across 1 5 ohms voice coil, we have 6·7 volts.
With VRI at 1 50 ohms in feedback loop, the voltage feedback will be
6·7
I ·675 volts. Therefore the input voltage for full output becomes
4
1 . 825 volts (1 ·675 + 0·1 5V). This degree of feedback may cause
instability due to phase shift in output transformer, but very pleasant
results can be achieved with only 20-30 ohms of VRI in the feedback
loop.
-
=
�
=
89
THE CATHODE FOLLOWER
�_47_K� 118M�_=_1
r-
__
__
'47
M
2·2
M
lOOK
6V6
':2
I<.
150
n
?;�-:VR.I
450
n
-,
__
FIG. 8/7.-Comparable amplifier to that of Fig. 8/5, but arranged for
tetrode output with variable voltage NFB overall.
CONCLUSIONS-OUTPUT STAGE
The use of the cathode follower in the output stage is judged to be
hardly worth while, but a trial of the system may intere;st the experi­
menter. The loudspeaker can be over damped, and the NFB is limited
to the output stage, leaving preceding stages to produce distortion
unless they have NFB overall.
INPUT STAGES
1. Using crystal microphones or pick-ups, the cathode follower with
its high input impedance is excellent as the first stage, in spite of
contributing less than no gain.
2. If a microphone or pick-up of high impedance is operated remotely
from the main amplifier, a long length of screened cable will intro­
duce excessive top cut due to its self capacity. A cathode follower
between pick-up and line will avoid this attenuation, and provide
a form of line matching which obviates the use of a step-down
transformer with its inherent distortion and unerring instinct for
finding any stray induction fields and producing hum.
3. As a device for mixing two or more inputs into an amplifier, a
cathode follower in each input channel provides ideal isolation and
great ease of mixing, using low resistance wire-wound potentio­
meters as cathode loads.
90
CHAPTER 9
PHASE
SPLITTERS
When discussing push-pull output stages in Chapter 6, it was seen
that the two grids of the output valves must be fed in anti-phas€ ;
therefore a voltage is required balanced to earth. This was obtained
in Fig. 6/2 by the use of a transformer, but a transformer as phase
splitter is not regarded with much favour in modern high fidelity
apparatus, due to its tendency to attenuate the upper frequencies by
shunt capacities. There is also the difficulty of maintaining adequate
primary inductance, although this problem is eased by shunt feeding
the primary, as shown in Fig. 3/7.
Other drawbacks are the high cost of a good component, and the
possibility of hum pick-up due to interaction with mains transformers
and smoothing chokes . A single valve may be employed as a phase
splitter, and, whilst it contributes little to amplification in some cases,
it is an elegant way of obtaining a phase split and is favoured both on
the basis of cost and efficiency.
When the output stages are driven into grid current, as in the AB2
and B2 modes of high power operation, it is essential to have a low value
of resistance in the grid circuit, otherwise the flow of grid current will
develop a voltage across the resistance and completely upset the biassing
condition. This requirement can only be satisfied by using a trans­
former with a suitable low-resistance secondary. To assist in this
direction, a step-down transformer is usually employed. The use of a
nickel/iron alloy core is precluded in this application due to the high
level of operation. Quite a large transformer will be required presenting
no small design problem. As we are more interested in good quality
than maximum noise, the RCC types of phase splitters have the
strongest appeal.
The desirable attributes of a phase splitter device may well be
summarised as follows :
(I) The two outputs should be of equal amplitudes, i.e. in balance
at all instants of time.
(2) The high frequency response should be well sustained.
(3) The two outputs should be in exactly 1800 phase relationship
over the frequency range involved.
(4) If possible some useful amplification should be obtainable, with
a sufficiently high voltage output to fully drive the succeeding
push-pull pair of valves.
(s) No hum should be introduced into the circuit.
91
PHASE SPUTTERS
(6) Initial adjustments for balance, if any, should not demand
complex apparatus.
(7) It should not be extravagant of valves or components. Six
possible types of phase splitters are described in this chapter and
a summary of their merits or demerits is made in the light of the
above requirements, before an explanation of their mode of
operation is attempted. It might be thought a waste of time and
space to include three types of phase splitters which are later
condemned out of hand, but these arrangements are so frequently
met with in published designs that an analysis of their working
and defects should prove of value in assessing the merits of a
particular circuit.
ANALYSIS OF S IX TYPES
The following list is not exhaustive, as other systems exist, and some
possible modifications in detail in those described are not discussed.
1. THE CONCERTINA O R SPLIT LOAD PHASE SPLITTER
(I) Good balance of outputs at all times, say 0. 125 per cent. error.
(2) Some unbalance of outputs at the highest frequencies (0·02 per
cent.) but amplitude well maintained.
(3) Some phase shift at the highest frequencies.
(4) Stage gain less than unity, for each output ; subject to limitation
imposed by permissible anode swings of the valve, equal to
0·9 of input to stage.
(5) Tends to introduce heater hum into the circuit, but this can be
counteracted.
(6) No initial setting up for balance apart from fairly close match
of load resistors.
(7) Economical of components and only one valve needed.
REMARKS
In order to obtain sufficient drive for succeeding stages, the tempta­
tion exists to increase input to stage to get greater output with risk of
introducing harmonic distortion due to operation over curved part of
valve's characteristic. NFB reduces distortion.
Whilst not quite perfect, can be recommended for use at low output
levels as giving excellent results with minimum trouble.
2 . THE
PARAPHASE SPLITTER
(I) Balance of amplitude of outputs usually poor at low frequencies.
(2) High frequency response falls off.
92
PHASE SPUTTERS
(3)
(4)
(5)
(6)
(7)
Severe phase shift at extremes of frequency.
Useful gain of a voltage amplification stage available.
No tendency to introduce heater hum.
Initial adjustment for balance not too easy.
Economical of components but two valves required glvrng
discount in gain obtainable.
REMARKS
Gives mediocre results, and two valves can be used to much better
effect.
3.
FLOATING PARAPHASE SPUTTER
(1) Never exactly in balance under dynamic conditions, and not
essentially stable.
(2) No initial balance adjustment required-in theory.
Other remarks as for Type 2.
REMARKS
To be avoided, although often used in cheaper apparatus for ease of
initial testing.
(I)
(2)
(3)
(4)
(5)
4. CATHODE COUPLED PHASE SPUTTER
Completely self-balancing in effect.
High frequency response well sustained.
Only very small phase shift effects.
Gives gain of approximately one-half a stage of amplification.
Some tendency to introduce heater hum which can be counteracted.
(6) No setting up for balance.
(7) Economical of components but two valves required.
REMARKS
Probably the best types to be employed where medium drive vohages
are required. Reduction of effective anode voltages make maximum
signal output somewhat reduced as compared to Type 5.
5. THE ANODE FOLLOWER PHASE SPUTTER
(r) To a very large degree self-balancing due to NFB employed.
(2) High frequency response well maintained.
(3) Phase unbalance can be held to I per cent.
93
PHASE SPUTTERS
(4) Gives gain of one stage of amplification.
(5) No tendency to introduce hum.
(6) No initial balance adjustment.
(7) Two valves required and somewhat extravagant of components.
REMARKS
The best all-round type, especially when the maximum drive voltage
is required.
6. THE H IGH GAIN C O NCERTINA PHASE SPUTTER
This type has all the characteristics of the Concertina Type I, but
has the advantage that the full gain of the preceding amplifier, a pentode,
can be realised, which offsets the disadvantage of no gain from the
normal circuit arrangement.
CIRCUIT DIAGRAM S
THE CONCERTINA PHASE SPUTTER
RL• RL2 =
of "ai'le
ra
s
FIG. 9 !1.-The single valve
Concertina Phase Splitter.
'---+-�---I
� OP 2
In the arrangement of Fig. 9/1 the valve is given two equal loads,
one in the anode RLl which is completely normal, and one in the cathode
RL2, which resembles the cathode follower output stage. The outputs
taken from anode and cathode will be equal and opposite in phase.
The circuit is interesting because of the NFB principles involved.
RL2 will provide current feedback for output 1 making it of high
impedance, whilst output 2 will appear to have a very low impedance
due to the effect encountered in the cathode follower. However, the
apparent internal resistance of the output will not influence the grid
circuit� to which the outputs are fed ; the only thing that matters is
that the voltages shall be equal. This specification is well met at low
frequencies providing the decoupling condenser Cd, if present, is of
adequate value with negligible reactance at the lowest frequencies ;
otherwise the anode load will rise and there will be inequality of output.
Cd should be at least 8 microfarads.
94
PHASE SPLlTIERS
The output at high frequencies is slightly out of balance due to the
presence of CgK and Cg•. The currents through these capacitances
(of unequal size) are not in phase with the space current, and they have
the effect of making the anode and cathode voltages unequal in ampli­
tude at the highest audio frequencies, and of giving them a small phase
difference not equal to the ideal 1 800•
Reference to Fig. 9 / 1 shows the arrangement for obtaining bias.
The value of CK must be made adequate or output 2 will become larger
than output I at low frequencies, although the degree of unbalance
will not be very serious as RK is small compared with RL2, say 1,000
ohms, compared with 50,000 ohms. In practice the two load resistors
are made roughly equal to the anode resistance of the valve. Rg, the
grid leak, should be as large as possible, otherwise the current flow
through Rg and RK tends to increase the cathode output, but with
Rg = 2 megohms, the degree of unbalance for a valve of mutual
conductance 2 rnA /V, is only o· 125 per cent., and the degree of unbalance
is not of practical importance unless Rg is below about lOoK ohms.
The output expected at each output can be reckoned as about 0'9
of the input for average stages.
The input resistance can usually be taken as about ten times the value
of the grid leak, thus the size of Cc, the coupling condenser, can be re­
duced accordingly, effecting a saving. This high input resistance is very
valuable when the preceding amplifier is a pentode, enabling more
gain to be obtained, and less attenuation of the higher frequencies.
A disadvantage of this circuit is that the cathode of the valve is
considerably positive with respect to its heater which is usually earthed.
This may give ri�e to a breakdown of insulation between heater and
cathode. Valve manufacturers usually give the maximum permitted
voltage that may be applied across heater and cathode, and care has
to be taken that this voltage is not exceeded. The difficulty might be
overcome by using a separate heater winding for this valve and raising
potential of the winding to some point above earth, as shown in Fig. 9 /2,
although this is an unwanted complication and expense.
HT+
INPUT
1
+ 90
95
FIG. 9/2.-Modifications to
Concertina Circuit. Arrange­
ment to make heater of valve
more
poslt!ve
than
the
cathode to avoid the effects
of heater emission, and to
avoid excessive voltage be­
tween heater and cathode.
Separate heater winding is
semetimes required.
PHASE SPUTTERS
Another way of tackling the problem is to raise the whole heater
circuits throughout the amplifier to a suitable positive potential. This
can be done quite successfully.
This circuit sometimes gives rise to mains hum, as the heater of the
valve may be producing thermionic emission, and the electron stream,
modulated with hundred cycle ripple, will be collected by the positively
charged cathode. Either of the artifices previously described may be
adopted to eliminate this trouble. Although with some valves heater
hum may be troublesome, the 6J5 and 6C5 seem to work well in every
respect, also the Mullard EF37A strapped as a triode. The use of a
pentode is inconvenient in this circuit, as with the cathode follower,
because a screen supply must be decoupled to the cathode if the valve
is not to become in effect a triode. The currents through the decoupling
condenser to cathode would materially affect the balance of output.
THE P ARAPHASE CIRCUITS
Figs. 9/3 and 9/4 show two circuits in which VI is a standard ampli­
fier and a fraction of the output of VI is fed into V2 so that the output
of V2 is equal to that of VI ;md of course in anti-phase. Fig. 9/ 3 shows
Rd
1 I\W.�
-
OPI
T
HT+
OP2
.---�--+ '- -'--'"
INPUT
R3
FIG. 9!3.-Paraphase Splitter.
a form of phase splitter in which V2 is fed with a fraction of the output
of VI from the potentiometer RI R2 so that the output of V2 equals that
R3 serving as grid leaks for succeeding valves.
of VI ' RI + R2
Fig. 9/4 is a similar circuit to Fig. 9/3 giving a somewhat improved
HF response as Rg can be made much larger than R2 in Fig. 9/3.
These circuits cannot be recommended as the phase unbalance at
extremes of frequency is too great (up to 12 per cent.). Their advan­
tages are that the system gives amplification and is capable of higher
=
96
PHASE SPUTTERS
output than the concertina type ; also the large difference in potential
between cathode and heater is avoided.
HH
�
+---I�
I
OPZ
,
r------.-� - - - -- - - - -
INPUT
Rg
FIG. 9/4_-Alternative feed to phase inverter valve in paraphase splitter.
FLOATING PARAPHASE
The resemblance here to Fig. 9/3 is marked, but the circuit has its
weakness, that of poor HF response. RI = Ra, therefore RI + R2
and R3 - R2 may be used as grid resistors for the succeeding valve.
This limits the upper values of RI> R2 and R3 in practice, and is one
reason for the weakness of the circuit.
The circuit of Fig. 9/5 shows the general form of the arrangement.
Resistors RI and R2 form a load across the output of V l' A portion of
S U I T A B L E VALUES
OF RI &
R3
=
'25 M
HH
OP2
OPI
X t-----'"----t
PHASE
INVERTER
VALVE
INPUT
FIG. 9/5.-Fioating paraphase.
the output of VI is fed to the grid of V2 whose output will cause a
voltage to appear across R3 and R2. The voltage developed by V2
across R2 will, of course, be in opposition to the voltage developed
across R2 by Vl' If the outputs of VI and V 2 are of the same order,
97
PHASE SPUTTERS
then. the point X �ill be virtually at earth potential. If the output of
V 1 rIses, then the mput to V 2 rises and so V 2 gives increased output
and the net result is that point X is balanced at earth potential and the
voltages across Rl and R3, the input voltages to the output valves,
are equal. The circuit has the advantages of the previous type, but in
practice the point X is never quite at earth potentia" but floating,
hence the name. Equality of output is rarely attained.
THE CATHODE-COUPLED TYPE
NOTE -Anode loads
dYe nor equal
IN PUT
Ral
I
�8}'F
'25), F
HT+
mini m u m
> OP2
BIAS
R E SISTAN C E
COUPL l fol G
R.ESISTA N C E
FIG. 9/6.-Cathode-coupled phase spliner.
The name of the circuit can readily be understood because the
cathodes of V1 and V2 (Fig. 9/6) are coupled due to the common
resistor RK. The mode of operation can easily be followed. If the
grid of V1 is made more positive, the anode current increases, and so a
larger voltage appears across RK. The anode of V1 will, of course,
be negative going . The increased voltage across RK makes the cathode
of V 2 more positive relative to the grid, which is another way of saying
that the grid of Vz becomes more negative than its cathode ; conse­
quently the anode current of Vz falls, and the anode of V2 is positive
going. We thus have two outputs in anti-phase.
A common bias resistor R3 is employed with suitable grid leaks
Rl and Rz returned to the junction of R3 and RK. The size of Cz is
important ; if it is not large enough it will introduce phase unbalance
at low frequencies, because then in effect the grid of V2 would not be
returned to earth but to a tapping on a potential divider formed by
Rz and Cz across RK.
There is a slight unbalance at all frequencies brought about by the
presence of Rl causing extra current in RK due to the input current
flowing through Rl and RK. This was also a defect of the Concertina
type, but if Rl is large, up to 2M being common, the unbalance due
to this cause is almost negligible. At high frequencies stray capacitances
PHASE SPLITTERS
greatly complicate the action of the circuit, but providing RK is not
large any unbalance is virtually cancelled out.
A further advantage of keeping RK small is that a large voltage is
not developed across it thereby seriously reducing the effective HT
voltage between anode and cathode, and of course not giving an
excessive cathode/heater voltage.
To preserve a good high frequency response the value of anode load
should not be too high or shunt capacities take their toll. As the anode
load should not be less than twice the anode resistance of the valve,
to secure good linearity, a valve of fairly low anode resistance will be
chosen.
The value of RK is usually chosen to equal the anode resistance of
the valve, but as a low anode resistance valve is needed for the reason
above, RK will be low which satisfies on all counts.
With the preceding conditions met there will be an unbalance if
equal values of anode loads are employed, but adjustment of these
values can put this right. The anode load of VI should be smaller
than that of V2 '
To obtain exact balance the equation below must be satisfied :
Ra2
R al t- ra2 + R az
1 - --RK (I + Raz)
Let us take two common examples of valves suitable for this application :
6SN7 or Mullard ECC32 or 6JS's.
14,000 ohms, Ra2 say 47K, RK = I SK, !1. = 32 .
ra
By substitution in formula :
47K
47K
47K
Ra l = 41K.
14K + 47K
1 ' 143
61K .
I+
1+
ISK (I + 32)
49SK
6SL7 or Mullard ECC3S.
ra = 34K, Ra z say 68K, RK = 33K, r
68.
By substitution :
68K
68K
Ral
----=-=
'-=--=-- --34K + 68K
I02K
I +
I +
33K (I + 68)
33K X 69
_
___
=
_
=
=
68K
68K
6SK.
K
I� 044
IOi
·
·
1 +
2,277 K
It can be seen �hat as RK and !1. become larger, and with a small value
for Ra2 the difference between RaI and Raz becomes less. Quite
.
satIsfactory results have been achieved using an ECC3S with Ra2 =
=
99
PHASE SPUTTERS
180K and RK = 47K with Ral = 180K as the difference in outputs is
negligible if the above formula is applied, the denominator being but
� ·oS6. The ideal value for Ral being approximately 1 70K, the error
IS only 0'5 per cent.
This circuit arrangement is by no means new, as it was used in the
Science Museum Receiver described in Wireless World in 1930. New
importance was given to the circuit during World War 11 by its suit­
ability for radar work, when it received its present name.
HH
Rl
'I7K
OPI
OP2
R4
lOOK
V1
Cl
° 1", F
INPUT
R7
' 2 5 1-.4·
Rs
2 10
Ra
o25M
FIG. 9/7.-The Anode Follower, which is the modified version o f the
" Floating Paraphase " of Fig. 9/5.
Fig. 9/7 shows one form of the circuit which will give phase unbalance
of less than 1 per cent. The cathodes of both valves are not at a high
potential relative to the heater, which is often an advantage. Also the
undistorted output given by each valve is greater than the arrangement
of Fig. 9/6 in which the effective HT voltage is reduced by the drop
across the common cathode resistor. Where high output voltages are
required to drive big output valves this circuit is undoubtedly the best
so far discussed, although for low level work it is somewhat costly and
cumbersome as compared with the Concertina type.
The valve V2 is fed from a potential divider comprising Cl> Ra, R5
across the output of VI so that the overall amplification of V2 is unity.
V2 is provided with negative feedback from anode to grid via C2 and
R4 thus giving it a low input impedance which is effectively one arm
of the input potential divider. As the input impedance depends on the
amplification and the potential divider ratio depends on the input
impedance, the circuit is largely self-compensating for changes of
amplification.
The greatest drawback in practice is obtaining values for Ra and R4
to ensure that near balance is achieved, but in Fig. 9/7 looK and 2M
would be suitable.
100
PHASE SPLITIERS
A HIGH GAIN PHASE SPLITTER CIRCUIT
The Concertina phase spliner, whilst an attractive circuit for its
simplicity, has the serious disadvantage that its gain is less than unity.
However, a circuit that overcomes the difficulty in a neat way appeared
in American literature some years ago and appears to be little
known.
R,
.-----�--r-��----�c-- H T +
.... F
�
R2,
'25M
lM
I1 .2S
:r:e....F
VI
2.4 K
47K
R4
'
R3
25,.. F
1·2K
SM
2K
47K
I
OPI
)
OP l
)
FIG. 9/8.-A typical pre-amplifier and Concertina phase splitter stage.
Fig. 9/8 shows a typical pre-amplifier and phase splitter stage in
which a gain of 94 might be expected as a maximum, using normal
values for a 6J7, RI
o'SM, R2 = 0'2SM, R4
o'SM with a peak
output of 94 volts . The outputs from the phase spliner would then be
about 84 volts peak for an input of I volt. The 6J 7 pentode actually
has an amplification factor of I,Soo with ra = I'S megohms, but full
use cannot be made of the valve because the anode load cannot be
made large compared with anode resistance, due to the excessive HT
voltage required. N<tw the input impedance of V2 is high, roughly
R4 X 10, and if this high input impedance is utilised as the actual load
for VI' say o'S X 10 megohms, a value bearing a sensible relatiJnship
to the anode resistance of VI is produced and the stage gain in theory
at least could be ;
( I ,SOO X (S X 106)
Stage gain = I.l. �
RL + ra
(S X 106) + (I'S X 106)
. I,SOO X (S X 106)
7,Soo
.
= · -= I,IS4 omes.
6 'S X 106
6'S
This is a handsome value which would raise the input sensitivity of
the system from I volt to 0'08 volts for equal outputs (not that 84 volts
of drive for output valves is ever likely to be needed in domestic
apparatus).
=
__
=
=
.
-
101
PHASE SPUTTERS
The circuit shown in Fig. 9/9 will give this result, but in order to
follow the exact working of the circuit it is necessary to analyse the
changes to the phase splitter.
Ra
R,
HT+
20 K
' '''
AHt-:=1---+
C2.
L--4-II--l----+---l---Ir-
OPZ
B
50.... F .vw
FIG. 9 !9.-High gain phase splitter.
Fig. 9/10 shows the interim stage. The load in the cathode has been
split into two parts RKI and RK2, but as each of these resistors is twice
the value of the normal cathode resistor the AC loading remains the
same. The presence of C3 in series with RKI across RK2 will have some
adverse effect on phase shift at the lowest frequencies unless it is
reasonably large. The bottom end of RKl connected to earth in Fig.
HT+
A
:i; R4
�iR"
1
(4
11
FIG. 9!Io.-Interim stage in
development of High Gain
Pha�e Splitter Circuit.
Rs
(3
Points A and B refer to the
same points in Fig. 9/9.
R
�
9/10 can be connected to HT + as in Fig. 9/9 with exactly the same
result in the AC sense, HT + being at earthy potential. R4 shown
dotted in Fig. 9/10 is a necessary addition to the circuit to provide
HT feed for VI and whilst the DC resistance is about 0'2SM the
impedance is, say, 1 ·6 megohms (0'2M + ' SM in parallel X 10)
because of the NFB (neglecting negligible reactances of Ca and Cs).
The grid circuit now has an effective " grid leak " of R4 and Rs
102
PHASE SPLITTERS
in parallel, but the impedance is ten times this value, also because of
the NFB. In this impedance calculation the multiplier of ten assumes
the values of anode and cathode loads to equal the anode resistance of
the valve. It also assumes that the amplification factor is around 27.
The actual multiplier for the 6J5 with ra = 7'7K, fJ. = 20 and RL =
20K works out at 9'8 so this guess of ten times is fairly close.
The anticipated stage gain for VI is now :
R in
1 '5M and fJ. 1,5 °0)
fJ.
(6J7 ra
A
ra + R4
1,500 X (1'6 X 106)
(1'5 X 106) + (0'25 X 106)
24 X 108
24 X 102
1
1,37 1 times approxImate y.
1 '75
1 '75 X lOG
It would be a reasonably safe assumption that a stage gain of approach­
ing 1,000 times can be obtained. This has, in fact, been confirmed by
bench tests. The 6J7 is not a very high fJ. valve and a Mullard EF37A
with fJ. = 4,500 and ra = 2'5 megohms would give higher gain. The
Mazda SP61 has also given good service. The Mullard EF50 could
also be expected to perform well in spite of being a semi-variable-mu
valve, as the grid swing would be very small, and the curvature of the
grid characteristic would not introduce much distortion.
A short list of suitable and tested valves for the position of VI is
appended. In every case it will be seen that fJ., the amplification factor,
is high.
=
=
------
-
•
- - =
Anode
Resistance
Valve type
Mullard EFso
(ARP3S, VR9 r)
Mazda SP6r
(VR6S)
Mullard EFS4
(VR r36 or CVI I36)
Mullard EF37A
Mullard EF36
(CVros6, VRS6)
Mullard EF42
Mullard EF40
ra
i ra
I
ra
ra
ra
ra
ra
=,
=
=
=
=
=
=
•
Amplification
Factor
f-I,
rM
=
6,soo
fJ. = S,9So
O'7M
IJ.
o'SM
2'SM
fJ.
fJ.
2'SM
O'44M
2'SM
i
=
=
3,800
4,SoO
S,OOO
fJ.
4,r 80
fJ. = 4,62S
=
=
Remarks
I
Very satisfactory.
_
Low noise valve.
Apt to be microphonic,
Miniature.
Miniature-Iow noise.
Very good.
The above table of valves must not be taken as exhaustive, but in
general a high fJ. is first choice with freedom from microphony and
heater hum.
r03
PHASE SPUTTERS
The value of cathode resistor must not be too high or the cathodel
heater voltage may be exceeded, but there is little point in going above
20K for anode load. A reduction to 10K reduces the gain from 0'892
to 0·881 per output, so that the values of RL
RK
r. will always
be safe.
=
=
CHECKING FOR BALANCE
If unbalance exists an audio voltage will be developed across R and so
fed to the phones . N.B.-The isolating condensers or transformer, or
HT+
t
OUT P U T
VALVES
�
Q
OUTPUT
T'lA"5Fo"",,,U .
"
r=
�
FIG. 9 / I I .-Checking push­
pull output stage for dynamic
balance.
both, are a very necessary safety precaution when wearing headphones
connected to the HT + supply.
DYNAMIC BALANCE
If a state of unbalance is suspected, the first step is to check the input
voltages to the grids of the pp output stage, it being assumed of course
that the valves and transformer are known to be in order. A steady
tone fed into the amplifier should produce equal voltages between the
grids of the output valves and earth as read by a valve voltmeter,
this being the only suitable instrument. In the absence of a valve
voltmeter it is virtually impossible to make the check ; and reliance
must be placed in the method of checking shown in Fig. 9/u . If the
AC components of anode current are not exactly equal and opposite
then a signal will be heard in the phones . Should this be the case the
constructor can do little about it unless the phase splitter device has
an adjustable element, as in Fig. 9/3. Suggested lines of investigation
are :
(a) Check the output transformer for lack of equality of inductance
in the two halves of the primary. This may be done by substitution
in the absence of available test gear.
(b) Check the two output valves for equal amplification factor (equal
anode currents do not eliminate the possibility of unbalance).
104
PHASE SPUTTERS
A visit to a friendly dealer equipped with valve tester for checking
mutual conductance is indicated.
Cc) Checking load values in the phase spliner, if not already done, is
about the last hope, barring the unfortunately not infrequent
chance of a leaky coupling condenser.
If muzzy quality still persists, it is almost certain that a parasitic
oscillation is present, usually only to be checked by valve voltmeter
or oscilloscope. It is interesting to note that one amplifier constructed
by the writer did not sound too bad, but a heterodyne on the Light
programme was traced to the 4th harmonic of the oscillations present
in the amplifier.
One last point that often escapes notice is that, in theory at least,
the anodes of a pair of push-pull valves may be fed with comparatively
unsmoothed HT from the rectifier and that no hum will result due to
the equal and opposite ripple currents in the two halves of the trans­
former primary cancelling out. Often no hum results, but severe
modulation of the speech frequencies by 100 cycle hum does occur.
The saving of cost by using a smaller smoothing choke may then-�­
fore prove dear in the end.
STATIC BALANCE
Notes were made on this problem in Chapter 6, and the only point
to add is that a final adjustment of the bias value of valves is helpful.
A small extra resistance added to one valve to give it extra bias of only
t volt may serve to equalise anode currents.
GENERAL NOTE
Where two valves are used in a phase splitter circuit it is quite
common to employ a duo-trio de for the sake of convenience, but it is
by no means essential to do so, as two separate triodes can be used
equally well. In fact, where a close match of valves is called for it is
probably a better solution to use separate valves as they are likely to
be better matched than the two sections of a duo-trio de.
105
CHAPTER 10
TONE COMPENSATION
The use of the familiar term " Tone control " has been deliberately
avoided here, because it has popularly become associated with the
mellow tone so often preferred by users of commercial radio sets.
Such mis-use of tone control was undoubtedly brought about originally
by a natural desire to eliminate HF distortion and spurious harmonics.
In many cases, improvements in quality have not been followed by a
more judicious use of the tone control by the average listener. It is a
well-known fact that as quality of reproduction is improved, the
frequency range can be extended and enjoyed, but extended " top "
should always be accompanied by improved LF performance, other­
wise in many musical items there is a distressing lack of balance.
In Chapter I, mention was made of scale distortion, and it must be
appreciated that scale distortion is not only associated with listening
to reproducted music or sounds, but with listening to the real thing.
A military band in a procession is first heard as the thin reedy noise
of the clarinet and the " edge " of the trombone and it is only when it
is quite near that the beat of the bass drum is heard. This is due to
the ear itself not having a straight line response, but tailing off in the
bass and the extreme top as the sound intensity is reduced.
For quiet domestic listening, it would seem that a considerable
measure of bass boost is desirable, together with some treble lift. In
fact, maximum flexibility in control of response is necessary to com­
pensate for deficiencies and variations which occur in radio, record,
pick-up, speaker, listening room and the human ear.
It should nevertheless be remembered that amplifier boost cannot
get to work on frequencies which are not present in the sound source­
radio or records-nor can it help the loudspeaker system if this cannot
reproduce them. Excessive top lift applied to a loudspeaker with poor
HF response may result in an unpleasant noise due to resonances and
increase of harmonic distortion, which becomes more and more
objectionable as frequency is raised. It is therefore most important
that the quality should be true and clean before top lift is indulged in,
so that high fidelity does not become unduly associated with high
futility.
GENERAL METHODS OF TONE COM PENSATION
The arrangement may include one or more of the following:
(I) Resonant circuits, employing inductances.
106
TONE COMPENSATION
(2) Resistance and Capacity networks.
(3) Selective Negative Feedback, i.e. feedback of differing amounts
at different frequencies.
Let us examine these in turn.
1.
RES ONANT CIRCUITS
This type of compensation includes values of inductance and
capacitance which resonate within the audio range. The circuit is
" selective ", giving a tuning effect in the same way that a desired radio
transmission is selected out of the many signal voltages induced in the
aerial. The values of inductance required will be high to resonate at
audio frequencies. This generally means the use of iron cores and
many turns of wire with comparatively high hysteresis and resistance
losses, which affect the " Q " of the circuit, as at radio frequencies.
The low " Q " means that the shape of the response curve is quite flat
and broad. This is undesirable when it is required to eliminate one
particular frequency such as a heterodyne whistle between two stations.
It is difficult to cut this one slice out of the frequency spectrum without
attenuating adjacent wanted frequencies at the same time.
HT+
Effect of tuned circuit
on frequency response.
{J)
on
o
Cl
u
«
N O RMI>.L RESPONSE
INPUT
i
I
F R E Q U E NCY
FIG. lO/ I .-Circuit tuned to one frequency. Typical values for resonance
at 9 Kcis would be:
CH= 0·6H.
C � ·0005 fJ.F.
Fig. 10/1 shows a circuit which could be employed to eliminate­
or greatly reduce-the undesirable 9 Kc/s whistle already mentioned.
This circuit will present the greatest impedance at resonance, and the
signal voltage appearing across R2 will be reduced as indicated.
A similar result might be achieved by employing the same choke
and condenser in series across R2, forming a series tuned circuit which
at resonance will have such a low impedance (i.e. act as an absorption
filter) that there will be a big reduction in the unwanted frequency
appearing across R2.
The inclusion of resistance in the tuned circuit will " spoil " the
107
TONE COMPENSATION
sharpness of the response curve and this effect can be utilised to give a
form of bass and treble boost at the expense of the middle frequencies.
If the broadly tuned circuit " resonates" and covers a wide band of
middle frequencies then top and bass pass to the next valve with little
attenuation.
The use of any resonant type of tone compensation circuit is to be
deprecated when other steps can be taken to achieve a similar result,
for the following reasons:
(a) A resonant circuit is an oscillatory system and any sudden change
in the circuit, i.e. a transient, can shock excite the circuit into self
oscillation at its natural resonant frequency. The lower the damping
the greater the likelihood of oscillation and the longer it will persist.
The likelihood can be reduced if the resonant circuit is incorporated
early in the circuit where signal levels are very low but this gives
prominence to the next objection.
(b) The inductance tends to interact with any stray electro-magnetic
fields in the vicinity, usually due to mains transformer and chokes,
giving rise to hum, or if interacting with the output transformer
instability can easily result. The cure lies in heavy iron shrouds, or
Mu-metal cases which become expensive items.
(c) The inductance of iron-cored chokes varies with change of direct
current, although usually the flow of direct current can be avoided,
but more serious still the inductance varies with change of A C flux
which cannot be dodged, except by once again incorporating the device
at the low level end of the amplifier. The use of air cored coils is
suggested, but they become very bulky to obtain adequate inductance,
much more wire being required in the absence of an iron core.
(d) For general tone compensation purposes, the tendency to produce
peakiness,
rather than a gradual curve, is undesirable.
a
(e) The production of considerable phase shift which may give rise
to instability especially when NFB is applied over a later part of the
circuit is a serious disadvantage.
2.
RES ISTANCE AND CAP ACITY CIRCUITS
The reactance of the condenser Cs will decrease as the frequency
goes up, resulting in HF attenuation. To take a practical example:
Amplification factor of valve=20.
Anode resistance, 10,000 ohms.
Anode load, 50,000 ohms.
Stage gain without shunt capacitance 1 6·6 times.
With 0·001 [J.F connected in parallel the gain at 400 c/s falls to 14
times, and at 4,000 c/s to 5 ·4 times. It will be seen that the gain is
108
TONE COMPENSATION
considerably reduced with increase of frequency, but it is also reduced
at all frequencies, even the lowest. This type of control has been used
in the past to boost the bass, although in reality it only cuts the response
more and more as frequency rises.
A variable resistance of say 100,000 ohms could be inc luded in
G E N E RAL
N AT U R E O F
R ES PO N S E
Rg
F R E QU E N CY
FIG. 10jIA.-RC Compensation.
Shunt capacity Cs across ano:le load RL reduces effective value of load
as frequency of input increases, giving reduced amplification.
series with Cs of Fig. IO/IA to control the degree of top cut. The
gr ater the value of the resistance, the more gradual would be the
a cenuation. The shunting condenser Cs could equally well be incor­
rorated across R g; the same effect would be produced.
HT+
VI
1!
..0
�
Y2 � �o______��
I-
./
__
__
__
__
__
__
__
G E N E RAL S H A P E O F
R E S P O N S E CURV E
F R EQU E N CY
FIG. Ioj2.-Bass-cut circuit.
Fig. 10/ 2 is the familiar circuit of an RC interstage coupling. The
coupling condenser C must be of adequate capacity, otherwise attenu­
ation of low frequencies will result, quite apart from undesirable phase
shift effects which would promote LF instability in an amplifier when
considerable NFB was applied. C and R are in series across the output
of V I and it is the voltage developed across R which provides the input
109
TONE COMPENSATION
to V2 at any particular frequency. Now when the reactance of C
equals the resistance R, the loss of output volts is equal to 3 db. This
is arrived at in the following manner:
Assume that R = I megohm and the reactance of C at a certain
_1_ also equals I megohm. The fraction of the input
27tfc
voltage appearing across R will be equal to the ratio of R to the total
frequency xC=
impedance (Z) of C and R in series, that is
:� = �.
Z in this case = VR.":f+xC2 megohms
= VI 2+ 1 2
= V-2
1 . 4 12 megohms
I
. ER
=
--
-
"
"
--
This ratio is equivalent to 3 db loss of volts. To spare the reader
any further painful calculations of this nature, the following table is
given. This shows that where the reactance of the coupling condenser
at a given frequency is equal to the resistance of the grid leak, the loss
increases at about 6 db per octave as the frequency goes down.
Table showing the loss occasioned by a value of coupling condenser
in which the reactance xC at frequency f is equal to the resistance of
the grid leak.
Frequency
f
2f
4f
if
if
if
Loss
Reactance of Condenser
xC = R
R
xC
2
R
xC = 4
xC = 2R
xC = 4R
xC = 8R
=
. -
Below i f the loss will be at a constant
per octave.
The foregoing argument has brought to
(I) In an RC coupled stage the size
must be large if the following grid leak is
bass response is to be obtained.
110
- 3
db
- I db
Negligible
-
7 db
- 12·5 db
- 18 db
rate of approximately 6 db
light two important points:
of the coupling condenser
small and vice versa, if full
TONE COMPENSATION
(2) By suitable choice of coupling condenser bass attenuation can
be obtained if desired, although in domestic listening the only useful
application is to avoid boom on speech reproduction.
CHOICE OF COUPLING CONDENSER
Bearing in mind the cost of condensers and the lower insulation
resistance which is probable as the capacity increases, the associated
grid leak should be large, although this is governed by the maximum
permissible grid/cathode resistance for the succeeding valve as quoted
by the makers.
The following table, calculated on the basis of approximately I db
loss at 12'5 c/s gives a useful combination.
Following
grid leak
Coupling
condenser
10,000 ohms
50,000
100,000
250,000
500,000
I megohm
2'5 f-lF
0'5
0'25
0'1
0'°5
0'025
"
"
"
"
Table showing values of
coupling condenser and grid
leak for r db loss at 12' 5 c/s.
"
"
"
"
"
CONTROLLED BASS CUT
CB is a large condenser with negligible reactance at the lowest
frequencies, say 0'1 rtF. C is a small condenser, 0'003, with consider­
able reactance at the lowest frequencies. xC = 0'25 M at 200 c/s, thus
HT +
Rc
FIG. rO!3.-A method of
controlling bass cut.
Rc is­
bass cut control say of rM.
R = 0·25M.
C " - 0·003.
there is a cut of 3 db at 200 c/s or I db 400 c/s. However, as the value
of Rc is reduced C is progressively shorted out, reducing the bass cut
to zero at Rc= zero.
In Fig. 10/4 is shown the basic form of a circuit giving top cut.
III
TONE COMPENSATION
H T+
db
I-0t--POINT AT WHICH
-3db
X,
R
U� t-=-'
'-''-- -----=:::,,.;.:.
R
_
_
•
0o
C
T
� R9
I-
G E N ERAL SHAPE O F
RESPONSE CURVE
FREQU E N C Y
FIG. I O/4.-Basic treble cut circuit.
Again there is 3 db loss at the frequency for which the reactance of
R
C=R, 1 db when xC= 2R, 7 db when xC= - and so on.
2
" LIFT " CIRCUITS
The use of this sub-heading really calls for an apology, because no
actual lifting of output level takes place in the circuits to be described.
All that can be done is to reduce the power over a range of frequencies
and so permit the excluded frequencies to proceed unmolested. This
has in point of fact already been done in the treble and bass cuts just
examined. A bass lift circuit could be more aptly described as a low
pass filter, but it is no part of the writer's function to rationalise the
terminological ambiguities currently used in radio circles, so we will
continue with the uplift.
TREBLE LIFT CIRCUIT
The circuit of Fig. 10/2 could be arranged so that the bass drooped
from 1,000 c/s downward. Referring to our table we could arrange for
the reactance of C at 500 c/s to equal the value of the grid leak R.
Thus at 1,000 c/s we should be I db down, with a loss of 24 db at
approximately 30 c/s. This would give a thin bass response. Even
the middle frequencies would be rather weak because at middle C we
should be about 7 db down. If the bass droop could be arranged so
that the tailing off could be slowed down, the circuit would be more
acceptable, giving a measure of top lift with a reasonably strong bass
response.
If the circuit of Fig. 10/2 is modified to that of Fig. 10/5, the action
of C is changed due to the presence of Ri in parallel with it. At very
low frequencies, xC is arranged to be so great that the loss in the circuit
is almost entirely a function of the ratio of Ri to Ri + R, but at the
highest frequencies xC has become very small, practically shorting out
Ri and so almost the whole gain is realised.
II2
TONE COMPENSATION
If RI is made to equal 3R, the attenuation at very low frequencies
is in the ratio of 4 : I , thus the lift available at high frequencies is
practically 12 db.
The value of C decides at what frequency the bulk of the lift begins.
HT+
R,
FIG. Io/s.-Basic treble lift
circuit. See text for values.
For example, with xC at 1,000 c/s equalling RI these two in parallel
equal 1 ' 5R and the attenuation will be 8 db, a ratio of I : 2 ' 5 . This is
equivalent to a lift of 8 db at 10,000 c/s where xC has fallen to
�-
and
10
gives very little attenuation. Remembering the attenuation at low
frequencies is 12 db, it will be seen that the bulk of the lift takes place
above 1,000 c/s.
If xC is made to equal RI at 500 cycles then the bulk of the lift would
be above 500 cycles. In this way the lift can be started at any desired
frequency, i.e. by choice of condenser. A list of reactance values will
be found at the end of the book.
BASS LIFT CIRCUIT
HH
FIG.
10/6.-Typical bass lift
circuit.
VI
V2
The circuit of Fig. 10/4, a treble cut circuit, can be modified to
that of Fig. 10/6 to give a bass lift effect.
The additional resistor RI is connected in series with C, thus smoothI I3
TONE COMPENSATION
ing out the rapidity with which the top is attenuated by the condenser
C.
The method of deducing the action is similar to that employed in
the treble lift circuit. At the highest frequencies the reactance of C
is negligible and the network will again show a range of control pro­
portionate to the ratio of RI to RI+R2• At the low frequencies the
reactance of C will be very large and the impedance of RI + xC will be
great compared with R, so that a large proportion of the input voltage
from VI is fed to V2•
The frequency below which the bulk of the rise takes place is again
determined by the reactance of C. If C is large its reactance is not
appreciable until a low frequency is reached; thus bass lift starts in
late. The rough approximation may be made that the rise begins at
the frequency where the reactance of C equals RI' The circuit is
particularly important because it is frequently used to correct for the
deficiencies of gramophone records when dynamic pick-ups are used.
Suitable circuits are given in the next chapter, so it will suffice to say
here that for 78 rpm, turnover frequency around 300 cls, R might
be 100,000 ohms and RI 20,000 ohms, so the condenser would be
0'025 fLF.
For an exhaustive treatment of the subject the reader is referred to
an article which appeared in Wireless World in June 1946, from which
Fig. 10/7 is reproduced.
VD ra= I5,000 n .
30 cls
40 "
50 "
II
fL=40. Stage gam 6'33 times.
10 " "
9 " "
RI 45K
R�
WIi
�
5 db lift
2 " "
100 cls
200 "
300 "
db lift
1
" "
HT+
R4
· 0 2.
69- 5 ><
VOLTS IN
From
U
Wireless World ", June 1946
FIG. rO!7.-Practical bass lift circuit with performance figures.
II4
TONE COMPENSATION
SIMULTANEOUS FIXED BASS AND TREBLE LIFT
This is considered desirable to offset scale distortion in low level
listening. Fig. 10/8 shows an interstage coupling with a circuit for
giving simultaneous bass and treble lift. RI and R2 and C will be seen
to constitute the bass lift circuit of Fig. 10/6, and Cl will provide a
path for top depending on its reactance. The presence of Cl across RI
would reduce the effective anode load for VI particularly at high
frequencies, but the inclusion of Ra overcomes this difficulty. The
higher the anode impedance of VI' the greater is the risk of reducing
the effective anode load to a value less than twice the anode resistance
with the possibility of amplitude distortion. The circuit has been used
most successfully with a 6C5 as VI ·
HT+
VI
FIG. lo/8.-Arrangement for
simultaneous bass and treble
lift. (Max. value max. top.
Min. value max. bass.)
:;;���
.... ,lltp1i
c
E. o. Powell
From
U
Wroreless World ", Dec. 1940
DUAL TONE COM PENSATION
It is sometimes desirable to include means of achieving both bass
and treble cut or lift at will, particularly in a general purpose amplifier
that may at one time be providing domestic listening and at other times
be doing duty as a small PA or home cinema amplifier. For the latter
application it is by no means difficult to improve on the average 16 mm
home cinema quality of reproduction, and a trimming of the bass
response removes some of the plums from the mouths of the speakers,
There are two circuit arrangements both of which have been used with
considerable success. Fig. 10/9 shows the first. This is the arrange­
ment of Fig. 10/8 with the addition of " cut " circuits.
The top cut condensers may be reduced in value if their effect is
found to be too severe. For click suppression, 10M resistors may be
connected between toggles of all switches and associated condensers;
otherwise shunting contacts between switch positions should be
employed. This circuit is taken from G .E. C. TP. I publication. Overall
gain of stages is approximately 20 db.
It should be noted that the circuit is so arranged that there is litde
variation in general output level whatever combination of switch
IIS
TONE COMPENSATION
positions is used. The turnover frequency where lift or cut starts is
about 800 c/s.
I
��
;> � �
�
52�
>:
0 °
��
�
�
�
0
0
..
61
�'
'"
.....
>'
�
�
0
+��,
� 'a. '!.
� � 0
.
�
�
�
-
e
u u u
... 'li
ro
a
�
i
ill
....
w
�
....
�
vvv
��m
� !1
vv
�
� �
�
u
"
V
(�( �(
FIG. ro/9-Bass
and treble lift!
cu t circuits with
switch comrols,
fr o m G . E . C .
Pamphlet T.P.!
o n Quality
Amplifiers.
I"
"
0
�
"'
�
",
N
f-
N ",
>"
0
z"
«w
n
,, �
,, 0
0 ·
�
�
� z
�
" "�
.
�
:>
ii!
"
�
">
....
0
0
..
....
::>
v
�
..
---vv
-�
�
..
�
�
� � �
u u u
..
The use of switched circuits is sometimes criticised because the
" ideal " tonal effect may lie between two positions, but they have the
advantage that the " level response " position is readily available when
required. Furthermore, the risk of noise from the more flexible control
given by variable resistors of the carbon type is avoided. Clearly,
both systems have their pros and cons, so a typical example of the
continuous control method now follows.
The circuit of Fig. Io/ro may be built arOl:md the two halves of a
6SN7 or two 6J5 or similar valves, and affords a range of nearly
rr6
TONE COMPENSATION
40 db of bass control at 20 cycles, and 30 db of treble control at 10,000
cycles, the turnover point being at 800 c/s. RT is the treble control
and RB is the bass control. This circuit is useful for record reproduction
50K
SO l'.
SOpF
�T
INPUT
2-2 M
'001
Rs
"f
2M�=--t--:----+
500pF
50K
'2210\
I I'.
Howard T. Sterling
From "Audio Engineering ", U.S.A., February, 1 9 49
FIG. 10/10.-Bass and treble lift/cut circuit employing variable resistances.
'01 to '005 will increase degree of
* Reduction of this condenser from
bass lift.
as the treble cut starts in at an octave higher than the top lift enabling
record surface noise to be reduced without reducing the medium high
notes and destroying the brilliance of reproduction.
3.
TONE CONTROL BY NFB
Selective negative feedback may be used to give tone compensation.
The simplest case is that in which the cathode resistor bypass condenser
is made smaller than usual, thus with increase of frequency the reactance
of the condenser becomes less, and has a bigger bypassing effect,
c
FIG. 10/ I I .-Out of phase
feedback, anode to grid. The
value of C determines the
amount of LF component
fed back.
IN PUT
red�.lCin� t�e current feedback and affording a greater stage gain.
This wIll gIve a form of bass cut and the exact amount will depend
on the relative values of RK and CK.
Fig. I OI I I shows a slightly more elaborate system in which negative
1 17
TONE COMPENSATION
feedback is introduced into the grid circuit from the anode. The
choice of value for C will determine the amount of feedback at a given
frequency. If C is small the feedback will be less at low frequencies
so modifying the stage gain as to give a form of bass boost.
FIG. IO/12.-Feedback circuit
for bass lift suitable for
correction of bass in record
reproduction.
For the mathematics of the circuit the reader is referred to an article
by J. Ellis, B.Sc., appearing in Wireless World for September 1947,
but suggested values for the circuit of Fig. 10/12 are appended, these
giving the requisite degree of bass compensation for record reproduc­
tion, with approximately a 6 db rise per octave below 300 cycles.
LOW LEVEL LI STENING
The tone compensation circuits so far described enable settings to be
made for the desired results from any media, at one particular volume
level. However, if the volume is turned down the reproduction tends
to become uninteresting, so much so that to the uninitiated it appears
that the volume control adversely affects the frequency response of
the amplifier, cutting both bass and treble.
This effect is due to the sensitivity of hearing, which is not so acute
at low volume levels at the extremes of the audio range. The Fletcher­
Munsen curves of hearing are already well known.
If the gain control of an amplifier is arranged to give increasing
bass boost as the gain is turned down much more satisfying results
47K
FIG. 1O/ 1 3.-Arrangement of
auxiliary gain control YR,
giving increasing bass boost
below 1,000 1\" as volume is
reduced.
HTt
100
K
IN
VRI
CT
.25,uF
+--lll---. - ····· � -(lOOS'
V Rz
.
Utlll,"
fill!, .. ,
R3
OUTPUT
7K
..L C B
. 02.f' F
1I8
0
TONE COMPENSATION
can be obtained. In general one rather neglects the treble boost as
the falling off of the ear is not so rapid as in the bass. The simple
circuit of Fig. 10/ 1 3 will afford a degree of bass compensation.
Juggling with the value of CB will give more or less bass boost, and
the value of R3 relative to CB will determine the frequency at which
boost starts, roughly when R3=xCB. The inclusion of the dotted
condenser CT will afford some treble lift if desired. In operation
VR1 is adjusted with VR2 about halfway, so that comfortable room
volume is obtained. Other tone control circuits set the balance, and
then as volume is turned down by means of VR2 extra bass boost is
provided.
Compensated attenuators for low level listening are quite difficult to
construct. A very useful LF Compensator, made by C. T. Chapman
Reproducers Ltd., of Chelsea, S.W. IO, is available at a reasonable
price. As the general volume level is reduced, the LF attenuation
becomes less severe, as indicated in Fig. 10/14.
o
-10
t-.
- t"::r---.-..
""�
"I-!;(
�
-20 ...
-c .
;) �
- 30 Z C
IU
- 40
- 50
FIG. I O ! I4 .-Response
characteristic of Chapman
Compensator .
...
40
100
C.P. S.
1,000
10,000
This unit may be installed between the existing volume control of
an amplifier and the grid ofthe next valve, or it could replace the existing
control.
NFB LOW LEVEL CONTROL
It is evident from section 3 that NFB circuits could be arranged to
provide high and low pass filters to improve realism at low volume
levels . Purists will object that the full benefits of NFB are not being
obtained at the extremes of the audio range, but as distortion is nor­
mally increased by increase of power, the low level case is to some extent
self protected. As NFB automatically reduces output from a given
voltage input, it is certainly sound economics to use the device for
subduing a range of frequencies where necessary. A typical circuit
was therefore included at the end of Chapter 7 on NFB.
EXLEY CIRCUIT
LOW LEVEL LISTENING
Reference to the design of Dr. Exley will not be out of place here. The
basic idea is to increase the impression of bass by deliberate production
of harmonics, so that difference tones help to bolster up the funda119
TONE COMPENSATION
mental frequency. The result, at very low volume levels, gives
remarkably good LF response. The writer's impression was that any
attempt to obtain more than about half a watt from the 4'S watt output
pentode produced unpleasant effects; but as the system is expressly
designed for quiet listening this criticism is rather beside the point.
The reason for this apparent overloading effect is that, whilst the
general level appears to be about half a watt, there is very considerable
power used to drive the loudspeaker at the low frequency harmonics
which are deliberately produced. The result is that the output valve
may be overloaded at low frequencies before the power level at medium
or high frequencies has been increased very much.
It should always be remembered that any attempt to produce heavy
bass from, say, an 8-in. speaker on a very small baffle will distress the
loudspeaker and generate non-linearity in the cone. Bearing these
fundamental principles in mind, the Exley Circuit gives very satisfying
results.
In view of the widespread interest which was shown in this rather
unorthodox system, we are reproducing Dr. Exley's latest circuit in
Fig. IO/IS along with his comments.
+
250
V.
10K
2L K
( l watt)
'IM
' 5 1.1
2K
VI
:l2)'F
350...
wKG.
z
;
"'0
Vol.
SOK
50....'
so.
�.,
2
K
· 5 0.\
lO K
IK
3S0...
W�G
'25tJ,
5'O�F
R
k
50....
wKG.
Reproduced by permission of
cc
Wireless World "
Fig. 10/15.-Ex!ey circuit.
COMPONENTS
V1 and V3, 6FS or H63·
V2, 6Js or L63·
V4, 6V6 (RK=240 ohms,
or
3 watts).
EL33 (RK= 180 ohms, 3 watts).
120
TONE COMPENSATION
Output transformer ; Wharfedale WI2 (22 : I ratio for 15 ohms
speech coil).
Power supply ; 250V, 60 mA, heaters, 6·3V.
The condenser shown in dotted lines is optional, but may often help
to remove instability (capacity, 0· 0001 mfd).
DES I GNER 'S REMARKS
The 4-watt negative feedback amplifier shown above (with minor
modifications suggested by the designer) was fully described in Wireless
World, April 195 1 issue. It was developed as an approach to the prob­
lem of achieving effective bass reproduction in the home without the
use of either a large loudspeaker baffle or a high power output.
The lower bass frequencies are converted electrically into their
respective harmonics by passage through a variable low-pass filter
followed by a non-linear (grid-distorting) stage V3. The output from
this is then mixed with middle and high frequencies which have passed
through the linear stage V 2. Since each harmonic is of shorter wave­
length than the fundamental, the resultant sound can be radiated more
efficiently from a small baffle. The human ear, presented with a
combination of harmonics such as these, tends to add the " missing
fundamental " subjectively, thus giving a sensation closely simulating
the fundamental.
The amplifier is provided with volume, treble and " harmonic bass "
controls. The latter should not be turned up too high, otherwise
unpleasant distortion products may become audible in the final output.
Dr. Exley has very kindly agreed to help experimenters who may
have difficulty in obtaining the desired results. His address is 146
Otley Road, Leeds 6. Letters should be brief and to the point and
should certainly contain a stamped, addressed envelope for reply !
GENERAL NOTES
It must be admitted that although the use of complex RC networks
can produce desired modifications of response, they should be used
as little as possible, particularly in feedback circuits, for the following
reasons:
(I) Angular phase displacements adversely affect transient response.
(2) Treble boost aggravates harmonic distortion.
(3) Bass boost by reduced feedback, if carried to the point of dis­
tortion, loses the cleaning-up effect of the feedback circuit.
(4) Bass boost systems often lead to instability due to phase shift.
It will be generally acknowledged that the best results are obtained
when level response can be adhered to throughout the entire system,
121
TONE COMPENSATION
. from microphone to loudspeaker. This has been demonstrated on
several occasions by Mr. C. E. Watts, of Sunbury-on-Thames, at
meetings of the British Sound Recording Association. Mr. Watts
records flat up to 20 Kc/s, or as near thereto as he can get. His realistic
results are due in no small degree to the absence of phase shift, helped
of course by the absence of surface noise when direct recordings are
played. It is a pity that such recordings cannot be bought, as many
people who are interested in sheer quality of reproduction would be
willing to pay quite a good price for lacquer discs which would give
up to fifty playings with suitable lightweight pick-ups.
I l2
CHAPTER 1 1
PICK- UP INPUT CIRCUITS
The gramophone pick-up is an electro-mechanical device for
converting the mechanical vibrations imparted to the needle by the
groove on the record into alternating EMF's of the same frequency
and relative intensity.
As explained later in the chapter, the problem so far as the amplifier
is concerned is to obtain sufficient voltage at the grid of the input
valve to produce the required volume level at the other end. Impedance
matching does not enter into the picture at this stage. If a transformer
is used, its sole purpose is to step-up the voltage. The question of
impedance does however assume importance when it is required to
control the response of the pick-up or match recording characteristics.
PICK-UPS
The principal types in use today are:
(2) Moving coil.
(4) Crystal.
(1) Moving iron.
(3) Ribbon.
As a chapter in Sound Reproduction was devoted to this subject, it
is only necessary here to remind readers that types I, 2 and 3 belong
to the magnetic or dynamic systems, where the voltage output is
proportional to the velocity, and therefore require bass lift to balance
the reduced output of records below the turnover frequency, whereas
the crystal response is proportional to the displacement of the stylus
and bass lift is not required, but the response of the crystal is not
necessarily the inverse of the recording characteristic.
RECORDING CHARACTERISTICS
The turnover point is about 250 cls on 78 rpm, and around 900 cls
on LP records . There is often pre-emphasis at high frequencies to
improve the signal to noise ratio. Typical characteristics are shown in
Fig. I I I I .
The EMI curve is similar to A without the rise above 4,000 c/s.
It is clear that a good deal of flexibility of control is required in
reproducing equipment to do justice to such varying conditions; a
switch is usually incorporated in high quality amplifiers to change the
turnover point to suit LP records.
123
PICK-UP INPUT CIRCUITS
20
15
+ 10
5
0
5
- 10
15
20
A-Decca 78 rpm.
100
20
15
+ 10
5
0
5
- 10
15
20
CD
o
t
u
10000
B-Decca 33� rpm.
20
10
100
-
_I
--
-20
20
1000
���.I� . A�EIRII ���I
0
g
w - IO
>
1000
/'
100
,/
FIG.
10000
....-
,;
1000
FREQUENCY c.P.S.
C-American NAB.
10000
! I lL-Recording characteristics.
OUTPUT LEVEL
The levels of output vary widely with the different types of pick-ups
and different makes of the same type. Full information is obtainable
from the makers but wme details are given later. It must be remem­
bered that the output obtainable from a pick-up will be reduced to a
seventh or tenth if compensation is effected immediately after the
pick-up. It is usually preferable to compensate after some amplification
to keep a good ratio between signal and circuit noise, although compen­
sation is best effected at quite a low level as most correction circuits
seriously reduce the load on the preceding valve and can give rise to
amplitude distortion if the signal voltages handled by this valve are high.
The highest voltage output is given by the crystal type, followed in
descending order by the high impedance magnetic, low impedance
magnetic, moving coil and ribbon.
1 24
PICK-UP INPUT CIRCUITS
CHOICE OF PICK -UP
There are other factors besides sheer performance to be taken into
account when making a choice. The obvious one of how much gain
will be required to obtain adequate output is the first, and is tied up
with the question of amplifier noise and hum. If it is necessary to use
a transformer with a very low output type, this transformer will couple
with any 50 cycle field in the parish. Therefore, the lower the output
level the greater the design troubles here. Astatically wound trans­
formers and Mu-metal boxes are of great help in reducing induced
hum.
Various methods of attacking hum in the early stages of a high gain
amplifier are treated in the next chapter dealing with microphones,
where the problem is still a low input to the amplifier.
Another consideration is the robustness of the design, particularly
if other people, probably completely non-technical, have occasion to
handle the apparatus. Compromise between performance, output and
robustness is very often necessary.
PICK-U P LOADING
The difference between a magnetic pick-up and a crystal is mainly
due to the fact that the former is partly inductive, whereas the latter
is capacitive. The reactance of an inductance increases with frequency
at the rate of two to one per octave, whereas the reactance of a capaci­
tance is reduced at the rate of two to one per octave. Therefore a
resistance in parallel with a magnetic pick-up cuts the top response,
but a resistance in parallel with a crystal pick-up cuts the bass.
It is always of supreme importance to ensure that a pick-up is
correctly loaded, and to ascertain what has been done to the input
circuit of an amplifier, before blaming the pick-up or loudspeaker for
peculiar results. Correct loading must be arranged before any form of
tone control circuit is considered.
Broadly speaking, a magnetic type works very well when loaded with
a resistance about twice the ohmic value of its own coil impedance.
As the value of the resistive load is reduced, the HF response of the
pick-up is cut down. The makers will always indicate the most suitable
value for maintaining a level response. For example, the 400 ohms
lightweight Connoisseur is happy when loaded with a 1,000 ohms
resistor. If a step-up transformer is used, say ratio 1 : 5, the load
required across the secondary will then be 1,000 ohms X the square
25,000 ohms.
of the turns ratio, 1 ,000 X 25
It is interesting to observe the effect of this loading on the production
of surface noise. Fig. I I/2 shows the difference produced with the
Connoisseur pick-up (without transformer) by using a resistance of
10K instead of the prescribed lK.
Difference in sound level approximately 6 db.
=
I25
PICK-UP INPUT CIRCUITS
Condition B is recommended by the makers, for level res ponse. It
is quite evident that Condition A woul d contain an undue proportion
of surface noise in relation to music. The photographs serve to illus-
B
A
FIG. I 1 /2.-0scillogram of needle scratch produced in wide range two­
speaker system. Shellac record, groove diameter 10 ins. 78 rpm, 400
ohms magnetic pick-up, without transformer.
A. Loaded by IOK resistor.
B.
"
-'-'
IK
-',
trate the importance of adopting correct working conditions with any
pick-up.
The voltage output of the Connoisseur model is approximately as
follows:
Direct With transformer ratio 1 : 6
400 ohms coil
35 mV
I75 mV
78 rpm
125 mV
LP
25 mV
Interchangeable heads are of course supplied.
The Decca pick-up is available in four types with different imped­
ances, three of them with interchangeable heads for LP. The following
table gives essential data, along with recommended value of load
resis tance.
Voltage with
I : 40
Load
Load
Impedance at Voltage
transformer
1,000 c/s
direct
30 ohms
Decca A 78
B
78
170
"
"
1 70 "
B LP
"
850 "
C 78
"
850 "
C LP
"
4,200
78
D
"
"
D LP 4,200 "
"
1 3 mV
36 "
18 "
70 "
30 "
1 80 "
80 "
IOK
IOK
IOK
IOK
0·7V
1 ·2V
0·75V
47K
47K
47K
Another illustration of variation in load requirements is the EMI
Type 14 with an impedance of only 2 ohms at 800 c/s and an output
of 6 mV direct. The matching transformer is normally ratio I : 1 10,
126
PICK-UP INPUT CIRCUITS
stepping up the output to l ' 5 volts and calling for a load resistance of
about looK.
These examples will serve to emphasise the importance of loading
a pick-up correctly to reduce subsequent tone controls to a minimum,
and to avoid distortion.
With a crystal pick-up, the situation is quite different as we have
already seen: the resistive treatment just outlined would result in severe
loss of bass. Broadly speaking, the crystal type is reasonably loaded by the
grid leak of the input valve provided this is not less than 250,000 ohms.
The response would be well maintained up to the resonant frequency
of the crystal, above which there would be severe attenuation. Fig. I I /3
shows a simple input circuit which could easily be adopted with a
GP20, and would provide correction for both 78 and LP records.
FIG. I 1 /3.-Input circuit for
a crystal pick-up (lightweight)
with switch for LP.
250
K
PICK- UP MATCHING
The term " matching " as applied to a pick-up is not strictly accurate,
as matching is a problem of power transference, e.g. matching the output
valve to the loudspeaker, or matching the output of a telephone repeater
(amplifier) to a line, whereas with a pick-up one is largely concerned
. with obtaining a reasonably high voltage between grid and cathode
of the first valve of the amplifier. Nevertheless, matching is generally
understood to apply to balancing the response of a pick-up with the
recording characteristic, and to damping possible resonances to ensure
a smooth response.
Mter making sure that a pick-up is being loaded correctly, we can
proceed to the general problem of matching by means of tone control
circuits. There are four main tonal varieties: dynamic and crystal
pick-ups, each to be adapted to 78 and 33t characteristics, further
complicated by the fact that the correction may be applied in the pick-up
circuit or-preferably-in a following valve circuit. Furthermore, as
the recording level on microgroove discs is 6-10 db lower than on
78 rpm, and as the correction network may reduce the output level by
90 per cent., the pick-up circuit is hardly the ideal stage in which to
apply compensation. In fact, if you put the correction there with some
types, you will find yourself in the position most of us are in when we
have paid our income tax-with little or nothing left! However, it is
in many cases easier to fiddle with control circuits which are located
externally than hidden away under the pre-amp or main amplifier.
Yet another difficulty is that the HF resonance of a pick-up may be
lower with LP head and stylus than with the normal 78 rpm type.
127
PICK-UP INPUT CIRCUITS
Complete matching circuits are readily available from all pick-up
makers, but acknowledgment should be made here of the information
and diagrams received from Connoisseur, Cosmocord, Decca and EMI
from which the following representative circuits have been compiled.
The intention is to furnish a range of tone control arrangements which,
in combination with those described in Chapter 10, will enable the
reader to achieve anything but the impossible. To simplify matters, let
us concentrate on the maximum circuitry with the minimum of
verbosity.
CONTROLS IN PICK-UP CIRCUIT
Figure I I !4 shows a bass lift circuit with LP switch suitable for the
Connoisseur lightweight 400 ohms model, which could also be used
lOOK
A
I
K
'03"uf
78/)
__-1
L!L
·o�F
'02I"F
FIG. I I !4.-Bass lift circuit,
with switch for 78 and LP
matching, suitable for 400
ohms Connoisseur pick-up or
other low impedance magnetic
types. Increase value of R, for
higher impedance coils or
transformer input.
IOK
B
with magnetic pick-ups of higher impedance by increasing the value
of the load resistance RI '
If used with step-up transformer with a ratio of I : S R I would be
2SK, across secondary.
NFB CONTROL
An interesting feedback circuit from EMI is given in Fig. I l !S.
As will be appreciated from an examination of the diagram, all fre­
quencies are fed back to the grid circuit from the anode, so that an
10k
FIG. I I !S.-NFB type bass
compensation circuit, with
separate brilliance control,
matching EMI pick-up types
1 3 and 14, at 78 rpm.
VOL(JM£ CONTROL
HII-�·50k
•
....AlUE
..
128
50
TO GRID
OF
NEXT STA (;E
DEPENDS ON VAlvt. USED
PICK-UP INPUT CIRCUITS
overall reduction in the stage gain is brought about, due to the out of
phase inputs to the grid. By adjusting the bass control however,
feedback of the out of phase lower frequencies may be reduced, thus
resulting in the greater amplification of these frequencies.
CRY STAL TYPES
As the voltage output here is comparatively generous, correction in
the pick-up circuit leaves a useful margin to be passed on to the grid
of the first valve. The GP20-quite a lightweight-has a direct output
of about '5 volt with 78 rpm head, or '25 volt with LP head . A simple
but effective 78/LP corrective circuit has already been given in Fig. 1 1 /3 .
MOVING COIL AND RIBBON
No rrention of these types has yet been made in this section, but no
disrespect to their elegant qualities is thereby implied. To ignore them
would be like preparing a Who's Who in audio and omitting Voigt
and Klipsch, or a What's What in motoring and omitting names like
Rolls-Royce and Bentley, although the latter are famous for high­
powered efficiency whereas the finest pick-ups and microphones are
conspicuous by their low output. It is for this reason that moving coil
and ribbon pick-ups would not have compensation applied until some
amplification had been brought in. The makers always supply the
required step-up transformer with instructions for use, but once such
a pick-up has been correctly matched to the first valve there is no
reason why the " fiuence " should not be applied-as outlined in the
next section.
V ARIABLE CONTROLS
We now come to variable bass and treble controls. Figs. I l /6 and 7,
which can be combined as shown in Fig. I l /8, give complete control
and LP switching in a form which can be applied to a pick-up circuit
or to an intervalve circuit, provided the requirements of input and
output impedance are reasonably met. These are followed by a pre­
amp circuit which employs NFB and conforms to the specified condi­
tions. The writer has to thank Mr. B. Marsden, chief radio engineer to
A. R. Sugden & Co . of Brighouse for valuable data and circuit diagrams .
The preferred arrangement for maximum results is as follows:
I . Pick-up correctly loaded.
2. Pre-amplifier to boost the level at least 40 or 46 db.
3 · Filter as outlined in Fig. I l /8 .
4· Flat response amplifier with NFB applied to cathode of first
valve to increase input impedance of first stage.
1 29
PICK-UP INPUT CIRCUITS
Fig. I I !6 gives the Connoisseur circuit for variable bass compensa­
.
tlon.
F Il. O M L O W Z S O U R C E
�
��
lOO K
...
· 02 F
·03.uF
�
_ _ _ _
FIG.
t
_
_
_
_
_
7J"'o-------fLP
'Olf'�
T
IOK
I '"
UN
E
200pF
I
lOOpF
�
OUT pur
•
Z
\\ I C I I
I I /6.-Variable bass control with 78/LP switching.
The condensers (200 pF each) are necessary to maintain constant
response at HF, as the wiring and grid capacity of following stage can
give top cut when potep.tiometer is in mid position.
Fig. I 1/7 shows a treble variation circuit which would have its uses
in coping with worn discs and LP characteristics.
LOW
INPUT
IM
Z
1>.5
250K
PU+T
T_
�II-_-+_O_U_
� -11.
POSSI B L E
'00 1
lOOK
H I GH Z
FIG.
1 I /7.-Variable treble
control.
The value of C in Fig. I 1 /7 determines the hinge point. Values
between '0002 and '002 mfd can be used to give compensation for all
discs, but '001 is a good compromise for LP and 78 variations.
As the input impedance of the bass control circuit is looK and output
load of the treble control is also looK, they can be joined together as
shown in Fig. 1 1/8, complete with 78 and LP switch. The addition
of a I M volume control should enable the user to make his pick-up
sit up and beg 01 lie down quietly, at will.
I N PUT Z
L O W I>. S
POSS I B L E
FIG.
1 "4
250
K
�����
'0W
O�
K-4���_
_ '�
���� _
-----.
1"1
'001
�F
LP
uN
10K
v
B
T
1 1 IS.-Pick-up circuit providing wide range of control, including
78/LP switch.
V ,- volume control.
B 0= bass control.
T treble control.
=
For a useful circuit with switched controls refer to Fig. ro!9 on page 1 1 6.
PICK-UP INPUT CIRCUITS
Since the vah e following the control is usually the first valve of
" flat " amplifier and has NFB applied to cathode, its input Z is very
high and one need not worry about Co existing from wiper of volume
control to ground. NFB applied to cathode does not of course remove
effect of any Co due to screened leads, wiring, etc., but reduces Co
due to Cgk and Cga.
PRE-AMPLIFIER
The following circuit employs a twin triode 6SL7 or 12AX7 in
cascade, and will give a stage gain according to the book of 50 or 60
per section: i.e. input G1 to output A2 about 3,000. The excess gain is
thrown away on feedback, giving a reduction in pre-amplifier noise and
distortion, with a reduction in output impedance which is useful if the
pre-amplifier is separated from the subsequent tone control.
100
lOOK
lo o K
K
HT 250- 300
... 11 fdt«ed
v
TO
250 K
)
FILTER
· 1 uF
22 K
R
LOAD
FORP.U.
._��
VA .......
3K
470 K
____
IJ2 AX 7 !
I� r'
3K
FIG. I 1 /9.-Pre-amplifier with low output impedance designed to feed
into filter circuit of Fig. I 1 /8 . Input circuit for magnetic pick-up.
R3
FIG. I 1 /9A.-Modification of
input circuit to pre-amp to
suit lightweight crystal pick­
up. Ra is made 10 times the
value of R., thus giving an
input of 50 mV to pre-amp.
·5M
Feedback is applied for each section as follows:
VAGI is connected to pick-up via a 22K resistor. Feedback from
anode to grid is via '1 and 470K. Amount of feedback is determined
by ratio of 470K to 22K (increasing 22K reduces gain).
In the second portion feedback anode to grid is via 250K and · ! .
Due to the high gain, care must be taken that coupling does not
exist between anode 2 and grid I , as this would be positive feedback
and would produce high frequency oscillation. To cancel the capacity
between the valve electrodes and wiring, a small condenser can be
connected between G1 and ground, say 50 pF or so. The loss of treble
introduced is negligible. (Connoisseur circuit.)
131
PICK-UP INPUT CIRCUITS
Other valves can of course be used. The 6SN7 gives less gain per
stage. Alternatively, two separate triodes could be employed.
The type of feedback used-anode to grid-reduces the input
impedance of the stage. A voltage in anti-phase to the incoming signal
is supplied by the anode. This is in series with the incoming voltage
and gives a larger circulating current, the exact effect that would be
present if the valve VA were replaced by a resistor between G1 and
ground. The lower input impedance will of course act as a load on
the pick-up feeding the first grid, and explains the presence of the 22K
resistor. This tends to isolate the low impedance pick-up from the
grid to allow the feedback to manifest itself, and also to reduce the
loading applied to the pick-up. With a crystal pick-up it is essential
to modify the input circuit, as indicated in Fig. I 1 /9A.
To conclude this section, it should again be stressed that tone control
circuits are a necessary evil; they cannot improve quality, but they are
necessary on account of the recording characteristics shown in Fig. I l l ! .
The Garner Amplifier includes a versatile and useful system. It is
hoped that this chapter will serve to throw additional light on a com­
plicated problem.
In any case there are wide differences in frequency characteristics
in commercial records, and the idea that they can be corrected by a
fixed network would appear to be a form of wishful thinking. It seems
inevitable that rec:>rding engineers must adjust their controls to suit
ambient conditions and the " tone " of the voice or instrument which is
being recorded. It is therefore necessary to have flexible controls in
the reproducing equipment, as depriving the user thereof would be
almost as bad as asking him to drive a motor-car with a fixed steering
wheel.
It is unfortunate that the big recording companies do not give us
more information on these points. One looks in vain for a really
interesting book on the subject of recording, written by a man of vast
experience.
MOTOR RUMBLE
Magnetic pick-ups are not seriously troubled by motor rumble,
which originates from vibration components, usually between 5 and
GPZO
10il. n
From " Wireless World
FIG.
",
Nov. 1950 (Kelly &
West)
I I / Io.-Complete compensating circuit for crystal pick-Up, including
.
rumble filter, and switched top control for FFRR recordings.
1 32
PICK-UP INPUT CIRCUITS
50 c/s. Their output is proportional to velocity, which falls with
frequency for a given amplitude, and consequently little output occurs
at these low frequencies. Crystal pick-Ups, on the other hand, are
particularly susceptible to the large amplitude at low frequencies and
show up the deficiencies of a motor in no uncertain way. With the
Acos GP20 a velocity type characteristic has been introduced below
about 30 cls and the trouble is considerably reduced. Where necessary,
however, a high pass filter may be employed, as shown in Fig. I I/IO,
consisting of the two condensers ·02 and · 01, and two resistors. A
switched circuit to give top control for FFRR characteristics is also
included in the diagram, for good measure.
ADAPTABILITY
For general use the ideal input circuit for pick-ups and microphones
is of course the Cathode Follower. Its input impedance is so high
that it accepts all that comes its way with equal impartiality. It catches
the ball, whether thrown high or low, and passes it on to the next
stage at a convenient height. As its stage gain in technical terms is
less than unity (which means that it has no gain and so we are referring
to something which does not exist), it is unlikely that the Cathode
Follower will have a big following in domestic circles; but it would be
an advantage if all amplifiers could be sold with high impedance input
and all pick-ups supplied complete with requisite loading. The fiasco
of connecting a crystal pick-up to an amplifier with low resistance
input originally designed for use with a magnetic type, or of playing a
400 ohms magnetic type into a load of 250,000 ohms instead of 1,000
would thereby be avoided.
Amplifiers and loudspeakers are now turned out ready for use and
are expected to be interchangeable within limits of 3 to 1 5 ohms.
It is about time that the input problem was tackled along similar lines,
so that a pick-up could be changed as easily as a loudspeaker.
It is the custom to stamp the output impedances on an amplifier
so that the user always knows where he stands when he connects
different loudspeakers . But how often do we find the same brand of
information stamped on the input terminals? The input impedance
should be clearly marked, or if the input circuit has been arranged to
match a certain pick-up the chassis should be labelled accordingly.
It is not suggested that all pick-ups should be made of similar
impedance, as this would clearly be impossible. They could, however,
be supplied complete with the necessary loading circuit so that they
could be plugged in to any amplifier, which in turn should have a
versatile high impedance input.
133
CHAPTER 12
WHI STLE AND SCRATCH F ILTERS
WHISTLE FILTERS
The use of whistle filters in association with radio feeders with a
wide frequency response is not uncommon, due to lack of elbow room
in the medium wave band and the production of objectionable whistles
in the 8-10 Kc/s region. When the interference is limited to a signal
of a definite frequency it is possible to incorporate a sharply tuned
resonant circuit to eliminate it, or at least to greatly reduce its nuisance
value ; but it must be emphasised once again that the elimination of
the one frequency will necessarily cut out any musical frequencies
of the same order. Suitable circuits and a discussion of their properties
were given in Chapter 10.
It is, of course, always a debatable point whether a more sharply
tuned radio feeder cutting off around 7-8 Kc/s is not a better proposition
than a whistle filter. In many districts, the use of a filter only removes
part of the " dirt ", and a tuner of the super-het type with variable
bandwidth comes as a boon and a blessing. It should always be
remembered that a filter removes some adjacent frequencies. If the
transmission cuts off at about 10 Kc/s and the filter is tuned to 9 Kc/s
there will be precious little left in the programme above 7-8 Kc/s .
The operation of such filters always strikes the writer as equivalent
to cutting out a thin slice of bread about t in. from the end of a loaf
and expecting the i in. crust to stand up on its own.
SCRATCH FILTERS
If examined under powerful magnification, shellac records will be
found to have a surface rather like fine glasspaper, due to the filler
A.
B.
D.
E.
C.
Combined output from IS-in. and 8-in. units.
I s-in. unit. crossover 1 ,000 c /s 68 db.
I s-in. unit, c rossover 3,000 c/s. 79 db.
8-in. treble crossover 1,000 c/s. 82 db.
8-in. treble crossover 3,000 c!s. 76 db.
.
8S db.
FIG. 1 2 / I .-Oscillograms of needle scratch. The approximate sound
levels above threshold are indicated. Note the increased level when the
8-in. treble unit is extended from 3,000 cycles down to 1,000, and the still
greater difference when the bass unit is taken up from 1,000 to 3,000 c/s.
1 34
WHISTLE AND SCRATCH FILTERS
which is mixed with the shellac. The irregularities in the surface cause
output across the pick-up terminals over a very wide range of
frequencies, almost equivalent to " white noise
The oscillograms
of Fig. 1 2/1 will illustrate to some extent the energy distribution of the
surface noise in relation to the frequency range.
It will be observed that there is very little power at frequencies below
H.
Top :
Second side of
Danse Macabre.
BOTTOM :
First side of
Beethove n' s
Fifth Symphony.
Standard shellac
records .
6
B
10
FREQU ENCY
12
(k.ch)
..;.,... ... . .
Courtesy Decca Record Co. Ltd.
I2!2.-Panoramic display of the audio spectrum from 0-20 Kc.s
showing energy levels with time exposures of I t and 2 minutes respectively .
250 c/s upwards.
FIG.
135
WHISTLE AND SCRATCH FILTERS
1 ,000 c/s. By comparison, it is interesting to note the energy distribu­
tion in music as shown in Fig. 12/2, which is the result of investigations
made by Decca.
The main peak would appear to be located around 1,000 c/s and with
FIG. I 2 /3.-0scillograms of surface noise.
A. Shellac.
B. Vinylite 78 rpm.
C. Vinylite 33t rpm.
D . Trace on 'scope without input.
a very much higher output below 3,000 c/s than above this frequency .
These illustrations serve to show why surface noise is most obtrusive
in the upper register, where it easily equals or exceeds the sound level
of the overtones of music.
Vinylite records, being without filler, are comparatively free from
surface noise. It is unfortunate that high cost precludes the general
adoption of this material for 78 rpm discs. Fig. 12/3 shows oscillo·5 m H .
I rnH.
2 ", H.
3 mH.
o
'"
N
MG
I
o
o
on
g
g.
FREQ U E NCY
A T T E N U AT O R
WITH
c
0
0
o·
15 O H M S PE A K E R
o db
- 5
- 10
- 15
0
0
0
0
N
-
20
FIG. I2/4.-Attenuation in response of 1 5 ohms speaker with tapped
inductance in series with voice coil. Values 0'5-3 mHo
grams of surface noise from shellac and Vinylite discs respectively,
taken under identical pick-up, amplifier and oscilloscope conditions,
with wide range two-speaker system.
The long-playing disc shows rather more vertical lines from dust, etc.,
WHISTLE AND SCRATCH FILTERS
than the 78 rpm equivalent. This is no doubt due to the finer stylus
point.
The elimination of scratch by means of a resonant circuit peaked at
any particular audio frequency is quite impossible. Such a " cure "
might appear to be successful if it eliminated a pick-up resonance
which was emphasising random noise.
As a general rule, the treble control on the amplifier will be effective
in reducing the output of surface noise by cutting HF response. If
an additional control is required, the simplest method is to use a tapped
inductance in series with the LS speech coil.
Figure 12/4 shows the extent of HF attenuation introduced by induct­
ance values up to 3 mH with 15 ohms speaker, and Fig. 12/5 shows the
corresponding reduction in surface noise up to 2 mHo
FIG. I2!5.-0scillograms to show reduction of surface noise from 15 ohms
speaker by series inductance, as used in Fig. 1 2(4.
A. Flat response.
B. 0'5 mHo
C. I mHo
D. 2 mH.
E. Oscilloscope line only.
The 2 mH condition obviously includes a good deal of treble cut and
would only be required with extremely worn and noisy discs.
The problem of surface noise may seem to be rather outside the
scope of a book on amplifiers, but it is so much wrapped up with the
question of frequency response and tone compensation that it cannot
reasonably be ignored.
137
CHAPTER 1 3
MICROPHONES AND MIXING CIRCUITS
The microphone is an electro-mechanical device for converting the
vibration of the air particles caused bya sound, into alternating electrical
currents of EMF's of the same frequency and relative intensities as
the exciting sound waves .
It is akin to the loudspeaker because its performance is vitally
affected by the conditions under which it is used ; in both cases the
design must take into account acoustical as well as electrical principles.
The sensitivity of a microphone is its electrical output for a given
intensity of input. Sensitivity varies greatly with microphones of
different basic types, and between different models of the same type.
The output is also greatly dependent on the distance of the sound source
from the microphone, both as to character of sound and intensity.
In free air, the intensity of sound from a point source is inversely
proportional to the square of the distance from the source. In an
enclosed space, reverberation plays an important part and ambient
conditions vary enormously. As the microphone is moved away from
the source of sound the balance between pick-up of direct and reflected
sound waves is varied, depending to some extent on the directional
properties of the microphone and the angle of incidence, which refers
to the direction in which the microphone is faced in relation to the
sound source.
The actual size of a microphone also affects its sensitivity in relation
to frequency, as diffraction effects become serious where the outside
dimensions are greater than the wavelength. It is stated in Microphones,
written by BBC engineers and recently published by Wireless World,
that for diffraction effects to be moderate at IS Kc/s the dimensions
of the outer casing (preferably spherical) should be less than 2 ins.
FREQUENCY RESPONSE
The frequency response of a microphone is its relative ability to
convert sounds of different frequencies into alternating currents or
potentials. With a fixed input of sound intensity the electrical output
may vary considerably as the sound frequency is varied ; thus, quite
a good microphone may be rated as having variations not exceeding
±6 db between 50 and 8,000 c/s. To claim that a given type has a
frequency response which extends from 50 to 8,000 c/s means nothing,
as the output at 8,000 c/s may be only a fraction of the level at 50
or 500 e/s .
MICROPHONES AND MIXING CIRCUITS
MICROPHONE TYPES
Microphones fall naturally into two groups:
1 . Low Impedance = Carbon and carbon granule, velocity or ribbon
types and dynamic or moving coil types.
2. High Impedance= Condenser and crystal types.
On the other hand, the ribbon is a velocity microphone in which the
electrical output follows the instantaneous particle velocity of the
impressed sound waves. The carbon, crystal, moving coil and conden­
ser microphones-shall we say the " c " types ?-all belong to the
pressure class, in which the electrical output substantially corresponds
to the instantaneous sound pressure of the impressed sound waves.
Microphones differ from pick-ups in that they are referred to the
acoustic property of the driving medium.
A brief description of the various types now follows, as a prelude
to the input circuits with which they could be used.
CARBON MICROPHONES
(a) SINGLE BUTTON TYPE
This consists of a case with a metal front plate or diaphragm
placed against the face of a cup containing loosely packed carbon
granules. One connection is made to the front diaphragm and the
other to the back of the container. A battery is placed in series with
the button and the primary of a suitable step-up transformer of about
As the diaphragm vibrates in sympathy with the
I : 100, Fig. 13/1.
sound waves its pressure on the carbon granules alternately increases
and decreases, causing a corresponding variation of current flow
through the circuit, since the change of pressure on the mass of
granules varies the resistance of the path.
p
5
OUTPU1
-=- 1'5- 4·5
vol ts
FIG.
Carbon
13/1.
Microphone.
--
1 : 100
The chief characteristic of this type is a high output level, 0'1-0'3
volt across the primary giving up to 10 volts on the secondary. This
is the normal post-office telephone type which without any amplifica­
tion will actuate a telephone headpiece. The frequency range is
restricted, but is enough to give intelligible speech. There is an
inherently high noise level, and variable behaviour with a tendency
for the granules to " pack " if the current through the microphone is
too high. This current is usually 50-1 00 mA.
This microphone is quite unsuited to any serious work.
139
MICROPHONES AND MIXING CIRCUITS
(b) DOUBLE BUTTON TYPE
OUTPUT
FIG. 1 3/2. -Double Button
Microphone.
This operates in a similar way to the single button, but in push-pull,
Fig. 13/2. The sensitivity is less, about 0·5 volt on the secondary
being an average value. The quality of speech is better, but not at
all useful for serious work.
(c) THE TRANSVERSE CURRENT TYPE
FIG. 1 3!3 .-Diagram of trans­
verse current microphone.
The current passes from one
electrode to the other across
the granules.
Figure 13/3 shows diagrammatically the principle of the transverse
current type. Carefully constructed, these can give quite good quality
of speech, but again are not to be taken very seriously for quality work.
RIBBON M ICROPHONE
The book Microphones (already referred to) contains a full descrip­
tion of the BBC-Marconi Ribbon, from which the following interesting
details are taken :
Ribbon-beaten aluminium foil, 2i ins. long, about ·00003 in.
thick, with corrugations to facilitate smooth adjustment of
tension.
Response-flat to 5 Kc/s, - 6 db at 10 Kc/s, - 1 6 db from 1 5 to
20 Kc/s.
Resonance-flexible suspension gives a fundamental resonance at
approximately 2 c/s.
Impedance--()·6 ohm-virtually a pure resistance.
Designed for high-quality broadcasting from a studio.
It is pointed out that ribbon microphones are being made in
America, in which the corrugations are confined to the end portions
of the ribbon, the centre portion being stiffened to ensure a more
1 40
MICROPHONES AND MIXING CIRCUITS
piston-like motion, free from harmonic modes. The " boom " experi­
enced when a velocity type microphone is used too close to the sound
source is due to the fact that the particle velocity follows a different
law for spherical waves compared with that for plane waves in �ree
space. This results in considerable accentuation of the low frequencIes.
This type of microphone is free from cavity resonance.
JT�t3l&1-
R.IBBON
CLAMP
I §I.-I!I--+- X-FOIL CORRUGATED RIBBON
-"'-+-_ POLE
PIECES
�;J��-I--t
I
if
RIBBON TENSION
AOJUSTMENT
(
I
,
I
� ----,-= �"=-�-=-p,:r;:e
r �����l,L:
:
I
I
From
L
_ _
I I
'
I I
\ I I I I
I I : : :
: !
...
I I r::: :::l � ...
r::
I L L .l. _ ,r�, _ _
� _ _ _ ....: � _ _ _ _ _ _ .J
\
:
___�_�
U Microphones,"
�
L ____J
courtesy BBC and
cc
Wireless World"
FIG. 13/4.-BBC-Marconi Ribbon Microphone.
The output from Ribbon types is very low, and hum pick-up can
be troublesome if long leads to the input of a high-gain amplifier
are required. In some cases, one stage of amplification is built into
the microphone head to raise the signal-to-noise ratio in the leads.
MOVING COIL M ICROPHONE
This is a form of moving coil loudspeaker in .reverse. Sound waves
impinging on the diaphragm cause the coil to move in the magnetic
field and thus an EMF is induced across the ends of the coil. The
141
MICROPHONES AND MIXING CIRCUITS
weight of the diaphragm and the stiffness of the suspension govern
the location of resonances in the musical scale, but with careful design
they need not be objectionable. The impedance of the coil may be
anything between o'S and 70 ohms, and a step-up transformer will
be employed. An average output might be 100 mV across secondary
for close speaking, or with reference level IV/Dyne/cm2 -83 db across
20 ohms speech
. coil or 43 db across transformer secondary at 200,000
ohms.
This is probably the most widely used type, as it combines reasonable
sensitivity with good frequency response and is not inordinately
expensive. Freedom from blasting and fluffing are other advantages.
S. T. & c. moving coil microphones, type 40 17 (now superseded by
403S-A) and type 4021A, are largely used by the BBC, especially on
outside broadcasts. The response is flat from 3S to 10,000 c/s.
Model 402 1A has a domed diaphragm only i in. diameter, with a
piston-like motion at all frequencies below I S Kc/s. The microphone
is used with the diaphragm facing upwards so that the angle of incidence
is the same for all sound in the horizontal plane. The complicated
structure of the instrument is well illustrated in Fig. 13/s. This model
today costs £17 10S. od.
-
DIAPHRAGM ANO COIL
ASSEM&LY(l1'ICmAm)
C E N T R E POLE
PIECE
.�.
'
1
OUS1-£.XCLUDING GAUZE
I
I
SCREEN
/
/
PERFOR ....TlONS IN CA50ING
GAuZE A� ",'L,!S
(A,
(RG,)
-AIR &£HINO DIAPHRAGM
(CG,)
COTTON WOOL
PACKING
POLE PIECE
&RASS RETAINING RINC
LOWER
(CG»)
COMPARTMENT
From " Microphones," courtesy BBC and " Wireless World "
FIG. 1 3 / 5.-S, T. & C. Microphone 402 1A.
Moving coil.
CONDENSE R MICROPHONE
The condenser microphone consists of two plates, one a rigid back
plate and the other, separated from the first by �bout a t�ousandth
of an inch a thin perforated metal membrane servmg as a dIaphragm.
This cond�nser is connected in series with a DC voltage source and a
MICROPHONES AND MIXING CIRCUITS
resistance which is useful to limit current in the not unlikely event of
a short, but is actually the load across which the output voltage is
developed, Fig. 1 3/6. When the membrane vibrates the change in
spacing causes a change in capacitance and a back and forth flow of
current. Q=CV with V constant, but C changing, Q will also change.
The first stage amplifier must be built adjacent to the unit or the
capacitance of the connecting lead will seriously impair both output
and frequency range.
HT+
FIG. 13/6.-Circuit of
Condenser Microphone.
CORRUGATED
D I A P H RAGM +
BAC K PLATE
FIXED
'-_ "'"I
R
HT-
This type of microphone has been developed in America to a much
greater extent than in Great Britain. Very wide frequency response
is achieved, but at the present rate of exchange the best instruments
cost about £75 each. Even at this price there is a good deal of interest in
obtaining specimens in this country, when dollar exchange is available.
More recently, reports are to hand on the performance of the
Neumann condenser microphone which is of German origin, costing
about £60 in England after payment of duty. The frequency response
is substantially flat at all audio frequencies; the physical dimensions
are small enough to avoid diffraction effects. The microphone can be
arranged to be omni-directional, or to have a cardioid characteristic
which reduces the sound acceptance at the back virtually to zero. It
is free from cavity resonance, and its behaviour is reported on good
authority to be more in keeping with the human ear in that it is not so
seriously affected by distance from sound source as most types of
microphone.
CARDIOID MICROPHONE
A combination of two types, such as moving coil and ribbon in one
instrument, enables a cardioid characteristic to be obtained, so that
the microphone is directional. This is extremely useful for recording
purposes in " live " rooms, as it helps in obtaining the desired balance
between direct and reflected sound, without placing the microphone
too near to the sound source. In other words, a more distant technique
is possible without recourse to acoustic treatment of walls, ceiling, etc.
The two units of the microphone work in series and are arranged
so that voltages are out of phase for sound reaching the back, resulting
in cancellation.
Switches and/or attenuators are usually fitted to enable the user to
143
MICROPHONES AND MIXING CIRCUITS
vary the characteristics as required, and to
use the velocity or dynamic
microphone on its own.
The output impedance is usually about 50 ohms, so quite long leads
FIG. 1 3 /6A.-C i r c u i t
0f
Stante! Cardioid Microphone.
Output impedance 50 ohms.
MOVING COil
MICRoPHONE
may be run to amplifier without fear of hum pick-up, loss of power or
attenuation of frequency response. These microphones today cost
between £35 and £42 each.
Fig. 1 3/6A illustrates the switching arrangements adopted in the
Stante! 4033 Cardioid.
CRYSTAL MICROPHONE
The sound cell type has a very flat response/frequency characteristic
up to the frequency of mechanical resonance, which can be arranged
3"
<- '!
diam .
c
FIG. 1 3 !7.-Crystal Micro­
phones drawn to scale.
I" wide
ILLUSTRATIONS ARE
HALF ACTUAL SIZE
r44
A.
Cosmocord.
Range 30 c!s to 8 Kc/s.
B.
Tannoy.
Range 30 c!s to r6 Kc/s.
C.
Cosmocord.
Range 5 Kc/s to 30 Kc/s.
MICROPHONES AND MIXING CIRCUITS
to occur at high frequencies by reduction of size. The small
size of the cells eliminates diffraction and phase-difference effects, and
the response at all useful frequencies is virtually independent of the
angle of incidence. With these omni-directional qualities, the sound
cell crystal is eminently suitable for use in response measurements of
loudspeakers. The above (Fig. 1 3/7) sketches illustrate the proportionate
size of three types used by the writer in taking sound level readings,
response curves, and oscillograms of various sounds.
Type A is used with Noise Meter which is effective up to 7,500 c/s.
Type B is used for response curves. The miniature Type C was
specially made for observing frequencies up to 30 Kc/s (not in loud­
speakers). These instruments cost about £12, £25 and £50 respectively.
As usual, size and sensitivity go down as price and performance
go up, and cavity resonance disappears.
Types A and B are of bimorph construction, which is adapted to
reduce the mechanical impedance of the crystal so that it more nearly
matches that of the medium in which it is working and thus increases
the efficiency-and hence the sensitivity-of the unit; thi;; limits the
upper frequency range of the microphone. Type C, with response
reaching up to supersonic regions, is an expander unit with sensitivity
some 20 or 30 db lower than the bimorph.
MICROPHONE INPUT CIRCUITS
When microphones are used for close speech, the output (with
associated transformer where required) is of a reasonable level which
could probably be connected to the input of an amplifier normally
working from a high quality pick-up ; but when distant sounds are
involved considerable extra amplification will be required. It is then
that the difficulty of obtaining low hum level, little microphony and
low valve hiss becomes apparent. The choice of the first valve is
limited, but certain types are recommended by the makers as being
less prone to microphony and hum than others : for example, the
Mullard EF37A and EF42.
Careful wiring of the first stage is essential, and attention to the
heater circuit as outlined in Chapter 1 5 will help to keep hum at a
low level. The use of DC for the heaters of early valves is often
recommended ; either accumulator or rectified AC may be used as a
source and often effects an improvement compared with AC.
If post-office jacks are used for mic-input circuits the outer casing
must be metal to afford screening. The bakelite cover types invariably
allow electro-static hum pick-up. All connections should of course be
screened : the Belling-Lee range of " Screenectors " and conventional
co-axial plugs and sockets are suitable. Any long length of connecting
cable must be screened, co-axial cable being superior to ordinary
screened leads for low capacitance.
1 45
MICROPHONES AND MIXING CIRCUITS
MICROPHONE LEADS
The technique of connecting microphone to the first valve of the
amplifier varies according to the type involved. The following methods
combine simplicity and efficiency :
Moving coil types.
The impedance here will probably be between 20 and 60 ohms .
The best plan is to place a balanced input transformer in the amplifier
and run low impedance lin�s direct from the microphone. Any
type of twin cable may be used, as there is no fear of hum pick-up
or loss of power, and the capacitance of the leads will have no effect
on frequency response.
Ribbon and low impedance MC types with transformer.
Owing to the very low impedance and the small output of a ribbon,
the transformer will be mounted in the microphone head. The same
course would be adopted with a moving coil type of very low
impedance as distinct from the 20-60 ohms models . The output
impedance from the transformer will probably be between 20 and
250 ohms. A further step-up transformer will be placed in the
amplifier. Twin leads should be run (in screened cover if 250 ohms)
which should be connected to the case ofthe microphone, but should
only be earthed at the amplifier end, as shown in Fig. 13/8.
:- - - - - - - - - - - - - - - -I
I
I
c£]
I
I
L
_ _ _ _ _ _
__ _ _ _
FIG. 13!8.-Diagram showing transformer input circuit balanced to earth.
The primary of the input transformer should be balanced to earth.
This could be done by using a centre tap on the winding, but it is
easier to find the electrical centre by using an external load which is
variable. A potentiometer with a resistance of at least ten times
the input impedance would answer the purpose, or two resistors
of similar value could be used.
As regards the microphone transformer, it is better not to balance
the centre of the secondary to earth, as this would create another
loop which could be a source of hum troubles.
Studio Technique.
Where long microphone lines are used under exacting conditions,
with high quality microphones of low sensitivity, the precautions
against hum pick-up must obviously be thorough and complete.
One system is outlined here, as it may prove useful to the amateur
MICROPHONES AND MIXING CIRCUITS
in overcoming problems of hum with a ribbon pick-up-a type
which often presents difficulties.
The circuit is outlined in Fig. 13/8A.
-
: Si,:_'_-!;_."!.:::
U iit �
r--- - - -
-
I I
1____
I
I
l
' .� -
_ _ ____
_
-
-
-,
�- - -
iL__
�::
I
1 -
_
-
;
-
�------
-, ---- -- - - --;-
- - -
5
-
- - --
S C RE E t-l
--
,· �
U
"
....
I
�1 . .J
::� I
:
i�
. .•
_ _ _
_
_
--4
- -- - �- - -
_
- �--l
_ _
.J.. -
,-'",
,
l
.J
FIG. I 3!8A.-Input circuit with stray capacities balanced to avoid hum.
Suitable line impedance would be 600 ohms.
Both transformers would be fitted with Mu-metal cases and would
have a screen between windings. The secondary of the microphone
transformer and the primary of the input transformer (to amplifier)
would be section wound so that the capacitance between ends of
winding and screen is uniform. Screened leads would, of course,
be used. Coaxial cable with two conductors would be very satis­
factory. By comparison with this system, with its balanced capaci­
tances, a centre tapped transformer would be looked upon as a
palliative.
Crystal Microphones.
These are of high impedance and are also capacltlve. There
will be loss of power in long leads to the first valve, but the
capacitance of the wires will not affect the response characteristic.
The best arrangement is to run a pair of wires above earth potential
and enclose them in a screened casing connected to the metal case
of the microphone. The screen should then be earthed only at the
amplifier end, as shown in Fig. 13/9.
.'
�"�"�'"�'-�--�--�--�-- -�--�-��1
C ' CRYSTAL
:...t----·
C .:---- ---t-'-- - -_:r:
MIC
CASE
SCREEN
_
:l M
2M
,r--,
FIG. 1 3i9.-Input arrange­
ment for Crystal Micro­
phone.
Ratio of voltage drop with
3M resistor is 5 : 2.
The writer uses a crystal microphone quite successfully with
18 feet of ordinary co-axial TV cable feeding into a pre-amp
147
MICROPHONES AND MIXING CIRCUITS
operated by dry batteries. Apparently this arrangement breaks all
the copy-book rules, b ut no difficulty is encountered in ob serving
very slight acoustic changes at frequencies over a range of nine
octaves, say 30 to 16,000 c/s.
Crystal microphones are usually supplied with a three-pin plug
and socket, the third pin being connected to the case.
Owing to the high impedance of the small sound cell type, the
input impedance to the valve could be of the order of 5 megohms,
say 2 megohms grid leak and 3 megohms resistor in series with the
grid, for ordinary use. For use with measuring type crystal micro­
phones, the circuit of Fig. 1 3/9A would be advised. This employs
double-shielded coaxial cable (Te1con K I 6-MYM) and a cathode
follower with an input impedance of 100 megohms .
.---- H T +
.
100 I(
OUTER S H I ELD
INNER S H I E L D
C E NTRE C O N D U CTOR
6n
6·3v
Courtesy S. Kelly, Cosmocord
FIG. 1 3/9A.-Cathode follower input circuit. Input impedance looM Q.
Reduced heater volts are applied to the valve to reduce the effects of grid
(gas) current flowing in the high value of grid leak . .
Cardioid Microphone.
The output impedance will be around 50 ohms, and the unit will
include the transformer for the ribbon. The procedure should be
the same as outlined above for ribbon types.
Condenser Microphones would probably have a small amplifier
incorporated in the head with 600 ohms output.
MIXING CIRCUITS
When the use of two or more microphones is contemplated, the
necessity of combining their outputs to feed into the main amplifier
arises. The most obvious way is to connect them in parallel, and this
system is often used successfully in spite of the fact that one micro­
phone is feeding into the other and may produce phase cancellation
effects, especially if the two microphones are not equidistant from the
sound source.
Another possibility is the series arrangement of Fig. 1 3/10, which
gives fairly satisfactory results, but the chief drawback is that both
MICROPHONES AND MIXING CIRCUITS
sides of channel A are above earth and the risk of hum pick-up is great.
Also stray capacities of channel A appear across the output of
channel B and tend to attenuate the high frequencies of the latter. It
FIG. I3jIo.-Series arrange­
ment of two inputs to one
valve.
is often desirable to control the individual levels of the two channels.
This can be done as shown in Fig. I3/ I I , in which the inputs are really
in parallel and one side of each channel is earthed.
The series resistors Rl and R2 prevent either control short-circuiting
R I - 4 EACH
' 5 M fl
I 3 j I I .-P r a c t i c a l a r ­
rangement of inputs in parallel
with isolating resistors and
volume controls.
FIG.
the other if of sufficient value, say 0'5 megohm, although there is still
some inter-action between controls.
In order to overcome interdependence of controls, the two inputs
may be fed into separate valves as shown in Fig. 1 3/12•
lOOK
1 3/12. - I m p r o v e d
circuit for mixing two chan­
nels, using separate anode
loads and isolating resistors.
FIG.
INPUT
A
'15'-2
'"
I N P UT
B '25-2
M
149
�
·5 M
100
K
'5 M
H T+
MICROPHONES AND MIXING CIRCUITS
The two valves might well be a duo-triode such as ECC35 or
6SN7 but with greater risk of mismatch. In order to avoi d limited
output due to low value of effective anode load, isolating resistors are
inserted so that the effective anode load for each valve is almost that
of the anode load resistor.
Figure 13/13 shows how pentodes can be incorporated in the same
HT +
I N PUT
A
'2 5 - 2
hi
OU)T
I PUT
FIG. 1 3 / 1 3.-Mixer circuit
e m p l o y i n g two p e n t o d e s
giving high efficiency due to
low input capacitance and
avoidance of unused valve
loading.
INPUT
B
mixer circuit and a high stage gain achieved, due to the high anode
resistance of the pentodes.
Crystal Mic. Mixer.
A circuit schematic of double microphone mixer suitable for crystal
input is given in Fig. 1 3/14.
HT +
25K
50 K
I
MIC
�
2M
2M
MIC
2
FIG. 1 3 ! 14.-Mixer circuit for crystal mics.
Cosmocord
There are several types of non-electronic mixers employed in studio
technique, normally with low impedance lines which give constant
input and output impedances ; thus they may be connected in parallel
with little ill effects. One type, the " T " attenuator, is shown in
Fig. 1 3/ 1 5 ' As RI becomes less, R2 becomes greater, giving constant
1 50
MICROPHONES AND MIXING CIRCUITS
input impedance. As Rz becomes greater R3 also becomes less, giving
constant output impedance.
FIG. 1 3 / 1 5 .-" T " Attenua­
tor for constant impedance.
The inclusion of carbon-track volume controls in the grid circuit
of an early valve is very questionable technique, due to the liability
of noise. When low-level microphones are used, with little possibility
of the input overloading the valve, the control could well be placed
later in the circuit where mixing takes place.
RECAPI TULATI ON
In the present chapter the performance of the best types of micro­
phone in each class has been examined, although it is realised that for
many uses a cheaper instrument may answer the purpose to better
effect.
The problem with microphones is an exact miniature of the loud­
speaker problem-to cover the audio range without resonance. It is
significant that moving coil microphones can be produced with a
piston-like action up to 1 5 Kc/s. If this could be done in a loudspeaker
(with reasonable acoustic output) there would be no necessity in future
to refer to the loudspeaker as the weakest link in the chain.
In the case of the crystal, the type fitted with a small diaphragm has
not been mentioned. This gives much greater voltage output but not
with the freedom from resonances associated with the sound cell type.
Some indication of the cost of high quality microphones has been
given, to which must be added the cost of the extra amplification that
follows from their low output, and the cost of the precautions which
must be taken against hum. There would be no point in buying a
microphone which is too good for the job.
It is interesting to note that BBC engineers are seriously concerned
with response up to 10 KcJs only. One can but agree that flat response
and good transients with little phase shift up to that frequency sound
very well.
Very few people can hear above 15-17 Kc/s. We understand that there
is no truth in the assumption that people with bats in the belfry can
hear ultrasonic frequencies .*
*
Not scientifically proven as yet !-H.H .G.
151
CHAPTER r4
POWER S UPPLIES
Supplies for powering receivers and amplifiers fall under three main
categories: Ca) Battery, both primary and secondary ; Cb) Mains, both
AC and DC ; Cc) Rotary generators, motor or hand driven.
BATTERY SUPPLIES
Unless the experimenter with no mains supply is an Atlas or a
Croesus, he will have to content himself with a modest power output,
as many watts mean either frequent carrying of dead weight secondary
cells or buying and running a petrol electric set. If a modest output
of seo milliwatts will satisfy, consideration of ways and means can be
undertaken. The technical service of Mullard Limited has recently
issued a design for a push-pull amplifier giving 450 mV output with
HT voltage of 90 and current at 7 mA, with LT at r · s volts 250 mA,
which is a reasonable load for the modern dry battery. This amplifier
employs DL94 pentodes. If a pair of DL92 pentodes are used the
output is 780 mW with 6 per cent. distortion for r7 mA of HT at
90 volts.
The lead/acid accumulator is the best answer for filament heating
and also provides a good solution for a stable HT supply, although the
initial expense is heavy and its life somewhat uncertain. The pre-war
scheme of hiring HT accumulators had much to commend it.
The use of vibrator HT supplies from 2-volt LT batteries is quite
a workable solution. The conversion efficiency is reasonably high, but
a large 2-volt accumulator is needed to cope with the current demands.
MAINS SUPPLIES
Cr) HT SUPPLY
Figure 14/r shows the elements of a rectifying circuit, a single diode
O UT P U T
I
---"
-
LOAD
.,--------'
FIG. 14/r .-Single phase Half
Wave Rectifier. The " load "
is the HT current drain of
the amplifier.
connected across an alternating voltage in series with a load. When the
polarity of input makes the anode positive with respect to cathode, the
POWER SUPPLIES
diode will conduct and current will flow through the load, producing
a voltage across it as shown graphically in Fig. 14/2b.
IN PUT
-" V O L T S
t
V O LT S
/ACROSS
LOAD
FIG. 14!2.-Voltage from Half
Wave Rectifier.
V O L TAGE
W\TH
CONOE N S E R
cc)
The voltage due to the pulses of unidirectional current will be quite
unusable as a source of HT for an amplifier. The addition of a large
capacity reservoir condenser across the load will help towards pro­
ducing a steady voltage, the condenser charging during the conducting
regime in the diode and discharging during the non-conducting period,
Fig. I4/2C. If this voltage is used to supply HT to valves, a steady hum
will be heard in the output, at the frequency of the input voltage,
usually 50 cycles. The addition of a filter circuit comprising L and C2
as shown in Fig. 14/3 will reduce ripple. The choke due to its self­
inductance will oppose any rise or fall of the current through it, and
the second condenser will further eliminate ripple by storing a charge
during the troughs.
The circuit of Fig. 14/3 is referred to as a " condenser input " filter
and is by far the most common form of filter for general use. The
10-20 I-l
L
R
FIG. 14!3.-Condenser Input
Filter.
condenser Cl will influence the peak emission delivered by the valve
during the charging period. If Cl is large, say 16 or 32 [LF, a very
heavy current will flow into the condenser, limited only by the internal
resistance of the diode and input voltage, both low. When using a
half wave rectifier as described, valve makers usually recommend
that a small limiting resistance of say 50 ohms be included in the anode
153
POWER SUPPLIES
circuit to act as a surge current limiter. This resistance has been known
to save a valve from complete destruction when Cl' usually an elec­
trolytic type, has developed a short-circuit or when the output has been
accidentally shorted. The omission of the resistance may lead to
premature ageing of the rectifier valves due to destruction of their
emission. The temptation to omit the resistor is strong when the
filtered output voltage is usually a bare 200 volts at full load for a
230 volts input, but a gain of 5 volts in this way is unwise.
If the circuit of Fig. 14/3 is modified by the omission of Cl> it
becomes a " choke input " type. The absence of Cl relieves the
onerous position of the rectifier having to supply high values of
instantaneous current, as the inductance limits the peak values. Thus
this circuit is of value when higher mean currents are required to be
drawn from the rectifier, but it has the disadvantage of being less
efficient, as the presence of the choke does not allow the voltage across
C2 to build up to so high a value. In short, it gives a smaller output
voltage than the condenser input filter.
v
'"
,. .
'
.
. . . . .. . .
CHOKE
.
FIG. 14/4.-R e g u I a t i o n
Curves of choke input and
condenser input circuits.
. .
.
I N PU T
. � "
"
"
. . . . . .
ma
Figure 14/4, showing the comparative behaviour of choke input and
condenser input circuits, reveals that the regulation of the former
is better, i.e. the output voltage does not vary so much for changes in
current drawn. This suggests that when an amplifier is using class
AB, AB2 or B output with large anode current swings, the choke input
circuit is preferable, but where the accent is on quality of reproduction,
the use of such output stages is not envisaged.
ACjDC TECHNI QUE
If the input voltage of Fig. 14/3 was DC instead of AC, and provided
the input terminal connected to the anode of the rectifier valve was the
positive leg of the mains, the valve would be permanently conducting
and would behave as a low value of series resistance. Should the mains
connection be the other way round, the valve would not conduct and
the apparatus it was expected to supply with HT would obviously
fail to work. However, the condensers Cl> C2, usually electrolytic,
would be guarded from destruction by being connected to a high
voltage source in reverse. On purely DC apparatus the inclusion of the
1 54
POWER SUPPLIES
valve is not usually considered foolish as its cost can well be saved and
elbow room easily provided by using electrolytic condensers instead
of bulky paper condensers of large capacity.
The type of half wave rectifier developed for ACjDC use will have a
fairly high wattage heater. This is due to the fact that a copious
emission from the cathode is required. The heaters of valves used in
ACjDC sets obtain the requisite heater wattage by high heater voltage
and low current, thus when they are connected in series across the
mains together with a suitable device to absorb surplus voltage, the
power consumption is not large and the power lost as heat in the
voltage dropping device is not very wasteful or embarrassing.
When heaters are series connected, all being of the same current
consumption, the voltage across each heater should automatically
adjust itself to the correct value, but a check should always be made.
Particular regard should be paid to the rectifier as it is most important
that the cathode of any rectifier should be maintained at its correct
working temperature to avoid damage to the emissive surface.
It will be appreciated that the use of directly heated cathodes would
not be practical due to the excessive length of fine and fragile filament
that would be required.
PRE · A M P.
A M P.
OUT P U T
RECT I F I E R
R
INPUT
M A I NS
FIG. 14/s.-Arrangement of Valve Heaters in ACjDC
Amplifier. Early valves at low potential end of chain.
Figure 14/5 shows the arrangement of heaters in series. The earliest
valve in the amplifier is connected to chassis or " earthy " end of the
chain to ensure the smallest possible chance of hum pick-up. R of
Fig. 14/5 may be of several types.
VOL TAGE DROPPING DEVICES
Wire wound resistor on ceramic or other heat resisting former,
I.
with tappings for different mains voltages and adjustable band for
adjusting resistance to value required. Known brands are made with
a low temperature co-efficient alloy and their " hot " resistance is not
much greater than their " cold ", but the " bargain " type wound with
any old wire can give wide resistance variations usually resulting in
serious under-running of the rectifier heater.
I55
POWER SUPPLIES
" Line Cords ". These comprise a length of resistance wire
spirally wound on an asbestos core, overlaid with asbestos and run in
the same sheath as the mains lead. Chief characteristics are that they
are completely unreliable, dangerous and a trap for the uninitiated.
Unreliable in that the internal wire soon breaks. It is often twisted up
again with lots of black tape, zinc oxide tape, corn plaster or boot lace
to complete the repair. Dangerous ; A short in the set will cause them
to overheat. They are hated by fire insurance actuaries, but beloved
of morticians . Several cases of " shortening the flex " by tidy-minded
people have given joy to local radio valve stockists. Verdict : Generally
nasty and should be illegal.
3· Barretters. Usual form is of iron or iron alloy filament in
atmosphere of hydrogen reminiscent of early drawn wire vacuum
lamps. Give excellent compensation for voltage changes over wide
limits, as much as 90 volts, but are fragile.
4· Household Lighting Lamps. In spite of the fact that there is a
momentary current surge on switching on (because the resistance of
the filament increases as it gets hot) a lamp gives excellent barretter
action for reasonable mains fluctuations, provided it is chosen to run at
approximately ·6 of its rated supply of volts. Many thousands of such
lamps were used by Ambassador Radio in a particular wartime ACfDC
receiver, without a single recorded failure of filaments over a period of
years.
Incidentally, a carbon filament lamp has a negative temperature
coefficient ; its resistance goes down as the filament becomes hot.
2.
GENERAL NOTES ON AC/DC WORKING
The danger of such sets without a transformer to isolate them from
the mains, is that a serious shock may be received on touching HT
minus. This is usually thought to be safe in conventional sets, as it is
connected to chassis and earth. The danger arises in this way. One leg
of the mains, AC or DC, is earthed and there is DJ guarantee that the
" earthed " leg is connected to chassis ; thus, in the absence of a
transformer, the chassis can easily be 230 volts " hotter " than the
floor, and a considerable shock can be experienced even through a
high resistance earth return made through your feet, or worse still
through the hot-water radiator that is likely to be too conveniently
to hand.
When constructing ACfDC sets, any metal part that can be touched,
such as a metal gramophone tone-arm, must be connected to a good
earth and isolated from the chassis . It should be noted that even if the
earthy side of the mains is connected to the chassis it is forbidden by
supply authorities to earth the mains at the consumer's end, so a
direct earth connection cannot be made to the chassis.
On some DC distribution systems the positive leg of the mains may
POWER SUPPLIES
be earthy on one side of a street, and whilst it is possible to treat HT +
as the low potential " earthy " part of an amplifier, the usual practice
of making HT - the " earthy " side precludes the earthing of a chassis
under these conditions, even if it were permitted.
If there is a guarantee that the exposed metal parts will be earthed,
these parts should not be connected to the chassis even by the smallest
of condensers, · because on AC mains, should the earth connection
come adrift quite an unpleasant shock can result through a condenser.
Bakelite tone-arms often have metal back bearings with a grub-screw
or pin appearing on the surface. These should be free from chassis.
The pick-up head may have a metal case connected to a common
earth connection on the output ; check this and isolate if necessary.
The screening braid of modern pick-ups is usually isolated from the
head with a view to ACfDC applications.
Cover all grub-screws on control knobs with a thick rubber band,
bind with tape, or fill the grub-screw hole with sealing wax. Any
space between knob and face of cabinet exposing a short length of
metal shaft should be filled in with felt washers, or better still the
metal spindle should be sleeved with rubber tubing.
This may seem grandmotherly advice, but the author has seen many
cases which amounted to a threat of sudden death waiting for an
unfortunate child or unwary person, and a great deal of thought must
go into the final version of any ACfDC apparatus.
A final word in this strain-is the third pin of your power point a
reliable and safe earth ?
2.
LIMITATIONS IMPO SED BY LACK
OF HT VOLTS
In ACfDC sets one can usually budget on having 200 volts of
ripple-free HT available. Can a quality amplifier be produced to
work at this voltage ? The answer is yes, as two Mullard UL41 output
pentodes with 200 volts HT will provide 12'5 watts with 4 per cent.
distortion. The distortion is, of course, less at lower output levels.
V OLT AGE DOUBLER
The use of voltage doubler circuits to obtain higher HT voltages
will probably occur to the reader, but it must be pointed out that the
regulation is generally not good and will only give an output voltage
twice that of the input voltage at moderate current drains .
The development of the Mullard PZ30 special voltage doubler has
produced a valve that is stated to give 480V at 200 mA from 240V
input, but earlier types have been prone to fail through overheating.
The application here is not ACfDC ; it is transformerless AC.
The circuit is undeniably useful for providing final anode voltages
157
POWER SUPPLIES
for Cathode Ray Tubes and Television Tubes where only low current
is called for.
AC MAINS TECHNIQUE
When designing apparatus for use on AC mains only, one has a
very much freer hand than when DC mains have to be taken into
consideration. First of all the use of double wound transformers to
give co�plete isolation from the mains supply reduces the problem
of ensurIng complete safety. Secondly, one is not tied down to any
maximum HT voltage, as any voltage can be obtained by the use of
transformers. Thirdly, the range of available valves is vastly greater.
The use of single phase half wave rectifying circuits is not con­
templated, and the half wave bi-phase rectifier, or full wave rectifier
as it is more commonly called, is used.
01
D2
A
+
LOAD
L. o ·)
(a)
FIG.
B
c
INPUT TO D I
.... I N PUT TO
G
D2
CHARGES I N T O
RESERVOIR COND�
FROM 0 1 S. 0 2
VOLTAGE O N
RESERVOIR CONO�
Cb)
14/6.-Basic bi-phase half wave Rectifier Circuit.
Figure 14/6a shows the basic circuit of a full wave rectifier. A centre
tapped secondary is used on the transformer. The voltage appearing
across AB will be equal to but in anti-phase to the voltage appearing
across BC. When point A is positive with respect to B then diode Dl
conducts and current flows through the load in the direction DB.
At the same time the point C is negative with respect to B therefore
the diode D2 is non-conducting, however half a cycle later the situation
is reversed and diode D2 conducts allowing current to flow through
the load in the direction DB which is the same as before. If a reservoir
condenser is connected across the load DB, it will receive a charge
every half cycle as shown in Fig. 14/6b. Comparing this with the
half wave arrangement it will be seen that the condenser has only
half the time in which to lose its charge, and the amount of ripple
superimposed on the DC will be correspondingly less . Furthermore,
being at double the frequency (100 cycles), it will be correspondingly
easier to filter out.
POWER SUPPLIES
The two diodes of Fig. 14/6a are more usually combined in one
envelope with a common cathode as shown in Fig. 14/7 ·
+
FIG. 14/7.-Usual form of
full wave Rectifier Circuit.
o
<!
o
-'
The inclusion of a series resistance in the anode circuit as in the
half wave system is no longer necessary as the inevitable resistance
of the transformer secondary provides the necessary limiting in practice .
There are two main types of valve rectifiers, the hard vacuum and
the gaseous. Generalising, the hard vacuum types are used for currents
up to 250 mA, beyond which it is usual to employ the gaseous types,
which are principally mercury vapour.
The vacuum types may be sub-divided into:
(a) Half Wave-indirectly heated as already explained.
(b) Full Wave-directly heated-verging on obsolete.
(c) Full Wave-indirectly heated, usually to be preferred and
practically universal now, because they heat up at the same rate
as the other valves in the set and avoid high voltages being
applied to the condensers which occurs with directly heated
valves. (The voltage applied to the condensers will be 1 ·4 times
the RMS value of the transformer voltage, until current is
drawn e.g. a 350 volts transformer will give 490 volts on no
load.)
(d) Full Wave-indirectly heated " Car Radio " types. When a
valve is used for car radio applications it is desirable to run the
heater from the car battery, yet the cathode is connected to
HT + so that heater to cathode insulation must be capable of
withstanding the full HT voltage. American 6X5 and Mullard
EZ35 are examples, both having 6·3-volt heaters. This is a
useful type of valve when one is short of a rectifier heater
winding .
THE GASEOUS TYPES
OPERATION
With the cathode heated, a gradually increasing positive voltage is
applied to the anode. At first a small anode current passes (Fig. 14/8),
1 59
POWER SUPPLIES
limi.ted by the internal resistance of the valve, exactly comparable to
a hIgh vacuum type.
fa
FIG. 1 4/8.-Anode Charac­
teristic of a gaseous Rectifier.
:15v approx.
Va
When about IS volts is applied to the anode the whole behaviour
of the valve changes ; the space current suddenly becomes great, as
great as the emission of the cathode will allow. The inside of the valve
begins to glow at this instant due to the ionisation of the gas molecules
by electron bombardment, the positive ions released neutralising the
space charge around the cathode and making the internal resistance
of the valve of negligible proportions. The commonest gas employed
is mercury vapour, and the valve exhibits the well-known ghostly
glow . The Osram GU 50 is probably one of the best known valves of
its type, being capable of passing 250 mA at 1,000 volts.
The chief characteristic of this type of valve is its low internal
resistance, lower than that of the high vacuum type. This reduces the
voltage drop across the valve, with of course better regulation, which is
also assisted by the fact that the voltage drop (IS volts) is practically
independent of the current.
Certain precautions have to be observed in the use of these valves :
(I) The anode voltage must not be applied until the cathode has
attained working temperature, usually done by using a thermal
delay switch. The reason for this is two-fold-the emission
may be destroyed by taking current before it is actively emitting,
and metallic mercury must be volatilised or an internal short
may occur.
(2) The temperature of the cathode must not fall below the specified
minimum value.
(3) Choke input filters must be used as the peak charging current
into a condenser input filter would be very high due to low
internal resistance and might result in cathode disintegration.
(4) The ambient temperature must be within specified limits,
rooC to 40°C is a usual range, or internal flashover may occur.
(5) Radio interference is produced, as with fluorescent lighting
strip, but this can be suppressed by the judicious application
of small condensers between anode and cathode of the valve.
r60
POWER SUPPLIES
METAL RECT IFIERS
The so-called " metal " rectifiers are divided into two classes, the
copper-oxide and the selenium types, and comprise a series of coated
discs assembled to form a stick of units.
These rectifiers can take the place of a rectifying valve in practically
all applications with an immediate saving in that there is no valve
heater to supply with current. Further advantages are that they are
very robust, and virtually everlasting providing they are suitably
protected against overload by the adequate fitting of fuses .
!
AC/DC
t
.f'oat.
8ll'O(
HALF WA,VE CIRCUli
REO
�
.F T
8-64
HT +
HT -
911I rn_
�: _�
:::
HT-
HT +
FIG. 14/9.-Typical Metal Rectifier Circuits.
The Selenium type is growing in popularity because of its relatively
small size, low cost and high efficiency. The growing shortage of
rectifier valves with the present nickel limitation is likely to accentuate
the position. As an example, a Selenium (STC) RM4 will provide
250 mA at 250V in a half wave circuit with less voltage drop than a
valve, with no limitation on size of reservoir condenser and no need
for limiting resistors .
Probably the most useful application of the metal rectifier is in the
low voltage, high current field for battery charging, etc., where a
simple arrangement will give IOA at 1 2 volts, which is a grossly un­
economic proposition using valves . Their use for supplying DC volts
for the heating of valves in early stages is admirable. Other useful
applications in audio work are for rectifiers in contrast and compression
units, peak limiters (crash limiters), and grid current protection devices.
They also have advantages to offer when used in conjunction with
vibrator packs in battery work, and will receive mention in this con­
nection. Their use in circuits for supplying fixed grid bias will also
be discussed.
THE ASSESS MENT OF VALUES FOR L & C
IN FILTERS
In a book of this character, it is only possible to give rule-of-thumb
guidance, because the calculation of residual ripple present after a
I6r
POWER SUPPLIES
filter is by no means easy. The professional engineer uses graphical
methods and the interested reader is referred to the Radio Designer's
Handbook edited by F. Langford Smith and published by !liffe & Sons
Ltd., for exact information.
The valve manufacturer will specify the value of condenser for the
reservoir, and after that a little intelligent guesswork, or trial and error,
will give the answer. With push-pull output, the HT supply to the
anodes need not be well filtered due to automatic cancellation in the
output transformer primary. Economy at no sacrifice of quality can
be effected by using 5H to smooth the anode supply, say 5H at 1 50 mA
(lIO mA for the output stages and 40 mA for the rest of the set).
The 40 mA supplying screens of output valves and all earlier stages
can be well smoothed by a 20H choke of modest size. The total cost
of a 5 H 150 mA choke and 20H 40 mA choke will be lighter than for
20H at 1 50 mA, and space will probably be saved. The value of capacity
for the condensers cannot be too large. It should be noted that the
screens of the output valves are supplied with well smoothed HT as
they are susceptible to hum.
The reservoir condenser can be a paper type with advantage, having
a better power factor and being generally more reliable and lasting.
The cause of electrolytics failing is often that they are used in positions
where too high a ripple current flows. The maximum value of this
ripple current is shown in makers' catalogues, and should not be
exceeded, or even approached in the writer's opinion. The lack of
information on permissible ripple current, which could at least be
coded on the condenser, is regrettable. The etched foil type of con­
denser permits of only a reduced ripple current flow. A point not often
made regarding ripple on a power supply is that the ripple voltage is
by no means sinusoidal, and the harmonic components may be cal­
culated by Fourier analysis. Observation of the ripple voltage by means
of an oscilloscope is illuminating. High order harmonics are best
eliminated by a mica condenser of up to 0'01 [LF across the filter
condenser, particularly when the latter is an electrolytic.
OBTAINING EXTRA HT CURRENT
Whilst the gaseous rectifier is a good solution for high voltage at
high current, its use is not always necessary. A single valve, the
Mullard GZ32 is rated to give 500 volts at 250 mA, but should more
than the 250 mA be required, the circuit of Fig. 14/10 can be employed.
A normal bi-phase rectifier is capable of passing 250 mA per anode,
therefore two of these valves in parallel can be used to produce 500 mA
in a bi-phase half wave circuit. It is considered good practice to
inter-connect the anode by a small resistance, then individual differences
in the two halves of the valve do not lead to one diode doing more than
its share of the work.
In transformerless circuits the same technique may be employed
1 62
POWER SUPPLIES
as shown in Fig. 14!I I . This device is often needed because the
average ACfDC rectifier is limited to 120 mA except in the case of
FIG. I 4 / Io.-C i r c u i t f o r
increased current output.
Mullard PZ30 and Mazda U801 which are rated to supply 200 mA
and 300 mA respectively, these having separate diodes in the one
envelope.
FIG. I4 / I I .-C i r c u i t f o r
increased current output on
transformerless circuits.
M ETH OD S
OF
OBTAINING
FIXED
GRID BIAS
Sundry methods of obtaining grid bias at the expense of the full
HT voltage have been described. The use of " self Bias," that is bias
which is dependent on the anode current of the valve, is normally the
one recommended by valve makers for the Class A condition of
operation when swings of anode current in a push-pull pair are equal
in either direction, producing a steady mean anode current, and in
consequence a steady value of bias across the common bias resistor.
To take two examples, a pair of Mullard EL 37's, 25 watt pentodes,
with 325 volts on the anodes, a 1 30 ohms bias resistor giving approxi­
mately 29 volts of bias will produce 35 watts of audio with 4"4 per cent.
total distortion. The same valves with 350 volts on the anode and 31
volts offixed bias, give 46 watts of output with 2 · 8 per cent. of distortion,
this for comparable HT supplies (325 + 29V), against 350V. Two
Osram output Tetrodes, type KT66 in push-pull, anode voltage 390,
self bias - 22'5V, produce 30 watts of output for 6 per cent. total
-
POWER SUPPLIES
distortion. The same valves with fixed bias of
-
40 produce 50 watts
with 5 per cent. distortion.
. Fro� the foregoing,. it can be deduced that if the maximum power
1S reqUlred from a pa1r of valves, they may be driven harder in the
Class AB} condition, but fixed bias must be employed. The variations
in self bias would produce increased distortion. The fixed bias may
�e produce� from a dry battery, which is perfectly satisfactory providing
�ts voltage 1S checked regularly, but human nature being what it is, it
1S safer to use a separate supply from the mains. This can be readily
achieved by using a separate small mains transformer and rectifier.
The additional cost of such a transformer is not strictly necessary.
Fig. 14/12 shows a possible circuit quite frequently employed in
American commercial apparatus, in which a tapping on one half of
the HT secondary is employed in conjunction with a half wave rectifier.
FIG. 14(I2.-Power Circuit
including a fixed GB supply.
Type 6X5 rectifier (or similar) is very convenient because its heater
supply can be the normal 6'3V line to the heaters of the other valves, or
a small metal rectifier can be employed. The latter is probably
preferable, since any failure of a valve would leave the output valves
without bias, and they would quickly come to grief. It is usual to
stabilise the output by means of a bleeder resistance. Really good
filtering of the circuit is essential, particularly as the rectifier is a half
wave type. This is easily done by the use of large capacity low voltage
electrolytics, with a word of warning : the negative of the condenser
is not connected to the chassis, so that it must be isolated or a cardboard
type used. A choke is by no means essential and may be replaced by a
resistance.
The use of a special transformer is a decided disadvantage of this
arrangement, but even this can be overcome by using the circuit of
Fig. 14/13, given in an article by G. R. Woodville (M.O. Valve Com­
pany), in Wireless World of December, 1948.
Independent bias controls for each valve provide a bias variable
from 30-60 volts for balancing purposes. The rectifier is fed from one
side of the HT secondary via a 0'02 IJ.F condenser of high working
voltage. The maximum bias voltage obtainable is determined by the
value of this capacity, and the value of the resistance network.
POWER SUPPLIES
It should be borne in mind that in either of the two bias circuits
described the source of voltage is of high impedance and � ill not be
suitable for use with stages in which grid current flows durmg part of
the cycle.
+��--T--71
50K
[ 21�
BIAS
- 60
SO K
T
-30
L-
'02J"F
lOOK 2 W
0
4)' F
- -- -..,.
60
---- 0-....
- r
3 to
B I AS Z
'WIRELESS WORLO'
OK " ••
�
_
_
_
--' G. •. w" d, '".
_
_
_
_
FIG. 14/ 1 3.-Alternative fixed bias circuit needing no special transformer.
POWER
S UPPLIES
EMPLOYING
VI BRATOR
UNITS
Vibrators consist essentially of a reed caused to vibrate on the
trembler bell principle. The speed of vibration can be adjusted to
50 cycles with quite good accuracy, or a higher frequency, say 100
cycles, may be employed with consequent economy of iron in trans­
formers and smoothing chokes .
FIG.
14!I4.-V i b r a t o r
Principles.
Readers familiar with the operation of the ignition coil of a car will
recognise in Fig. 14/14 the same principle. As the reed vibrates
against the fixed contact a pulse of current will flow through the
primary of the transformer, creating a magnetic field which will cut
the secondary winding of the transformer, comprising many turns
of wire, and a high EMF will be produced across the winding. On
the break of the contacts the field will collapse rapidly and a higher
back EMF will be induced in the secondary. A back EMF will also
be produced in the primary and will attempt to jump the gap between
the contacts, causing burring and pitting of the surfaces. The fitting
of a condenser of suitable capacity across the contacts will absorb this
POWER SUPPLIES
back EMF and so prolong the life of the contacts. As car users will
know, the failure of this condenser reduces the spark almost to ex­
tinction. This is because the presence of the condenser increases the
rapidity with which the field can collapse and the greater rate of
cutting of the conductors by the lines of magnetic force increases the
EMF produced.
The choice of value of the condenser on a car ignition system, usually
about 0'2 [LF, is quite critical, and this is the case in vibrator power
packs . A balance must be struck because on the closing of the contacts
the charge on the condenser will be discharged again in opposition
ELECTRO­
MAG NET
FIG. 14!14A.
Photograph
showing working parts of
vibrator.
-
to the battery. Too large a value of condenser would produce excessive
sparking on discharge and would also hinder the build-up of the field
again. It is usual to fit a buffer resistance across the primary to supply
some damping of this oscillatory circuit due to the presence of both
inductance and capacity.
The system so far described will produce a form of AC across the
secondary and after rectification this could be smoothed and used as
an HT supply.
166
POWER SUPPLIES
Figure 14/IS shows another arrangement sometimes employed in
which the vibrating reed causes pulses to flow through the two halves
s
VIBRATING
REED
)00----1\ 1\ 1 \
'x--<l
.
FIG. 14/15.-Vibrator system
employing
centre
tapped
primary.
of the primary in opposite directions, giving an increased efficiency and
better waveform of AC in the secondary. In the two arrangements
so far described, the rectification system on the secondary would have
FIG. 14/16.-Vibrator system
with valve rectifier.
to be single phase half wave, but the addition of a centre tapped
secondary as in Fig. 14/16 would enable bi-phase half wave, " Full
Wave " to be employed. A normal rectifier valve could be employed.
The types 6XS or EZ3S already mentioned would be admirable,
having 6-volt heaters which could be connected across the battery
actuating the vibrator, and being rated to withstand the full HT
between cathode and heater. The heater current Q'SA imposes an
additional drain on an already severely tried battery and the use of a
metal rectifier is indicated. In some arrangements the rectifier valve
is heated from a winding on the vibrator transformer. This imposes
extra loading on the transformer and vibrator, but is convenient when
the apparatus is intended for battery or AC mains operation. The
addition of a mains primary to the transformer makes the change-over
to mains easy, but again the use of a metal rectifier is the best solution.
Another type of vibrator is self rectifying or synchronous. An
additional pair of contacts is added to the reed together with another
pair of fixed contacts.
The extra pair of contacts will give greater heating of the vibrator
and will lirnit the total wattage it can handle. The enemy of the vibrator
reed, which is made of spring steel, is heating with loss of temper.
When it ceases to function, much generation of heat takes place unless
adequate fusing is fitted. The trend of design is against the self­
rectifying type as it proved fickle under Service conditions. Manu-
POWER SUPPLIES
fa�uring �ifficulties in contact alignment have undoubtedly con­
tnbut:d to. its fall, as the usual types are most reliable, providing correct
buffenng is fitted to reduce contact sparking to a minimum (inci­
dentally improving output waveform).
A complete circuit for suppression of sparking and elimination of
radio interference is given in Fig. 14/17.
�T
+
IO-20H
=
75 loo n
f
RFC
75 loo n_
· 1,.. F
I1I1
LT+
LT-
·ll"F
(CLOSE TO
V I B RATOR
BASE.)
j
FIG. I 4/I7.-Complete vibrator pack. Metal rectification
is employed for reasons of valve heater economy. Note
RFC's + bypass condensers to eliminate RF ripple.
VIBRATOR MAINS CONVERTORS
A range of vibrator convertors has been produced for any value of
DC input between 6V and 250V to give 50 or 60 cycle AC output at
I IO or 230 volts. They have a high conversion efficiency and appear
to give long service before the renewal of the vibrator unit is required.
Prices are eminently reasonable and give the sufferer of DC mains or
house lighting plant a chance to compete with AC mains.
ROTARY CONVERTORS
I.
HT SUPPLY
A wide variety of rotary convertors were produced for Service use,
and providing brush gear and commutators were given attention
proved exceedingly reliable, but of lower conversion efficiency than
the vibrator (60 per cent. against 85 per cent.). The usual input
voltages are 6, 12 and 24V. Models were produced for outputs up to
1,100 volts DC. A type is available for I IOV input which may offer a
crumb of consolation to those on 1 I0V DC mains . A small condenser
·001 flF between each brush holder and frame is usually fitted and
2 flF across the output is sufficient to remove ripple. RF filters may
have to be fitted on both the low and high voltage sides, but the prob­
lem is vastly easier than with vibrators.
A point worth mentioning for those on DC mains is the use of a
1 68
POWER SUPPLIES
230 or I IO volts output machine as a battery charger. The output
side can be connected to the mains, the low voltage side to the battery
which will supply the excitation of the shunt field, and receive a charge.
A limiting resistance is necessary on the mains side.
On the care of commutators the best rule is no oil and no abrasives .
Keep clean with carbon tetrachloride and in extreme cases a piece of
the very finest grade of glasspaper, 00, may be applied-never emery.
Do not scratch out between the sections or a lip will be raised that will
act like a rotary planing machine against the brushes. Oil bearings
regularly.
2.
MAINS CONVERTORS
Rotary convertors to produce AC are available for those on DC
mains of any voltage, and are completely satisfactory except that if
loaded too lightly the nominal 50 cps output is very nominal, and
induction type gramophone motors behave accordingly. The initial
cost is high, but their life with care is very great.
For those on 24-volt supplies a useful ex-Government convertor
giving 230V 50 cycles 100 VA is available. Samples have been given a
frequency check against the normal grid supply with every satisfaction
-especially in cold weather!
THE PROTECTION OF POWER S UPPLIES
FIG. I 4 / I 8 .-A rectifier circuit showing fitting of fuses.
In Fig. 14/ 1 8 a power pack is shown with suggested positions for the
insertion of fuses to afford protection to the rectifier valve and mains
transformer. F1, included between the rectifier and the reservoir
condenser, is an insurance on the valve in case of a breakdown in the
reservoir condenser, particularly if an electrolytic type is used. F2 is
an alternative position affording equal protection and with the probable
advantage that it protects the transfonner should an internal short
develop in the valve. The fuses Fa and F4 are a fire insurance should
an internal short develop in the transfonner. The use of a fiashlamp
bulb in positions Fl and F2 is deprecated due to danger of shock when
searching for trouble. The type of fuse-holder recommended is either
the screw-in type common on Service apparatus, or the type where the
fuse is contained in the lid of the fuse-box. Open type fuse-holders
are dangerous in moments of stress.
The value of HT fuse can usually be rated to blow at two to three
POWER SUPPLIES
times normal current. The lower value will sometimes blow if the
apparatus is switched off and on again immediately, due to charging
surge whilst the valve is still warm.
Circuits employing metal rectifiers should receive exactly the same
attention as those for valves.
Circuits employing vibrators should receive the same attention on
the HT side, and the primary side must also be carefully fitted with
fuses to afford protection when a vibrator sticks. Due to the com­
paratively high current in these circuits, the fuse fitting must be
capable of affording really sound contact to avoid serious voltage drop
(0'6 volt on one 6-volt circuit examined). The gauge of wire fO! the
battery lead should always err on the generous side ; a little investiga­
tion with a voltmeter across the ends of the lead will sometimes yield
surprises.
A very neat little device containing two fuses and two neon
indicators is being brought out by Belling & Lee, Ltd., and is
illustrated in Fig. 14/ 19.
FIG.
I4/I9-Twin neon In­
dicator fuse-box.
The fuse-box takes standard cartridge fuses I t InS. x t in.
The photograph is approximately actual size.
CHAPTER 1 5
HUM AND NOISE IN AMPLIFIERS
As this book has progressed, mention has been made of precautions
that should be taken to avoid hum and noise when the circuit described
was known to be prone thereto, but a thorough recapitulation would
not be out of place here, particularly as it may serve as a useful guide
should trouble-shooting be in progress.
The main sources of hum are dealt with under the following eight
headings, in the order of probable likelihood.
I . Lack of sufficient filtering of the power supply, due to inadequate
induct'ance or capacity, or both. The remedy is obvious, starting with
the addition of further capacity across the filter condenser, but not
across the reservoir as the rectifier may be damaged. If the addition
of this capacity reduces hum but does not cure it, the next step is to
find out where the hum is getting into the amplifier. Remove the
penultimate valve, or valves if a push-pull driver is in use, and if there
is a large reduction of hum or almost complete silence, the trouble does
not lie in the output valve or valves or the HT filtering. If the trouble
proves to be in the output stage and the valves are pentodes or tetrodes,
add extra capacity between screens and ground. If the valves are
directly heated, such as PX4's, the centre tap on the heater winding
may not be accurately made, and should be artificially located by the
use of a potentiometer or " humdinger " as shown in Fig. 15/1. Adjust­
ment of slider will find a point where hum is at a minimum.
lO-loon
Wlr\! wound'
H E AlER
'W I N O I N G
O N MAIMS
TRAN.SFORMER.
--..
FIG. 15/I .-Artificial location
of centre tap on heater
winding.
If the output stage is push-pull, unbalance of current in output
transformer windings due to bad mis-match between the valves may
be the cause of hum. Check anode currents with both grids earthed.
If removal of earths on grids causes change in anode currents suspect
.
gnd-earth return paths. If removal of earth on grids increases hum
HUM AND NOISE IN AMPLIFIERS
and an inter-valve transformer is used, pass on to section 2. If no cure
is found now, pass on to sections 3 and 4.
If the output stage is blameless, replace penultimate valve and earth
its grid. If hum reappears, add filtering to anode circuit say 20K and
8 flF, or increase capacity of condenser or value of resistance if filtering
is present. If no cure, change valve as fault may be one described in
section 3. If penultimate stage is concertina or cathode coupled type
of phase splitter, suspect heater hum as described in section 5. Troubles
of sections 3 and 4 may be present.
If penultimate stage with earthed grid is blameless, remove earth ;
if hum reappears suspect HT filtering of previous stage, or leaky
coupling condenser. Increase or add filtering, replace coupling
condenser.
Replace pre-amplifier and earth grid if hum now appears, check
HT filtering to this stage, add C or R. If pentode, increase screen
bypass condenser up to 8 flF. If removal of earth on grid produces
hum, suspect input wiring. See Fig. 15/2, for schematic of wiring to
an early stage.
F I G . I 5 ! z . -All " earthy "
points returned to single
point on chassis.
It is imperative that all " earthy " points should be bonded to chassis
at one point only, otherwise small ripple voltages will be introduced
into the amplifier, due to eddy currents in the chassis producing slight
differences in potential between two points on the chassis surface.
The chassis material can influence the spread of electro-magnetic
hum. Magnetic materials are commonly employed and strong eddy
currents exist throughout such substances. These can be virtually
eliminated by u&ing a dia-magnetic material such as aluminium,
duralumin, or best of all copper which is even silver-plated in the highest
grades of professional apparatus.
The use of an earth bus bar to replace the common technique of
soldering down to the chassis at various points has much to recommend
it. It is common practice to earth one side of the heater pins on the
valve base, Fig. 1 5/3a. This latter earth connection should not be the
same earth point as the one described in the preceding paragraph.
HUM AND NOISE IN AMPLIFIERS
Modification of the valve heater circuit to the arrangement shown in
Fig. 1 5 /3b sometimes reduces the residual hu� in an amplifier, but in
the writer's experience is rarely necessary wlth mo�ern valv.es. �he
device shown in Fig. 1 5/3c is sometimes helpful m removmg hlgh
order harmonics of hum in an early amplifying stage .
a
FIG.
1 5 /3.
Ca) One side o f heater wind­
ing earthed.
Cb) Centre
tap on heater
winding to reduce heater
hum.
b
Cc) Bypass condenser at each
valve holder to avoid
hum due to high order
harmonics o n mains
supply. Also avoids interstage coupling.
It has been assumed that the wiring of the valve heater circuits has
been properly carried out by careful disposition of the heater wiring
close to the chassis, avoiding close proximity to grid and anode pins
on valve bases. The use of a twisted pair of wires for heater wiring
reduces stray external fields .
If attention to all these points fails to reduce hum, sections 3, 4 and 5
should be referred to.
2. Hum may be introduced by electro-magnetic coupling from the
mains transformer to other iron cored apparatus such as choke, output
transformer, inter-valve or input transformer for pick-up or micro­
phone, iron cored scratch filter inductances or even the air cored variety.
This is a problem of design layout. The filter choke and mains trans­
former are usually not far apart. They should be arranged in such a
way that their external fields are mutually at right angles, and also
the output transformer, as in Fig. 1 5 /4.
In one piece of commercial apparatus tested by the writer, the hum
level was only 28 db below the full audio output. Removal of the
rectifier valve reduced the hum level to - 34 db, but re-positioning of
the output transformer reduced the hum to - 42 db, which was toler­
able. This is a bad case and should never have happened.
1 73
HUM AND NOISE IN AMPLIFIERS
An inter-valve transformer should be placed well away from the mains
transformer and smoothing choke ; it usually pays to attach flexible
leads and orientate this component for the position of minimum hum.
M A I N S T R A N S FO R M E R
~
-'"
VE R T I C A L
OUTPU,
TRANSFOR MER
MOUNT E D
C
••.
/'
CHOKE VERTICAL
I
FIG. I s/4.-Cores of com­
ponents arranged mutually at
right angles to avoid interaction of fields.
Transformers used with microphones or pick-ups and a high gain
amplifier are particularly liable to pick up stray fields and must be
located well away from any other apparatus. It is usually necessary to
employ a Mu-metal screening box to afford electro-magnetic screening.
Mu-metal is a specially treated nickel alloy which has a particularly
high value of permeability. The permeability of a vacuum is taken to
be unity but that of any other " non-magnetic " material such as air,
I is accepted for
wood, copper, etc., is so nearly the same that fL
practical purposes. As indicated in an earlier chapter, magnetic
permeability is analagous to electrical conductivity and, just as the
perfect insuiator does not exist, there is no such thing as a " flux
insulator ". But flux will always tend to flow within or through a high
permeability material, i.e. take the path of lowest " magnetic resistance "
(Reluctance). Thus ifa transformer is enclosed in a Mu-metal box it is well
protected against external fields. The thicker the protective cover the
length
.!
greater will be the shielding effect as reluctance S =
X
.
area
fL
It should be noted that Mu-metal is a proprietary name of Telegraph
Construction & Maintenance Co. Ltd. Another brand is Permalloy
C and both are 76-79 per cent. nickel with small percentages of copper
and molybdenum or chromium or manganese.
BBC amplifiers introduce a negligible amount of hum, yet the gain
employed between a ribbon microphone producing but a few micro­
volts and the final modulators producing kilowatts of audio output is a
fantastic figure, possibly one billion times or 240 db. But all BBC pre­
amplifiers use transformers due to the necessity of standardising
outputs and inputs to 600 ohms for connection to telephone lines.
This proves that hum pick-up in transformers can be avoided by
adequate precautions.
3. Hum may be introduced into a circuit by leakage between the
heater and cathode of a valve. Most valve testers have facilities for
checking this insulation, which should be at least I megohm. The
=
174
HUM AND NOISE IN AMPLIFIERS
leakage may not be and often is not apparent unless the valve is warm
so any tests must be made bearing this in mind, and also the leads of
an ohmmeter should be so connected that the cathode is negative with
respect to heater or the issue may be clouded by heater emission,
see section 5.
Internal leakage in valves is not very common, but a baffling fault
that has convinced the writer that it is unwise to assume any piece of
apparatus is perfection, is that of leaky valveholders. Certain moulded
valveholders show leaks between adjacent pins which reduce the
insulation to a figure as low as 2 megohms. Thus a cathode pin,
often close to the heater, may inject 50 cycles AC into the circuit.
This can and does cause bad hum in a stage at the beginning of a high
gain amplifier.
4. If a valve, particularly an early one in an amplifier, is placed
in the magnetic field of a mains transformer, choke or energised
loudspeaker, it is possible for hum to be induced in the amplifier due
to the velocity modulation of the electron stream.
5. Thermionic emission can take place from a valve heater, and
if the adjacent cathode is positively charged with respect to the heater,
it will behave as an anode and collect electrons modulated with a
50-cycle ripple. As mentioned earlier, this effect can be noticed
particularly when using concertina and cathode coupled phase splitters,
but the likelihood is always present when cathode biassing is employed,
raising the cathode positive with respect to heater. Cures have been
discussed when dealing with the circuits in Chapter 9.
6. Induction can readily take place between a high impedance
circuit and a neighbouring conductor. If the mains lead is taken near
to the grid of an early valve, the effect will be readily observed. This
gives point to the strictures on careful disposition of valve heater wiring.
The general line of approach is the screening of " hot " leads, such as
early grid leads, and the employment of electrostatic screening cans
around valves not already metallised, screened top cap connectors, etc.
This type of hum is easily recognised since it usually contains a large
proportion of higher order harmonics to which the ear is more sensitive
than to the low fundamental frequency of 50 or 100 cycles.
7. Capacity must necessarily exist between the heater and cathode
of a valve, usually kept to a low value, but it is one way in which a
trace of hum can be introduced into a stage.
The idea of " neutralising " hum injected in this way, or any other
way, is worth mentioning, although it is a " brute force and ignorance "
method. A small capacity, usually consisting of a pair of wires twisted
together for a few turns, is used to inject into cathode, grid or screen
circuit a trace of 50-cycle ripple from one side and then the other of
the heater circuit.
8. It is not common practice to use electro-magnetic speaker fields,
1 75
HUM AND NOISE IN AMPLIFIERS
now that permanent magnets have reached such a high degree of per­
fection, but the introduction of hum from this source should be
mentioned. The speaker field may be energised in one of three ways.
Firstly, by having its own rectifier, when hum would be heard without
connecting the speaker to the amplifier. Secondly, the field coil may
be connected across the rectifier output of the amplifier. To check if
hum is due to the field, the speech coil is disconnected from the second­
ary of the output transformer which should be loaded temporarily with
a resistance about equal to the nominal speech coil resistance. Any
hum now present is due to the field. Thirdly, the speaker field may
be used in place of or supplementing the choke in the filter circuit of
the HT rectifier. To check for field hum the speech coil should be
disconnected and a PM speaker connected to the secondary of the
output transformer.
A hum-bucking coil of a few turns is frequently wound adjacent to
the field winding and connected in series with the speech coil in such
a way as to introduce a hum voltage in opposition to the hum induced
by the field. This is only a partially successful device as phase differ­
ences and the presence of harmonics rarely allow of complete
cancellation.
NOISE IN HIGH GAIN AMPLIFIERS
Random noise quite apart from normal hum due to mains apparatus,
various " frying ", " breathing " and sizzling noises, can be present in
a high gain amplifier. This is due to a variety of causes, some of them
quite outside the control of physicists. Thermionic emission is not a
gentle spraying of electrons from the surface of the cathode like the
spray from a rose-spray hose, but rather like the spray of fat from the
frying-pan when father is cooking the breakfast; a fine haze with
random spurts. This irregular emission accounts for the background
of " breathing " even in the best of amplifiers . In any body that is
warm, i.e. at any temperature above that of absolute zero, electrons in
the atom are in a state of continual agitation and this gives rise to
thermal noises in any component. The use of a refrigerator is not a
complete solution to the problem ; however, neither random emission
nor thermal agitation is a really serious problem.
One of the biggest offenders is the common moulded carbon resis­
tor and its variants, and many an elusive sizzle is removed when resis­
tors are changed, particularly in the grid circuit of the first amplifying
stage. Carbon track variable resistances will also add to the noise,
quite apart from noisy operation when operated.
Leaky decoupling and interstage coupling condensers will add their
quota of noise, and the writer never takes a condenser into service
until the megger has given it a clean bill. Screen decoupling condensers
of 2M " insulation " resistance used in conjunction with a 2·2M screen
HUM AND NOISE IN AMPLIFIERS
resistor will completely upset the working conditions of an amplifier,
and will introduce noise if the leakage is not constant.
Another source of frying can arise in a bad connection to the metal­
lising of a valve. The only cure is to scrape off the metal and use a
shielding can, or change to metal valves. Government surplus valves,
stored under damp conditions, are bad offenders in this respect.
One bitter lesson is that the top cap connection to a valve can give
rise to scratching noises, as the formation of oxide on the cap and
connector, aggravated by heating, produces a partly intermittent
contact. The soldering iron provides the best cure by soldering on
direct.
The use of electrolytic condensers to decouple early stages can
introduce random " noise ", or even instability, due to high power
factor. A new condenser may well have a power factor of 10 per cent.
Naturally this rises with age and may lead to trouble, particularly in
amplifiers incorporating heavy NFB, where a phase shift in the system
at low frequency may introduce a rise in the responsive curve to the
point of low frequency instability.
NOISY VOLUME CONTROLS
Carbon track volume controls are disliked as, besides becoming
noisy in use, the contact is normally so uncertain and the track itself
so unstable that noise is inevitable. Good quality carbon track potentio­
meters fitted to the input of an amplifier with 100 mV input for 25
watts output have proved to be a great source of trouble after only a
few months' use. The practice of using carbon track controls when
DC exists in the circuit is worse, as they become noisy in operation
even more quickly. In certain sets the detector diode load resistor is
the potentiometer controlling audio gain, and noisy operation always
results. When this type of volume control is used it is better to use it
fairly late in the amplifier, certainly after the pre-amplifier.
Audio gain controls cannot reasonably be much lower than 1 00,000
ohms in a grid circuit and whilst this value has been produced in wire­
wound controls, the wire is so fine that the slider rapidly destroys it.
Switch type attenuators are very much more reliable and quiet in
operation.
The professional type of attenuator, as made by Painton & Co.,
comprises a stud switch making certain, noiseless contact, and wire­
wound resistances, controlling gain in steps of 3 db. This type consti­
tutes a virtually perfect control.
A simplified version can easily be constructed, using a good type of
" make before break " wafer switch and good quality carbon resistances .
Such a device was described in Wireless World of February 1950.
Preferred-value carbon resistors are used which are readily obtainable.
With 5 per cent. tolerance, the error at any step of the attenuator is not
likely to exceed about 0'1 db.
177
CHAPTER 16
THE MEASUREMENT OF HARMONIC
DISTORTION IN LF AMPLIFIERS
Although no amateur is likely to have at his disposal the laboratory
equipment necessary for tracing and measuring small amounts of
harmonic distortion, the question cannot be entirely ignored, and a
brief description of some of the methods used is included here.
Perhaps the simplest method whereby the introduced harmonic
distortion of an amplifier may be observed, is to display the input and
output waveforms on a double beam oscilloscope. The time base of
the oscilloscope is adjusted so as to exhibit two or three complete
waves when injecting a steady input frequency signal to the amplifier.
The output display is then moved so as to nearly coincide with the
input display, and the difference in wave shape will be discernible.
A variation of this method involves the use of a simple single beam
oscilloscope without a time-base generator. The input and output
of the amplifier under test are connected to the respective X and Y
plates of the oscilloscope and the resultant diagonal line (at say 400 c/s)
is observed. The introduction of harmonic distortion by the amplifier
will produce a departure from a straight line in the display. The
position of the kinks in the line can be used to determine the order
of the harmonic content.
Both the above methods are only satisfactory where a total distortion
of approximately 8 per cent. is permissible, since at lower figures than
this the kinks do not show up well enough on most commercial
oscillographs.
The standard laboratory method used for measuring harmonic
distortion calls for the use of two pieces of precision apparatus. The
first is a low frequency signal source of sufficient output to load up
the amplifier under test. Naturally the signal from such a source must
be beyond suspicion in so far as harmonic content is concerned. A beat
frequency oscillator is commonly selected for this purpose and the
greatest care is taken to ensure purity of waveform-a figure of o · I
per cent. being about the maximum permissible distortion.
A Distortion Factor Meter measures the total harmonics content
at the output of the amplifier. This consists of a frequency discriminat­
ing bridge network to enable the fundamental test frequency to be
balanced out. Following the bridge comes an amplifier and a valve
voltmeter, the whole being housed in a single container. At the com­
mencement of measurement the bridge network is switched out of
circuit and the amplifier gain adjusted to give a reading of 100 per cent.
THE MEASUREMENT OF HARMONIC DISTORTION IN LF AMPLIFIERS
on the valve voltmeter scale. This then is a measure of the total output
of the amplifier fundamental and harmonic content combined. Now
the bridge is switched into circuit and adjusted until the fundamental
frequency is completely balanced out, the harmonic content alone
carrying on to the amplifier and valve voltmeter. Consequently the
reading on the scale of the valve voltmeter will now be considerably
reduced (we hope). It will be seen that careful adjustment of the
bridge results in a minimum reading. This is the point at which the
fundamental frequency is virtually removed, and the percentage
indicated by the valve voltmeter represents the total harmonic content
of the amplifier output.
Sometimes, where it is desirable to know the magnitude of individual
harmonics, a device known as a Wave Analyser replaces the Distortion
Factor Meter. In this instrument the bridge network is replaced by a
frequency selective amplifier having an extremely narrow bandwidth
(of the order of a few cycles per second) . A superheterodyne type of
tuner and mixer is used to select the individual harmonics, and the
magnitude of each is read on a valve voltmeter scale, as before.
Another method is to measure the intermodulation product. Two
pure tones at widely separated frequencies are fed into the amplifie r,
the output being set at the required power. One of the AF components
is then filtered out and the extent of modulation which remains is a
measure of the distortion produced in the amplifier. As the inter­
modulation product may be four times the percentage of distortion,
it is possible to trace very small amounts of non-linearity. Oscillograms
showing the visual effect of intermodulation were reproduced in
Chapter I .
F.H.B.
1 79
CHAPTER 17
GARNER AMPLIFIER
By G.A .B.
The rather ambiguous heading to the penultimate chapter in the book
does not imply that the Garner Amplifier was made by G. A. B. ;
it simply means that I am writing about the equipment designed by
my colleague, to ensure as far as possible an impartial line.
I suppose I am both well and badly equipped for the task. In my
favour is the fact that I am not commercially interested in amplifiers,
but have nevertheless at my disposal quite an array of test equipment,
ranging from enough AF oscillators to produce a travesty of a string
quartette, to almost perfect free-field acoustic conditions. On the debit
side is the fact that I do not really know anything about amplifiers, *
and I am actually the mysterious correspondent who admitted, in
Chapter 7, his inability to multiply by more than ten.
About half-way through the chapter I pass the ball to H. H. G.,
so that he can score a few goals in outlining the technical merits of his
designs, or kick into touch if necessary.
My opinion is that the best way to obtain a first-class amplifier is
to buy one of the reputable makes. I do not think it is possible to
improve the standard of performance by home construction. But I
also realise that many amateurs like to experiment and build at home,
and it is necessary to round off the book by giving circuits which employ
some of the principles which have been investigated. Our object is
neither more nor less than this, and the designs should be viewed in
the light of this declaration.
A
B
c
FIG. 17!I .-A. 100 cycle note showing burst of oscillation.
B. Square wave at 1 ,000 c!s.
C. Distortion at 15 Kc!s at 5 watts output.
*
Was it A. E. Housman who said that a dog can recognise a rat although
it cannot define one ?
180
GARNER AMPLIFIER
The amplifier and pre-amp were assembled in Chelmsford and were
sent to Bradford for test. A first inspection by oscilloscope, AF
oscillator and square wave generator revea�ed a �light triggered osci�la­
tion a transient over-shoot, and some distortIon at 5 watts at high
freq�encies, which were photographed and are reproduced in Fig. 17/1.
Having recently devised a method of taking loudspeaker response
curves by oscilloscope fitted with camera and moving film (which
also shows up frequency doubling and non-linearity), the system was
applied to the amplifier, and the result appears in Fig. 17/2.
FIG. I7/2.-0scillogram of response of amplifier, exposing distortion at
IQ- I S Kc/s.
It was interesting to find that the distortion snapped at 1 5 Kc/s
in Fig. 1 7/ I C was exposed in the response curve.
The value of this type of response record in exposing instability
and distortion was further tested by running off a curve of the strange
amplifier (not the Gamer or any renowned make !) which had furnished
the pictures reproduced in Figs. 5/2, 5/3 and 5/5 in the chapter on
Instability.
1'!,'-.l���M � � �
Cl
� � � �
50"
1)0
SvJ
"
I
..,
"" "...
",,,
--.c'(
_
"
�
�.
�
.,
... -
� �__
- .I ':"
FIG. I7/3.-Response oscillogram of nameless faulty amplifier. Effects of
motor-boating, HF oscillations and harmonic distortion are included.
Amplifier as used in Figs. s12, S/3 and sls .
�
)
The February 1952 issue of Wireless World contained a brief technical
de� cription of this oscillographic response-recording technique. A
prmt of Fig. 1 7/3 was sent to the Editor as being of likely interest, but
it was returned with the alacrity one might expect if a seditious article
�ere subJ?itted to The Times . They I?ointed out that a full interpreta­
tIon of this complex curve would prOVide someone with several months'
work.
Even so, it was possible to use this amplifier-once the motor-boating
was stopped-without any concrete evidence of trouble beyond an
impression of peculiar tone quality. Removing the treble speaker from
181
GARNER AMPLIFIER
a crossover . network always re-started the motor-boating . CA really
stable amphfier should stand up to any such activities in the speaker
circuit.)
Revenons cl nos moutons. The Garner Amplifier received some post­
natal treatment, including the loading shown on the primary of the
output transformer, and the following actual photographic records of
response, square wave results, LF and HF waveform, all taken by the
writer, should give a reasonable indication of performance to be ex­
pected from the circuits employed. Very good stability is achieved
with 25 db of NFB, but personally I liked the " tone " of the reproduc­
tion with about 15 db feedback ; the bass was warmer as a result of
the lower damping factor. The margin of stability is of course much
greater with the reduced NFB. (Bass lift was later increased.)
It will be observed that the areas of trouble pictured in Fig. 1 7/1
have been cleaned up.
All the oscillograms of Figs. 1 7/4, 5, 6, 7 and 8 were taken with
15 ohms resistance across the secondary of the output transformer.
If'''I'''�''''
\\IIl!wa
o
.....
. ��-
" ��
.. ..,. ..
� ..---.-.� �
� � ....
50 �
100
500 �
1« ,
...... � _
5
,0
2
·
FIG. 17/4.-Response characteristic of final version of Garner Amplifier
(without pre-amp). Note absence of peculiarities exposed in Fig. 17/3
which was taken under similar conditions, which proves that the Garner
Amplifier is not as bad as it could be. The response is flat up to 35 Kc/s
(see Fig. 1 7/ I I for 20-40 Kc/s).
FIG. 1 7/5.-Actual photographs to show waveform at 30 c/s and 20 Kc/s.
A. Reasonable linearity at 30 c/s, 10 watts, with 22 db negative feedback.
B . 10 watts output without NFB.
harmonic distortion.
Note rounding of waveform indicating
C. First signs of rounding-over at 20,000 c/s occurred at 7! watts.
D. Serious distortion at 10 watts at 20,000 c/s.
When considering 7t watts at 20 Kc/s, it should be remembered
that the power produced at very high frequencies is not likely to
exceed a fraction of a watt in the loudest musical item.
GARNER AMPLIFIER
FIG. 17/6.-Response of original version of Garner Amplifier wi th pre­
. .
.
There are mdlCatlOnS
amp. Controls set for maximum bass and full top lIft.
of phase shift usually associated with tone control circuits.
FIG. 17/7.-Controls set for minimum bass and minimum top. Otherwise
same as Fig. 17/6.
The width of trace in the previous three figures is proportional to
voltage output.
A
250 c/s.
B
500 c/s.
FIG. 17/8.-Square wave photographs of final amplifier. The slope of lower
line of the wave should be ignored as it arose in the original wave as
generated.
My impression was that the bass lift shown in Fig. 1 7/6 was not quite
adequate for equalising recording characteristics with a magnetic
pick-up. The ratio of lift from 250 to 60 c/s is about 2 ' 5 to 1 or 8 db,
whereas a lift of 12 db is required. The pre-amp was then modified
by the designer and the final circuit as shown at the end of the book
gives a lift of 14 db at 50 c/s in relation to the level at 250 c/s.
Major Garner informs me that he uses the Chapman compensated
tone/volume control in his own equipment, with very satisfactory
results. This made up for any lack of bass in the original design.
The performance of the tuning units struck me as being very good.
Where I live, I should plump for the superhet.
GARNER AMPLIFIER
O UTPUT TRANSFORMER AND MAIN AMPLIFIER
Analysis of the circuit of the Williamson amplifier reveals that it is
literally designed around its output transformer. The circuit originally
appeared in the Wireless World, and some of the faults to be guarded
against were summarised as follows:
(I) Low winding inductance, giving rise to frequency distortion and
intermodulation and harmonic distortion at low frequencies.
(2) High leakage reactance which, resonating with self-capacity,
produces sufficient phase shift to cause parasitic oscillation when
NFB is applied over all.
(3) Excessive flux density, which will greatly aggravate harmonic and
intermodulation distortion normally present due to the non­
linearity between flux produced in the core material and the
magnetising force.
(4) Harmonic distortion introduced by excessive resistance in the
primary winding.
The design of a practical transformer must be a compromise between
these conflicting requirements ; but the Williamson transformer, built
on a generously proportioned core of superior laminations with two
coils, each containing five primary sections interleaved with four
secondaries-i.e. eighteen sections in all-gives the following result :
Primary inductance looH measured at 50 cls with 5 volts RMS on
primary=2'5 mW.
Leakage inductance 'c22H measured at 1,000 c/s. Primary resistance
250 ohms each half.
As there is a booklet on this amplifier, published by Iliffe & Sons
Ltd., no useful purpose would be served by offering a similar design
in these pages. It was therefore decided to produce a less ambitious
unit with tetrodes for economical 10 watts output, and a good margin
of stability with fairly generous NFB. The 6L6 valves are well within
the drive capabilities of the concertina phase splitter employed, in
which balance is readily achieved by no more than a careful selection
of top and bottom load resistors.
Due to the high input impedance of the phase splitter the preceding
stage can be a pentode, giving a high stage gain and yet retaining an
excellent top response. The pentode employed in the original model
is a Mazda SP61 (Service type VR65), but a Mullard EF37A gives good
results with anode load of 0'22M, screen resistor 0·68M and bias
resistor 2·2K. A 6J7 has been used with RL=0'25M, Rsc = I '45M,
RK= I '3K, with entirely satisfactory results.
GARNER AMPLIFIER
The output transformer is a Wharfedale WI5 with the following
specification :
Primary inductance 70H.
Primary resistance 375 ohms (the two sections reading 185 and
190 ohms respectively).
Leakage inductance 0 ·083H.
The primary is wound with 34 gauge Conysil ; the secondary with
23g enamelled copper, and this can be adapted for 15 ohms or
3 ohms speaker without affecting the leakage inductance.
When using the 3 ohms output, it will be necessary to modify the
values of resistance in the feedback loop for the same fractional . feedI
.
. 250
.
back. MaxImum feedback with the 1 5 ohms output IS
of out1,500 6
12 y
put volts. At 10 watts this is roughly
2 volts. With 10 watts at
6
3 ohms we have approximately 6 volts, so the fraction fed back should
be about one-third. This means reducing the series resistor from 1,250
250
to approximately 500 ohms
750 3
There are seven sections as shown in Fig. 1 7/9.
--
=
_
=
( = !.) .
p
�
IIOO Tur
p ---../j\---,
ItOO Turn�
W 15
--
�
[1 1 0 :
?
c-
-A
S5T",n5
Turns
1I0 0T"" n5
5 5 T","s
B
FIG. I7/9.-Winding data of
W I 5 transformer as used in
Garner Amplifier.
Ratios
I and
1.
20: 40:
1 5 ohms Speaker : Connect
B to C. Use A and D.
3 ohms Speaker : Join A and
C, also B and D. Use A
and D.
The core is I � ins. stack of size 4A laminations '014 in. thick in
Stantranis No. 1 . Any output transformer of equivalent or superior
specification may be used. An inferior type should not be entertained.
(The W 1 5 transformer is inferior to, and cheaper than, the Williamson
model, and is not suitable for use in the Williamson Amplifier.)
NFB is applied from the secondary of the OP transformer into the
grid/cathode circuit of the pentode amplifier and is adjustable between
�ero and 22 db. Whatever the amount of NFB up to 22 db, no valve
IS called upon to handle more voltage swing than with zero feedback­
a most important point. Of course, a larger input to the main amplifier
is required as NFB· is increased.
GARNER AMPLIFIER
MAIN AMPLIFIER
FIG. 1 7/Io.-Main amplifier. Chassis 14 ins. by 10 ins. by 3 ins.
A. First stage (Phase splitter 6} 5 to right).
B. Output valves.
C.
"
transformer.
D. Mains transformer.
E. Choke.
F. Feedback contro\.
Power take-off plug on end of chassis.
Circuit diagram and list of components will be found on pages
214/5.
HF
RESPONSE
There is no loss of top response in the pentode first stage.
Fig. 17/1 1 shows the overall response of the main amplifier between
20 and 40 Kc/s, with a similar curve of the output from the tone source
for comparison. There is a drop in the amplifier of about 2 db at
36 Kc/s, and the outline is not so clean and smooth as the oscillator,
but apart from these minor differences the curves are identical. In
short, the HF response is good enough to avoid undue phase shifts.
The square wave results indicate that transient response is also good.
186
GARNER AMPLIFIER
B
A
Amplifier Output.
Oscillator Output.
FIG. 17/1 I.-Oscillogram of response at 20-40 Kc/s.
N.B.-The white line at about 27 Kc/s in curve A should be ignored, as
it was due to a slip in the clutch driving the film.
DELIBERATE D I STORTION
A few tests were made on this amplifier with one of the output valves
not working. Naturally, some peculiar results were observed.
Fig. 1 7/I IA shows the waveform at 30 c/s with 5 watts and 1 0 watts
output, and also the nature of the response between 20 and 40 Kc/s.
It will be observed that the output falls off suddenly at about 28 Kc/s
with signs of distortion. This sudden drop of HF response would
cause severe phase shift effects in the feedback circuit.
FIG. I 7 / I IA.-Oscillograms to show effect of failure of one output valve
on LF quality and HF response. Primary of output transformer loaded
20K and 400 pF on each half.
The curve gives a further illustration of the effectiveness of these
oscillograms in providing a clear picture of amplifier response at high
frequencies, together with evidence of distortion that may be taking
place.
STABILITY
Precautions have been taken so far as possible to ensure that stability
will be achieved by anybody desiring to build the amplifier. The
anode load of the first stage is modified, the two halves of the primary
of the OP transformer are shunted, and the feedback circuit is selective
at high frequencies. As a check, a second amplifier was assembled
exactly to the circuit diagram and proved to be completely stable,
and gave expected results with unmatched output valves (one Brimar
and one Ferranti).
GARNER AMPLIFIER
TEST FIGURES
The second Garner amplifier to be built-as mentioned in the previ­
ous paragraph-was tested at different NFB settings for output
resistance, hum level, input sensitivity and harmonic distortion. A
Wave Analyser was used for the distortion measurements. The results
appear in the following table :
TEST OF MAIN A MPL IFIER
Maximum Control Control set Minimum
Feedback half way for 10 db Feedback
Output resistance ohms . .
1'3
4' 3
57
47
" Volts r.m.s. . . 0. 018
0'057
Hum level db below IOW
"
Input voltage for
output
IOW
Harmonic distortion at
I OW output
Harmonic distortion at
5W output . .
I IO
37
0'182
qV
I '5°
I
, v
,80 /
, U
2 ' 5 01
4'2"
1'3°/
2' 5(1
, t)
, 0
, ()
, 0
6'9" '
. 0
Mains supply voltage was 230V, 50 c/s.
Hum, sensitivity and distortion tests were carried out using a 1 5 ohms
resistive load.
For the distortion test, the input waveform was purified by means of
a filter. The harmonic content of the input was below 0'1 per cent.
The test frequency was 1,000 c/s.
These figures are very satisfactory for an amplifier which does not
claim to compete with the Williamson or the highest grade of com­
mercial equipment.
The hum level could be reduced by using a larger smoothing choke.
A mains transformer with an electrostatic screen between primary and
secondaries reduces the highest order harmonics which are all included
in the hum and noise figures quoted.
Tests with a large number of unmatched pairs of 6L6 valves in the
output stage revealed that-at the comparatively low level of output
demanded from them-there is no need to worry about a close match.
The other positions do not call for carefully selected valves. U seful
equivalents should be easily obtainable in any part of the free world,
the whole of the equipment having been designed with this end in view.
The amplifier is stable with or without loudspeaker load ; with any
type of crossover network ; and with any number of loudspeakers
attached thereto.
188
GARNER AMPLIFIER
PRE -AMPLIFIER
FIG. I7! I2.-Pre-amplifier. Chassis 8t ins. by sI ins. by 2! ins.
8 by 8 [J.F condenser at rear. VI left, V2 right. Note plug to main amplifier.
The pre-amplifier is a general purpose device and in this respect
has considerable limitations. The author (H. H. G.) was rather
reluctant to produce a finalised design because he feels that every pre­
amp is an individual problem intimately connected with the frequency
characteristic of the particular pick-up to be used. For this reason,
he was even reluctant to include 78/LP compensation. However, the
circuit published has performed well with a lightweight magnetic and
also with a lightweight crystal pick-up. With the crystal, it is recom­
mended that the correction be applied directly in the pick-up circuit,
as indicated in the diagram, the output being applied to the input of
V2. Good domestic volume will then be obtained with about 12 db of
feedback on the main amplifier. If more NFB is preferred, or more
gain, the corrected output from the crystal may be fed into V I, but
the 78/LP correction circuit which follows VI must then be cut out.
When a magnetic pick-up is used, the input is applied to VI, and a
suitable load resistor for the pick-up must be included, as outlined in
Chapter I I . The gain is adequate for a moving coil pick-up with its
associated transformer.
GARNER AMPLIFIER
If the range of bass boost seems to be less than expected, this is due
to the fact that treble boost begins at a frequency lower than the start
of bass lift. This is thought to be desirable as it avoids reproduction
that is all top and bottom with no middle. Some juggling with the
value of the capacitor 0'02 fLF can always be tried-a reduction in
value affording more bass lift, starting at a higher frequency. Or, the
capacitors in the treble network may be adjusted to give more or less
boost or cut as desired. In all cases the controls will be found to be
interdependent.
The 6J7's in the pre-amp were given a clean bill of health on the
score of hum, but certain specimens in the first position were somewhat
prone to microphony.
CIRCUITS AND COMPONENTS
Pages 208-2 1 5 show full circuit diagrams of the main amplifier,
pre-amp, TRF and superhet feeders, with details of components
required for construction.
THE TRF FEEDER
This simple feeder is capable of good results within about thirty
miles of a main BBC station, and will give reasonable selectivity. It
comprises a high gain RF amplifier and a so-called "infinite impedance"
detector-actually a cathode follower. This detector imposes very
little loading indeed on the associated tuned circuit, and accounts for
the good degree of selectivity achieved. Great care must be taken
with the layout and screening of the unit, as with the high gain available
instability can easily result due to coupling between the aerial coil and
HF transformer. The neutralising device of Fig. 1 7/14 will help in
balancing out the effects of feedback due to C.g, and two turns of
fine wire wound over the transformer, with a coupling condenser of
3-5 pF, are usually adequate. The sense of the winding must be
determined by trial and error.
This feeder, with pre-amp, is illustrated in Fig. 17/13.
ANODE CURRENT
The radio feeder and pre-amp draw their HT and LT current
from the main amplifier. The total anode current for the three circuits
under the actual working conditions is 142 mA. The mains transformer
and choke are rated at 150 mA.
Attention is drawn to this point because a calculation of anode
current-on paper-would indicate a higher value. Any alteration to
the working conditions of the valves which resulted in higher anode
current-particularly with the 6L6 output valves-would obviously
call for a more generous source of supply.
190
GARNER AMPLIFIER
THE TRF FEEDER
FIG. I7/I3.-TRF Feeder and Pre-amp. Chassis 8t ins. by si ins. by 2t ins.
EFso front left, RF amp.
6}S rear left, Det.
6}7 back centre, 1st stage pre-amp.
6} 7 front right, 2nd stage pre-amp.
8 by 8 [J.F back right.
8 [J.F rear.
H T+
::::1111
: Ca9
191
FIG. I 7/I4.-Stabilising RF
amplifier by introducing out
of phase feedback to cancel
effects of inter-electrode capacity and stray couplings.
GARNER AMPLIFIER
THE
SUPERHET
FEEDER
FIG. 17 ' I 5 .-Superhet Feeder and Pre-amp.
by 3 ins.
Chassis I I ms. by 8 ins.
Frequency changer, front left.
I F amp, centre left.
Det. AVC, at rear.
8 iJ.F behind gang condenser.
Pre-arnp on right.
Note variable selectivity knob, front left.
This consists of a triode-hexode frequency changer, a variable-mu
pentode IF amplifier, and a double diode for signal detection and
delayed automatic volume control. It should be noted that reduced
AVC is applied to the IF amplifier as distortion results if full AVC is
applied to this valve when a very strong signal is received.
Variable selectivity is applied to the first IF transformer by changing
the trimming values on the wide position, thus giving a " staggering "
of the resonant frequencies of primary and secondary windings and
increasing the pass-band. The second IF transformer is not treated
in this way as the loading imposed by the diodes ensures a wide pass­
band.
This unit, complete with pre-amp, is illustrated in Fig. 17/1 5 .
CHAPTER 18
A FEW QUESTIONS ANSWERED
The following questions have been drawn up with a view to helping
the large number of quality enthusiasts who live out in the country
or in remote corners of the world, where facilities for radio service
and repairs are a long way off or non-existent.
Questions by G.A.B.
Answers by H.H.G.
Q. I .-In the chapter on Valves, internal resistance, amplification factor,
mutual conductance, etc., are discussed. When a valve is growing old
and is due for replacement, are all these qualities affected? In other
words, does it suffer from lameness, shortness of breath, indigestion,
etc., or only from one illness at a time?
A. I .-The formula is fL = gm X ra. As the cathode emission falls off
with advancing years, less anode current will flow for a given anode
voltage, thus SEa will not cause such a big change in la, and anode
resistance will rise. Similarly, SEg will not cause such a big
change in la, and gm is less. The effect in the case of a power
output valve is a reduction in available output ; in a voltage
amplifier the output voltage will be restricted.
Q. 2.-In the absence of valve-testing equipment, can you outline a method
of checking the state of a valve with the aid of an instrument like the
Avo Meter?
A. 2.-A valve with lowered emission will pass less anode current,
therefore the voltage across the bias resistor will be less. Providing
HT voltage is correct and the resistance of the anode load has
not changed, the voltage developed across the bias resistor is a
sure guide to the state of the valve's emission. It is always a good
plan to record these voltages when the set is new. With battery
valves it is necessary to break the anode circuit and check the
anode current-of course making sure that the value of load
resistance has not changed, and that the bias voltage is normal.
Q. 3.-With push-pull output, let us assume that we have two valves
which are grossly mismatched and are therefore causing trouble.
What sort of distortion would result?
19 3
A FEW QUESTIONS ANSWERED
A . 3 ·-The general effect of gross mismatch is to restrict the power
.
output aVaIlable. At low and moderate levels, distortion would
not . be n�ticeable, but if full output is attempted harmonic dis­
tortlOn wIll be produced due to the inequality of contribution
fr �m each valve into the two halves of the output transformer
pnmary.
Q. 4·-Could the quality in the case of No. 3 be improved by equalising
the anode current to the two valves by introducing extra resistance
in one anode circuit?
A. 4.-The quality would most certainly be made worse by such
treatment. But extra resistance can be introduced into the cathode
of the valve taking the higher anode current, thus biassing it
back until equality of anode currents is obtained.
Equality of mutual conductance is a very desirable feature in
a pair of output valves, so that equal grid swings produce equal
anode current swings.
Q. 5.-In the case of No. 3, would you say that the effects were a step
in the direction of .removing one valve entirely, as illustrated in
Fig. I7!I2A.
A . 5 .-Yes; the removal of one valve is but the ultimate case of a
mismatch.
Q. 6.-Phase Splitters. It is clear from Chapter 9 that good balance is
essential. Can you give a brief outline of the steps which are normally
taken to obtain this, or would they depend on the type of splitter in use?
A. 6.-The question of adjusting for equality of output from a phase
splitter is entirely governed by the type of splitter used. The
most reliable check is made with a valve voltmeter, which can be
easily strung together as it does not need to be calibrated for a
comparative test. Radio Laboratory Handbook by Scroggie and
Radio Designer's Handbook by Langford-Smith (both published
by Iliffe & Sons Ltd.) give several simple designs. An oscilloscope
can also be used-again of the simplest type.
Q. 7.-Pre-Amp. Assuming that a particular valve proved to be noisy or
microphonous in position I, is it likely that it could be used in position
2 with success ?
A. 7.-Yes. It is always worth while changing over VI and V2 to
check which is the less subject to rnicrophony.
Q. 8.-When dealing with hum pick-up we are always told to encase
1 94
A FEW QUESTIONS ANSWERED
the transformer at the input end in Mu-metal. If the mains trans­
former is the source, would emission of hum be avoided by encasing
the offending unit in Mu-metal ?
A. 8.-Yes, but the cost would be absolutely prohibitive. A micro­
phone transformer case only costs about 10/-. The Admiralty
use heavy cast-iron cases-but what a weight ! (They probably
save ballast.)
Q. 9.-In some cases an amplifier begins to motor-boat and show signs
of distress when crossover networks or tapped inductances are used
in the voice coil circuit. Would you agree that a really stable amplifier
should stand up to any variation of load conditions ? Could you
explain btiefiy why motor-boating is induced in some cases ?
A. 9.-A truly stable amplifier remains stable whatever the type of
load placed across the output. If motor-boating or instability
results when a crossover network is used it is due to phase shifts
introduced by the L and C of the crossover system reflecting back
into the NFB circuit and producing in-phase feedback by an
addition to existing phase displacement.
In one case, a capacity of 2 mfd placed across the output terminals
of an amplifier produced a display of fireworks in the output
pentode which did not fail to draw admiration from the onlookers,
particularly as the amplifier had only been received on approval.
Q. lo.- Taking the Garner Pre-amp and Amplifier as a typical case,
which valve or valves would be the most likely to deteriorate first,
and how would such deterioration affect the performance ?
A. lo.-The order of deterioration would be rectifier first, output
tetrodes next, and the remaining valves a long way behind.
During three years' actual experience with more than 100
amplifiers of a similar type, the replacements for failure have
been as follows :
1 5 per cent. HT rectifiers
4 "
"
Output tetrodes
2 "
"
Pentode amplifiers
No phase splitters.
A failing rectifier will give reduced HT with lower available
power output. Failing tetrodes will give reduced power, but not
necessarily distortion at low levels.
Q. 1 I .-H� oscillation in an unstable amplifi�r often burns out spots of
.
wIre
m an OP transformer, resultzng zn shorted turns, without
producing an open circuit. The drop in inductance leads to loss of
195
A FEW QUESTIONS ANSWERED
bass and thin, reedy tone. In the absence of an Inductance Bridge
for checking the condition of the transforme r, is there an alternative
test which could be carried out with simple equipment ?
It is interesting to note here that a single turn of 38's copper wire
shorted round a typical output transformer will reduce the inductance
from 40H to about 36H. Three turns of the same wire will drop the
inductance by 25 per cent.
A. I I.-The thin reedy tone is not wholly caused by the loss of
inductance. The effect of shorted primary turns is analogous to
a secondary winding with a dead short on it ; thus a very low
value of load is presented to the valve which will naturally cut
bass. An inductance bridge will, of course, show a reduced
inductance as it measures the impedance. A transformer with a
loaded secondary takes more primary current ; i.e., its primary
impedance has gone down. One can make a rough check of the
impedance by connecting to a source of AC voltage and reading
E
current passed Z=r.
A rough check of the inductance of the primary can be made by
connecting this in series with the 6'3 volt heater winding of the
mains transformer. The current which then flows, as indicated
on an AC milliammeter, will be proportional to the inductance ;
thus an inductance of IOH would pass practically 2 mA on a
50 cycle supply, ignoring the DC resistance which will be less
than 5 per cent, of the impedance in normal cases.
20H would pass I mA.
'5 mA
40H "
to estimate the inductance by
matter
easy
an
is
it
ately
Proportion
this means.
It is clear from this simple test that the value of current for any
transformer will be doubled if the applied voltage is doubled.
"
.
CONCLUSION
According to C. E. Montague, a wizard in the use of words, the
common adversary of good writing is formlessness-a kind of lumpish
and sluggish recalcitrance, a hugger-mugger fecklessness-an inveterate
halfness . . . .
If this is true of the cult of writing, how much more does it apply to
semi-technical work, where it is necessary to elucidate countless prob­
lems with the minimum of recourse to the mathematics and electronic
symbols on which the problems are based ! To a large extent, words
have to be used to replace figures. Clarity of meaning is therefore a
first essential, or in the modern idiom, a " must ".
It is, unfortunately, impossible to avoid running into some degree of
error in a book of this description, but it is important that such mistakes,
whether of fact or deduction, should stand out clearly and should not
be hidden or confused by equivocation.
The modern amplifier is of necessity a highly technical and elaborate
piece of equipment. The purpose of this book has been to clarify some
of the problems for the benefit of those who like to experiment and
construct at home; and also to clear the air for the ever-growing army
of "hi-fi " listeners who use professionally made equipment but still
like to know " how the wheels go round ".
Should a reader experience trouble in building the Garner Amplifier,
the designer is willing to give reasonable technical help by post.
His address is :
Major H. H. Garner,
393 Baddow Road,
Chelmsford,
Essex.
Letters should be as short as possible and must contain a stamped
addressed envelope for reply. It is always a great help when queries
are listed and numbered instead of being hidden in a maze of verbosity.
Arguments about the relative merits of alternative designs cannot be
entertained.
I think this is a very noble gesture on the part of H. H . G. I have
already received well over 2,000 letters from readers of my L.S. and
S .R. books, and letters are still rolling in at the rate of about twenty
a week from all English-speaking countries. I have office and secretarial
facilities for dealing with this friendly correspondence, but H. H. G.
will be working at home in his spare time, 80 will correspondents
kindly temper the wind accordingly.
G. A. B.
197
SUPPLEMENT
USEFUL FORMULAE
OHM' S LAW
I
=
� or R
=
� or IR
E
I
E2
POWER IN A CIRCUIT
Watts = PR or ­
R
RESISTANCES IN SERIES
RT = RI + R2
etc.
RI X R2
RESISTANCES IN PARALLEL
RT =
RI + R2
Cl X C2
CONDENSERS IN SERIES
CT =
Cl + C2
CONDENSERS IN PARALLEL
CT = Cl + C2
etc.
LT = LI + L2
etc.
INDUCTANCES IN SERIES
LI X L 2
LT
INDUCTANCES IN PARALLEL
LI + L 2
RESONANT FREQUENCY OF A
_� /�
r2
f
TUNED CIRCUIT
27tV LC 4V
Where f - resonant frequency in c/s
L
Inductance in Henrys
C
Capacity in farads
Series resistance in Ohms
r
2
The term at the right �2 is usually so small with radio frequency
4L
coils as to be negligible. Only ' 005 per cent. error results in neglecting
it with average coils. At audio frequencies, the term may be included,
although the degree of accuracy required seldom warrants it.
R
=
=
_
-
-
-
,
TIP
RESISTOR COLOUR CODE
•
BAND
(or S P OT)
I III
I
BODY
I s t --'"
The body colour determines the first
figure of value, the tip the second,
and the band or spot the number
Tolerance
20%
of noughts.
o
FIGURE
2�d NUMSE R.
FIGURE
Black
3 Orange
6 Blue
=
r Brown
4 Yellow
7 Violet
9 White
SELECTED RESISTORS
Silver Band
Red Band
=
ro% Tolerance
2%
"
O F o's
Gold Band
=
Brown Band
=
5%
r%
2
5
8
Red
Green
Grey
Tolerance
"
SUPPLEMENT
REACTANCE OF A CAPACITOR
AT AUDIO FREQUENCIES
Xc
=
10·
cuC
-
0
CU =
27tf = 2
CAPACITY IN
MICROFARADS
000005
°0001
°0005
°001
°005
°01
°05 Coupling
°1
Condensers
°25 R C C
Stages
°5
I
HT
8
16
Smoothing
Cathode
25
Bypass
HT
32
Smoothing
Cathode
50
Bypass
X
3014f, and C
=
capacity in microfaradso
REACTANCE IN OHMS
30 c/s
50 c/s
400 c/s
100 c/s
1000 c/s IO,OOOC/s
3° 18M 318 >450
3 1 0 8M
7°96M
63°7M
106M
1 059M 159,200
3 ° 98M
3 1 08M
1 5°9M
53M
3 1,850
3 ° 1 8M 796,000 3 1 8,500
6°37M
1006M
15,900
3°18M
1 059M 398,000 1 59,200
5°3M
31 ,800
3,180
79,600
l ooM 637,000 3 1 8,500
15,920
1,590
39,800
5 3 1,000 31 8,500 1 59,200
318
3,180
3 1 ,850
7,960
63,700
106,100
3,980
15,920
1,590
1 59
31,850
53,100
63°7
6,370
637
12,700
1,590
21,200
3 1 °8
318
3,180
796
10,600
6,370
398
15°9
159
3,180
1 ,590
5,310
1 °99
49°7
199
19°9
663
398
°99
24°9
99°5
199
9°95
332
I
212
127
63°7
166
99°5
49°8
106
63°7
3 1 08
i
1
15°9
6°37
064
12"4
4°98
"498
7°95
3°18
°318
These reactance values follow a simple ruleo As the capacity is doubled the
reactance is halved ; as the frequency is doubled the reactance is again halvedo
REACTANCE O F INDUCTANCES
REACTANCE IN OHMS
0°5 mH
50 c/s
500 c/s
I
1 ,000 c/s
10,000 c/s
5,000 c/s
°
°
°157
1 °57
3 ° 14
1 5 °7
3 1 °4
°
°
°314
3°14
6028
3 1 °4
62
I
"
2
"
°°
0628
6°28
12°56
62
125
3
"
°°
°942
9"42
18084
94
1 88
4
"
°°
1 025
12°56
25° 1 2
125
5
"
°°
1 °57
15°7
3 1 °4
157
314
0°1 H
,
251
°
3 1 °4
314
628
3,140
6,280
0°5 "
°°
1 57°0
1 ,570
3,140
15,700
31,400
1 ·0 "
°°
3 14°0
3,140
6,280
3 1,400
62,800
°
199
SUPPLEMENT
DECI BEL RELATIONSHIPS
( 2) POWER AND VOLTAGE
RATIOS EXPRESSED
DECIBELS
Ratio : I
1 '0
1'1
1 '2
1'3
1 '4
1'5
1 ,6
2'0
2'5
3 '0
3'5
4'0
4'5
5 '0
5'5
6'0
6'5
7'0
7'5
8'0
8'5
9'0
9'5
10'0
db
(Power
Ratio)
I
I
I
I
I
I
i
II
,
i
I
I
0
0'414
0'792
1 ' 139
1 '461
1 '76 1
2'041
3'010
3'979
4'771
5 '441
6'02 1
6'532
6'990
7 '404
7'782
8' 129
8'45 1
8'751
9'031
9'294
9'542
9'777
10'000
I
I
I
I
I
I
I
I
!
I
i
I
I
I
I
DECIBELS EXPRESSED AS
VOLTAGE RATIOS
IN
Voltage
Ratio
db
(Voltage
Ratio)
1 '0000
'8913
'7943
'7079
'63 1 0
'5623
'5012
'4467
'398 1
'3548
' 3 1 62
'28 1 8
'2512
'2239
' 1995
' 1778
' 1 585
'1413
' 1 259
' 1 122
' 1000
'056
'03162
'01778
'010
'0056
'003 162
'001
'0003 162
'0001
'00003 162
'00001
0
0,828
1 ' 584
2 '279
2'923
3'522
4'082
6'021
7'95 9
9 ' 542
10,881
12'04 1
13'064
13'979
14' 807
1 5 '563
16'258
1 6'902
17'501
1 8 '062
1 8'588
1 9 '085
19'554
20'000
MEMORY AID
3 db power ratio nearly 2
6 db voltage ratio nearly 2
I
: I
:
-,
db
I
I
I
I
I
i
I
I
I
I
I
I
I
I
I
II
i
I
I
i
I
I
I
0
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
25
30
35
40
45
50
60
70
80
90
100
I
Voltage
Ratio
1 '000
1 ' 122
1 '259
1 '413
1 '585
1 '778
1 '995
2'239
2'512
2,818
3' 162
3'548
3'981
4'467
5 '012
5'623
6'310
7'079
7'943
8'913
10'000
17'78
3 1 '62
56'23
100'0
177'8
3 1 6'2
I
1 ,000
I
3,162
I
I
10,000
3 1,620
100,000
I
I
I
I
I
I
For power ratios, the db figure is
half the above,
It is required to find a ratio corresponding to 23 db, or any other value not
included in Table, Take the next lowest multiple, in this case 20 db and note the
corresponding voltage ratio, in this case 10, Take the difference between the
specified levels and the next lowest multiple, in this case 3 db, and note the
corresponding voltage ratio, in this case I '413, Multiply the two ratios so
determined and the answer gives the voltage ratio, in this case 10 x 1 '4 1 3
14' 13 ,
=
Cross checking, 14' 1 3 lies between 20 and 25 db,
20
times the common log of voltage ratio
10 times the common log of power ratio
200
=
=
db.
db.
SUPPLEMENT
TRANSFORMER RATI O S FOR
LOUDSPEAKER MATCHIN G
VOICE COIL-OHMS
LOAD
REQUIRED
IN OHMS
2
1,000
24/ 1
,
3
6
10
15
19/1
13/1
10/1
8/1
22/1
1 6/1
12/1
10/1
1 ,500
28/1
2,000
32/1
26/1
1 8/1
14/1
12/1
3,000
38/1
32/1
22/1
17/1
14/ 1
5,000
50/1
4 1 /1
29/1
22/1
18/1
7,500
62/1
50/1
35/ 1
28/1
22/1
10,000
70/1
58/1
41/1
32/1
26/1
15,000
85/1
70/1
50/1
38/1
32/1
80/1
58/1
100/1
20,000
I
I
I
36/1
45/1
I
,
FORMULA
.
. /Load required
Load required
.
Ratio squared =
or Rauo = V
LS impedance
LS impedance
The impedance of a moving coil loudspeaker may be assumed to be 50 per
cent. higher than the DC resistance of the voice coil.
The impedance tends to rise with frequency above about 1,000 c/s, and
reflex loading often results in a rise in impedance at frequencies below 100 C/S.
Perfect matching of load by careful choice of transformer ratio is, therefore,
rather outside the realms of possibility. One of the main benefits of the use
of NFB is the provision of a low impedance source which greatly reduces the
problem of load matching and so removes the necessity of trying to achieve
the impossible.
FREQUENCY AND WAVELENGTH
F.
I
(c/s) 27·5
40
I
50
60
70
80
90
14'
12·5'
500
1,120
I
Wavelength (ft.)
40'
28'
22 " 4'
18'
16'
F.
120
150
200
300
400
Wavelength (ft.)
9'
6·8'
5"6'
3"4 '
2·8'
2°2'
F.
(c/s)
2K
3K
4K
5K
10K
15K
Wavelength (ins.)
5"6#
3·7"
2·8 "
2'2#
ro IN
·78"
(c/s)
I
I
I
'
20K
. 56"
Formula for Wavelength in feet : 1,120 divided by frequency.
201
100
II·2
'
SUPPLEMENT
LOUDSPEAKER WATTS EXTENDED T O NEAREST
USEFUL DECIMAL POINT
VOICE COIL IMPEDANCE IN OHMS
2
0 ' 1 25
0'5
WAITS
x ' 125
2'0
IN
3 ' 125
4'5
VOICE
6'12
8'0
COIL
10'1
12'5
15 '0
1 8' 0
3
0'083
0'3 33
0'75
1 ' 33 3
2'083
3'0
4'08
5'3
6'7
8'3
10'0
1 3 '0
14 '0
x6'3
I
6
10
12
15
0'041
0' 166
0'375
0·666
1 '041
1'5
2'0
2·6
3'4
4'2
5'0
6'0
7'0
8'2
9'4
xo·6
1 3 '5
16·6
0'025
0'1
0'225
0'4
0·625
0'9
1 '23
1·6
2'0
2'5
3'0
3'6
4'2
4'9
5'6
6'4
8'1
10' 0
12'1
14'4
16'9
X9'6
22'5
0'02
0'08 3
0 ' 1 87
0'3 33
0'52
0'75
1 '02
1'3
1 '7
2'0
2'5
3'0
3'5
4'0
4'7
5'3
6'75
8'3
10'X
12'0
14'1
x6'3
18·8
-
202
0'066
0 ' 15
0'266
0'41
0·6
0'8x
1 '0
1'3
1 ·6
2'0
2'4
2·8
3"3
3 '75
4'3
5'4
6·6
8'0
9.6
I I '25
1 3'0
15
VOLTS
ACROSS
COIL
0'5
1 '0
x'5
2'0
2'5
3'0
3'5
4'0
4'5
5'0
5'5
6'0
6'5
7'0
7'5
8 '0
9'0
10
IX
12
13
14
15
INDEX
c
A
Capacity, Bypass
29, I I I
Coupling
"
Inter-electrode 33, 54, 86
"
139
Carbon Microphone
Cardioid "
Cathode Bias
Coupled Phase Splitter
"
93, 98, 99
Follower
.. 85, 87, 90
"
21
Indirectly
Heated
"
119
Chap man, C. T.
30
Choke as Load
153
Smoothing
"
Circuit Noise . .
9, 17
59
Class A Amplification
59
B
"
"
59, 60
"
" C
16
Combination Tones
Concertina Phase Splitter
92, 94, 95, 184
Condenser Coupling . .
I11
Input
Filter
.
.
154
"
. . 142
Condenser Microphone
126, 130
Connoisseur
128, 144
Cosmocord
Crossover Network
12
.
. 144
Crystal Microphone
.
. 129
Pick-up
"
37, 68--9
Current Feedback
Cut-off Grid
23
6, 1 8
Abbreviations
31
AC Component
154
AC/DC Technique
20
AC Valves
25
Amplification Factor
1 80, 207
Amplifier, Garner
9 et seq.
Quality
"
63, 181
Response
"
188
Test
"
20
Anode
Characteristics
"
21, 22, 24, 34, 35
19, 20, 26
Circuit . .
"
19
Current
"
Follower Phase Splitter
"
93, 100
22-5, 35
Impedance
"
28, 29
Load
"
54
Stopper
"
19
Atom
177
Attenuator
151
Attenuators " T " Type
1I7
Audio Engineering
1 35
Audio Spectrum
40,
192
Automatic Gain Control
B
. . 176
Background Noise
60-2, 104-5
Balance, Push-Pull
155
Ballast Resistance
. . 1 56
Barretters
Bass Cut and Lift
80, 109-16, 1 I 8-2 1, 128, 130, 212
Bass Synthesis . .
1 19
. . 152
Battery Supplies
43
Beam Tetrode . .
Beaumont, F. H.
80, 178
Bias, Various Types
27, 28, 36, 37, 103
32
" Point
38
" Resistance Calculation
Variable
"
BBC
D
Damping Factor
47
Decca
124, 128, 135
Decoupling
48, 52, 53
DB Tables
. . 200
Diode
20, 21, 23
Direct Coupling
31
Directly Heated Valve
19, 20, 37
Bias
37
"
"
"
Distortion
27, 182, 187
Frequency
.
.
II
"
.
Harmonic .
9
"
Intermodulation
15
"
203
INDEX
H
D
Distortion Phase
"
Scale
"
Transient
11
12
13
. . 152
10
9, 178-9
Harmonic D istortion
171, 173
Heater Winding
.
210
Heterodyne Receiver
Whistle Filter
. . 1 34
"
HF Oscillation
50, 180
. . 1 3, 14, 1 86
HF Response
High Gain Phase Splitter
101-2
High Tension Battery
. . 152
"
"
Supplies
152 et seq.
Hum
17, 171 et seq.
. . 176
Hum-bucking Coil
Half Wave Rectification
Harmonics
.
E
Earth Connection
"
Point
Eddy Currents
Electrons
Emission
E . M . I . Ltd.
Equivalent Valve Circuit
Excel Sound Service
Exley Circuit
. .
External Fields
1 57, 172
172
. . 172
19
1 8, 19
128
43
49
1 19
173, 175
I
45, 48, 90
21
Indirectly Heated Cathode
. . 1 36
Inductance, Tapped
.
75, 90
Input Impedance
Instability
17, 48 et seq., 54, 77
Insulator
19
15
Intermodulation
. .
Impedance
.
F
. . 175
19, 20
20
"
Battery
20
Coating
"
153
Filter, HT
1 34
Scratch and Whistle
"
198
Formulae
71
Fractional Feedback . .
11
Frequency Distortion . .
201
and Wavelength
"
1 58
Full Wave Rectification
169
Fuses
Field Winding, LS
Filament
K
13
148
Keen, A. W.
Kelly, S.
L
10
Leak, H. J .
Leakage
Current
Electrolytic
Condenser
Leaky Grid Bias
G
Lift Circuits
72
Gain Reduction Factor
15, 180, 207- 15
Garner Amplifier
1 59
fier
Gaseous Recti
II6
GEC
23
Grid
28
Bias
"
23
Characteristic
"
27
" Current
30,
I
II
" Leak
54
Stopper
Load Line
Load Matching
177
39
I 12
45
43, 44
Loudspeaker Characteristic
Curve
Loudspeaker Damping
"
Resonance
46, 64, 65 -8
Transients
"
14
Low Level Listening
79, l I8
Low Pass Filter
55
204
INDEX
p
M
20
Mains Valves
161
Metal Rectifiers
Phase Shift
1 1 , 70, 108
Split
58, 59, 9 1 et seq.
"
Pick-up Circuits
145-6
. .
Mixing Methods
103
"
Crystal
33, 76
Equalization
. . 128
"
Loading
. .
19
Motor Boating . .
" ': .', Rumble
Magnetic
"
Matching
132
"
NFB Control
"
Output Level
"
Transformer
. . 129
Mullard DA90, EC52, ECC33
Mutual Conductance . .
22-4
125
125-7
"
50, 5 1
. . 142
Pick-up . .
127
"
Molecule
Moving Coil Microphone
123 e t seq.
. . 148
Miller Effect
"
41
Output Stage
138 et seq.
Circuits . .
"
Microphony
"
"
1 38-44
Microphone Response
Microphones
11
Pentode Distortion
. .
126
. .
128
124, 126
126
19
Plate
Potential
24, 25
19
152
Power Supplies
Pre -amplifier for Pick-up
N
1 16, 1 3 1 , 189, 2 1 2
Needle Scratch
Pre- emphasis
16, 126, 1 34
12
5 7 e t seq.
Push-Pull
Negative Feedback
16, 37, 48, 63 et seq., 1 85
Negative Feedback Current
68, 69
"
"
Selective 79, 107
"
"
Summary 7 3, 76
"
"
Voltage
Quality
Questions Answered
69
Neumann Microphone
143
Neutralisation of Hum
175
Novice's Corner
Q
R
81
Radio Designers' Handbook
Frequency Interference
160
"
Reactance, Condenser
. . 108
o
Oscillation, Continuous
"
Intermittent
"
Parasitic
Output, Power
"
"
9, 32
9, 17, 50
"
Output
43
Parasitic Oscillation
"
Valve
. .
159
. .
161
1 52 et seq.
Resistance, Anode Load
67
26, 29
"
Capacity Coupling
"
Damping . .
54
"
Grid
30
29, 35
31
57
Paraphase Phase Splitter 92, 96, 97
Pentode
"
Metal
Reflex Loading
p
Parallel Feed
123-4
Rectifiers, Gaseous
49, 50, 180
41 et seq.
Transformer
1 10, 199
Tables
Recording Characteristics
32
Oscillator
44, 194
Resonance of Loudspeaker
49, 50
46, 64-8
35
Resonant Circuits
205
1 06-8
INDEX
T
R
Ribbon Microphone
Pick-up
"
Tone Control
140-1
"
129
Rotary Convertor
. . 168
1 I7
Transformer Coupling
30
s
Saturation
22
Scale Distortion
Scratch Filter
9, 12
. .
108
Transformer
"
. .
"
146
Valves . .
175
Sine Wave
13
Smoothing Circuits
Choke
"
13, 14, 56, 183
Stage Gain
28, 102, 103
Cathode Follower
"
"
Standard Telephones & Cables
85
142-4
. .
Step-up Transformers
30
. Suppressor Grid
35
Surface Noise
23, 24
"
Output Stage
42
Oscillator
32
TRF (Garner) . .
190
26, 41
"
Inter-electrode Capacity
"
Noise
9, 17
"
Diode
20, 2 1
"
Tetrode . .
34, 43
"
Theory
"
Triode
"
Pentode . .
..
18
23, 24
35, 4 1 , 54
39
Resistance
177
165
Voltage Feedback
69
Dropping
155
"
Doubler
"
Volume Controls
Suspension Loudspeaker
Diaphragm
23, 26
Vibrators
16
..
13, 17
"
"
54, 55
1 92, 210'
13
..
Variable Mu Valve
Stoppers, Anode, Grid, Screen
Superhet Feeder
43
33, 54
22
. .
20 I
. .
Valve as Amplifier
48, 188
61
Characteristic . .
153
Source Impedance
Square Wave
Step-down
v
44, 194
Space Charge
Ratios
"
153
Smith, F. Langford
Ltd.
"
35
Shielded Microphone Leads
"
Triode
54
Secondary Emission
Output 43, 61, 62, 184
80, 109, 1 1 I, 1 12, 1 16, 130, 2 1 2
147
Screen Stopper
Load
"
Distortion
"
Treble Cut and Lift
34
Screening
"
Transient Definition
134
Screen Grid Valve
37
by NFB
"
157
177
14
w
T
Watts, C. E.
..
" T " Attenuator
. . 151
Watts
Tannoy Microphone
. . 144
Wavefront and Wave Envelope
Tetrode
"
Beam Type
Thermionic Emission . .
Tone Compensation
"
"
34, 89
Wavelengths (c/s)
43, 44
Whistle Filter "
18, 19
Wireless World
106 et seq.
122
42, 202
12
201
. .
134
1 I4, 1 1 5, 120, 1 38 , 1 64, 1 8 1 , 184, 194
Dual
Waveform
1 I 5, 1 1 6, 1 1 7
Williamson Amplifier . .
206
182
. . 1 84
GARN ER
C I RCU I T S
T R F
H T+
R
2
T,
'0
R
I
o
d
t
C3
Cs
,
.
._ - - - - - - -- - - - - -
- - - - - - - - - - - _ .
_ - - - - - -,
R
3
d
II
(6
To
Rad
on
>
Pream p.
H T-
TRF COMPONENTS
Cl & C2
2-gang condenser 0 ·0005 fLF each
section
C3
0 · 1 fLF
C4
0·1 fLF
C5
100 pF
C6
0·25 fJ F
RI
100 n
R2
5K
R3
lOoK tw
TI
Wearite PA2 for medium waves ·
T2
"
PHF2
"
"
"
VI
EF50 or 6AC7
V2
6J5, Mullard EBC33 or Osram
L63
Note: (r) With an EBC33, the top cap grid is
used for triode connection.
(2) It is sometimes desirable to decouple
V2 with 47K and 8 fLF to earth, to
avoid motor-boating.
Chassis 8t ins. X 5i ins.
die-cast or similar.
209
X
2t ins. Eddystone
S UPERH ET
IF
FILTER
�
R
3
!L
{
"
-=f
9
- - -\- - a
TI
R
I
RII
,
1.. - - _ . _ - -
a
VI
R
9
t-t-I 11
R
2.
H T ....
I�*
-
-
C
C3 =!=
Cl
�i �i �i
ic·�
R
4
C IO
a
T2.rl---r�
Cl
3
R
8
RI6
RIR
10 13
1<. 1 2.
e1 S/6
a
C l7
R
15 s
(IS
CI9
:
I
'20
To
on
Ra<!
Prearnp�
HT-
S UPERHET PARTS
0· 1 fLF paper
RI
0·0005 fLF 2-gang
R2
200 n
C3
0·1 fLF paper
R3
soK
C4
0·1 fLF paper
R4
SK
IF trimmer, say 200 pF
RS
soK
Cl
C2.A & C2B
Cs & C7
.
soK
C6 & C8
3,300 pF
R6
90K
C9
0·1 fLF paper
R7
SK
CIO
50 pF mica
R8
CII
450 p F padder
300 n
Cn
50 pF mica
RII .
o · SM
CI 3
0·05 fLF paper
RI2 .
IM
R13 ·
S K (variable)
lOoK
R9 & RIO 470K
C14, CISA & CISB 0·1 fLF paper
CI6
50 pF mica
CI 7
0·1 fLF paper
R14 ·
RI S .
270K
CI8
100 pF mica
R1 6 .
47K
CI9
100 pF mica
C20
0·2S fLF paper
TI
T2
Wearite PA2
P02
"
VI
ECH3 S
V2
EF39
V3
6H6
Chassis 1 1 ins. X 8 ins.
or slightly smaller.
2II
X
3 ins.
8 uF
C
"T
·25
M
1 '3
�
M
-
:r:
Ra d
.5 �
M
. 1 1l1tL-O:A S.W �
B
0
Id
f
�F
SW I - T 78
'001 TRE B LE
AAF
BOOS T
lOO K
•
P.U .
'0051
.}.I F
78
(R'tSTAL
P.. IJ .
LP
·s
M
· 002
pF
50 1,uF
' 51�F .OII,u F
M"�
��� _ _ n _ ll_ nn[j.n nnl� nnn--n- --�;'�\178
__
1 00
K
::;: ·O !,u. F
500
pF
100
p I'
I�O
PIN 3
CUT
�I TIO�i I '5r.�
E
INPUT
HT
47K
8)-<£
100
pF
CORREC TIO N C I RC U I T
F O R C R.Y S TA L P. U .
PIN 5
�
PINI6
a
B
or
VI
PRE-AMPLIFIER
Component values are given in circuit diagram.
The two electrolytic condensers, 50 fLF, should be I2V working.
SW2 is a 3-position switch. The tuner would be connected to
terminals Rad. and E.
A crystal pick-up with its associated correction circuits (as shown)
would be connected to B and E if there is adequate gain ; otherwise
to input of VI with the following 78/LP correction cut out.
A magnetic pick-up-with or without transformer-must be fitted
with its appropriate resistive load before it is connected to input of VI .
A moving coil pick-up with transformer may be connected to input
of VI without further ado.
V ALVES-GENERAL NOTE
It should be pointed out that it has not been considered necessary to
outline all the possible alternative makes of valve throughout this book,
as it is not intended to serve as a valve guide. No brief is held for
any particular make, and readers should not hesitate to use equiva­
lents to those mentioned, provided the specification is the same in
essentials.
O SCILLOGRAMS
It will no doubt have been appreciated by readers that the original
oscillograms reproduced in the book are untouched photographs.
Although the interpretation of the various phenomena may in some
cases be at fault, the accuracy and authenticity of the records as shown
are beyond question. In this respect the camera is more convincing
than a free-hand drawing.
213
MAIN AMPLI FIER
R
16
R
r
��'.
(3
. 2.
0
'" I
5,\
6
)
8 /·
1
r-I
O CTAL SOC KET
TO S U PP LY
POWER TO
P RE A M P &
RECEIVE I N PUT
a
a
�
C I3
e-e
/4
�
19 (141 CI61
r
+C2.1
R
25
� i7
s�!��rq I \;;)
2� 2�� ��
�
a
I �R
IS
-' 1
.
J33
R
R.
16
�
I
R 2. 9
0
3
l,p
2� � ��
a�
b
_
Y7
( 23 11
R
b�
5V2A
b
R3
1\
,
�
�
R37
a
a
:?
6V A
F U S E S O O ", A
R
36
OJ5
MAIN AMPLIFIER PARTS
RI5 2M
RI6 0'91M
RI7 20K
RI8 47K
RI9 220K
R20 I K
R21 250 ohms Potr.
R22 0'91M
R23 2K
*R24 47K
R25 20K
*R26 47K
R27 0'33M
R28 0'33M
R29 15K
R30 250 ohms 3W
R3 1 15K
R32 IK
R33 47 ohms
R34 20K
R35 20K
R36 47 ohms
R37 1 5 ohms 1 5W
R38 1,250 ohms
*Close match essential.
CI2
CI3
CI4
CI5
CI6
CI7
CI8
CI9
C20
C21
C22
C23
0'25 mfd 350V
200 pf mica
4 mfd 350VW electrolytic
0'25 mfd 350V
4 mfd 350VW electrolytic
'25 mfd 350V
'25 mfd 350V
400 pf mica
400 pf mica
8 mfd 500VW electrolytic
4 mfd 500VDC paper
'002 mfd
VALVES
V3 SP61 Mazda (Alternatives :
EF37A, 6J7) See page11 84
for circuit values if EF37A
or 6J7 is used.
V4 6J5
V5 & V6 6L6
V7 GZ32 (must be indirectly
heated type)
Note : The SP61 requires a
Mazda Octal base. All
other valves are Interna­
tional Octal.
MAINS TRANSFORMER
350--0-350 1 50 mA, 6V 5A and
5V 2A
All resistors are t watt unless
otherwise stated, and close
tolerance is not required except
for R24 and R26.
CHOKE
N.B.-R37 (dummy load) and
Jack (5) may be omitted
according to discretion of
user.
OUTPUT TRANSFORMER
IOH at 150 mA
Wharfedale W I 5
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement