Colour Television NTSC Carnt Wireless World 1961

Colour Television
N.T.S.C. SYSTEM, PRINCIPLES AND PRACTICE
P. S. CARNT, B.Sc.(Eng.), A.C.G.I., A.M.LE.E. and

G. B. TOWNSEND, B.Sc., M.I.E.E., A.K.C.
LONDON ILIFFE BOOKS LTD

First published in 1961 for
WIRELESS WORLD
by Wife Books Ltd., Dorset House,
Stamford Street, London S.E.1.
© P. S. Carnt and G. B. Townsend, 1961
Printed in England by
The Chapel River Press Ltd.,
Andover, Hants
BKS 3773
Contents
PREFACE ix
ACKNOWLEDGEMENTS xi
1 INTRODUCTION 1
About this book - Introduction - Why N.T.S.C.? - Outline of

N.T.S.C. system - Transmission of three signals - Colour sub -
carrier - Reforming the colour picture at the receiver - Recep-
tion on ordinary black and white receivers - Limitations of the
human eye - The acuity of the human eye - Contrast and
adaptation - Colour vision - Colour blindness - Colour per-
ception - Subjective colours - Some relevant monochrome re-
vision - Field - Phase - Fourier components - Distortionless
transmission - Frequency spectra - Summary
2 COLOUR MEASUREMENT 22
Introduction - Characteristics of light - The nature of light -
Non -spectrum colours - Colours of objects - Measurement of
brightness - Colorimetry - Tristimulus values - Trichromatic
units - Luminosity coefficients - Chromaticity co-ordinates -
Maxwell colour triangle - Chromaticity of colour mixtures -
General colour triangle - Colour space - Chromaticities of the
spectrum colours - Colour mixture curves - Transformation to
different reference stimuli - C.I.E. standard reference stimuli
X*, Y* and Z* - Light characteristics in colorimetry - Addition
of colours in the C.I.E. system - Chrominance - Perceptibility
of changes on the chromaticity diagram - Summary
3 COLOUR PICTURE TUBES, CAMERAS AND FILM

SCANNERS 54
Introduction - Display primaries - Three -tube displays -
Dichroic mirrors - Rotating -disc displays - Lawrence type tubes
- Sensing tubes - Shadow mask tubes - Colour cameras -
Vidicon colour camera - Double and single tube cameras -
Camera circuits - Chromacoder - Colour film scanners -
Colorimetric requirements - Electronic masking - Summary
CONTENTS
4 TRANSMITTER CODING 78
Introduction - Early methods of colour transmission - Field
sequential systems - Line sequential systems - Dot sequential
systems - Bandwidth saving in simultaneous systems - Non -
engineering considerations - Choice of primary colours and
white - Receiver white - Studio white - Gamma correction -
Constant luminance principle - Colour difference signals - Con-
stant luminance - Reduction of colour difference signal ampli-
tudes - Choice of colour axes - Modulation of the sub -carrier
- Representation by a single sub -carrier - Zero phase reference
- Video signal - Bandwidths of the chrominance signals - Time
matching of signals - The final chrominance signal - The com-
bined luminance and chrominance signal - Camera signals in
terms of transmission signals - Transmission signals in terms of
camera signals - Chrominance transmission signals in terms of
colour difference signals - Colour difference signals in terms of
I' and Q' - Colour difference signals in terms of camera
signals - N.T.S.C. tristimulus values in terms of C.I.E. values -
The colour burst signal - Frequency relationships - Asynchro-
nous operation - Vision -sound carrier spacing - Frequency inter-
leaving and the energy spectrum - Delay characteristics - Trans-
mitter tolerances - The colour test stripe signal - Summary
5 COLOUR SPECIFICATION IN N.T.S.C. SYSTEMS 117
Introduction - System white and primary colours - Y' lumin-
ance signal - Choice of burst phase - Colour bar signal -
Chrominance signals and colour triangle for unity gamma -
Chrominance signals on the C.I.E. chromaticity diagram -
Chromaticity diagram on normalized colour difference axes -
Chromaticity transitions at colour changes - Luminance
transitions at colour changes - Summary
6 TRANSMITTER CODING CIRCUITS 156
Introduction - Complete block diagram of transmitter - Encoder
- Vision modulator and carrier oscillator - Sound carrier mixer
- Sound modulator - Divider chain - Colour bar generator -
Summary
7 INTRODUCTION TO COLOUR RECEIVER DESIGN 182
Introduction - Block diagram of typical receiver - Asynchro-
nous operation - Crosstalk between chrominance signals -
Types of chrominance channel response - Choice of demodula-
tion axes - Cross -colour and parc - Sound -sub -carrier beat and
chrominance buzz - Colour controls - Power supplies and heat
dissipation - Summary
8 COLOUR RECEIVER AMPLIFIERS 196
Introduction - R.F. tuners - I.F. stages - Sound rejection -
Vision 1.F. detectors - Luminance amplifiers - Chrominance
amplifiers - Summary
9 COLOUR RECEIVER DECODING CIRCUITS 213
Introduction - Synchronous detection - Diode detection -
Matrixing - Decoding circuits in general - Equiband working
- Choice of axes - Mixer detection - High level detection -
Symmetrical detection - XZ detection - Calculation of the X
and Z axes - Chrominance signal matrixing - Summary
vi
CONTENTS
10 COLOUR RECEIVER REFERENCE FREQUENCY
GENERATORS 259
Introduction - colour burst signal - Specification of reference
generator performance - Burst gating - Burst amplification -
Classification of reference generators - Passive integrators -
Dynamic integrators or A.P.C. loops - In -sync performance -
Static phase error - Dynamic phase error - Pull -in performance
- Crystal oscillator A.P.C. loops - Two mode loops or quadri-
correlators - Colour killing - Automatic chrominance control
- Reference generator circuits - Passive/dynamic integrator -
Simple A.P.C. loop - Quadricorrelator circuit - Alternative
possibilities for reference generation - Summary
11 OPERATION OF THE SHADOW MASK TUBE 308
Introduction - Electrode voltages - Deflection yoke - Line
time base - Field time base - Purity - Convergence - Grey
scale tracking - Summary
12 COLOUR RECEIVER TEST EQUIPMENT AND
PERFORMANCE MEASUREMENTS 335
Introduction - Servicing equipment required - Performance
measurement equipment required - Crystal controlled sub -carrier
oscillator - Calibrated phase shifter - Luminance channel per-
formance - Chrominance channel performance - Reference
generator performance - Colour killer performance - Perform-
ance of time bases - e.h.t. supply performance - Convergence
accuracy - Purity and colour balance performance - Summary
13 RECEIVER INSTALLATION 354
Introduction - Aerial requirements - Monochrome adjustments
- Purity - Time bases - Convergence - Grey scale adjustment
- Colour adjustments - Reference frequency generator - Satura-
tion control - Hue control - Effect of other adjustments on
saturation and hue - Summary
14 COLOUR RECEIVER FAULT FINDING 367
Introduction - Safety precautions - Initial examination with
no signal input - Examination with monochrome signal input -
Examination with colour signal input- Miscellaneous- Summary
15 MONOCHROME RECEPTION OF N.T.S.C. SIGNALS 377
Introduction - Asynchronous working - Carrier frequencies -
Sub -carrier dots - Sub -carrier dot rectification - Transient
effects - Burst visibility - Synchronizing effects - Sound buzz
- Summary
16 SHORTCOMINGS OF N.T.S.C. SYSTEMS 391
Introduction - Constant luminance failure - The elliptical sub -
carrier - The circular sub -carrier - Effects of non-linear encoding
of chrominance parameters on displayed chromaticity - Sub -
carrier amplitude control of displayed purity - Sub -carrier phase -
control of displayed dominant wavelength - Effects of gamma
correction on transient response - The I' Q' fallacy - Effects due
to multiplexing luminance and chrominance signals - Transmitter
notch filter - Effects of gamma characteristic on cross -colour -
Spurious colour in white areas - Single sideband distortion -
Effects of method of modulation of vision carrier and sound
carrier - Summary
vii
CONTENTS
APPENDIX 1 ANALYSIS OF BURST WAVEFORM 434
Calculation of Fourier coefficients - Burst amplifier bandwidth
APPENDIX 2 RELATION BETWEEN DYNAMIC PHASE
ERROR, S/N RATIO AND NOISE BAND-
WIDTH 440
APPENDIX 3 IN -SYNC PERFORMANCE OF AN A.P.C.
LOOP 445
Static phase error - Phase transfer ratio - Dynamic phase error
APPENDIX 4 PULL -IN PERFORMANCE OF AN A.P.C.
LOOP 452
The loop parameters - The loop differential equation - The
simplified loop - Quantitative analysis of the loop - Pull -in
range - Pull -in time
APPENDIX 5 PHASE DETECTOR 466
Sensitivity of the phase detector - Effect of gated nature of burst
- Justification for approximations
APPENDIX 6 REACTANCE VALVE 471
Value of the impedance --va - Sensitivity of reactance valve
Ia
APPENDIX 7 QUADRICORRELATOR EQUIVALENT

FILTER 475
APPENDIX 8 EFFECT OF RESISTIVE MISMATCH ON THE
VALUE OF THE PHASE SHIFT ALONG
A CHARACTERISTICALLY TERMINATED
CABLE 477
BIBLIOGRAPHY 479
INDEX 481
i
viii
Preface
The object of this book is to provide a comprehensive account of the
American N.T.S.C. colour television system, and the British and Con-
tinental versions of this system, from the point of view of the colour
receiver engineer. Since the British reader will probably find it easier
to think in terms of the monochrome system with which he is most
familiar, N.T.S.C. colour television systems are explained with particular
reference to the 405 line version, but wherever there are differences
between the 405, 525, and 625 line systems, these differences are fully ex-
plained. A proper knowledge of the receiver demands, in turn, familiarity
with the make-up of the signal and the principles of colour vision,
and these aspects of N.T.S.C. have also been described in some detail.
No book on colour television can do other than draw heavily on the
original work of the American National Television System Committee,
as reported in references 2 and 4 of the bibliography, and here due
emphasis must be placed on the major part played by the Radio Cor-
poration of America in the development of colour television. A con-
siderable body of opinion considers, indeed, that the present American
system of colour television would more appropriately be called the
R.C.A. colour television system.
The book as a whole is concerned with colour television broadcasting,
but the same principles, cameras, display tubes, etc., are used in those
industrial and closed circuit colour television applications in which only
one cable is used between camera and monitor. In those few cases
where the monitor is very close to the camera it may be more economic
to use three connecting cables carrying the three primary signals. Moni-
tors for this latter case present special problems in stability and are not
specifically dealt with in this book.
It will be found that four chapters are of more than average length.
Of these, one is concerned with transmitter coding and two deal with
receiver decoding, while the fourth discusses shortcomings of N.T.S.C.
systems. These subjects warrant thorough description since they include
ideas and techniques quite foreign to monochrome practice.
Every attempt has been made to keep mathematics to a minimum,
but an exception has been made in the appendices which are mainly
concerned with the quantitive behaviour of phase locked loops. The
latter are an essential feature of the N.T.S.C. system, and call for adequate
treatment.
April, 1961 P. S. CARNT
G. B. TOWNSEND
ix
Acknowledgements
The authors are indebted to the management of the Hirst Research
Centre of the General Electric Company Limited, for permission to draw
on the information and knowledge they have gained in their employ.
As they have been engaged full-time on colour television work, the
preparation of this book has been necessarily a part-time task and has
been made possible only by the understanding co-operation of their
wives, who have tolerated grass-widowship for rather more than a year.
Much encouragement and help have been received from the authors'
associates at Wembley, and in particular from Dr. A. J. Biggs, H. B. S.
Brabham, R. Harman and E. Ribchester. Most of the colour photo-
graphs, which are actual television pictures, were taken on Kodak film
by their colleague F. H. Brittain, who is perhaps better known for his
work in the field of acoustics. They are also grateful to Mrs. Ibbott
who converted their manuscript into typescript.
Finally, the authors would like to gratefully acknowledge the ready
assistance given to them by many organisations, colleagues and friends,
in industry and broadcasting. In particular they are indebted to the
following organisations for their kind permission to reproduce diagrams
and information from various reports and publications:
The Hazeltine Research Corporation

Figs. 9.21 and 9.23-A Color Decoder for Television Receivers. Part I.
Theory and Measurements. No. 7192-1. June 24th, 1958. Fig.
10(b) and the demodulator section of Fig. 13.
Fig. 9.30-Some Notes on the Subcarrier Matrix. No. 7166, March
23rd, 1955. Modification of Fig. 6.
Fig. 10.8-Lectures on the Design of Color Television Receivers. No.
7149. October 29th, 1953. Lecture 6, Figs. 4 and 5.
Fig. 10.17-A Color Synchronizing Circuit Design. No. 7155, June
9th, 1954. Fig. 6.
Figs. 16.9 and 16.10-Directions of Improvement in N.T.S.C. Color
Television Systems. No. 7184, August 22nd, 1956. Figs. 4 and 8.
The Institute of Radio Engineers

Proc. I.R.E. January 1954, Vol. 42, No. 1. (2nd Color Television Issue.)
Fig. 16.1-The Constant Luminance Principle in N.T.S.C. Color Television
Systems. W. F. Bailey. Fig. 5.
xi
ACKNOWLEDGEMENTS
Figs. A.3, 10.10, 10.11 and A.6 - Color -Carrier Reference Phase
Synchronization Accuracy in N.T.S.C. Color Television. D. Richman.
Figs. 17, 18, 19 and 23.
Fig. 5.21 -Transients in Color Television. P. W. Howells. Fig. 4.
Fig. 5.16-Colorimetry in Color Television. Part HI, F. J. Bingley.
Fig. 5.
Fig. 5.19 -Transfer Characteristics in N.T.S.C. Color Television. F. J.
Bingley. Fig. 6.
Fig. 10.6 -Color Television Receiver Design -A Review of Current
Practice. R. G. Clapp, E. G. Clarke, G. Howitt, H. S. Beste, E. E.
Sanford, M. 0. Pyle, R. J. Farber, Proc. I.R.E., March 1956. Fig. 10.
B.B.C. Engineering Division Research Department
Fig. 4.15 -Specification of Colour Television Standards for Experimental
Transmissions from Alexandra Palace. June 1955. Fig. 1(a).
Compatibility of the N.T.S.C. Colour Television System Adapted to the
405 Line Standard; Field Trials. Report No. T-062 (1956/53).
Fig. 5.21(a) was suggested by Dr. R. D. A. Maurice.
Laboratories RCA Ltd.
Industry Service Laboratory, Zurich, Switzerland.
Figs. 9.27, 9.25 and 10.19 -Engineering Description of the RCA Color
Television Receiver Chassis CTC-5N. Figs. 16, 17, 23 and 24.
Figs. 13.3, 13.1, 13.4, 13.2, 13.5 and 13.6 -RCA Service Data for CTC7
Color Receivers. 1958. No. T10. Figs. 6, 7, 8, 9, 12 and 13.
Fig. 10.20 -RCA Service Data for CTC9 Color Receivers. 1959, No.
T6. Phase locked loop section of the circuit schematic diagram.
Figs. 9.13 and 9.17 -High Level Triode Color Demodulator. Report No.
LB -959. September 16th, 1954. Figs. 1 and 3.
Fig. 11.6 -Large Screen Color Receiver, 1955. Fig. 5.
Fig. 3.10 -Live Color Camera Chain, Type TK-41. Catalog B.3850.
March, 1955.
EMI Ltd.
Fig. 3.12-A Vidicon Camera for Industrial Colour Television. I. J. P.
James. Journal of Brit. I.R.E., Vol. 19, No. 3, March, 1959. Fig. 4.
Marconi W.T. Ltd.
Murphy Radio Ltd.

Figs. 8.5, 8.9, 8.14 and 11.2 -The Development of an Experimental 21"
Colour Television Receiver. H. Fairhurst, Journal of the Television
Society, Vol. 8, No. 2 and Vol. 8, No. 3. Figs. 2, 5 and 10.
xii
ACKNOWLEDGEMENTS
McGraw-Hill Book Company Inc.
Figs. 1.10 and 1.11-Television Engineering Handbook. Editor:
D. G. Fink. Figs. 8.42 and 8.43. D. C. Livingston.
Fig. 5.18-Television Engineering Handbook. Editor: D. G. Fink.
Fig. 9.10. Bailey and Loughlin.
Fig. 2.12-Colour Television Engineering. Wentworth. Fig. 2.12 based
on Colour Mixture data given in tabular form in Appendix B.
John Wiley and Sons Inc.

Fig. 2.15 and 2.18-Colour in Business, Science and Industry. D. B. Judd.
Figs. 31 and 75.
,
CHAPTER 1
Introduction
1.1. About this book
Colour television is a practical reality in America and Japan and
a British version of the American N.T.S.C. (National Television
System Committee) system has been successfully developed in this
country.
This new field of modern communications makes an absorbing
study. It is the culmination of large scale, co-ordinated research
which has succeeded in reconciling several conflicting requirements.
Colour television today is complex, but the step from black and
white or monochrome television to colour is smaller than that from
sound broadcasting to vision.
It is suggested that it is preferable if the reader acquires a broad
outline of the subject before proceeding to a detailed examination of
each part. This book has been written in a form suitable for such
a study, since each chapter begins with a preview of its contents and
ends with a summary of the more important points.
Colour television transmitters and receivers are both dealt with
but there is a special emphasis on the latter. Familiarity with black
and white television is assumed but a physical explanation of colour
television is presented whenever possible rather than a mathematical
description of it. In general, only elementary mathematics are
used, and always with explanations in the text. Difficulties and
intricacies have not been glossed over and some matters have been
dealt with mathematically when this is the only satisfactory way.
The reader who is not mathematically inclined can understand the
conclusions reached, however, since advanced mathematics have
been deliberately avoided in the descriptions of the operating
principles and the circuits of transmitters and receivers.
The service engineer has been much in the authors' minds and
after the basic principles have been covered in the earlier chapters,
there is a considerable amount of practical information on the
operation, installation and servicing of colour receivers.
Student and practical man alike will find a study of colour tele-
vision rewarding, for it will become a universal form of entertainment.
2 COLOUR TELEVISION
1.2. Introduction
This chapter describes in simple phrases an overall picture of the
basic N.T.S.C. colour television system. This system makes use of
certain limitations in the ability of the human eye and brain to
register what they are actually seeing, and so the performance of the
eye under various conditions is described. Finally, to prepare the
ground for a more detailed discussion of the N.T.S.C. system, there
is a short résumé of some relevant points in monochrome television.
1.3. Why N.T.S.C.?
Black and white television, which was a practical public service
nearly a quarter of a century ago, has had a surprisingly long
popularity considering its stunted representation of reality. Bereft
of colour, it would have remained a laboratory curiosity had there
not been very considerable difficulties in broadcasting coloured
images. The extent of the growth of monochrome television has,
by using so many channels in our overcrowded ether, now become a
major problem in instituting a colour television service.
To tackle the problems which were holding back a satisfactory
colour broadcasting service, the American Radio Industry (in 1950)
placed its resources, facilities and knowledge at the disposal of its
own National Television System Committee. Most of the larger
American radio companies co-operated in this enterprise, and the
colour television system finally evolved has been given the initials
of the organizing committee, N.T.S.C. It is achieving world-wide
acceptance as a practical and economic way of adding colour to our
existing television services.
Simpler methods of transmitting colour have been rejected for a
variety of reasons-they used too much bandwidth, gave unsatis-
factory pictures, or were likely to restrict the development of future
advances in the science. The N.T.S.C. system can use the same
channel width as monochrome systems of similar resolution and will
produce on existing monochrome receivers a good black and white
version of the transmitted colour picture. Any such system is said
to be compatible. Colour receivers for the N.T.S.C. system can
also be designed to produce good black and white pictures from the
existing monochrome transmissions. This is called reverse com-
patibility. A disadvantage of the N.T.S.C. system is that black
and white receivers cannot be satisfactorily modified to colour
reception, but such ordinary television receivers can receive the
colour transmissions in monochrome.
The American N.T.S.C. system has been adapted to suit the
various transmission standards used in other parts of the world.
INTRODUCTION 3
The use of positive modulation and A.M. sound in our British 405 -
line system avoids some of the difficulties which arise with the
original American system.
1.4. Outline of N.T.S.C. system
Fortunately, in colour television it is not necessary to transmit
separate signals corresponding to every distinct colour. Just as an
artist can mix a complete range of hues from three primary paints
so can the television receiver give the impression of fully coloured
scenes from three primary coloured lights. But the artist uses paints
which appear coloured due to the reflection of light from them
whereas the coloured effect in a receiver comes from transmitted
light. Equal parts of primary paints can produce a black effect but
a suitable mixture of three coloured light primaries, red, green and
blue, produces the sensation of white.
The colour camera splits the scene it is viewing into a red, a green
and a blue image, using an arrangement of mirrors and coloured
light filters. At the receiver, the red, green and blue images can be
recombined to form the full -coloured picture. Fig. 1.1 (a) shows
a simple picture of a red tulip against a blue sky, with some white
clouds. If the colours were very pure, the three component images
would be as shown in Fig. 1.1 (b), (c) and (d).
Three separate camera tubes convert the three coloured images
into three electrical signals, in the same way as an ordinary black
and white camera tube converts the brightness of the scene into one
video signal. In colour television each video signal is proportional
to the amount of either red, green or blue in the picture.
1.5. Transmission of three signals
To transmit three separate signals, each in a manner similar to
black and white transmission, would require three television chan-
nels, so the signals are modified before transmission in order that
they can be sent in only one channel. This modification of the red,
green and blue signals is referred to as coding. The main part of
the coded signal is similar to a normal black and white television
signal and consists of sync pulses and a video waveform. The video
waveform is made up of a signal formed by adding a fraction of
each of the red, green and blue signals together, so that it is rep-
resentative of the brightness of the original scene.
The remaining part of the information, the colouring informa-
tion, is transmitted with a much reduced bandwidth of between
and of the monochrome bandwidth. This means that at the
receiver an unsharp colouring picture is superimposed on a sharp
2
(a) FULL COLOUR PICTURE (b) RED COMPONENT PICTURE
BLUE
Yift.1101.1. GREEN
=re
ANOWNOO
WM.010.
.101.1110MOWN/
1.1110WWk.
.1.1%
4.0.0010.
=01101.'0
Ww.
WHITE
(C) GREEN COMPONENT PICTURE (d) BLUE COMPONENT PICTURE
Fig. 1.1. Analysis of a scene into its red, green and blue components
INTRODUCTION 5
black and white brightness picture. The resulting effect is similar

to the best quality photographic colour transparencies.
The electrical signal representing the scene brightness is sub-
tracted from the signal representing the red component image so
that the resultant electrical signal is proportional to the difference
between the red image and a black and white version of the scene.
This signal is called the red difference signal. A blue difference signal
is formed in a similar manner by subtracting the brightness signal
from the blue image signal. These two colour difference signals
together with the brightness signal contain sufficient information to
enable the receiver to reconstruct the full colour picture. Two
colour difference signals are not unduly repeating the information
which is already being conveyed by the brightness signal.
1.6. Colour sub -carrier
The colouring information is transmitted by modifying the colour
difference signals-which are also restricted in their bandwidth-
and then modulating both signals onto a carrier with a frequency,
SUBCARRIER WITH
NARROW BAND
SOUND CARRIER COLOURING VISION CARRIER
INFORMATION
41 43 44 45 aR
FREQUENCY (MCA)
Fig. 1.2. Sub -carrier in vision band
for our British 405 -line system, of about 2.7 Mc/s. This 2.7 Mc/s
carrier is called the sub -carrier. The modulated sub -carrier is
added to the brightness signal and both are modulated onto the
vision carrier, and tranmsitted together (Fig. 1.2).
1.7. Reforming the colour picture at the receiver
At the colour receiver the signals are separated out again, or
decoded, into their red, green and blue values, and applied to the
colour display. The colour display may be a combination of three
cathode-ray tubes viewed by a system of mirrors so that the three
6 COLOUR TELEVISION
images are superimposed on each other (Fig. 1.3). More probably
the colour display will consist of a single colour cathode-ray tube
which can reproduce all the three primary red, blue and green images
at once. In both cases a full colour picture is seen by the viewer,
including such colours as yellow, purple and brown.
To enable the colour receiver to separate the electrical signals
which are carrying the colouring information, the transmitter sends
HALF SILVERED MIRROR
HALF SILVERED MIRROR
Fig. 1.3. Three -tube colour display
a special colour synchronizing signal during the back porch of the

line blanking pulse. This colour synchronizing signal is called the
colour burst.
1.8. Reception on ordinary black and white receivers
The presence of the sub -carrier signal in the normal video band
would cause interference with black and white reception, were
precautions not taken in the design of the transmitting system. These
precautions are not completely successful and on good quality
monochrome receivers an interference pattern of small dots can
be seen on close inspection in those parts of the picture correspond-
ing to areas which are both strongly coloured and bright.
On the British 405 -line system these dots appear to move slowly
upwards. Large scale tests have shown that most viewers do not
find these dots annoying, and a simple modification to the receivers
will eliminate most of them.
1.9. Limitations of the human eye
The limitations of the human eye can sometimes be turned to
good account. In the cinema the eye is presented with a succession
INTRODUCTION 7
of stationary pictures and persistence of vision leads the brain to
register the sensation of movement. Television carries this process
of deception still further by constructing each stationary picture
element by element.
The performance of the human eye varies considerably with the
viewing conditions: for example, most people need a strong light in
order to read fine print. For the average conditions of television
viewing it is possible to draw broad conclusions about the average
behaviour of eyes, although it is necessary to remember that there
are variations between one person and another.
1.10. The acuity of the human eye

There is a limit to the smallness of the detail which most people
can see without magnification. Thus the average limit of resolution
corresponds to seeing the separate interlaced lines of a 21 in. 405 -
line receiver at a distance of about 10 ft, and the maximum acuity
of the eye is usually taken as 1 minute of arc (o of a degree). The
eye can only resolve such fine detail when it is looking directly at
an object.
Objects only slightly off the direct line of vision are less sharp and
the colours of such objects are less clear, while at the extreme edge
of the field of vision, objects are only seen as varying shades of grey.
To make up for this limitation, the eye is very sensitive to move-
ment which occurs at the edge of the field of view and a slight change
seen out of the corner of the eye attracts attention so that the eye
turns to look directly at the moving object to see what it is. Hence
very bright television pictures appear to flicker more when we are
looking to one side of them.
Monochrome television standards assume that people will not
sit too near to the receivers, and bandwidth is saved by not trans-
mitting all the fine detail in the picture. Interlacing was adopted
to save further bandwidth and the ragged edges which occur on
objects rapidly moving across the picture are accepted in the same
way as the line structure of the raster. Still further use is made of
the limitations of the eye in colour television, to save bandwidth.
If fine detail is seen by white light, red light or green light of the
same brightness, our eyes have about the same acuity, but in blue
light of the same brightness there is some evidence that our eyes are
not quite as acute. However, the brightness of the blue component
of most pictures is much less than that of the green or red compon-
ents and the sharpness of this blue component is then two or three
times less critical than that of the other two images.
8 COLOUR TELEVISION
1.11. Contrast and adaptation
Switching on the room lighting on a bright summer's day makes
little difference to the appearance of the room. The change in
brightness which can be discerned varies with the brightness of the
scene and it is found experimentally that if the smallest noticeable
change in a brightness B is AB, then
AB
-B = constant
This is the Weber-Fechner Law and holds fairly well over the range
of brightness used in television.
It means that if an interfering signal is producing a brightness
change AB in a television picture it will be more noticeable in the
dark parts of the picture where B is low, than in the highlights where
AB
B is large. Typical values of the Fechner fraction -B which varies
'
with the conditions under which the eye is viewing and with the
colour of the light, are 0.005 for large bright areas, to 0.02 for small
areas against a dark background.
Although the human eye can adapt iself to see in dark cellars and
the glare of sunlit snow, it cannot accommodate such a large bright-
ness range at one and the same time. Thus although the screen of a
television receiver may look white before the set is operating, as
soon as the screen fluoresces the contrast between the brightly
glowing areas of the picture and the non -fluorescing areas makes
the non -fluorescing areas look black instead of white.
A similar sort of adaptation occurs when the eye looks at colours
which are nearly white. After looking for a few minutes at a scene
with an overall blue cast, the eye corrects its impression of the
scene so that the mind assumes that the scene has no overall colour
cast. This is why artificial tungsten lamp lighting is acceptable by
itself at night time, but looks very yellow in daylight. It is as if
there were separate " automatic gain controls " in the red, green
and blue responses of the eye, all operating to keep things looking as
white as possible.
Similarly, when the eye is adapted to an overall blue cast a normal
white colour appears to look yellowish by comparison. This colour
contrast adaptation effect has been studied since Goethe's investiga-
tion (about 1800) and several attempts have been made to use it as
the basis of a two-colour reproduction process in photography.
Recently Land, in America, has produced very satisfying colour
INTRODUCTION 9
pictures by photographing a scene on black and white film, once
through a green filter and once through a red filter. The two black
and white pictures are projected in register using white and red
light respectively. With suitable scenes the eye sees a range of
tones from red through pink and white to the complementary colour
of greenish -blue, by the adaptation process already mentioned.
Further, colours such as yellow and blue are seen by many people
in areas of the picture which subtend a small angle at the eye.
However, such ideas appear not to be consistently applicable to the
wide range of pictures needed for colour television.
When the eye is adapted to normal levels of brightness, it is said
to be adapted to photopic vision, that is colours are discernible. At
low light levels such as moonlight, the eye changes over to scotopic
vision and loses its ability to discern one colour from another although
it can still distinguish between shades of light and dark.
1.12. Colour vision
No one as yet understands exactly the mechanism by which the
eye distinguishes one colour from another. The lens of the eye
focuses an image of the scene onto the back interior surface of the
eye, the retina. There the variations in light and colour are con-
into electrical impulses and the information is conveyed to
the brain through an intricate interconnection of nerve paths by a
system of pulse code modulation.
The retina contains two distinct types of light sensitive receptors
called rods and cones after their appearance under the microscope.
The rods and cones are about 500 in. in length and very thin, about
the wavelength of light in diameter, and stand on end with their
small cross-section facing the light. The rods are colour-blind but
are more sensitive than the cones, whereas the cones resolve detail
better and appear to be responsible for colour vision.
The cones are concentrated near the centre of the retina, at the
fovea. There are no rods here and this is the only area of the retina
which will resolve fine detail; it corresponds to vision over an angle
of less than 2°. The rods are more numerous at the outer edges of
the retina and are responsible for vision at low light levels, i.e.,
scotopic vision. In very dim light things are often better discerned
by looking to one side of them so that their image falls on this more
sensitive part of the retina and not on the fovea. There are about
120 million rods and 6 million cones in the human retina.
It has been popularly supposed that there are three distinct types
of cones, each type sensitive to either red, green or blue light, but
physiologists have so far been unable to distinguish any difference
10 COLOUR TELEVISION
between the cones. At the time of writing new experiments are
being made by shining light into the eye ball and studying the light
which is reflected back from the retina. There is some evidence
from these experiments that there may be at least two distinct
types of cones.
1.13. Colour-blindness
A few people, 0.003 % of the population, see all colours only as
variations of light and dark; other people are deficient, com-
pletely or partially, in red, green or blue sensitivity, or combinations
of these.
The various kinds of colour-blindness have different names.
People with normal colour vision are called trichromats, and those
(6 %) who are partially down in sensitivity to one of the three
primary colours, red, green or blue, are anomalous trichromats.
About 1 % to 2 % of people see only two primary hues instead of
three and of these only 1+% are women, for although colour-blind-
ness is in general inherited it is essentially a male complaint. There
are various forms of this dichromatism. About of the popula-
tion are colour-blind people who confuse reds with greens and see
both bluish green and purple light as a neutral grey colour. This is
quite certain because some people are only colour-blind in one eye
and are normal in the other eye. Another % of the population see
only blues, yellows and greys and are completely blind to hues at
the extreme red end of the spectrum. There is a rare form of colour-
blindness in which the eye is relatively insensitive to the blue end of
the spectrum and sees yellows as greys.
Some 5 % of the population are slightly lacking in green sensitivity.
A few observers are weak in red sensitivity, whilst in rare cases the
blue vision is weak.
Anomalous trichromats may disagree with normal people over
the correct setting of the colour balance of a colour television
receiver, whereas people with other forms of colour-blindness will
generally find the pictures acceptable.
1.14. Colour perception
The colour of an object depends not only upon that object but
also upon the conditions under which it is observed, such as the
colour of the light shining upon it, the brightness of the light,
the size of the object, the colour of other things near the object, the
colours which the eye has just previously been observing and so on.
Staring fixedly at a coloured green patch for a short time causes
the eye to become less sensitive to green, and transferring the gaze
INTRODUCTION 11
to a plain white background will then produce an apparent image

of a purple patch on it. These after images are coiiiplemirifarTiri
colour-terthe--original colour.
A piece of blue material laid on a larger area of red looks a different
shade of blue from a piece of the same material laid on a green back-
ground. This is a form of " crosstalk " between the signals from
adjacent areas of the retina.
A small piece of material looks a slightly different colour from a
large piece of the same material and women are careful to match
cotton to material by using the whole reel rather than a single
thread. With fine threads it becomes difficult to see any colour in
them until they are grouped together. The eye in fact becomes
colour-blind to small objects and the objects are seen only as shades
of grey.
Close inspection of a number of differently coloured cotton or silk
threads attached to a piece of white card will reveal the colour of each
individual thread, whereas from a suitable distance only the position
of each thread is distinguishable. At intermediate distances some
colours can be discerned and not others. As the apparent size of
the coloured area decreases, the eye begins to lose its ability to
distinguish between blue, blue -greens and greens of the same
brightness, and then between yellow and grey. All colours become
more pastel, but red and orange are plainly distinguished from blue-
green, although not from themselves at the same brightness. Finally,
as the apparent size decreases, reds and then blue -greens also become
indistinguishable from grey.
This inability to see the exact colour of a small area should not be
confused with the acuity of the eye in various colours, which is the
ability to see fine brightness detail under one particular kind of
illumination.
The eye is very easily deceived, and if a sharp brightness change is
superimposed over a fuzzy colour change, a sharp transition from
one bright colour to a different dark colour can be seen. The
converse is also true. A slow brightness change from one area to
another looks sharp if it is superimposed over a sharp colour
transition. The first effect is used in colour television to save
bandwidth in the colour channels. Fig. 1.4 illustrates a simple
experiment to demonstrate the point.
1.15. Subjective colours
In 1826 a French monk, Benedict Prevost, observed and recorded
an unusual phenomenon. A piece of board moved rapidly to
interrupt a ray of light in a darkened room produced a colour
effect. The eye sees various pastel colours if it is presented with
flashes of black and white light in certain sequences. Benham used
the pattern of Fig. 1.5 (a) pasted onto a top; if this particular pattern
is rotated clockwise about its centre at 8 rev/s, just not fast enough
for the pattern to merge into a uniform grey colour, most people
BRIGHT A SHEEN RED
NEUTRAL FILTERS TRANSPARENT FILTER

OVER SOURCE OF WITH FUZZY CHANGE
WHITE LIGHT FROM GREEN TO RED
BRIGHT DARK
GREEN RED
VISUAL IMPRESSION OF
COMBINATION IS
SHARP COLOUR CHANGE
Fig. 1.4. Sharp brightness change superimposed on an

unsharp colour change
see a blue colour at the outside, then green near the middle, and
finally yellow and brownish red near the centre. Reversing the
direction of rotation causes blue to appear at the inside and red at
the outside. Some people describe the colours as bright, others
have difficulty in discerning them at all.
The sequence of on -off waveforms needed for various colours can
be deduced from Fig. 1.5 (a), but the necessity for such a slow
repetition rate as 8 rev/s, and the paleness of the colours, makes
the phenomena unsuited to a television system. Attempts have
INTRODUCTION 13
been made to use the effect in televised advertisements. Two of the
other patterns which have been used are shown in Fig. 1.5 (b)-
Fechner's pattern-and 1.5 (c)-Rood's pattern.
The colours are generally explained on the assumption that the
three supposed types of red, green and blue receptors in the eye
have different build up and decay times when light is switched on
and off, but there is some doubt about this explanation, for example,
monochromatic light can be used to produce such effects. Similar
effects occur when coloured light is shone on to Benham's rotating
disc.
Rapid movement of a mixture of colours across the field of
view will sometimes show up the individual component colours,
whereas under practically all other conditions the eye consistently
integrates the mixture into one sensation.
1.16. Some relevant monochrome revision
It is assumed that the reader of this book has a broad under-
standing of monochrome television. Nevertheless, since colour
television places emphasis on one or two aspects of black and white
television to which little attention is normally paid, the next sections
are devoted to topics which the reader may wish to revise.
1.17. Field
In British monochrome television the word "frame " means the
2024 lines which constitute half a complete interlaced picture of 405
BENHAM'S TOP FECHNERS PATTERN ROOD'S PATTERN
WHITE LIGHT VIEWED

(a) (b) (C) THROUGH SLOTS
Fig. 1.5. Production of subjective colours. Each disc should he rotated at just
below grey .fusion frequency which is about 8 c/s
lines. In American nomenclature such a half picture of 2024 lines

would be called a field, and they use the word " frame " in the
cinematograph sense of a complete picture. To avoid confusion it
is becoming customary for both nations to drop the use of " frame ",
+A0
I I
I
I II I I I
I I I TIME (sacs)
0 t2 /3 tS to t7 t8 1E9 tIO tII tI2 tI3
to
9=0
9= 2 if
(a)
SINE WAVE B
t3
1
t4
/I/ /
I
5
t
6
I
7
ICI
III
t8
tit
9 10 tlI
tI2
TIME (secs)
ROTATION
-x
(b)
Fig. 1.6. Phase for a sine wave oscillation

INTRODUCTION 15
and to refer to 2021 lines as one field, and to two interlaced fields,
405 lines, as a picture.
1.18. Phase
A sine wave such as that shown in Fig. 1.6 (a) may be considered
as being generated by an anticlockwise rotating line OC, called a
phasor, when the projection of OC on the y axis represents, at any
given time, the amplitude A of the sine wave, and the projection on
the x axis represents the amplitude of the cosine wave. For the
triangle ODC, by definition
sin 0 = -12-C cos 0 = 9--D tan 0 = 13C

OC OC OD
so that A = OC cos 0 = Ao cos 0
If the phasor OC rotates at a frequency f cycles a second, it rotates
through 360° = 27r radians, every cycle, or
B=2irft=wt
where t is the time in seconds from the phasor starting from the
position shown for t = 0. The time might have been chosen from
any other position, of course.
The phase of the line OC is measured by the angle 0 with respect
to the x axis. Phase is always relative, and it is necessary to specify
the position from which the angle 0 is being measured. The
diagrams of Fig. 1.6 (a) show the positions of OC for phases of 0,
7T 3ir
2' 2and 277 radians from the x axis, 277 radians repeating the
original position at time t = 0.
Fig. 1.6 (b) shows a second sine wave of the same frequency f
which goes through its maximum and minimum amplitudes at a
different time from that shown in Fig. 1.6 (a). At time t = 0
the phasor OE has a phase -(1) with respect to the x axis, so that
at time tn, when OC is at a phase angle 0, OE is at a phase angle
0 - 1 with respect to the x axis.
The sine wave B goes through its maximum value at a time -
4:11
w
seconds after sine wave A has gone through its maximum, so that
sine wave B may be derived from A by delaying it and, if B is smaller
than A, reducing its amplitude. Sine wave B is said to have a
phase lag of t with respect to A, and sine wave A has a phase advance
of 4:1) with respect to B. Similarly, sine wave A has a phase difference
of 4:1) radians with respect to B, and B has a phase of -D radians
with respect to A. Phase has no meaning unless it is referred to
some kind of standard. Since sine waves A and B have the same
frequency they have a constant phase difference, 4:1), which is related
to the time delay between the two waves by
1:130 = oitd = 27r ftd
where f is the frequency in cycles per second, w is the angular fre-
quency in radians per second, and td is the time delay in seconds.
A curve of the phase shift suffered by a sine wave in transmission
through a receiver channel may be plotted for various frequencies,
1000°-
/
800°-
600°-
4°2
/
/
TYPICAL RECEIVER PHASE RESPONSE
400°-
DOTTED LINES ARE TANGENTS AT wI ANDu)2
200
1 2 3
VIDEO FREQUENCY (MC/S)
+02
+01 IDEAL GROUP DELAY RESPONSE
-01
-0 2
2 3
VIDEO FREQUENCY 04C/0
Fig. 1.7. Phase response curves

INTRODUCTION 17
D.C. LEVEL AO
FUNDAMENTAL fi
SECOND HARMONIC
'2= 2f1
THIRD HARMONIC
I = 3f
3 I
Fig. 1.8. Fourier sine wave components of a repetitive waveform
in the same way as the amplitude response of the channel. Some-

times the slope of the phase response curve is plotted rather than the
phase curve itself (see Fig. 1.7). The slope of the curve at any fre-
quency is given, of course, by the tangent to the curve at that point.
This function is called the group delay, or envelope delay response
and it shows up variations in the linearity of the phase curve. For a
linearly increasing phase response the group delay curve is a hori-
zontal straight line; that is to say, the group delay is a constant with
frequency, or
dcl)
Group delay = -dco = constant
The group delay at an angular frequency w is the quantity which

matters when we are considering a carrier at frequency w together
with the group of sidebands on either side of it. The group delay
response determines the shape of the modulation waveform, the
carrier envelope, and often the absolute value of the delay is
immaterial.
1.19. Fourier components

Any repetitive waveform, such as in Fig. 1.8, can be considered
as made up of a series of sine waves, called the Fourier components
of the waveform. The frequencies of these sine waves are multiples
of the repetition frequency of the waveform, and the amplitudes and
relative phases of the sine waves depend on the shape and amplitude
of the waveform. If at a particular value of time t, the instantaneous
amplitudes of each of the components are summed the correspond-
ing instantaneous amplitude of the original waveform is obtained.
In general the amplitude of each component decreases as the
frequency increases. Abrupt transitions in the waveform give rise
to very high frequency Fourier components, and conversely, any
network which will not pass the very high frequency components
will not transmit an abrupt waveform change but will increase the
time taken for the fastest waveform transition.
1.20. Distortionless transmission

For an electrical waveform to be transmitted without distortion,
all the Fourier components must be transmitted with their ampli-
tudes increased or decreased in the same ratio, and each component
must suffer the same time delay as the others. The output wave-
form then has the same shape as the input waveform. Since the
time delay must be constant, - must be constant with frequency, so
that the phase shift (13 must increase linearly with frequency, as in
Fig. 1.9. In this instance phase shift refers to the phase of the
Fourier component after it has passed through the network com-
pared with its phase before it entered the network.
For the type of circuits normally used in television receivers,
called minimum phase circuits, the amplitude response determines
the phase response; attention is usually concentrated on the ampli-
tude response since this is easier to measure and interpret. If the
amplitude response is flat at all frequencies, the phase response is
linear as required. Any changes in the amplitude level produce
corresponding departures from linearity in the phase response but
these are spread over a much wider frequency band. Any band-
width restrictions produce phase distortion over frequencies at and
INTRODUCTION 19
adjacent to the ends of the bands, with a corresponding distortion

in the transmitted waveform.
1.21. Frequency spectra
If the transmitted picture is a stationary scene or pattern, every
complete 2 field picture is the same as the other pictures, and the
video waveform repeats itself every A of a second. The video
waveform must therefore consist of harmonics of 25 c/s, and there
should be no frequencies in between these harmonics.
The amplitudes of the numerous harmonics of 25 c/s which make
up the frequency spectrum of a video waveform depend upon the
AMPLITUDE RESPONSE
2
FREQUENCY (MC/S)
PHASE RESPONSE
FREQUENCY (MC/S)
Fig. 1.9. Frequency characteristics for a distortionless transmission
particular scene being transmitted. Usually there is a marked

similarity between the waveform of any particular line and that of the
adjacent lines. If each line were an exact repetition of its neigh-
bours, and if there were no frame blanking signals, then the complete
video waveform would be composed only of harmonics of the line
frequency, 10 kc/s, 20 kc/s, 30 kc/s etc., or fL, 2fi, 3fL, etc., where
fL = 10.125 kc/s. This does not happen, of course, but there is
3
FREQUENCY fC
Fig. 1.10. Frequency spectrum for a typical stationary scene (Reproduced from
"Television Engineering Handbook" (Editor D. J. Fink), by D. C. Livingstone.
Copyright 1957, McGraw Hill Book Company, Inc.)
sufficient correlation between line signals for those 25 c/s harmonics

which occur at and close to the 10 kc/s harmonics to be larger than
those which occur half -way between these harmonics. Half -way
between the 10 kc/s harmonics is described as being at the odd
multiples of half the line frequency, at +A, 24-fL, etc., or at
(2n + 1) (4-fL) where n = 0, 1, 2, 3, etc.
Measurements on actual television waveforms give frequency
spectra similar to the one shown in Fig. 1.10. The only 25 c/s
kAAA 4A AAA Adkv
r,f, (n+ I) ft.
Fig. 1.11. Frequency spectrum for a picture which is changing (Reproduced

from "Television Engineering Handbook" (Editor D. J. Fink), by D. C. Living-
stone. Copyright 1957, McGraw Hill Book Company, Inc.)
INTRODUCTION 21
harmonics which have any appreciable amplitude are those close to
the 10 kc/s harmonics, and there are gaps between these 10 kc/s
harmonic groups where there is very little energy.
Because the two fields of which a picture is composed are very
similar in character, there is a certain amount of repetition in the
waveform from one *, second to the next, with the result that those
25 c/s harmonics which are also 50 c/s harmonics are larger than the
others.
If the transmitted picture is changing then other frequencies are
introduced into the waveform, and the spectrum becomes less
discrete, as in Fig. 1.11; however, the general character of the
spectrum is unaltered.
1.22. Summary
The colour camera analyses the image of the scene into its red,
green and blue components. The three electrical signals which
arise from these primary images are then encoded into a form
suitable for transmission. In the N.T.S.C. system the transmitted
signals are a brightness signal similar to the one used for mono-
chrome television, and a sub -carrier which is modulated with the
colour information.
Monochrome receivers operate satisfactorily on the brightness
signal and ignore the sub -carrier, although the interference pattern
which the latter produces is noticeable on close inspection.
Colour receivers decode the transmitted signals into the original
red, green and blue voltages and generate red, green and blue
images which are combined into one full colour picture.
Since the eye is not a perfect recording instrument, bandwidth
can be saved by not transmitting those picture variations which the
eye would not be able to see from a normal viewing distance.
CHAPTER 2
Colour Measurement
2.1. Introduction
A television system is not capable of transmitting colour as
colour, it can only transmit electrical signals proportional to certain
characteristics of colour which can be measured. Fortunately,
the measurement of colour has been the subject of considerable
scientific study and it has been given the name colorimetry. This
chapter outlines the basic principles of measuring colour.
White light when split by passing it at a small angle through a
prism, separates out into a band of rainbow coloured lights called
a spectrum. By making a mixture of two or more of these primary
colours all shades of colour can be reproduced. White light can be
imitated by mixing primary red, green and blue, as can all shades of
colours, except for a few very pure hues. Similarly, each shade can
be identified by finding the red, green and blue components which
when mixed together will produce the same sensation in the eye as
the original colour. In the colour television receiver, three primary
colours, red, green and blue are varied in strength to produce the
fully coloured scene.
These ideas must be elaborated in some detail in order to appre-
ciate how the colour television camera may analyse a colour and how
the receiver may then reproduce the same colour with some precision.
In particular, a very useful, if rather theoretical way of plotting
colours on a diagram will be developed.
2.2. Characteristics of light
Before a colour can be measured quantitatively, the variable
qualities of colour have to be defined. Various characteristics
can be used but three qualities are always necessary to specify a
colour exactly. Different terms are used to describe the subjective
sensation caused by a particular quality of colour and the corre-
sponding objective measurement of that quality.
2.2.1. BRIGHTNESS
Brightness is the quality of a colour which is concerned with the
" power " of the colour and is best explained by an example. If
22
COLOUR MEASUREMENT 23
the light from a 500 watt lamp passes through a piece of red glass,
then the resultant colour is brighter than the dull red colour given
when the same piece of glass is placed in front of a 15 watt lamp.
The sensation of brightness is a variable factor and has to be specified
in order to define a colour accurately. It will be shown in this
chapter that brightness has a slightly more accurate meaning if it is
described as luminance.
2.2.2. HUE
This is the most noticeable quality of colour. The light from a

500 watt lamp passing through a red glass will give a colour which
has a red hue, but when the light is passed through a green glass
the resultant colour will have a green hue. Colours of different
hue may or may not have the same brightness, so the quality of
hue is a second independent variable. Later it will be shown that
the hue of a colour depends on the dominant wavelength of the light.
2.2.3. SATURATION
The third quality of colour, saturation, is the variation in depth

of colour of the same hue and it must not be confused with bright-
ness which is a measure of the power of the light itself. If the
light from a 500 watt lamp passes through a red glass and is allowed
to fall on to a sheet of white paper, then the colour reflected from
the paper is a clear bright red. But if at the same time white light
is also allowed to fall on to the paper the reflected colour, although
still of the same hue, i.e., red, appears to the eye as a paler shade of
red and is said to be pink. The more white light there is falling
on the paper, the paler or less saturated the reflected colour becomes.
Corresponding to the sensation of saturation, an objective term
purity is defined, which can be measured precisely.
2.3. The nature of light
Light consists of electromagnetic waves and has the same nature
as other electromagnetic waves such as radio waves, X-rays, gamma
rays and heat rays. The difference between these various radiations
is one of wavelength as shown in Fig. 2.1. The eye is sensitive
to radiations which have a wavelength in the range 0.000015 to
0.00003 in., i.e., the eye responds to wavelengths from 380 to
780 mil. Frequency instead of wavelength is often used for radio
waves and light waves can be similarly described
Af -= c
where A is the wavelength in metres, f is the frequency in c/s and
C is a constant equal to the velocity of light, i.e., 3 x 108 m/s.
So visible waves correspond to a frequency of about 109 Mc/s.
The eye and the brain together recognize each wavelength as a
distinct hue (see Fig. 2.1) so that light of wavelength 700 mil creates
the sensation of red and light of wavelength 510 niµ looks green.
VISIBLE
RADIATION 03 co
700M/.1 600mµ 500mµ 400 My

WAVELENGTH, MILLIMICRONS
E LECTRO MAGNETIC
RADIATION
RADIO
HEAT AND
INFRA -RED
ULTRA
:VIOLET
X-RAYS y RAYS
1 1 I 1 1 I I 1 1
WAVELENGTH, MICRONS 3O0µ 3µ 30 M µ 0.3 M µ
FREQUENCY. MC/5. I 102 104 106 108 106) le 1014 1016
WAVELENGTH, 300m 3m 3CM 0.3 mm 3.I0 8M 3.108M 3.16110 3.ici'in 3.16'm
Fig. 2.1. The spectrum of electromagnetic radiation
The colours mentioned in Fig. 2.1 are called spectrum colours

and these merge into each other so that, for example, light of wave-
lengths 480 miL and 470 rap, will have a slightly different blue
hue.
Light consisting of only one wavelength or of a group of wave-
lengths up to a bandwidth of about 5 mil is said to be 100 % saturated
or is called monochromatic. A sodium discharge lamp emits light
of this type and the energy distribution graph (Fig. 2.2) shows
how all the energy is concentrated at one wavelength. The line (1)
represents the energy curve of a signal lamp light and shows how
the energy, though not concentrated in only one wavelength, is dis-
tributed in the red end of the spectrum.
2.4. Non -spectrum colours
Certain colours, such as black and white, are not present in the
spectrum and since these colours are present in a fully coloured
scene they have to be considered separately.
2.4.1. WHITE LIGHT
White light consists of a mixture of wavelengths. Exactly equal
energy at every wavelength gives a white referred to as equal energy
white. Fig. 2.2 shows how certain lights, though approximating
to white light, vary in energy distribution. The examples given
are the light from a north sky, sunlight, and the light from a gas -
filled tungsten lamp.
The sensation of white can be produced by various simple mix-
tures. Thus blue and yellow lights mixed in the right proportion
look white, as does an appropriate mixture of green and purple.
Pairs of colours which produce white light are called complementaries,
so that green is the complementary colour to purple, and purple is
the complementary colour to green. The sensation of white can
15 ....
.
NORTH SKY
ENERGY WHITE
0
cc
,,,° SUNSHINE
r/
VGAS-FILLED
...........
..
05 TUNGSTEN LAMP
BLUE GREEN YELLOW RED

0
400 500 600
WAVELENGTH X (me)
Fig. 2.2. Spectral energy distribution of lights
also be produced by mixing amounts of the red, green and blue

wavelengths.
2.4.2. BLACK
Black is the absence of light of all wavelengths. But the range
of brightness which the eye can register at the same time is limited
(see Section 1.11), so that if there is a marked difference between
bright and dull areas on a television screen, for example a white
LIGHT LIGHT
'A' 'A'
LOOKS LOOKS
YELLOW BROWN
(a) b)
Fig. 2.3. The sensation of brown. (a) has a black surround

and (b) has a bright surround
patch near a dark grey area, the dull areas will appear black to the
eye because of the limitation of its range.
2.4.3. PURPLE
The spectrum in Fig. 2.1 contains no mention of purple colours.
Such colours are called magenta in colorimetry and
duced by lights which consist of a mixture of red and blue wave-
lengths. These colours are not shown in the spectrum because the
red and blue shades are at the extremes of the visible wavelength
band and so do not merge into each other.
2.4.4. BROWN
Just as the eye is deceived into recording a dull area with a bright
surrounding as black (see Section 2.4.2) it is also deceived into
recording a dim area of yellow or orange light with a bright surround-
ing as brown. Brown light is therefore not a true colour but is the
effect of seeing dim yellow or orange light in contrast with a brighter
area (see Fig. 2.3).
2.5. Colours of objects
An object appears coloured by virtue of the light which is reflec-
ted from it. A white object lit by white light looks white, but lit
by red light, it looks red. A red object absorbs light of all colours
except red, and reflects the red light.
Paints produce the effect of colour by absorbing some wave-
lengths and reflecting others. This is fundamentally different from
adding one coloured light to another coloured light. Paints sub-
tract colour from the incident light and we see what is left. This
method of producing colour sensation is called the subtractive
process. Adding lights together, as in colour television, is called
an additive process. To be more precise, while photographic
colour transparencies are purely subtractive, paints do produce a
slight additive effect as well as a subtractive one.
2.6. Measurement of brightness
The eye is not equally sensitive to light at different wavelengths.
For example, a yellow -green light of wavelength 555 mµ will appear
brighter to the eye than an orange light of wavelength 600 mitt,
when both lights have the same energy level.
The eye is also deceived by the conditions under which it views
a light source. A light source seen in a darkened room appears
to be much brighter than when it is seen in sunshine.
Although the intensity of a light could be measured simply by
the energy per unit time at each wavelength, this measure of radia-
tion would not bear a simple relation to the sensation of brightness.
So a different method of measuring the brightness of a light source
has been devised which takes into account the wavelength response
of the eye, whilst still providing an objective measurement of
luminous intensity.
WHITE
DIFFUSING SURFACE
Pi - P2
MIRROR
GLASS PHOTOMETER
HEAD
(a) (b)
Fig. 2.4. Photometer. (a) shows two types of visual field

and (b) typical construction
2.6.1. RELATIVE LUMINOSITY
If the viewing conditions were standardized, it would be possible
to measure the relative brightness or luminosity of two colours.
By using an instrument called a photometer, the principle of which is
shown in Fig. 2.4, it is possible to view two adjacent patches of
coloured light against a black background. The radiant energy
of each light source can be adjusted to make the two areas appear
equally bright.
If the first colour has a power P1 watts and the second colour a
power P2 watts
colour 1 P2
Relative luminosity of
colour 2 P1
Since light at a wavelength of 555 MIL appears brighter per watt

than at any other wavelength, it is used as the reference wavelength
to which all other wavelengths are compared and a curve of the
relative luminosity of every wavelength is shown in Fig. 2.5. This
1.2
BLUE GREEN YELLOW RED

10
08 680
LUMENS/ WATT
u
0 06
04
cc
02
0
400. 450 500 550 600 650 700
WAVELENGTH a (mid)
Fig. 2.5. The curve of relative luminous efficiency (VA) for

monochromatic radiation
curve varies from one human to another, so an average curve called

the standard observer's luminosity function has been fixed by inter-
national agreement.
This curve is valid for photopic vision (i.e., when the brightness
of a scene is great enough for the eye to distinguish colour) but for
scotopic vision (i.e., when the level of brightness in a scene is so
low that the eye can only register different shades of black and grey)
the whole curve moves bodily down to wavelengths which are 40
shorter.
2.6.2. THE LUMEN
Luminous flux is the name given to the total quantity of light

emitted in unit time, when measured by its ability to produce the
sensation of brightness; it is measured in terms of a unit called the
lumen. At 555 miL wavelength, a watt of light power corresponds
to 680 lumens of luminous flux, whereas at other wavelengths a watt
of light gives correspondingly less sensation and (see Fig. 2.5)
gives 340 lumens at 510 mp.
In order to define a unit of luminous flux, a theoretical standard
source has to be defined and this has been fixed by international
agreement.
This standard source is a point source which radiates evenly in
every direction so that if the point were the centre of a sphere, an
even amount of radiation would fall on each part of the sphere's
surface. The power of the source is equal to 0 of the total radiated
power from one square centimetre of a perfectly black body at
2042°K.
A sphere can be divided into 4ir unit solid angles or steradians.
A solid angle is the angle subtended by a spherical surface at its
centre, and is measured by the ratio of the area of the surface to the
square of its radius.
The lumen is the luminous flux emitted from a standard source
down a unit solid angle.
2.6.3. UNIT FOR AN EMITTING SURFACE
The objective measurement of the brightness of an emitting

surface is called its photometric brightness, or luminance. There
are three variables which must be taken into account when defining
luminance and these are the type of surface, its area and the amount
of light being emitted. The foot lambert is the unit of luminance and
is defined as the luminance of a perfectly diffusing surface emitting
one lumen per square foot, where a perfectly diffusing surface is
one which looks equally bright from whatever angle the surface
is viewed. Most direct -viewed cathode-ray tubes satisfy this
requirement. The maximum luminance of a monochrome tele-
vision screen is usually in the region of 50 to 80 foot lamberts.
The overall luminance of a mixture of lights of several wave-
lengths is the sum of the individual luminances. If the light has a
continuous spectrum as in Fig. 2.2, then the total luminance is the
sum of each small group of wavelengths SA. If the spectrum distribu-
tion is given in terms of energy then the energy at each wavelength
must be corrected by the luminosity factor (see Fig. 2.5) to take
into account the variations in reception of the eye. This correction
can be denoted by 9.
The total luminance can then be written
L= IBA CIA
where BA is the mean luminance over the wavelength range A to

A + SA, and the integration need only be between the limits of 380
and 780 mp. Then also
L = KfEAy d.A
where EA is the energy at the wavelength A and K is the maximum

luminous efficiency of radiation, or 680 lumens per watt.
If this light source shines on an object which reflects differing
amounts of light at different wavelengths, the total luminance of the
object will be
Lr = Kf RAEA9 dA
where RA is the proportion of the incident light reflected at the

wavelength A.
2.7 Colorimetry
It is possible to specify any particular colour by its spectral
distribution of energy, as in Fig. 2.2, and to reproduce the same
colour by generating light at a great number of wavelengths through-
out the spectrum to simulate the original spectral distribution.
Fortunately for television engineers, a simpler method is possible.
Most colours can be matched visually by mixing together only
three coloured lights. These three primary colours, or reference
stimuli as they are called, are chosen in the red, green and blue
regions of the spectrum since it is found that this enables the widest
range of other colours to be matched. The instrument used to
carry out such visual matching is called a colorimeter. The observer
looks into an eyepiece and sees the colour to be matched in one
half of a split circle as in the photometer (Fig. 2.4) and in the
other half sees the mixture of reference stimuli, that is, a mixture of
red, green and blue lights. The amounts of these lights can be
adjusted until the mixture matches the test colour.
The red light can be described as an R* light, the green as a G*
light, etc., where * indicates that the particular colour is being dis-
cussed without specifying its luminance. Precision colorimeters use
as reference stimuli a narrow band of spectrum wavelengths, the
international standards being 700, 546 and 436 rap. The controls
which vary the amounts of red, green and blue light in the mixture
are calibrated in arbitrary units. If an unknown colour C* has
been matched by u units of R* plus v units of G* and w units of B*,
then the result is written
cC* ---- uR* + vG* + wB*
where _-_-- means, in this context, that the colours match.
Cl, It*, G* and B* have no numerical significance in this equation
but merely indicate the kind of light being used, whilst c indicates
the amount of C* being matched.
2.7.1. GRASSMAN'S LAW
It is found experimentally that if
c,C,* = u,R* + v,G* + w1B*
and if also
c2C2* = U2B* + V2G* + w2B*
then the mixture of c,C,* and c2C2* is matched by
c,C,* + c2C2* = (u1 + u2)R* + (v1 + v2)G* + (w1 + w2)B*
In fact the equations representing such colour matches obey the

simple algebraic rules of addition and subtraction.
The numbers c, u, v and w could be expressed in terms of watts
per square foot, but this leads to large differences in the numerical
values of v and w and even if luminance units such as foot lamberts
are used, the mean luminance of typical primary colours may
differ by a factor of twenty.
2.8. Tristimulus values
To make the numbers involved approximately equal, the units of
R*, G* and B* are specified in terms of an agreed reference colour,
such as the equal energy white. This agreed reference colour is
matched in the colorimeter and the scales are adjusted so that all
three read unity for this match
wW* IR* ± 1G* ± 1B*
w=I+1+1=3
Using these arbitrary units, the match to colour C* may be
CC* = RR* + GG* + BB*
where
C=R+G+B
The numerical quantities R, G and B, are called the tristimulus
values of the colour CC*. Tristimulus values are a complete
specification only if the primary colours R*, G* and B*, and the
reference colour W* are known, and if w, or the luminance of the
W* reference light, is also given.
2.9. Trichromatic units

To overcome the variations due to observers using the same
reference light at different luminances, it is usual to divorce from
the equations the amount of light used, by dividing throughout by
the total amount, which is R-FG-FB,
when
R-FG-PB C*_-= R+G±BR*H-R-FG-I-BG*+R-FG-FBB*
1 C* R* B*
R G B R G BG* R G B
gG* bB*
where
R
r-R+G+ B g-RH-G+ B b- R-FG-FB
so that
rd-gd-b= 1
The amount of C* in this equation is then said to be one trichro-
matic unit, abbreviated to one T -unit.
2.10. Luminosity coefficients
There are two ways of expressing a match to 1 foot lambert
of the reference white W*. Using foot lamberts as units:
1W* /R* mG* nB*
from the VA curve (Fig. 2.5), it is clear that 1, m and n are by no

means equal. Alternatively, trichromatic units may be used, when
by definition:
3W* 1R* + 1G* + 1B*
or
1W* kG*
so that for this particular reference white

3 T -units of white = 1 foot lambert
1 T -unit of red = 1 foot lamberts
1 T -unit of blue = m foot lamberts
1 T -unit of green = n foot lamberts
In general, for
1C* ------ rR* + gG* bB*
in T -units, the luminance of 1 T -unit of C* is given by

rl + gm + bn foot lamberts
1, m, n, are called luminosity coefficients since they give the relative
luminance of equal T -units of the red, green and blue primaries.
2.11. Chromaticity co-ordinates
In converting to T -units any indication of the absolute magnitude
of the luminance of the colour is lost. The remaining information
must relate in some way to the hue and saturation only, and is
called the chromaticity of the colour.
r, g and b, are known as chromaticity co-ordinates since they carry
the chromaticity information. As
r g b=1
for any colour, any two chromaticity co-ordinates are enough
to determine the chromaticity. Fig. 2.6 shows a chromaticity
diagram on which the chromaticity of colours can be plotted, in
this case in terms of r and b. When r= 1, then g=b----0 so
that r = 1, b = 0 must be the chromaticity of the red primary R*.
Similarly r= 0, b = 1 is the chromaticity of B* and when r = b = 0,
then g= 1 and the origin of the chromaticity diagram must
1.0 Fig. 2.6. A typical chromaticity diagram
REFERENCE
WHITE
-02
represent the green reference stimulus G*. A point such as C*

represents a colour with chromaticity r = 0.45, b= 0.35 and
hence g= 0.2. This colour C* can be written as
1-C* = 0.45R* + 0.2G* + 0.35B*
and represents a desaturated purple colour.

Notice that since r + b cannot be greater than 1, chromaticities
measured in terms of R*, G* and B* must lie inside the triangle
R*, G*, B*, providing r, g and h are positive. But some very
saturated colours cannot be matched in the colorimeter with the
particular primaries chosen, notably the blue-green spectrum
colours. In such cases it is always possible to obtain a match by
adding a particular amount of one of the reference stimuli to the
sample colour in one half of the colorimeter field of view, and
matching this combination with a mixture of the remaining two
reference stimuli in the other half.
DD* + RR* = GG* + BB*
which can be written as
DD* -RR* + GG* + BB*
whence
r - -R -RG B
The colour D* then has a chromaticity which lies outside the

colour triangle, and D* cannot be reproduced in its full saturation
by any real mixture of R*, G* and B*.
2.12. Maxwell colour triangle
It is not necessary to use the rectangular co-ordinates of Fig. 2.6
for plotting chromaticities and R*, G* and B* can be located at
I 0-
r-0.8
r=0.6
0
05-
r-0.4
M102
0+V4W = HEIGHT OF TRIANGLE
Fig. 2.7. The Maxwell colour triangle
the vertices of an equilateral triangle, as in Fig. 2.7. The chro-

maticity of a colour C* is then given by the lengths of the per-
pendiculars to the sides of the triangle, u, v and w. Since the sum
of the areas of the small triangles R*G*C*, R*B*C* and B*C*G*
is always equal to the area of the main triangle R*G*B*,
u v w = constant
For a triangle R*G*B* of unit height, this constant is unity, and
= r, v = g and w = b are the chromaticity co-ordinates. At the
4
reference white point, r = g = b = +, which is at the geometrical
centroid of the triangle. Fig. 2.7 shows lines of constant values
of r, g and b on such a triangle.
2.13. Chromaticity of colour mixtures
A colour formed by mixing r units of R* with g units of G*
has a chromaticity which lies on the straight line joining R* to G*.
The exact position of the resultant chromaticity depends on the
relative amounts of R* and G*. If r and g are measured in
T -units, then the chromaticity is at the point D*, Fig. 2.8,
where
R*D* g
D*G*
with a total magnitude given in T -units of r g. The reader may
prove this for himself by calculating the chromaticity of the mixture
in terms of the tristimulus values of the separate colours. This
process of finding the chromaticity of a mixture is similar to finding
the centre of gravity of two weights.
Similarly, mixtures of (r g)D* with bB* lie along the line
D*B*, at a point such as 0',
where
D*C* b
C*B* r g
Normally, data on the luminance of colours are required in foot

lamberts or similar absolute units and not in T -units. The con-
version is made by using the luminosity coefficients and knowing
the luminance of the reference white used in the colorimeter.
2.14. General colour triangle
A more general chromaticity representation is when R*, G* and
B* are at the vertices of any triangle as in Fig. 2.9. If the reference
colour for which r = g = b is given as the point Y*, which in
general is not the centroid of the triangle, a simple geometrical
construction will give the lines of constant r, g and b. The lines
joining the vertices of the triangle to 17* meet the opposite sides
of the triangle at the points S, T and U. The points T and U are
points at which r = and the side G*B* of the triangle represents
the line for which r = 0. The two straight lines TU and B*G*
meet at a point 0 through which all the lines of constant r pass.
Rae 9 0.2
VG* - - 045
ogc*= b 0.35
CAB*r+g = 0.65
0.5 1.0
b
B*
Fig. 2.8. The chromaticity of colour mixtures
of,
okt';0,1..-.:6 10
.,.
-
----6:4
0b os , 1 e-
. ''''
1/4 (7
1/4 . , v
--- 0.1 ICI.

0 0 o
Fig. 2.9. The general colour triangle

2.0
By
1
A
(0)
\ 2.0 (G) 1.0
R L
y I0
\ 1-
\ J-
(b) 20
Fig. 2.10. Two views of colour space
Similarly, TS meets R*G* at P, through which all the lines of

constant b pass and SU meets B*R* at N, through which all the
lines of constant g pass. The points 0, P and N, also lie on a
straight line and any straight line parallel to PN has the useful
property that the lines of constant r make linear intercepts on it,
proportional to the value of r, and similarly for the lines of constant
g and b.
2.15. Colour space
By divorcing luminance from chromaticity, chromaticity can be
plotted as a two-dimensional quantity. It is quite practical to use
three-dimensional space to plot all the three qualities of a colour,
including luminance.
Thus, if three axes are taken to represent the directions of unit
vectors (R), (G) and (B), which are conveniently taken to be mutually
perpendicular, then the projections of a vector OC on to these three
axes represent the tristimulus values of the colour OC, as shown in
Fig. 2.10.
Using vector notation

OC = R(R) G(G) B(B)
where the direction of the vector OC represents the chromaticity

of the colour. OC varies as the luminance of the colour C* in-
creases, but is only directly proportional to luminance for any
given fixed direction of OC, since luminance is the scalar or arith-
metical sum R G ± B. The direction of the vector which is
equidistant from the R, G, B axes represents grey colours of varying
luminance.
On such a colour space diagram the Maxwell chromaticity
diagram is the triangle formed by joining the ends of the unit vectors,
and the intersection of the plane of this triangle with the colour
vectors gives the chromaticity of that colour.
PRIMARIES R
700
1.0
Rt = 700 mid 610
60
G = 546 mAt 590p
Bt = 436m/1 sao,
REFERENCE:EQUAL ENERGY WHITE 570
560
550
0.1,540
/ -0.2
\b 0 460
ssort
-0.6
.0, //
./6'490
s2of -1.0
-1.2
510 -1.4
Fig. 2.11. Chromaticities of the spectrum colours

The chromaticity diagram of Fig. 2.6 is the projection of this
Maxwell triangle on the ROB plane.
Since the units in which the primaries are measured have different
luminosity coefficients, the vectors which terminate on the Maxwell
triangle do not have the same luminance, and this plane is not a
plane of constant luminance.
It is generally more convenient to work with a two dimensional
chromaticity diagram.
2.16. Chromaticities of the spectrum colours
The chromaticities of the spectrum colours for an equal energy
white may be measured and plotted, when they are found to form a
curve as in Fig. 2.11. As in all colour matching experiments, it is
necessary to define the primary lights used and also the reference
or normalizing white. For most of the spectrum wavelengths a
colour match can only be obtained by transferring one of the
primary colours to the spectrum colour side of the colorimeter, so
that one of the chromaticity co-ordinates is negative.
At the extreme ends of the spectrum locus the wavelength scale
becomes very cramped and a large wavelength change produces
little or no change in the chromaticity.
2.17. Colour mixture curves
The tristimulus values obtained by matching the equal energy
spectrum colours with a particular set of primaries may be plotted
on a relative scale as in Fig. 2.12, for primary units chosen to match
a given reference white. Such colour mixture curves are used a
great deal in colorimetry and average curves have been agreed for
the standard observer and are referred to as I, g and b curves. The
curves in Fig. 2.12 are not the standard international curves but a
set based on the primaries used in colour television.
The amount of red primary required to match any colour whose
energy distribution with wavelength E(A) (see Fig. 2.2) is given,
may now be found by integrating the product of f and E(A) at each
wavelength throughout the visible spectrum
f.
fE(A) dA
R= o
This is usually done graphically and practical limits 380 to 780 miL
are sufficient. For instance, at every 10 mil the values of E(A) and
f may be taken and multiplied together and a new curve plotted
with these values. The total area under the new curves is the
required tristimulus value. Areas below the zero tristimulus axis
are negative. Similarly G and B may be found. Once the colour
mixture curves are known the colorimeter may be discarded and
purely physical measurements of the spectral energy distribution
of a colour made, from which the tristimulus values may be cal-
culated. Such energy measurements can be made with an instru-
ment called a spectrophotometer.
The colour mixture curves themselves may be calculated from the
chromaticity co-ordinates of the spectrum, and the V5 curve.
BLUE GREEN RED
I0
.. .
.... ....
0
i i/ i \\ i1
:
F
..%
,..
/ V ...
.
0.6
A
04
/
, .-
I
- \
\
\ .
02 , i
..
.
\ ..
,,,
........... ,-
.: \ \----....- ...
..
.
.....
0
, ..... ..._......-
02
\ , ......
. ..-
400 450 500 550 600 650 700
Fig. 2.12. Colour mixture curves for an equal energy spectrum, using the
N.T.S.C. reference stimuli and Illuminant C as normalizing white (Based on infor-
mation from "Colour Television Engineering" by Wentworth. Copyright 1955.
McGraw Hill Book Company, Inc.)
The areas under each f, g and b curve must be equal. These

areas are
f: dA, f: dA and f b dA
2.18. Transformation to different reference stimuli

If a colour C* has been matched with one set of primaries R*,
G* and B* in terms of the unit amounts of R*, G* and B* which
match the chosen reference colour, it is possible to calculate the
tristimulus values of C* in terms of any other set of primaries,
say, X*, In' and Z*. It is necessary to know the specification of
X*, Y* and Z* in terms of the R*, G* and B* reference stimuli.
WO) = kiR* k2G* k3B*
1(Y*) = k4R* k5G* k6B*
1(Z*) = k7R* k8G* k9B*
The brackets round X*, Y* and Z*, indicate that the units of
measurement are the T -units of the R*G*B* system. From these
equations, in which k1 to k9 are known constants, it is possible but
tedious to calculate the values of a1 to a9 in these equations
1R* = a1(X*) a2(Y*) a3(Z*)
1G* = a4(X*) a,(Y*) a,(Z*)
1B* = a7(X*) a8(Y*) a9(Z*)
from which a colour 1C* which is given by

1(C*) = TR* + gG* bB*
can be converted to
1(C*) cj(X*) c2(Y*) c3(Z*)
This will give the match to 1 T -unit of C* in the T -units of the old
R*G*B* system. This unit trichromatic equation for C* can be
converted to the T -units of the new X*Y*Z* system if the new
reference colour W*2 is given, since
1W*1 = +R.* +B* in the old system
and 1W*2 1-X* IY* i-Z* in the new system.
Usually the reference whites are the same. So finally

1C* = xX* yY* zZ*
Fig. 2.13 shows a set of 3 such primaries, X*, Y* and Z*, on a

chromaticity diagram. The chromaticity axes shown are the g and
r axes rather than the r and b axes which have been used up to now.
r and b axes are usually more convenient for colour television, whilst
g and r axes are normally used for colorimetry. The reason for
changing the chromaticity diagram at this juncture will become
clearer in the next section. The positions of X*, Y* and Z*,
shown in Fig. 2.13, are outside the spectrum locus which encloses
the chromaticities of all real colours, and hence X*, y* and Z*
do not correspond to any physically recognizable lights. They may
perhaps be thought of as super -saturated colours. They are mathe-
matical abstractions only but there are marked advantages in
referring colours to such primaries.
If the point C marks the chromaticity of a particular colour C*
on the R*, G* and B* chromaticity diagram, let the line from X
ky0 PRIMARIES
\ R4`= 700nvi
G*= 546mp,
436mg
REF = EQUAL ENERGY WHITE =SE
2.0
\sl\
s
Nu.5.20
\\
\
1 11 X Y Z ARE C.I.E. PRIMARIES
\
11
X = 1.275 -0.278 0.003
\ 1.4
500
Y= -1.739 2.767 -0.028
1.2
D G Z = -0.743 0.141 1.602
\\ N
Nk550
\ \ 0.8
560
\ 490\
, \\480()
0.6
\ \N \ 570
N
\ o.4,ScoosE \
A1TCHNE
-__ \ \59600
--- -- _.4Z 02 C
*
-- 470., R
00 ,
-1.6 -1-4 -1.2 -I.0 -01 -0.6 -04 380 0-2 0.6 0.8 0\ 1-2 1.4
-0.2
X*
Fig. 2.13. Transformation to new reference stimuli, showing the red -green
chromaticity diagram spectrum locus for a standard observer
through C meet YZ at D. Then C* will be matched by a mixture

of X* and D* in proportions DC to CX, in T -units. Similarly D*
will be matched by a mixture of Y* and Z* in the proportions ZD
to DY.
2.19. C.I.E. standard reference stimuli X*, Y* and Z*
An international committee, the Commission Internationale de
l'Eclairage or C.I.E., have adopted a set of non-physical primaries,
X*, V` and Z*, to simplify colorimetric calculations (see Fig. 2.13).
These non-physical primaries have the following chromaticities:
X* r = 1.2750 g = -0.2778 b = 0.0028
Y* r = -1.7394 g = 2.7674 b = -0.0280
Z* r = -0.7429 g = 0.1409 b = 1.6020
The units of X*, Y* and Z* are such that a mixture of 1 unit of
each matches equal energy white.
Since the chromaticities of all spectrum colours lie inside the
triangle formed by the points marking the chromaticity of X*,
Y* and Z*, now all real colours have only positive chromaticity
coefficients. Any colour C* is matched by
CC* = XX* + YY* + ZZ*
1C* = xX* + yY* zZ*
where
x - X+ XY±Z' y - X+ Y±Z and z - X+ Y-FZ

which are the C.I.E. chromaticity co-ordinates (see Section 2.11).
It should be noted that although X*, Y* and Z* are non-physical
colours, it is possible to make a colorimeter which will measure
the chromaticities of real colours directly in terms of the tristimulus
values, X, Y and Z of the colours X*, Y* and Z*. For example, to
measure the amount of X*, a spectrum colour of the same dominant
wavelength as X*, i.e., a red colour, is displayed in the known side
of the colorimeter field and reference white is added to the other
side. The amount of reference white added desaturates the unknown
colour to the same extent that the chromaticity of the chosen spec-
trum colour is less than X*.
The standard reference stimuli have been carefully chosen. In Fig.
2.13 it can be seen that the line joining Y* and X* runs along the
spectrum locus for an appreciable range of colours at the red end
of the spectrum so that all the colours along this line have zero blue
coefficients. X* and Z* are located on the line where all the points
have zero luminance, that is the line
Ir+mg+nb=0
where 1, m and n, the luminosity coefficients of the R*G*B* pri-
maries, are in the proportion 1 to 4.5907 to 0.0601 and this line
(X*Z*) nearly passes through the origin.
620
630 --
630
700
470
I "ISO I I I I I
01 400 0'2 0.3 04 0.5 0.6 0.7 0.8
CO -EFFICIENT OF X
Fig. 2.14. The C.I.E. chromaticity diagram
This line is called the zero luminance line or the alychne and is
used as the x-axis in the chromaticity diagram shown in Fig. 2.14.
This is a convenient arrangement since the luminance of any colour
is then equal to the number of Y* primary units needed to match
the original colour. This rotation of the axes will be clear if the
axes of Figs. 2.13 and 2.14 are studied.
The spectrum locus is plotted using the new axes on Fig. 2.14
and the chromaticity locations of the C.I.E. standard illuminants
are also shown.
Illuminant A is the light from a gas -filled tungsten lamp.
Illuminant B is similar to sunlight.
Illuminant C is similar to the light from a north sky.
Equal energy white is located at the point y = = x. The various
groups of colours of similar chromaticities are shown in Fig. 2.15.
2.20. Light characteristics in colorimetry

The three characteristics of light which give rise to the subjective
sensations of brightness, hue and saturation can all be measured
objectively in colorimetry.The corresponding terms are luminance,
dominant wavelength and purity. These objective terms are only
used to describe the measured qualities and are not synonymous
with the words used to describe the sensation which they produce
in the observer.
09
520
515 5
0.8
540
510
550
0.7
SOS 560
0.6 GREEN
1 570
/YELLOW-,
`500
0 / GREEN/
05 500
/ / YELLOW
--.. .., / -- - .... ..---
04 49S
N.
\ \ / / fr "/ / ORANGE
590
\^ 600
/ /_
CYAN 1
F
ti i
,t1ss
4-r
+B
/
/ PINK /
610
62
0.3 490 OR / -
/ /
//\\
o
BLUE-
//
650
1
RED 700
GREEy \
/ / N7' - \ \\
0.2 465
/ / MAGENTA
OR
BLUE / PURPLE
0.I
480
/
470
460
430
0 0.1 400 0/ 0.3 0.4 05 0-6 0.7 013
CO -EFFICIENT OF X
Fig. 2.15. Approximate boundaries of colours on the C.I.E.

chromaticity diagram
2.20.1. LUMINANCE
Luminance has already been defined in Section 2.6.3. It corres-

ponds to the subjective appreciation of brightness, but whether a
given luminance looks bright or dark to the observer depends on
the adaptation of his eye.
Luminance is denoted by the tristimulus value Y of the Y*
primary and is usually measured in foot lamberts. In transforming
tristimulus values to chromaticity co-ordinates, information about
the luminance of the colour is lost.
2.20.2. DOMINANT WAVELENGTH
A straight line from a reference white such as Illuminant C in

Fig. 2.14 to a spectrum colour F* passes through all the colours
formed by a mixture of spectrum colour F* and the reference
white. All these colours are said to have the same dominant wave-
length as F*, but whether they produce the same hue sensation
depends on the colour adaptation of the observer's eye. Strictly, the
point marked F* in Fig. 2.14 should be called the chromaticity of
the colour F*.
2.20.3. PURITY
The spectrum colour F* has 100 % saturation or purity and the

line CF* represents the desaturation of the colour F* with the
reference white. The purity of the colour with chromaticity rep-
resented by H will be equal to -C HC%. To be precise, this is the
definition of excitation purity. The reference white is completely
desaturated or is at zero purity. The point which marks the
reference white on the chromaticity diagram is referred to as the
achromatic point.
2.20.4. DEFINITION OF COMPLEMENTARY COLOURS
If a line is drawn from a spectrum colour F* through the reference

white to the other side of the spectrum locus, the point J marks the
complementary colour to F*. All complementary colours can be
found in this way by drawing a line through the point C.
2.21. Addition of colours in the C.I.E. system

Colours are added to form mixtures of different kinds of light
in the C.I.E. system in the same way as in the RGB system, but now
the calculation of luminance in absolute units is much easier.
Colours are normally specified in terms of their x and y chroma-
ticity co-ordinates and their luminance, thus
Colour 1 =
Colour 2 = x2, Y2, -Y2
These must be changed to tristimulus values before adding them
together.
Let
+ Y1 I
Z1 =
Then
X1 Y1
x1 = and y1 =
and hence
Y1
D1=
and X1 =
=
Yi
further Z, = z1D1 = (1 - x1 - AD,
1 - x1 - y,
.Y1
Similarly for X2 and Z2.

Then the tristimulus values of the mixture arc
X1 + 2 = + X2
= Y1 + Y2
Y1+2
Z1+2 Z1+ Z2
and the chromaticity co-ordinates are
+ X2
X1 + 2
(XI + X2) + + Y2) +(Z1+ Z2)
+ X2
D1+ D2
Y1 + Y2
Y1 + 2 =D1 + D2
As an example, consider the mixture of two colours given in Table
2.1.
Table 2.1
ADDITION OF COLOURS
x Luminance
Colour 1 0.2 0.6 18 foot lamberts
Colour 2 0.3 0.1 8 foot lamberts
18
Di
yi 06 = 30
Therefore
Xl = x1D1 = 6
z1= 0.2, so Zi = 6
also
8
D2= -=01 -= 80
Y2
Ya
Therefore
X2 = 24
z2 = 0.6, so Z2 = 48
Therefore from these equations
X1+2 = 30, Y1+2 = 26 and Z1+2 = 54

SO
D1+2 = 110
therefore
30
x1+2 - 027
110
26
Yl+ 2 -110 0.24
Y1+ 2 = 26 foot lamberts

This chromaticity may be found graphically as in Fig. 2.16. It
must lie on the line connecting x1y1, the point P*, to x2y2, the point
Q*, at a position which divides the distance between P* and Q*
0.3 04 0.5 0.6 07 0.8

CO -EFFICIENT OF X
Fig. 2.16. Graphical construction for additive mixtures
inversely as the ratio of the number of T -units in colour P* to those

in colour Q*. Therefore erect a line perpendicular to P*Q*,
at the point P* with a length proportional to D2 equal to kD2.
Erect also a line perpendicular to P*Q* at point Q* on the opposite
side to the perpendicular at point P*, with a length proportional to
D5 equal to kD,. The line joining the ends of these perpendiculars
crosses the line P*Q* at the required chromaticity.
2.22. Chrominance
As the luminance of a colour increases, so the tristimulus values
XYZ or RGB increase, but the chromaticity co-ordinates do not
change. Another quantity called chrominance is used in colour
television, and this can now be defined as the colorimetric difference
Effect of removing each primary colour from a correctly balanced
picture; (above) red absent from picture; (centre) green absent
from picture; (below) blue absent from picture
Two types of colour television camera. The top photograph
shows an EMI camera, while the lower one a Marconi three-
vidicon colour camera. (Courtesy EMI Ltd. and Marconi Ltd.)
between a colour and a reference white light of the same luminance
as the colour. This quantity can only be illustrated in a three-
dimensional colour space, such as the X YZ space of Fig. 2.17.
Since only Y controls luminance, planes parallel to the OZX
plane are of constant luminance. Thus the vectors OC and OW
representing a colour and a white of the same luminance must end
in a plane parallel to OZX through the appropriate value of Y.
The vector WC then represents the chrominance of the colour C.
The chrominance increases as the lengths of the vectors OC and
OW increase with increase in luminance. The chrominance also
increases as OC moves away from OW, that is, as OC represents
colours of increasing saturation.
2.23. Perceptibility of changes on the chromaticity diagram

The specification and plotting of chromaticities follow linear
laws, but the eye does not distinguish equally between chromaticity
changes at one side of the diagram and changes at the other side.
A very small change in chromaticity at the blue end of the diagram
is noticeable, while the chromaticity of a green colour must move
quite a long way over the diagram before the eye can notice any
change in the corresponding colour.
Fig. 2.18 shows the relative susceptibility of the eye to chromaticity
changes. The chromaticity change in going from a marked dot
to any point on the surrounding oval produces a colour change
which is about equally noticeable for each dot. The smallest
PLANE OF
CONSTANT LUMINANCE
Fig. 2.17. Two views of chrominance vector in colour space

5
noticeable change depends on the viewing conditions but is approxi-
mately one hundredth of that shown. The x, y diagram corresponds
more closely to the subjective discrimination of the eye than do the
x, z, and y, z diagrams, because the eye is not as sensitive to the
blue component.
2.24. Summary
Any colour can be matched by amounts of three reference pri-
maries, thus
CC* =. RR* + GG* + BB*
09
520
530
0.8
550
0.7
560
06
570
>-
05 580
590
1.4
0.4
0 600
610
620
.3 630 --
650
700
0.2
0.1
03 04 05 08
CO -f IF ICIFNT Of X
Fig. 2.18. Approximate perceptibility of chromaticity changes. These are

Judd's equally perceptible ellipses (100 minimum perceptible change in
normal viewing)
or
1C*=----= rR* + gG* bB*
where
R*, G* and B* are measured in T -units, so that for a given reference

white
3W 1R* + 1G* + 1B*
R, G and B are called tristimulus values and increase with lumin-
ance. The chromaticity co-ordinates r, g and b, are independent
of luminance. Equal numbers of T -units of the primaries R*,
G* and B* contribute to the luminance of the mixture in the ratio
of the luminosity coefficients 1, m and n.
The tristimulus value of a mixture is the sum of the corresponding
individual tristimulus values of the separate colours. The lumin-
ance of a mixture is the sum of the luminances of the separate com-
ponents. A colour specified in terms of one set of primaries can be
specified in terms of another known set of primaries without further
measurement. The most convenient representation is in terms
of the C.I.E. non-physical primaries X*, Y* and Z*. The chroma-
ticity of a mixture of two lights lies on the straight line joining the
chromaticities of the constituent colours, at a point which divides
the line in the inverse ratio of the T -units of each colour.
A quantity, chrominance, is defined which is the colorimetric
difference between a colour and a reference white of the same
luminance.
CHAPTER 3
Colour Picture Tubes,

Cameras and Film Scanners
3.1. Introduction
The simplest way to reproduce a colour picture is to reproduce
separately the red, green and blue primary images and then to super-
impose them so that the eye merges the three images into one full -
colour picture. The fusion of the three images may be accomplished
in several ways. If a separate display tube is used for each primary
image and the pictures are combined optically, then at any instant all
three primary images are being traced out and the reproduction is
said to be made by a simultaneous display.
The three primary images may be presented to the eye in quick
succession and the persistence of the eye's vision will merge the red,
green and blue pictures into one coloured whole. This is a sequential
display.
The eye will also fuse colours together if small areas of different
hue are placed adjacent to each other. When small areas are
viewed from such a distance that the eye has difficulty in resolving
each individual area, the overall colour sensation is that of a normal
mixture of the colours involved. The small areas in such a reproduc-
tion may be lit either sequentially or simultaneously.
All these methods and many ingenious combinations of them have
been used to produce colour television displays. Only a few will be
described in brief detail in this chapter, and of these the only colour
tube which is made in quantity is the R.C.A. shadow mask tube.
This tube will be discussed at length in Chapter 11 and the short
outline given in this chapter must not be taken as a measure of the
practical importance of the shadow mask tube.
Considerable inventiveness has likewise been shown in the sugges-
tions put forward for the design of colour cameras, but here also
only one type of camera is in general use. It uses three normal
monochrome camera tubes, each of which produces a signal propor-
tional to one of the primary images. For the transmission of colour
film, flying spot scanners are used as in monochrome television but
54
COLOUR PICTURE TUBES, CAMERAS, ETC. 55
three photocells are now necessary, each of them sensitive to one
primary colour.
In general both transmitting and receiving equipment aims at
reproducing a colorimetric match to the original scene although the
viewing conditions and the absolute luminance levels are different.
Departures from this principle are only made on the grounds of
expediency and cost. There is some doubt as to whether such an
exact match gives the viewer the most pleasing reproduction but it is
considered to be better practice to leave such subjective considera-
tions to the artistic control of the studio producer.
3.2. Display primaries

The choice of chromaticities for the primary lights used to produce
the receiver's colour picture determines the range of reproduced
chromaticities which are practical. As large a range, or gamut,
of chromaticities as possible is obviously desirable, although the
efficiency of light output and practical manufacturing limitations
restrict the choice. The primary chromaticities which have been
standardized for use in the N.T.S.C. system are discussed in the next
chapter and are plotted in Fig. 3.1.
Gelatine colour absorption filters for converting white light to the
primary colour lights, and phosphor powders for use in cathode-ray
tubes, can all be manufactured to produce approximately the
N.T.S.C. primaries.
Typical phosphors are zinc phosphate for red, zinc orthosilicate
for green and a zinc sulphide activated with silver or copper for blue.
The red and green phosphors usually have a medium after -glow or
persistence, which falls exponentially, still giving -1 or 37 % of the
e
original light output about 2 or 3 ms after the electron bombard-
ment ceases. The blue phosphor has a shorter persistence, falling
to 37 % in about 0.1 ms. Objects moving across the screen may
therefore leave a yellow trail behind them as the red and green
phosphors continue to glow after the blue phosphor has ceased
phosphorescence.
The luminous efficiency of these phosphors is low, only a few
per cent of the electrical energy being converted into radiant energy,
with the green phosphor having the highest efficiency. Ten or so
lumens per watt can be obtained at 20 kV, but the light output
may saturate at current densities of a few µA/cm2, which is often
reached for the red and blue phosphors in projection type tubes.
Because of the relatively low efficiency of the red phosphor used in
0 01 400 0-2 03 04 0.5 0.6

CO -EFFICIENT OF X
Fig. 3.1. Typical phosphor characteristics. P22 screen phosphor chro-

maticities after screening are shown as x and N.T.S.C. reference stimuli
are shown as 0
Table 3.1
TYPICAL PHOSPHOR CHARACTERISTICS
C.I.E. co-ordinates Persistence

00
Dry After Elapsed
Colour Composition powder screening time brightness
x y x y msec left
Red Zn0(PO4)2:Mn 0.674 0.326 0.650 0.322 33 12.7

40 8.2
Green Zn2SiO4:Mn 0.203 0.728 0.200 0.718 33 8.9
40 5.1
Blue ZnS:Ag 0.146 0.052 0.151 0.073 40 0.01
direct viewed tubes, a phosphor which glows more orange than the
N.T.S.C. red is sometimes employed.
3.3. Three -tube displays
Perhaps the simplest type of colour display, in conception,
consists of three projection type tubes arranged so that each is
throwing a picture on to the same screen. Each tube phosphor
glows in one of the primary colours and gelatine colour absorption
filters may also be used to modify further the chromaticity of the
screen face. Video signals proportional to either the red, green or
blue tristimulus values of the scene are applied to each tube.
The input-output characteristics of the three channels must
match exactly if the same ratio of red, green and blue signals is to
produce a dark grey at low brightness levels and a white of the same
neutral colour at high brightness levels. A neutral grey scale such
as the central gradation blocks of the Test Card " C " is very useful
for testing the match between the three colour channels at different
input levels. Not only must the amplifier channels match but the
voltage input -light output characteristics of the cathode-ray tubes
must also be identical. The production of a good neutral grey
scale is one of the major problems in any three -gun display. Typical
errors are the dark greys looking reddish in tone, the middle greys
appearing pinkish and the whites looking neutral.
Care is needed to ensure that all three images are the same size
and shape, and the three identical deflection yokes are fed from the
same timebase circuits. The tubes may be offset from the projection
lens axes so that the images coincide, or each tube may be mounted
on the inclined axis of its own lens and close to the next tube, when
the resulting geometrical distortion may be corrected by modifying
the scanning waveforms.
The images on three direct viewed tubes may be combined optically
by mirrors so that the observer sees only one composite image, as
in Fig. 3.2. This type of arrangement is sometimes called a irino-
scope or a jumbo. The optical path length from the observer's
eye to each tube is made the same for all three tubes so that the three
images appear the same size. Such arrangements are bulky and
only one observer can view directly on the axis. A rather more
compact arrangement may be made by placing the red tube on the
other side of the viewing axis, turning its mirror through 90° and
moving it back so that if forms a cross with the blue mirror, but
this arrangement gives a dark line across the centre of the picture.
The mirrors used may be semi -transparent silver mirrors but these
prevent a large part of the light emitted by the tubes from reaching
the eye. Very efficient mirrors called dichroic mirrors are normally
used.
3.4. Dichroic mirrors

A dichroic mirror transmits light at all wavelengths except for a
group at one end of the spectrum, which it reflects. Thus a blue
reflecting dichroic mirror may transmit all wavelengths above 500nit
BLUE REFLECTING RED REFLECTING

DICHROIC MIRROR DICHROIC MIRROR
Fig. 3.2. Principle of direct -viewed three -tube display
and reflect all wavelengths below 460mµ, with a gradual changeover

from transmission to reflection for the region between 460mp, and
500mii. Similarly, a red reflecting dichroic mirror may change over
to transmission at about 580mp. The wavelengths at which the
changeover occurs depend on the angle of incidence of the light, the
changeover moving to longer wavelengths as the angle of incidence
decreases. An increase of perhaps 30mµ may occur in changing from
45° to normal incidence.
Dichroic mirrors are sheets of glass which have been coated not
with silver but with alternate layers (normally seven but may be as
high as twenty) of materials with high and low refractive indices.
Each layer is usually a quarter of a wavelength of light thick at
the rejection wavelength. At each interface some of the incident
light is reflected and the thicknesses of the layers are adjusted so
that the reflected rays cancel at some wavelengths and add at other
wavelengths. This is the way in which a thin film of oil on water
produces coloured patches.
To correct for the change in angle of incidence which can occur
across an inclined dichroic mirror with wide angle optical systems,
it is possible to taper the dichroic film from one side of the mirror to
the other.
Because some light is also reflected from the other side of the
glass, very bright objects in the field of view may give rise to spurious
images which are just noticeable.
3.5. Rotating -disc displays

The difficulties which arise with three -tube displays over registra-
tion of the three images and the matching of the three channels to
ensure good grey scale reproduction, can both be avoided by the
use of only one amplifier chain and one cathode-ray tube which
sequentially amplify the red, green and blue signals. This was the
colour system used by Baird in 1928, a more sophisticated form of
which was officially adopted in America for a short period from
1950 to 1951.
At the instant when the channel is amplifying the red signal, a red
colour filter is mechanically placed in front of the white glowing
cathode-ray tube. When the signal changes to the green signal, the
red filter is replaced by a green filter and so on. The mechanical
change of the coloured filters can be achieved by rotating a disc with
coloured sections (see Fig. 3.3) in front of the cathode-ray tube.
The shape of each coloured filter depends on the relative position of
the tube face and the centre of the disc and on the decay time of the
C.R.T. phosphor. Drums of filters which revolve around the
complete C.R.T. have also been used.
The mechanical difficulties of such arrangements for high rates of
rotation limit the device to field sequential systems in which the
Fig. 3.3. Rotating colour filters, (a) disc and (b) drum
colour is changed only once every field. With reasonably pure
primary filters there is also a considerable loss of light (typically
5 to 1) from the white cathode-ray tube, but good colorimetric
performance can be obtained.
All such field sequential receivers suffer from two defects called
colour flicker and colour break-up. Since the luminosities of the red,
green and blue pictures are not in general the same, a brightness
R = RED
Y = YELLOW Fig. 3.4. Colour break-up of
W = WHITE a moving object, due to dis-
C = CYAN
placement of primary images
B = BLUE
during sequential scanning
G =GREEN
flicker occurs unless a complete cycle of 3 colour fields occurs in, at

most, II- of a second. If each colour field occupies Ks of a
second, len when any area of the transmitted picture contains only
one of the primary colours, e.g. a red patch, it is only lit for K,
second in every -A-a, second or just over 16 times a second.
If any object in the picture changes its position during the cycle
of 3 colour fields it is seen as a series of red, green and blue images
as it moves across the screen (Fig. 3.4). This is called colour break-
up or colour fringing. It can also occur if the viewer moves his head
quickly so that the red, green and blue images fall on different parts
of the retina. However, if the colour field repetition rate is high
enough to overcome the brightness flicker, then colour break-up
effects are usually negligible.
Since a simultaneous transmission provides all the colour informa-
tion all the time, a sequential display can be used at the receiver for
either simultaneous or sequential transmission.
3.6. Lawrence type tubes
Various types of colour display tubes using a phosphor screen
made up of strips of red, green and blue phosphors have been
proposed (Fig. 3.5). The electron beam can be made to fall on
any particular one of the three primary phosphor strips by means of
a wire grid placed between the electron gun and the screen. Such
tubes are called Lawrence tubes after the American physicist who
suggested this arrangement, but particular laboratory models have
also been called " Chromatron ", " Chromaton ", focus mask tubes
and post -deflection focus (P.D.F.) tubes.
The wire grid is spaced about a centimetre behind the phosphor
screen and the wires are connected together in two sets. All the
wires of one set are in line with the red phosphor strips and all the
wires of the other set are in line with the blue phosphor strips. There
are no wires in line with the green strips, of which there are twice
as many as there are red or blue strips.
If there is no voltage between the two sets of grid wires, the
electron beam passes through the wires and is focused on to the green-
glowing phosphor strips. This focusing action is obtained by keeping
the mean voltage of the grids many thousands of volts below the
potential of the phosphor screen.
By placing three to five hundred volts between the two sets of
grid wires, the electrons are bent towards the nearest positive wire
and alight on the corresponding red or blue phosphor strip depend-
ing on the polarity of the applied voltage. In this way the grid
wires control the primary colour which is being reproduced at any
instant and the video signal must be switched so that the correspond-
ing red, green or blue signal is applied to the electron -gun grid -
cathode space.
RED
GREEN
BLUE
GREEN
RED
GREEN SINE WAVE
SWITCHING
(a)
WIRE GRIDS PARALLEL

TO PHOSPHOR STRIPS
4.
I -I.
500 VOLTS
BETWEEN GRIDS
- 3 NO VOLTAGE
BETWEEN GRIDS
(2
ELECTRON
- CI
3
ELECTRON BEAM
+
(c)
(b) GRID 5 KILOVOLTS PHOSPHOR SCREEN
HIGHER THAN GUN 13 KILOVOLTS
HIGHER THAN GRID
Fig. 3.5. Operation of Lawrence tube. (a) shows phosphor stripes viewed
from the front; (b) and (c) are side views
The switching is most conveniently done at field frequency but
can be done at either line or picture element frequency. The sub -
carrier frequency itself can be used to synchronize the video and
wire -grid switching.In such cases the tube is operated with the
phosphor strips horizontal and parallel to the scanning lines. By
applying a sine wave voltage between the wire grids the electron
beam is swept across the phosphor strips and the intensity of the
beam is controlled by a signal gated out from the video information.
Since the beam traverses the green strip twice in each cycle, the
phase of the gating signal must either be adjusted to miss the green
strip on one crossing per cycle or the gating width for green must be
half that for red and blue. Differences in efficiency between the
blue, green and red phosphors can also be compensated by altering
the gating width. About seventy volts of video drive are needed.
Various ingenious ways of deriving the gated video signal have been
proposed but at the time of writing no commercial receivers using
such tubes have been produced.
The two sets of grid wires have appreciable capacity, between
1,000 and 2,000 pF, and some 20 to 30 watts of power have to be
generated for the wire -grid switching, which necessitates effective
shielding of the generator circuits. The two grids themselves do
not radiate appreciably as they are only a fraction of a wavelength
apart (10-5 for sub -carrier switching). Ideally, each strip and
the focused electron spot should have a width of at most that of a
picture element. Such close spacing of the strips has not yet been
achieved and there is some doubt as to whether adequate cathode
life could be obtained with the required small intense spot.
Three -gun versions of the focus mask tube have been made in the
laboratory. The three electron guns are aligned so that the wire
grids focus each electron stream on to only one colour phosphor,
and no switching voltage is required.
3.7. Sensing tubes

Another promising type of colour tube, which has been demons-
trated but is not yet commercially available, also uses a phosphor
screen of red, green and blue strips. Instead of directing the electron
beam on to the required colour phosphor the beam (as it traces out
its normal raster) scans over each vertical strip in turn. In order
that the video circuits may select the right primary colour signal
for the single electron gun, it is necessary for the display tube to be
able to tell the video circuits which colour phosphor the electron
beam is falling upon at any particular instant. Various ways of
PHOSPHOR BLACK
STRIPES f -------AREAS
ALUMINIUM
COATING
111
(c)
SECONDARY
EMISSION
LAYERS
(a)
TUBE FACE
VACUUM
(b)
ALUMINIUM COATING 27 kv ABOVE CATHODE
CARBON COLLECTOR 30kV ABOVE

CATHODE ATTRACTS SECONDARY
ELECTRONS
WRITING OR SIGNAL BEAM
PILOT OR INDEX BEAM

(d)
Fig. 3.6. Philco Corporation "Apple" or sensing tube. (a) is the back elevation
and (b) the plan. (c) shows pincushion arrangement of phosphor strips and
(d) is a cross-section view of the tube
producing this sensing information have been patented but the most
successful method appears to be that used by Philco Corporation
in their " Apple " tube.
In the " Apple " tube the phosphor screen of vertical red, green
and blue strips has dark non -luminescent areas between each
phosphor strip, each dark area being as wide as a phosphor strip.
This ensures that the scanning spot can illuminate any primary
colour without desaturation. The efficiencies of the phosphors are
adjusted by the inclusion of non -activated material so that the
unmodulated scanning beam produces a white raster. The reference
white and the grey scale balance are thus not functions of the
circuitry or of the tracking between three electron guns.
Behind the normal aluminium screen backing another layer of
vertical strips of high secondary emission material (magnesium
oxide) is deposited, with one strip to each triad of phosphor strips,
the magnesium oxide overlapping 40 % of the total width of each
triad (see Fig. 3.6) which is 0.06 in.
As the scanning beam passes over a secondary emission strip or
index strip as it is called, it emits a burst of secondary electrons which
are collected by a carbon coating on the inside of the tube bulb.
The index strips lie behind the red phosphor strips and the bursts of
secondary emission can be used to synchronize the gating of the
red colour signal, and hence of the green and blue signals if the
linearity of scan over this short width of a triad is reasonable.
However, if an averaging process is used in the generation of the
green and blue signals, then the overall linearity of the line scan
must be good.
Since the writing electron beam disappears for the black parts of
the picture and varies in intensity with the picture colour, a second
electron beam, i.e. the index beam, is provided and scans with the
writing beam. The index beam has no picture modulation and does
not disappear in the black parts of the picture but remains as a just
visible background of about 4- foot lambert. The amplitude of the
index beam is varied at a high frequency outside the video band so
that tuned circuits can extract the index beam information from the
signals produced by the video information.
The " Apple " tube has usually been operated so that it acts as
its own decoder. The chrominance sub -carrier frequency is con-
verted by a heterodyning process to the indexing frequency, which
depends on the stripe geometry and line scan. The conventional
chrominance signal then goes through its colour phase sequence (see
Chapter 4) in step with the scanning of the writing beam over the
phosphor strips.
The width of the triads is graded across the tube face to allow
for average scanning non -linearity, and the pattern of triads is made
slightly pincushion in the corners to allow for average scanning coil
distortion. As the transit time of the secondary electrons from the
index strip to the collector varies from the centre of the tube out-
wards, the phase of the index strips with respect to the phosphor
strip is varied to compensate.
As with the Lawrence tube, narrow strips and very small spot
sizes are necessary to achieve high definition and good saturation.
3.8. Shadow mask tubes
The most successful colour display tube is the R.C.A. shadow mask
tube. This tube has been manufactured and sold for a number of
years in some quantity and is capable of producing a very good 405 -
line colour picture. It is a three -gun tube and is basically suitable
for simultaneous systems although it can be used for sequential
displays. The phosphor screen of a 21 in. shadow mask tube is made
up of just over a million phosphor dots, arranged in triads of red,
green and blue (Fig. 3.7). About in. behind the phosphor screen
A DIRECTION OF
SCAN FOR
MINIMUM MOIRE Air" PHOSPHOR DOT
SCREEN
0A0A0VM
00003rit ---!),111E 8,
0000000
000000 GREEN BEAM
CritYAMPOO
11):940949 09, ---
00a$100P.A0
theIVAIOA,110, SHADOW
MASK
ISaill
(a) (b)
PURITY
MAGNETS
DEFLECTOR
COILS
3 ELECTRON
GUNS
BLUE
LATERAL
SHIFT
CONVERGENCE
COILS
(c) COLOUR
EQUALIZER
ASSEMBLY
Fig. 3.7. R.C.A. 21 in. shadow mask tube. (a) shows an enlarged
view of the shadow mask with the phosphor dot screen behind it and
(b) the principle of the geometrical separation of three electron
beams. (c) is a cross-sectional view of the tube
is a metal plate with 357,000 holes in it, each hole being aligned with
one of the triads of phosphor dots. The three electron guns are
mounted symmetrically about the tube axis and the geometry of the
whole arrangement is such that electrons from any one gun can only
reach one colour of dots, the metal mask shadowing the dots of the
other two colours.
The focused electron beam of each gun covers an area of two or
three triads and the majority of the electrons are absorbed by -the
metal plate, only some 20 % reaching the phosphor screen. Despite
this absorption of electrons, a high light brightness of 20 foot lam-
berts is usual with an E.H.T. voltage of 23 kV.
The phosphor dots are deposited on the inside of the face plate
glass, which is curved with a radius of 26 in. The shadow mask
plate is made of copper -nickel alloy or, in later tubes, of steel about
0.006 in. thick and is curved to match the phosphor dot screen
curvature.
The mounting of the shadow mask is designed so that as the
mask heats up the expansion moves the apertures in the direction
of the beam travel.
Several auxiliary devices are needed outside the shadow mask
tube in order to obtain good colour pictures. The three electron
beams have their centres of deflection (see Section 11.6) inside the
deflector coils, at a distance of approximately 15 in. from the
centre of the phosphor screen. The electron guns are aligned so
that the undeflected spots approximately converge at the centre of
the screen. To achieve exact convergence in the undeflected position,
or static convergence, the electron beams are moved by means of
external magnets or electromagnets placed round the tube neck.
The convergence magnets move the three spots radially, the flux
from the external magnets being carried around the electron
beams by pole pieces built into the gun structures (see Fig. 3.8).
Thus any two spots can be made to overlap. The lateral move-
ment required to make the third spot coincident is provided by the
blue lateral shift magnet at the end of the tube, where there are
further pole pieces inside the tube neck around the blue beam.
As the three beams are deflected, the point of convergence moves
away from the phosphor screen to a position inside the tube itself.
The three electron -beam spots therefore diverge (Fig. 3.9) and must
be brought into coincidence again by varying the convergence fields
dynamically.
This dynamic convergence is achieved by applying parabolic
current waveforms to electromagnets on the convergence magnet
assemblies. These waveforms are derived from the frame and line
REMOTE CONTROL
IRIS MOTOR
MIRROR EITAD 3242
MOUNTING CONTAINER
DICHROIC FILTERS
RELAY LENSES
EITAD 32413 REAR
E/TAD 3195
EITAD 3249 FRONT
VERTICAL ASTIGMATISM
COMPENSATOR BLUE FILTER
EITAD 3235 EITAD 3246
HORIZONTAL ASTIGMATISM
COMPENSATOR WAD 3239
-MIRROR
0,0 3242
RIS HOUSING
-SUPPLY LEAD
TO IRIS MOTOR
The Marconi image orthicon colour camera with the covers re-
moved. The optical system which is shown in the top photograph
is in the front part of the camera. (Courtesy Marconi Ltd.)
A typical scene in the operating theatre at the Hammersmith Hospital. The camera, with lights mounted
on the front, photographs the actual operation via a 45° mirror vertically above the patient and is transmitted
by a closed loop television circuit to a lecture room remote from the theatre. (Courtesy E.M.I. Ltd.)
STATIC CONVERGENCE MAGNET
MOVES RADIALLY
DYNAMIC
CONVERGENCE COILS
BLUE BEAM MOVED

RADIALLY IN TUBE
T NECK BY
0- t 'OR TRANSVERSE FLUX
(a)
BLUE LATERAL
SHIFT MAGNET
ROTATES IN HOLDER
BLUE BEAM
MOVED TRANSVERSELY
ACROSS TUBE BY
RADIAL FLUX
(b)
KO 0
(c)
Fig. 3.8. Control of beam convergence in a shadow mask tube.

(a) shows dynamic convergence coils, (b) the blue lateral shift
magnet rotating in holder and (c) the arrangement of three guns
viewed from the socket of the tube
time base waveforms since the convergence correction required

depends on the position of the spot on the raster.
In order that the red electron beam shall land on only red phosphor
dots, its point of origin must be located with some precision with
respect to the shadow mask. Purity magnets are used to move all
three electron beams about in the tube neck until they appear to
6
GO OR Go R
Go oR
0 0
B 0 B
GO OR Go 0R
o o
B Go oR B STATIC CONVERGENCE
CORRECT IN
o
UNDEFLECTED POSITION
Go oR Go Go oR Go oR Go oR GO OR
0R
°B 0
B
Go oR
0
B
GO OR Go oR
0
B OR 13
Go
Go oR Go ,,R
0
0 0
B B
e)
Fig. 3.9. Convergence of three -gun tube. (a) shows the relative
flatness of the tube face and the tilt on the guns for dynamic conver-
gence control, and (b) shows the approximate separation of spots
as three beams traverse the screen
originate from the correct positions. The purity magnets may

consist of two annuli magnetized transversely to the tube axis so
that rotation of the two rings relative to each other alters the strength
of the field, while rotation of both rings as a unit rotates the direction
of the magnetic field across the tube neck.
Unwanted magnetic fields which vary across the picture area will
upset the alignment of the electron beams and spoil the colour
purity.
Upon installation, and occasionally after the receiver has been
moved to another position, the shadow mask and associated ferrous
material in the receiver must be demagnetized (or degaussed).
Finally, cancellation of any remaining field may be achieved by
adjusting a number of small colour equalizer magnets around the
bulb, although in later tubes it has been possible to dispense with
such equalizer magnets.
Moire patterns can arise through beat effects between the line
scanning structure of the raster and the shadow mask pattern but
these are negligible if the scanning lines are parallel to one of the
lines into which the shadow mask holes fall, and reach a maximum
at + 30° from these directions.
The three electron beams are focused electrostatically, all three
focus electrodes being connected together. The electron guns are
of tetrode construction and the three grid characteristics can be
made to match over the grey scale by using high screen potentials
to keep the characteristics relatively linear, and then adjusting the
screen potentials individually. Typical operating conditions are
Gun screen grid to cathode potential 200 to 500 V

Black -out grid to cathode voltage -50 to -150 V
Focus voltage 4 to 5 kV
Final anode voltage 20 to 25 kV
The holes in the shadow mask are etched by a photo -engraving

process from a master negative. Each shadow mask is then used
to lay down its own phosphor screen by a photographic process.
The tube face is coated with a photo -resist layer containing one of the
colour phosphors, say the red. The mask is placed in its correct
position and a point light source, situated at the position from which
the red electron beam will eventually appear to be emanating, is
used to expose the photo -resist layer. The coating is then washed
in a suitable solvent when the exposed dots will have been hardened
by the light so that only the unexposed parts are washed away. This
process is repeated for the other two colour phosphors.
3.9. Colour cameras
Both image -orthicon and vidicon tubes have been used in colour
cameras. Vidicon tubes make a much smaller and lighter camera
but suffer from the usual vidicon lag, which produces smear on
moving objects at low light levels. The image -orthicons must be
operated below the knee of their characteristics so that all three
tubes operate over a similarly linear part of their input-output
curve. The input-output characteristic of vidicon tubes is never
linear; however, its curvature is just about right to correct for the
input-output characteristic of display tubes so that no gamma
correction (see Section 4.11) is necessary.
The red, green and blue camera tubes are identical in design
and are mounted alongside each other in the camera housing (Fig.
3.10). The optical system is arranged so that the light from the
objective lens in the turret is split into its primary red, green and
blue images by dichroic mirrors and care is taken that the image on
each tube is identical in shape and size with its neighbours. Colour
filters are used to match exactly the combined spectral responses of
the tubes and dichroic mirrors to the required colour character-
istics. Neutral density filters are inserted to adjust the red and
FRONT SILVERED
MIRROR
OBJECTIVE CONDENSER RELAY
LENSES LENSES LENS
A
CAMERA TUBES
N.D.
A
'4--- GREEN
ASTIGMATISM IRIS
CORRECTOR
( 2 FLAT PLATES)
FRONT SILVERED
MIRROR
CORRECTOR PLATE
N.D.. NEUTRAL DENSITY FILTERS

C . COLOUR CORRECTING FILTERS
Fig. 3.10. Schematic plan of an R.C.A. colour camera
green camera sensitivities so that all three tubes operate over

similar portions of their transfer characteristics.
Since the dichroic mirrors have a finite thickness, the refraction of
a ray of light through them introduces an asymmetry between the
vertical and horizontal directions (Fig. 3.11) which is corrected by
the insertion of two further plain sheets of glass of the same thick-
ness as the dichroic mirrors, and at the same angle, but rotated
about the light axis by 90°. An additional corrector is needed in
the blue channel so that the blue light passes through the same
thickness of glass as the red and green light. Sometimes the
Fig. 3.11. Asymmetry produced by

dichroic mirrors. Astigmatism cor-
recting plates are rotated through
90° from the dichroic mirrors about
the light axis. (a) shows a horizon-
tal and (b) a vertical section through (a)
the dichroic mirror
.......,
.......i......
-.... . INN.13
:M. . WI/.
..---- * r. A
I
(b)
dichroic film of the mirror is itself sandwiched between glass to

ensure equality between transmitted and reflected rays.
To leave room for all these mirrors and filters it is necessary to
use a relay optical system. The real image produced by the objective
lens is relayed to the mosaics of the camera tube by an auxiliary
lens system, with a magnification of unity. A turret of condenser
lenses is mounted in the plane of the real image produced by the
objective lenses in order to collect as much light as possible. The
Fig. 3.12. Schematic plan of the E.M.I. vidicon

colour camera
DICHROICS
GREEN
VIDICON
TUBE
COLOUR FILTER
TURRET
LENS
FRONT
SILVERED MIRROR
turret of condenser lenses rotates with the objective lens turret but
only the objective lens turret moves longitudinally for focusing.
The iris diaphragm is mounted in the relay lens system and is
remotely controlled. The temperatures of the three image orthicons
are kept within the range 85-120°F.
Studio lighting levels are commonly of the order of 400 to 1,000
foot lamberts, several times the usual monochrome level.
3.10. Vidicon colour camera

Fig. 3.12 shows a block diagram of an E.M.I. colour camera
which uses small vidicon tubes. The optical system is novel since
each vidicon camera tube has its own narrow angle (9°) objective
lens. The dichroic and reflecting mirrors are placed in front of the
objective lenses where the narrow angle of field keeps the angle of
incidence over the dichroic mirrors fairly constant and astigmatic
errors are avoided.
The turret lenses are special colour corrected doublets which
convert the angle of view to either 6°, 18° or 29°.
3.11. Double and single tube cameras

Suggestions have been made for single tube cameras employing a
coloured filter strip over the mosaic, similar to the phosphor screen
structure of an " Apple " display tube. Other suggestions have
included two -tube cameras, the first tube producing a high definition
monochrome picture while the second tube uses a colour grid to
provide a low definition colour picture. So far only the three -tube
cameras appear to be practical.
3.12. Camera circuits

The viewfinder on colour cameras is usually a small monochrome
cathode-ray tube accepting either the green signal output or the
luminance signal.
Camera amplifier circuits are fairly normal in conception but of
necessity are in triplicate. Special correcting waveforms may be
added to the scanning waveforms of one or two of the camera tubes
to aid the registration of the three images. Aperture compensa-
tion, shading correction if required, blanking insertion and gamma
correction must be carried out as in monochrome practice but always
the three signals must be treated similarly. Particular care must be
taken that the three channels black out together. The standard
voltage output is 0.7 volt peak -to -peak across 75Q for red, green and
blue, or 1.0 volt peak -to -peak if synchronizing pulses are added.
3.13. Chromacoder
In this type of studio equipment a single camera tube is used with
a rotating disc to provide a field sequential colour signal. The
sequential signal is then standards converted to a simultaneous signal
for transmission (Fig. 3.13). This type of pick-up has been called
simulquential.
The field sequential signal does not need to conform to the band-
width of the transmission standards and much of the objection to
field sequential operation can thereby be overcome. The pick-up
camera operates at three times the field frequency of the transmission,
with three times the line scanning frequency, so that three times the
video bandwidth is required in the camera circuits. The resulting
STORAGE TYPE
CAMERAS
RED SIGNAL
ROTATING
COLOUR DISC WIDEBAND
CAMERA
HR
CIRCUITS
IGREEN SIGNAL
STUDIO
SCENE
PICK UP
HG
CAMERA
FIELD BLUE SIGNAL

GATING
CHROMACODER FIELD SEQUENTIAL

CIRCUIT
mAl HB
CAMERA OPERATING AT THREE TIMES CONVERTER
NORMAL FIELD AND LINE RATE CAMERAS ON
NORMAL SIMULTANEOUS
STANDARDS
Fig. 3.13. Chromacoder arrangement of cameras and converter tubes
red, green and blue signals are gated out and displayed on separate
monitor display tubes. The monitor display tubes are mono-
chrome with a relatively long afterglow and each monitor is viewed
by a separate camera operating on the normal simultaneous trans-
mission standards. The combination of long persistence phosphors
and storage in the converter camera tubes, which are typically C.P.S.
Emitrons, enables the information provided in one high repetition
rate chromacoder field to be read off at the slower simultaneous
field rate. Thus for 405 -line standards the chromacoder would
provide a complete red field, scanned in o of a second, once
every second. This red field would then be read off by the
converter camera in ir, second whilst the other two converter
cameras were also reading off their fields.
To overcome moire patterns the pick-up camera and the monitor
tubes may be scanned vertically whilst the converter cameras scan
horizontally, as normal.
3.14. Colour film scanners

These are basically similar to monochrome film scanners in the
methods used to present the scanning apparatus with a stationary
film frame. The film may move continuously and an optical
device used to maintain a projected image of a frame stationary in
the gate. The long pull -down time required may be offset by
only illuminating the film frame for a fraction of a field period and
using a storage type camera. The camera viewing arrangements
are then similar to the beam splitting arrangements used in colour
cameras.
For flying -spot type scanners which have no storage the film
image must be kept stationary during the entire scanning process.
A very bright image of an unmodulated cathode-ray tube raster is
focused on to the film frame, the light passing through the film is
collected by a condenser lens system and is then split into red, green
and blue beams by an arrangement of dichroic mirrors and filters.
Each beam is focused on to a photocell which transforms the light
into an electrical signal. Since the phosphor of the cathode-ray
tube producing the original raster has an appreciable afterglow,
each of the three photocell amplifiers must have decay time correc-
tion.
3.15. Colorimetric requirements

Colour television systems are engineered, as far as possible, to
reproduce a colorimetric match to the original scene. The colori-
metric analysis which the camera performs must therefore be
related to the primaries which the receiver is going to use. These
primaries and the reference white fix the relative spectral response
curves which the red, green and blue cameras must have.
The colour response of a camera channel is the net response of the
optical system, the dichroic mirrors, the gelatine colour correcting
filters and the camera mosaic. Typical mosaic responses are
shown in Fig. 3.14 for the vidicon and image -orthicon colour
camera tubes. The resultant overall responses are required to
have the same shapes as the colour mixture curves I, g, b, for
the receiver primaries (see Section 2.23). Thus if the input to the
1.0
- -. - -
08
1......
N.. ...
. .....
.."-IMAGE ORTHICON
0 "s
.
04
VIDICON
0.2 `..
\..
0 . .
400 450 500 550 600 650 700

WAVELENGTH A MA
Fig. 3.14. Spectral response curves of two typical colour camera tubes.
Illumination is equal energy white
camera is a spectrally pure radiation, of wavelength A, the receiver

will reproduce a match to this colour if the camera outputs are
proportional to the ordinates on the colour mixture curves at the
wavelength A. For non -spectral colours the camera mosaics will
effectively integrate the product of energy distribution in the light
reflected into the camera and the transmission through the camera's
colour filters, over the spectrum
r,80
E(A) dA = R
J 380
so that the camera output is proportional to the required tristimulus

value.
The colour mixture curves for the N.T.S.C. primaries and
reference white are given in Fig. 2.12. It will be seen that they have
appreciable negative lobes. These negative lobes can be simulated
by using more than three camera tubes but this is not yet a practical
proposition. It has been suggested that the camera spectral
responses should follow the C.I.E., x, y and 2 colour mixture curves
which have no negative lobes, and the camera outputs matrixed
from X, Y and Z, to give the required RGB outputs. Such matrixing
results in a worse signal-to-noise ratio.
The present practice is to match the camera response characteristics
to the positive parts of the colour mixture curves, with some trim-
ming of the sides of the responses to offset the loss of the negative
lobes, as in Fig. 3.15.
The colour mixture curves of Fig. 2.12 and the camera curves of
Fig. 3.15 assume that the scene lighting is equal energy white. In
1.2
/ N.\
BLUE GREEN RED
1.0 ...-.
.. ..
/
08
.. ..
.
/ .
06
:
.
.. ..
.
.. i \
. .
. .
\.
. \
04
. \
0
:
i N.
0 i.
400 450 500 550 600 650 700
WAVELENGTH A mit
Fig. 3.15. Practical camera taking responses
principle the filters should be modified for other types of lighting

so that the combined effects of the spectral distributions of the
lighting and the responses of the camera mosaics, filters, dichroics,
etc., total the curves of Fig. 3.15.
3.16. Electronic masking
Colour films have certain residual colour errors of a systematic
nature which arise from imperfections in the dyes used. These
errors are equivalent to a form of crosstalk between the red, green
and blue signals and can be partially corrected at the transmitter
by introducing a cancelling crosstalk between the three colour
signals. The film errors have a logarithmic variation with bright-
ness but the correction applied at the transmitter is based on linear
signals and is only partially effective, although it can be applied
after gamma correction has been carried out.
Colour films usually have a higher contrast range than the tele-
vision system can accommodate and the extra gamma correction
needed has the effect of reducing the saturation of the reproduced
colours. Correction can also be made for this desaturating effect.
The process of correction is called electronic masking by analogy
with the photographic process of masking in which a correction
transparency or mask is printed in register with the original film.
The electronic masking signal takes the form of a colour difference
signal (see Section 4.13) which is added to each primary signal.
Thus the masked green signal would be
E" G = K4E' G 1E" R K 2E' G K 3E' 13)
where Ki. + K2 + K, = 0, so that the correction signal disappears
for greys when ER=EG= E' B. The constants can be computed
from colorimetric measurements on the film but are usually adjusted
on the basis of subjective assessment of the resulting picture quality.
3.17. Summary
The colour camera or film scanner analyses the scene to be
transmitted into its three primary colour components and develops
an electrical signal proportional to the integrated distribution
coefficient of the corresponding receiver primary. The electrical
signals undergo various correction processes before being trans-
mitted. These processes take account of deficiencies in the original
film stock, transmitting equipment, and receiving equipment.
The most successful type of receiving tube is the three -gun shadow
mask tube, despite all the external correcting adjustments which are
necessary.
CHAPTER 4
Transmitter Coding
4.1. Introduction
After the scene to be televised has been analysed by the colour
camera, the red, green and blue outputs have to be transmitted in
an agreed form. In the N.T.S.C. system this agreed form is deter-
mined partly by engineering considerations and partly by more
general requirements.
The main part of the signal is similar to a monochrome television
signal and the modulation waveform contains sync pulses combined
with a video signal representing the brightness of the scene being
televised (Fig. 4.1 (a) ). This luminance signal will produce a
satisfactory black and white picture on either a monochrome or a
colour receiver and is said to be compatible.
To produce a colour picture on a colour receiver the transmitted
signal must contain further information. In the N.T.S.C. system
this further information represents the difference between the colour
of the picture element being transmitted and a white light of the same
luminance or photometric brightness, and this difference becomes
smaller as the colour in the scene approaches a pastel shade or a
grey tone and also becomes smaller as the luminance of the scene
decreases (Fig. 4.1 (b) ). This extra information is the chrominance
of the scene, described in Section 2.32, and needs two quantities
or numbers to specify it.
The two chrominance parameters are added to the main lumi-
nance signal in such a way as to cause minimum interference to
monochrome reception. They are first modulated onto a carrier
whose frequency falls in the upper part of the video frequency spec-
trum, at 2.7 Mc/s for the 405 -line system. The combined signal at
2.7 Mc/s, with sidebands, is then modulated onto the main vision
carrier, which may be 45 Mc/s for example, along with the luminance
signal (Fig. 4.2). The first carrier at 2.7 Mc/s is then called a
sub -carrier. To modulate two quantities onto the one sub -carrier
frequency requires special techniques at the transmitter, and the
detection of such a signal at the receiver necessitates circuits which
are unlike those normally used in monochrome television. Such a
78
TRANSMITTER CODING 79
relatively complex process was developed only after more obvious
methods of transmission had been shown to be unsatisfactory.
4.2. Early methods of colour transmission
Since there are three quantities to be transmitted, the simplest
solution would be to use three separate television channels (Fig.
4.3), in the same way as the monochrome vision and sound signals
TRANSMITTED SCENE
Z ,,,' Z
Z ,,,
tai
, lc.
CC
W
C.,
,...
). ...
La
,..,
,c
5 0-J
IA1 0
CC
LLI W Cr IA./
be
(..,
_,
a,
3 X 6- g- ¢ ....
0 = cc w
.cc
en
..,
ffi133
cc c:,
Fig. 4.1. Modulation wave-

forms for monochrome and
colour. The signals are:
(a)
(a) monochrome video
and colour luminance;
(b) chrominance;
(c) complete colour
signal, luminance plus
chrominance
are transmitted separately. This necessitates an increase of rather

more than three times in the bandwidth occupied by the vision infor-
mation. The three transmitted images must be registered accurately
at the receiver so that the red picture lies exactly over the green
picture and both must coincide geometrically with the blue image.
Further, if the colour balance of the resultant picture is to be con-
sistent, the three channels, including the transmission paths, must
have the same gain and identical input-output linearity, for long
periods.
4.3. Field sequential systems
To overcome these difficulties, many systems have been proposed
using a single channel only and transmitting the red, green and blue
SOUND VISION
SUBCARRIER CARRIER SUBCARRIER CARRIER
2 3 42 43 44 45
VIDEO FREQUENCY (M C/5) R.F. FREQUENCY (M C/5)
(b)
Fig. 4.2. Addition of sub -carrier to vision signal
information sequentially along the one channel. Baird demons-

trated such a system in 1928 using two rotating discs, each with
segments of red, green synchronism
before the camera tube and the receiver tube (see Section 3.5).
Each primary coloured filter remained over the tube face for the
period of one field. A more advanced system, developed by the
1.3 SOUND RED GREEN BLUE
CARRIER CARRIER CARRIER CARRIER
5E! Fig. 4.3. Three simultaneous

cc
colour channels
40 45 50 55
FREQUENCY (MC/S)
Columbia Broadcasting System, was adopted in the United States

by the Federal Communications Commission as the official American
Colour System in 1950, but was discarded in 1951.
Such field sequential systems have two major drawbacks. If the
picture being radiated consists mainly of tones of one hue, such as
green, then the red and blue fields are black or nearly so, and each
green field is succeeded by two black fields. This increases the
brightness flicker of the reproduced picture. Even when the
transmitted picture consists of grey tones, the luminance of successive
fields varies since the system primaries have different luminosity
coefficients (see Section 2.16). To overcome such flicker effects
the field scanning rate must be increased by a factor of three, in

which case even when only one primary colour is being transmitted,
the time between active fields is the same as for monochrome
television. Such an increase in field scanning rate requires an
equivalent increase of three times the bandwidth, which makes the
total channel width similar to that for three -channel simultaneous
operation.
A second disadvantage is colour fringing, described in Section
3.5, although the increase in field rate needed to overcome bright-
ness flicker is normally also enough to overcome colour fringing.
4.4. Line sequential systems
To avoid these defects, attempts have been made to send the three
primary colours at line sequential rate, that is to say, one line in red,
the next in green and so on. This reduces the large area flicker and
fringing effects to line effects, but they are still annoying.
4.5. Dot sequential systems
A logical development was to send the colour information at
picture element sequential rate, or one picture element in red, the
next in green and so on. Each line is then made up of a sequence
of red, green and blue dots. Such a dot sequential method of
sending the information within the restricted bandwidth of a normal
monochrome channel was found to be similar to sending the infor-
mation simultaneously and led to the use of the sub -carrier technique.
4.6. Bandwidth saving in simultaneous systems
This dual character of the dot sequential system renewed interest
in simultaneous systems and,' particularly in America, considerable
engineering effort was devoted to reducing the wide bandwidth
previously required for simultaneous three -channel operation.
4.7. Non -engineering considerations
A public colour television service must provide more than an
academic solution to transmitting coloured images from one
laboratory to another; it must be economically suited to society.
The present national production per day of monochrome receivers
sells for about the price of one complete transmitter, including its
site and buildings. Simplicity and cheapness of the receiver can be
seen to be of paramount importance while complexity and cost at
the transmitter are secondary considerations.
The ether space is already overcrowded with communication
services of one kind and another and ideally the colour system
should allow colour broadcasting to take place with only a few
minor modifications using the same channel frequencies and band-
widths and the same cable and radio links as are in use for mono-
chrome. Such a colour broadcast service ought not to interfere
with other radio services. This ideal system should give good black
and white reproduction on the millions of present-day monochrome
receivers and should allow them to be cheaply converted to colour
reception. The quality of the colour picture should satisfy public
demand for a few decades, reception difficulties should not be
increased and studio and domestic equipment should be reliable
and rugged.
Above all, the system adopted should be such as to encourage the
development of new techniques and devices and to be capable of
absorbing advances in the art as they are made. It is difficult to
know when this criterion has been met.
Any change in a widely used public service tends to cause con-
siderable financial expenditure and such changes should, in the
public interest, be as infrequent as possible. If the advent of colour
broadcasting necessitates any major upheaval in reception equip-
ment it is clearly politic to revise the system on all scores such as
resolution, aspect ratio, compatibility with foreign programme
sources, etc. Such changes are only worth while if they procure
major improvements in picture quality. Systems so far demonstrated
represent only compromise solutions to these conflicting require-
ments.
The N.T.S.C. system is the most successful one yet field tested.
It makes use of certain properties of the eye to reduce the amount
of information which must be transmitted. The extra information
which is needed to change a monochrome picture into a colour
picture is transmitted inside the normal channel bandwidth in such
a way that it hardly upsets the monochrome picture. Although the
amount of this extra colour information is small, the resulting
picture does not look blurred as the eye is deceived by the sharp
luminance picture which is transmitted.
4.8. Choice of primary colours and white
In any colour television system it is necessary to standardize the
primary colours, or reference stimuli, which are used in transmission.
The primaries adopted in the N.T.S.C. system have been chosen to
allow a wide range of colours to be reproduced and they are colours
which can be achieved at the receiver with cathode-ray tube phos-
phors which are reasonably efficient. The position of the N.T.S.C.
red, green and blue primaries on a C.I.E. diagram are shown in
CO -EFFICIENT OF X
Fig. 4.4. N.T.S.C. system reference stimuli
Table 4.1
N.T.S.C. PRIMARIES AND WHITE
x y z X y z
Red 0.67 0.33 0.00 Cyan 0.17 0.31 0.52
Green 0.21 0.71 0.08 Magenta 0.35 0.18 0.47
Blue 0.14 0.08 0.78 Yellow 0.45 0.51 0.04
The reference white-C.I.E. Illuminant C 0.3101 0.3163 0.3736
7
Fig. 4.4. Any colour which falls inside the triangle R*G*B* can
be reproduced; colours outside the triangle can be represented by the
correct dominant wavelength but at lower purity (see Chapter 2).
The complementaries of these primaries are represented by the
points where the line from one primary drawn through the white
point Y meets the opposite side of the colour triangle. These com-
plementary colours are called cyan, magenta and yellow.
4.9. Receiver white
The white point Y determines the relative size of the units in which
R, G and B are measured, these units being such that one unit of
red plus one unit of green plus one unit of blue light produces 3
units of the chosen white light Y, which is called the normalizing
illuminant. The single unit of red light gives rise to a signal voltage
of 1 in the red channel, and similarly for green and blue. The
combination of these three lights produces 3 units of white, but the
corresponding luminance signal voltage is unity, as will be seen in
Section 4.15.
The particular white chosen for N.T.S.C. receivers is C.I.E.
standard Illuminant C, a rather bluish white representing the light
from an overcast north sky at midsummer, and not unlike the colour
of many modern monochrome television tubes. When the camera
output signals are equal, no sub -carrier is transmitted and the
receiver must reproduce Illuminant C.
4.10. Studio white
It is cheaper and easier to use lighting in the studio which is more
red than the equal energy white theoretically assumed. The practice
is to use whatever illuminant is available at the camera and to
normalize the camera outputs so that when a nominal white card
is held in front of the camera the three outputs are equal and no
sub -carrier is transmitted. On such a signal the receiver reproduces
Illuminant C, and the producer monitors the transmission on this
assumption.
4.11. Gamma correction
The light output from cathode-ray tubes is not directly proportional
to the modulating grid -cathode voltage, even when constant voltages
such as bias and signal set-up have been taken into account. This
non -linearity distorts the tone gradation in monochrome television,
and a typical graph of input voltage against light output is shown in
Fig. 4.5 (a), where the peak amplitudes of both have been normalized
to unity. The law connecting effective signal voltage Ve-that is
to say, voltage above the black -out bias on the tube-with L, the
light output in foot lamberts, is approximately
L = k(Ve)7
where normally y = 2.2 2.7
By analogy with photographic work, the exponent is usually
represented by the Greek letter gamma (y).
The transmitter is predistorted to correct for the transfer character-
istics of the cathode-ray tube and this applies to both black and white
and colour television. The camera output signal is passed through
1.0 I.0
08 'S 0.8
RECE VER
y =2-2
r-- 0-6 06
0 0.2
5 04 gl 0.4
TRANSMITTER
y =2.2
0.2 0.2
I
0 0
02 04 046 08 10 02 04 06 08 10
GRID -CATHODE VOLTAGE LIGHT INPUT
(a) (b)
Fig. 4.5. Input-output curves for receiver and transmitter
a gamma corrector circuit with an input-output characteristic,

Fig. 4.5 (b) which is the inverse of Fig. 4.5 (a).
The gamma circuit is sometimes called a rooter since it carries out
the mathematical operation of taking the gamma root
which is approximately the square root, of the incoming voltage.
In N.T.S.C. television, this transmitter distortion leads to some
departure from the ideal performance of the system (see also
Chapters 5 and 16); but it does improve the signal-to-noise ratio in
the darker parts of the picture by about 12dB.
N.T.S.C. gamma correction is based on a value of y = 2.2.
Each red, green and blue camera output is corrected separately,
and this distortion is performed after the camera outputs have
been normalized on the studio white and before any coding of the
signals has taken place, that is to say, before the luminance or
chrominance signals have been formed. To indicate that gamma
correction has taken place the symbols representing signal voltages
are marked with a prime if they refer to circuit points after the
gamma correctors; thus G represents a voltage in the green camera
channel before the gamma corrector and G' a voltage which has
passed through the correcting circuits.
4.12. Constant luminance principle
Everyday experience as well as scientific measurement indicates
that the eye is more conscious of changes in luminance than changes
in hue or saturation. Similarly, experiments in colour television
have shown that noise which is altering the luminance of the picture
is more noticeable than noise which is only altering the chromaticity,
and indeed that observers can tolerate an increase in the amount
of noise in the colour channel of some 8dB if this is not affecting the
brightness of the picture.
As far as possible modern colour systems and receivers take
advantage of this by making variations in colour signals produce
only variations in chromaticity of the reproduced picture. The
luminance of the picture is constant however the other colour
signals vary and whatever interference is superimposed on them.
This is called the constant luminance principle. If one of the three
signals is a function only of the luminance of the colour then the
receiver can be designed to obey the constant luminance principle.
In the N.T.S.C. system the main signal, termed Y', carries the
monochrome or luminance information and is formed by adding
fractions of the red, green and blue signals together so that these
signals contribute to the brightness signal in proportion to their
luminosity coefficients. For the N.T.S.C. primaries and illuminant
C' these luminosity coefficients are red = 0.299, green = 0.587 and
blue = 0.114. For most engineering purposes it is sufficiently
accurate to write
Y' = 0.30R' + 0.59G' + 0.11B'
Since the quantities Y', R', G' and B' represent voltages they are
often written E'y, E' G and E' B.
The camera outputs are normalized or adjusted so that they give
equal outputs on the studio white, thereby fixing the relative units
in which R, G and B are measured. The maximum value of each
signal on peak white is 1, so that 0.30R' equals 0.30 whilst peak
white is being transmitted. For a grey picture which gives half the
maximum R', G' and B' signals
Y' = 0.30(0.50) + 0.39(0.50) + 0.11(0.50)
=-- 0.15 ± 0.295 + 0.055
= 0.50
In general, if R' = G' = B', then
Y' 0.30R' + 0.59R' + 0.11R'
= R' = G' = B'
If we give R, the camera signal before gamma correction, the

value of 0.5 then, for a grey picture
Y' = 0.30(0.50)2'2 + 0.59(0.50)2'2 + 0.11(0.50)2'2

= 0.30(0.73) + 0.59(0.73) + 0.11(0.73)
= 0.73
which is different from the value of Y' when R' = R" and = 0.50.
The actual value in volts of any of these signals depends on the
point in the system at which they are measured. Thus a peak Y'
signal which may be 1 volt when it is formed from R', B' and G'
may be represented by a signal of 1 mV at the receiver aerial, by a
few volts at the detector, and by a hundred or so volts at the cathode-
ray tube. The relative values of Y', R', G', and B', are carefully
maintained, however, and it is convenient to relate them to the
corresponding peak signal value so that the amplifier gains can be
ignored.
For a pure red colour, G' = B' = 0, and R' will vary between
0 and 1 depending on the luminance of the red; the corresponding
Y' signal will vary between 0 and 0.30.
4.13. Colour difference signals
The information which must be transmitted apart from Y' is
- Y') and (B' - Y'),
first derived as two colour difference signals (R'
obtained by subtracting the derived Y' signal from the red and blue
camera output voltages.
In the three signals, Y', (R' - Y') and (B' - Y') there is sufficient
information for the receiver to calculate R', G' and B', which are the
voltages it needs to apply to the display tube, since
R' = (R' - Y') Y' [4.1]
B' = (B' - Y') Y' [4.2]

and since
Y' = 0.30R' + 0.59G' + 0.11B' [4.3]
= 0.30Y' +0.59Y' +0.11Y'
and hence
0 = 0.30(R' - Y') 0.59(G' - Y') 0.11(B' - [4.4]
that is
0.30 0.11
G' - -059 (R' - -059 (B' - Y')
and
0.30 0.11
G' = Y' -(57§ (R' - Y') -IT§ (B' - Y') [4.5]
= Y' - 0.51(R' - Y') - 0.19 (B' - Y') [4.6]
From both Equation 4.3 and 4.6 it is clear that

G' = 1.70Y' - 0.51R' - 0.19B' [4.7]
Notice that with these three signals, Y', (R' - Y') and (B' - Y'),
for a peak white signal in which
R' = G' = B' = 1
then Y' = 0.30(1) + 0.59(1) + 0.11(1) = 1
so that B' - Y' 1-1=0
and R'-Y'=1-1=0
and the colour difference signals disappear, as indeed they do for
any amplitude of white such as
R' G' B' = I
This illustrates one advantage of using colour difference signals;
for a monochrome scene only the luminance signal needs to be
transmitted and for desaturated colours the colour difference signals
are small, which improves the compatibility of the signal. In practice,
for 90% of the time it is found that the colour difference signals
have less than one-third of their maximum values. Similarly,
colour difference operation also has the advantage that changes
in relative gain between the three transmitted signals do not affect
the colour balance of the grey scale, since the colour difference
signals are zero for greys. Before they are finally transmitted, the
colour difference signals are reduced in bandwidth and the receiver
has to derive a low -definition colouring picture to superimpose on
the sharp luminance picture. Colour difference signals have an
advantage here in that the receiver design is simplified compared
with transmission of a wide band Y', and, say, narrow band R' and
B' signals. In the latter case, the derivation of a combined high
definition luminance signal and a low definition colour signal for
each primary colour is more involved, because, for example, if the
R' signal were applied directly to the red display, the high frequency
luminance components carried by the Y' signal would be absent.
One way to derive the required signal would be to add to the low
definition R' signal those high frequency components of Y' which
fall outside the R' pass -band, but this requires careful tailoring of
the frequency response shapes in the overlap region. A better
method of obtaining the combined signal would be to form (R' - Y')
and (B' - Y') signals at the receiver, bandwidth -limit them, and
then apply them to the displays, together with the high definition
Y' signal. Obviously, the transmission of bandwidth -limited
(R' - Y') and (B' - Y') signals avoids this complication at the
receiver. It would be possible to transmit any two difference signals
such as (R' - Y') and (G' - Y'). However, it will be found that
the green difference signal for any colour is always smaller than, or
at most equal to, either the red or the blue difference signal for that
colour. Hence, the use of (G' - Y') in transmission would require
gain at the receiver for the derivation of the third difference signal,
with a corresponding deterioration in noise performance.
The values of the colour difference signals are not always those
which one might superficially expect for a particular colour. For
a saturated red signal, in which G' = B' = 0, and R' = 1
= 0.30
and (B' - Y') = - 0.30
and (R' - Y') = 0.70
Table 4.2 includes example values of the three colour signals
for various typical cases; the I and Q values will be explained in
Section 4.16. The reader will find it instructive to check the
arithmetic in this table and to work out a few examples for himself.
Notice that the blue difference signal (B' - Y') does not vanish
for colours which contain no blue component, but is negative. For a
bright pure blue colour, the (R' - Y') signal does not disappear but is
(0 - 0.11) - 0.11. Note also that the desaturated red taken as an
example in Table 4.1 may be considered as a grey signal of
value R' = G' = B' = plus a saturated red in which
Table 4.2
TYPICAL VALUES
Camera Colour Trans-

Voltages Difference mission Sub -carrier
Voltages Signals
Colour
Amplitude Phase
R' G' B' Y' (R'- Y') (B'- Y') I' Q' 'W (f)2-1-(0)2
(tan -1 lo -,)-1- 33°
White 1 1 1 1 0 0 0 0
Grey 0.2 02 0.2 0.2 0 0 0 0 0
Black 0 0 0 0 0 0 0 0 0
Bright Pure Red 1 0 0 0.30 0.70 -0.30 0.60 0.21 0.63 103°
Less Bright Pure Red 0.5 0 0 0.15 -0.15 0.30 0.11 0.32 103°
Bright Pastel Red 1 0.5 0.5 0.65 -0.15 0.30 0.11 0.32 103°
Dark Pastel Red 0-2 01 0.1 013 -0.03 0.06 0.02 0.06 103°
1 0 1 0.41 0.59 0.28 0.52 0.59 61°

Bright Pure Magenta
Dark Pure Magenta 0.3 0 0.3 012 018 1
0.08 016 018 61°
Bright Pure Green 0 1 0 0.59 -0.59 -0.59 -028 -0.52 0.59 -119°
Bright Pure Blue 0 0 1 011 -011 0.89 -0.32 0.31 0.44 -13°
Bright Pure Cyan 0 1 1 0.70 -0.70 0.30 -0.60 -0.21 0.63 -77°
Bright Pure Yellow 1 1 0 0.89 0.11 -0.89 0.32 -0.31 0.44 167°
R' = 1, G' = B' = 0, and that the colour difference signals are the
same as for the example of a medium bright saturated red in which
R' = I- and G' = B' = 0. Notice that a saturated red of high luminance
gives rise to larger colour difference signals than the same red at a
lower luminance, although the chromaticity of the colour has not
altered.
4.14. Constant luminance
If there is no gamma correction this type of colour difference
signal satisfies the constant luminance principle. If a noise signal
voltage is added to the colour difference signals it does not affect
the luminance of the reproduction but only the chromaticity. As
an example, consider the case of a colour represented by
0.70R* + 0.40G* + 0.50B*
which gives rise to camera voltages of R = 0.70, G = 0.40 and
B = 0.50 and to transmitted signals, which for clarity in this
particular case are indicated by a subscript t, of
Ye = 0.30R ± 0-59G ± 0.11B
= 0.30(0.70) + 0.59(0.40) + 0.11(0.50)
= 0.50
(R - Y)e = (0.70 - 0.50) = 0.20
(B - Y)e = (0.50 - 0.50) = 0
The addition of a noise voltage of, say, 0.10 to either or both
colour difference signals will not affect the luminance of the received
colour.
The new signals may be
Ye = 0.50, (R - Y)e = 0.30 and (B - Y)e = 0.10
R = (R - Y)e + Ye = 0.80
B = (B - ne + Ye = 0.60
G = Ye - 0.51(R - Y)e - 0.19(B - Y)e
= 0.50 - 0.51(0.30) - 0.19(0.10)
--- 0.33
These give rise to a received colour of luminance
0.30R + 0.59G + 0.11B
= 0.30(0.80) + 0.59(0.33) + 0.11(0.60)
= 0.50
which is the luminance of the original colour without the noise
voltages added.
In general, for transmitted signals Yt, (R - Y)t and (B - Y)t,
the receiver can deduce that
R = (R - + Ye
B = (B - Y)e ± Yt
G = Ye - 0.51(R - Y)t - 0.19(B - Y)t
from Equation 4.6
and these values give rise in a linear system to a colour of luminance
0.30R + 0.59G ± 0.11B
or
0.30(R - + 0.30 Yt + 0.59 Yt - 0.30(R - Y)t - 0.11(B - Y)t
+ 0.11(B - Y)t 0.11 Yt
= Ye
whatever variations occur in (R - Y)t and (B - Y)t.
It is possible to obtain constant luminance operation without
using colour difference signals. For example, if the three trans-
mitted signals were Ye, Re and Be, then for the same colour as used
in the previous case, Yt = 0.50, Re = 0.80 and Bt = 0.60 if a
noise voltage of 0.1 is added to Re and to Bt.
The receiver then works out that
G = 1.70(0.50) - 0.51(0.80) - 0.19(0.60) [4.7]
=0.85 -0.41 -0.11
= 0.33
Then as before the reproduced luminance on a linear system would
be
0.30(0.80)+ 0.59 (0.33) + 0.11(0.60) = 0.50
The essence of constant luminance operation is thus to transmit
a monochrome signal which is accurately a function of luminance,
and of luminance only.
In practice the system departs from this ideal constant luminance
operation because of the need to gamma correct the signals. The
subjective addition of the brightness of the three primary components
is only carried out after they have been raised to the power y, and the
10
RED
MAGENTA
0.5
YELLOW
-1.0 -0.5 0.5 1.0

BLUE
GREEN
CYAN
POSITIONS OF PRIMARY
COLOURS AND THEIR COMPLEMENTARIES
Fig. 4.6. Colour difference axes
examples quoted are therefore only approximately accurate. This

departure from constant luminance operation is discussed further
in the next chapter.
The chrominance signal which is finally transmitted in the N.T.S.C.
system is obtained after the (R' - Y') and (B' - Y') signals have
both been reduced in amplitude.
4.15. Reduction of colour difference signal amplitudes
Describing the chrominance information by two quantities
(R' - Y') and (B' Y') makes it convenient to represent the
colour being transmitted by a diagram such as Fig. 4.6. The point
C indicates a colour whose chrominance is measured by the pro-
jections of OC on the (R' - Y') and (B' - Y') axes.
(R' - Y') = OC sin 96
(B' - Y') = OC cos
The particular colour illustrated is a magenta colour. From
Table 4.1 the chrominance signals corresponding to the other colours
of the spectrum can be plotted and the positions of these primary
colours are indicated in Fig. 4.6. The angle the line OC makes
1.14
1.0 -
RED.
.MAGENTA
0.5 C
YELLOWI.
-1.0 -05 0 1.0
BLUE
BLY'
2.03
-0.5r
GREEN
CYAN
POSITIONS OF PRIMARY
Fig. 4.7. Weighted colour difference axes
with the (R' - Y') and (B' - Y') axes indicates the dominant wave-
length of the colour, but the length of OC increases with both
luminance and the purity of the colour.
It is expedient to change the scale along the (B' - Y') axis by
plotting -13(B' - Y') rather than (B' - Y') itself, in order to
2. 0
reduce the amplitude of the transmitted chrominance signal.
Similarly, the scale along the (R' - Y') axis is changed by plotting
1
1 14
(R' - Y') instead of (R' - Y'). Such a reduction in chromi-
nance. signals avoids undesirable overload effects at the transmitter.
The considerations which led to the choice of such reduction factors
1
as
2- rather than a half are discussed in Section 5.4.
The diagram of Fig. 4.6 is replotted using these new scales in Fig.
4.7. Notice that the positions of the primary colours are now
different although they correspond to the same values of (R' - Y')
and (B' - Y'). It is important to check the scale along any dia-
grams with (R' - Y') and (B' - Y') axes to see whether the reduc-
tion factors - and -- have been used or not, since all the angles
and lengths on the diagram are generally different for the two cases.
This reduction in amplitude of (R' - Y') and (B' - Y') is called
the weighting of the difference signals.
4.16. Choice of colour axes
As well as reducing the amplitude of the chrominance signals it is
also necessary to reduce their bandwidth in order that they may be
superimposed on the luminance signal without causing undue
interference. Section 4.21 deals with the choice of bandwidths for
the two chrominance signals. It is advantageous to reduce the
width of one chrominance signal more than that of the other signal.
/ 114
a1
MAGENTA
0.5
Fig. 4.8. I' and Q' axes YELLOW
BLUE 2.03
,
O'
0yo.
GREEN
-0.5(
CYAN
-1.0
POSITION OF PRIMARY
The (B' - Y') axis could be made the narrow band axis but from
the data in Section 1.14 it is reasonable to make the narrow band
axis lie amongst those colours which the eye is first unable to
distinguish as the size of the coloured area decreases. If the eye
cannot see changes in these colours in small areas, then it will not
notice, from a reasonable viewing distance, that such changes are
not being transmitted. The narrow band axis is chosen to lie among
the green -magenta colours. The exact choice of position is not
critical and the ideal axis varies with the luminance changes associated
with the small coloured areas; it has been standardized as an axis
which makes an angle of 33° anticlockwise from the -..11T-)i(B' - Y')
axis, as in Fig. 4.8. This new axis is called the Q' axis. The
second chrominance axis is chosen at right angles to the Q' axis
and is called the I' axis. The I' axis then lies among the orange cyan
colours for which the eye has maximum chromaticity discrimination
in medium -small areas. This sensitivity to chromaticity changes
does not vary quickly with the position of the I' axis and the choice
of the I' axis perpendicular to the Q' axis is quite acceptable. On
the (R' - Y') and (B' - Y') axes of Fig. 4.6 the I' and Q' axes
make angles of 139° and 20° with the (B' - Y') axis and have scales
of 1.51' and 1.8Q'; the importance of checking the scale used for
(B' - Y') and (R' - Y') is thus underlined.
The colour C of Fig. 4.8 can be represented either by its projec-
tions1 on the old
2.03
(B' - Y') and -1.14 (R' - Y') axes, or by its
projections on the new I' and Q' axes, and there is a direct relation
between the values of I' and Q' for a colour C and the values of
(R' - Y') and (B' - Y') for the same colour.
The chrominance component along the new Q' axis can be con-
sidered as the sum of the projections of the (R' - Y') and
1-14
1
(B' - Y') components along the Q' axis.
2 .03
(R' Y')
.*. Q' - 1.14
sin 33° +
Y)1.14 sin 2.03
cos 33° [4.8]
Similarly
(B' - Y')
- (R' - Y') cos 33°
1 . 14 2 03
sin 33° [4.9]
The negative sign occurs because the (B' - Y') component lies
along the negative part of the I' axis.
4.17. Modulation of the sub -carrier
The two I' and Q' chrominance signals are transmitted by modulat-
ing them onto a sub -carrier. Two sine waves are generated at the
same frequency but differing in phase by 90° so that the phasors
representing these two sine waves (Fig. 4.9) could well be drawn
along the I' - Q' axes in Fig. 4.8. One phasor is then amplitude
modulated with the I' information, and the other phasor is ampli-
tude modulated with the Q' signal. This modulation is carried out
in balanced modulators (see Chapter 6) so that if there is no I' signal,
RESULTANT
SUBCARRIER
SUBCARRIER
d SUBCARRIER
33°
o° -o°
(a) (b)
Fig. 4.9. Resultant of two sub -carriers

RESULTANT
O' SUBCARRIER
SUBCARRIERS
+
REDDISH
MAGENTA
(a) o°
+1'
ORANGE
o° -o°
(b)
-o° -o°
BLUISH
GREEN
(C)
Fig. 4.10. Equivalent sub -carrier representations
the sub -carrier at /' phase has zero amplitude, and if Q' = 0 as
well then both sub -carriers disappear. In this way the D.C. value
of the I' signal is represented by the mean of the peak amplitude of
the sub -carrier at I' phase and the modulation sidebands of this sub -
carrier represent the variations of I' about this D.C. level. Negative
values of I' correspond to a reversal in direction of the phasor or a
phase change of 180° in the I' sub -carrier.
This method of operation
in which the carrier has no standing value when the modulation is
zero, unlike the modulation of the sound carrier, is sometimes called
suppressed carrier working.
4.18. Representation by a single sub -carrier.
Since the I' and Q' sub -carriers have the same frequency, they can
be represented by a single resultant phasor of the same frequency,
as in Fig. 4.9. Variations in the amplitude of the I' and Q' com-
ponents make the resultant sub -carrier vary in both amplitude and
phase (Fig. 4.10 gives examples of reddish magenta, orange and
bluish -green), and this type of modulation can be considered in two
ways. A single sub -carrier can be imagined as conveying amplitude
modulation proportional to V(//)2 + (Q')2 and phase modulation
proportional to tan -1
/' Trigonometrically the resultant sub -carrier
OC can be written as the sum of a Q' signal represented by
Q' sin (cot + 33°),
and the I' signal represented by
I' sin (tot + 33° + 90°) or I' cos (cot + 33°)
i.e. OC = I' cos (cot + 33°) + Q' sin (cot + 33°)
= V(r)2 + (Q')2[vv,)2
r cos (cot + 33°)
+ v, 02 +Q' sm vut + 330)1

sin (wt
=V(1')2 + (Q')2 [sin 0 cos (cot + 33°) + cos 0 sin (wt + 33°)]
=V(r)2 + (Q')2 sin (cot + 33° + 0)
I'
and tan 0 =
Alternatively, the signal can be considered as composed of two
sub -carriers each of which carries only amplitude modulation, and
these two sub -carriers are 90° apart in phase.
4.19. Zero phase reference
As the phase of the sub -carrier is important, it is necessary to
standardize on a reference phase. The phase of the positive
direction of the (B' - Y') axis is called zero phase, and angles are
measured anticlockwise from this phase, so that the positive
1' axis is at phase 123° (see Fig. 4.8). The (B' - Y') and
2.03(B' - Y') axes lie along the same direction.
4.20. Video signal

The sub -carrier signal is added to the luminance signal and the
mixed sync signal so that the complete video signal is a combination
of all three, as is shown in Fig. 4.1. The sub -carrier signal thus
rides up and down on the luminance signal. Viewed on an oscil-
loscope synchronized to the line scanning frequency the sub -carrier,
where it is present, appears as a blur, which is represented in Fig. 4.1
by the rectangular outline of the chrominance waveform. The
sub -carrier is a high video frequency and even if the colour it
represents is constant from line to line, the phase of the sub -carrier
relative to the line sync pulse reverses on every line (see Section
LUMINANCE
LEVEL
-),..
LINE n
(a)
LINE n+2
(b)
SUPERIMPOSED
WAVEFORMS
(c)
Fig. 4.11. Video waveforms for vertical

colour bar
4.32) so that only the envelope of the waveform is clearly visible

unless a high speed scan is used to pick out the chrominance, in
which case the detailed picture of the waveforms is as represented
in Fig. 4.11.
8
4.21. Bandwidths of the chrominance signals
To reduce the visibility of the interference caused by putting the
sub -carrier in the frequency band which is normally occupied only
by luminance information (see Fig. 4.2), the sub -carrier frequency is
chosen to be a high luminance frequency. This necessarily implies
that the sideband frequencies of the sub -carrier are limited on the
sound carrier side. Thus for a sub -carrier frequency of 2.7 Mc/s
LUMINANCE Y
W 034
0 MC/S
1:71 Fig. 4.12. N.T.S.C. bandwidths.
a0 Dotted line is upper limit
of transmitter tolerance on I
bandwidth
I 2
FREQUENCY (MC/S)
Table 4.3
N.T.S.C. BANDWIDTHS
Q bandwidth I bandwidth
525 and 405 525 and 405
625 lines lines 625 lines lines
400 kc/s 300 kc/s <-2dB 1.3 Mc/s ; 1 Mc/s < -2dB
500 kc/s 340 kc/s <-6dB 3.6 Mc/s 2.5 Mc/s ?- -20dB
600 kc/s 450 kc/s > -6dB
on the 405 -line system, the sub -carrier sidebands can only extend
up to about 0.3 Mc/s in the high frequency direction, i.e., 3.0 to
2.7 Mc/s.
If the sub -carrier is to be modulated double sideband, this limits
the rate at which the chrominance information can be changed to
0.3 Mc/s for both colour difference signals. This means that there
can be no sharp vertical colour boundaries and the sharpest change
from an orange colour to a cyan colour of the same luminance will
spread along about in. on a horizontal line of a 21 in. receiver.
On most pictures, the colour changes will occur in conjunction with
luminance changes and the sharp edge of the luminance change will
camouflage the unsharp chromaticity change (see Section 1.14).
Nevertheless the wider the bandwidth which can be allotted to the
chrominance signal, the sharper the final picture will appear. This
bandwidth restriction does not affect the sharpness of colour changes

in a vertical direction.
On the low frequency side of the sub -carrier there is room for a
much wider modulation bandwidth but this would mean using
asymmetric sideband modulation as is used for the main vision
carrier. Such unbalanced transmission produces an unwanted
signal which has a phase at right angles to the wanted signal, that
is to say, quadrature crosstalk occurs. This is discussed further in
Chapter 8. The result is that an asymmetric Q' signal would produce
a spurious signal at right angles in the I' phase.
To avoid this crosstalk the Q' signal is sent double sideband. The
I' signal is sent asymmetric (see Fig. 4.12) but is double sideband
over the whole frequency band of the Q' signals and does not there-
fore crosstalk into the Q' signal over this frequency band. Outside
this frequency band the I' signal is asymmetric but the quadrature
crosstalk components due to this part of the signal fall outside the Q'
pass -band and are rejected by the Q' channel. In this way the two
signals can be received without mutual interference.
The transmitter response specifications are
1' channel Q' channel
at 1.0 Mc/s less than 2dB down at 300 kc/s
at 2.5 Mc/s at least 20dB down at 340 kc/s less than 6dB down
at 450 kc/s at least 6dB down
The system characteristics so far discussed are common to the
American N.T.S.C. and the versions adapted (A.N.T.S.C.) to both
405 and 625 lines. The bandwidths, however, vary from one
system to another. Thus, for 525 lines
I' channel Q' channel
at 1.3 Mc/s less than 2dB down at 400 kc/s less than 2dB down
at 3.6 Mc/s at least 20dB down at 500 kc/s less than 6dB down
at 600 kc/s at least 6dB down
For 625 lines the suggested chrominance bandwidths are the same
as for the American system. If an 8 Mc/s channel is available for
each station, as in the U.S.S.R., the optimum utilization of the
extra megacycle of bandwidth is probably to increase the luminance
pass -band by 0.5 megacycle and use the other 0.5 megacycle to
increase the vestigial sideband of the vision carrier to 1.25 Mc/s.
The latter reduces the effects of vestigial sideband distortion,
particularly on negative modulation. If the sub -carrier frequency
is left unaltered, the extra luminance bandwidth can be used to
increase the chrominance bandwidth and to ease the problem of
providing a sharp luminance and chrominance cut-off at the sound
channel frequency.
4.22. Time matching of signals
Since the bandwidths of the Y', I' and Q' signals are different, a
sharp transition in the picture will give rise in general to a sharp
transition in the Y' signal, a slow transition in the I' channel, and a
PICTURE TRANSITION
Y' SIGNAL
f = 3MC/S
1. SIGNAL
=Iticls
Q SIGNAL
1=0.3MCIS
DELAYED S' SIGNAL
DELAYED I SIGNAL
Fig. 4.13. Time matching of signals. Build-up time,

10-90%, 112f. Transmitter tolerance is ± 0.07µs
(405 lines) and ±0.05µs (525 lines)
still slower change in the Q' channel, as in Fig. 4.13. The fastest
possible rate of rise, measured from 10 % to 90% of the amplitude
1
change, is given by T
2 f wheree f is the bandwidth of the channel
concerned. The time delay of each signal depends upon the band-
width of the channel through which it is passing, and the attenuation
slope at the cut-off frequency.
For these three transitions to appear to be coincident in time, the
centres of the transitions should coincide. This is achieved by
delaying the I' and Y' signals with respect to the Q' signal. The
BBC transmitter specification for this time matching is that the
signals as transmitted shall be coincident in time; but see Section 6.3.
4.23. The final chrominance signal
The chrominance signal as it is finally transmitted consists of two
components, I' and Q', amplitude modulated onto two sub -carriers
which are 90° apart in phase and can therefore be represented as in
Fig. 4.8.
The amplitudes of the I' and Q' signals are calculated from the
reduced values of the colour difference signals
- 1") (B' - Y')

1 .14 and 2 .03
according to the Equations 4.8 and 4.9, which are
and
I'03
Q' -
(K - 1")
1 .14
- (R'1.14
- Y')
sin 33° +
cos 33°
since cos 33° = 0.839 and sin 33° = 0.545

(B' - Y')
2.03
(B' - Y')
I' = 0.736(R' - Y') - 0.268(B' - Y')

2.
cos 33°
sin 33°
[4.8]
[4.9]
[4.10]
Q' = 0.478(R' - Y') 0.413(B' - Y') [4.11]
The combined resultant of these two sub -carriers, (Section 4.18)

/'
has an amplitude \/(I')2 (Q')2 and makes an angle 33° +
(B' - Y') (B' - Y')
with respect to the axis. The axis lies along the
2 03
. 2.03
old (B' - Y') axis but the scale is different.
Table 4.1 gives sample values of (/')2 (Q')2 and the phase angle
tan-i
/' 33° = ¢ for various colours. The reader should work
through some of these examples.
A sub -carrier resultant of specified amplitude OC and phase 4, can
be resolved into its components along any chosen pair of axes, as
(R'1-.141r') (B'2-.03Y')
well as along the I', Q' or and axes, provided
the modulation frequencies being considered fall inside the region
where I' and Q' are both double sideband (see Chapter 8).
4.24. The combined luminance and chrominance signal
The preceding sections of this chapter have laid the basis for
describing the final formulation of the N.T.S.C. video signal.
The luminance signal is first formed from the red, green and blue
camera outputs, after the camera amplifier gains have been
normalized on studio white and the signals have been gamma
corrected
Y' = 0.299R' + 0.587G' + 0.114B'
where the luminosity coefficients have been given to 3 decimal places.
The colour difference signals (R' - Y') and (B' - Y') are next
(R' - Y') B Y')
formed and these are reduced in amplitude to and
1.14 2 .03
to limit the peak excursions of the transmitted signal.
The reduced colour difference signals are now converted to their
I' and Q' equivalent signals by resolving them along the I' and Q'
axes, that is to say, I' and Q' are formed thus, from Equations 4.10
and 4.11
I' = 0.74(R' - Y') - 0.27(B' - Y')
Q' = 0.48(R' - Y') + 0.41(B' - Y')
Up to this point the signals have been wideband but the frequency
bandwidth of the Q' sub -carrier is now limited. The 1' signal is
also bandwidth limited but not so severely as the Q' signal, and
is time delayed.
The Q' signal is next amplitude modulated on to a sub -carrier,
which can be represented as
Q' sin (ad + 33°)
where (.0 is the angular frequency of the sub -carrier and equals
2irf where f .-' 2.7 Mc/s for the 405 -line system, t represents time
in seconds, and 33° represents the phase of the sine wave with
respect to the zero phase of the (B' - Y') axis.
The I' signal is amplitude modulated on to a sub -carrier which has
the same frequency as the Q' signal, but is shifted in phase by 90°
and thus can be represented by a cosine function I' cos (wt ± 33°).
During the final modulation on to the main vision carrier the I'
sub -carrier bandwidth is made asymmetric as in Fig. 4.12, by the
conventional frequency characteristic of the main modulator.
Both these signals are added to the delayed luminance signal to
form the combined video signal M'
M' = Y' Q' sin (wt + 33°) I' cos (wt + 33°)
This is sometimes written more formally to emphasize that it is
voltages which are being considered.
E'M = E'y k {E'Q sin (wt + 33°) + E'1 cos (cut 33°)}
where k is normally unity, but can vary between 1 and 0.3 if it is
desired to make the signal more compatible by decreasing the
amplitude of the sub -carrier. Decreasing k decreases the signal-to-
noise ratio of the chrominance signal at the receiver.
For the slow modulation changes which correspond to frequencies
where I' and Q' are both double sideband, there is no need to refer
to the I' and Q' axes and the signal can be written in terms of
(R' - Y') and (B' - Y')
Em = E'y
E' - E' y
sm wt
R - 'y
cos
2.03 1.14E
To help in transforming from R', G' and B' signals to I' and Q'
signals, etc., the following identities can all be derived from the
equations already given.
4.25. Camera signals in terms of transmission signals
R' = Y' 0.956/' 0.621Q'
= Y' - 0.2721' - 0.647Q'
B' = Y' - 1.106/' + 1-703Q'
R' = Y' 114S'R
= Y' - 0.581S' R - 0.394S'B
B' = Y' 2.03S' B
where
R' - Y'
Sin - 1.14
and
B' - Y'
S" B -
2.03
These equations can be neatly written in matrix form, thus
[R' 1 0.956 0.621 Y'
G' = 1 - 0.272 - 0.647

B' 1 - 1.106 1.703 Q'
and the circuits which perform such calculations are called matrix
circuits.
4.26. Transmission signals in terms of camera signals
Y' = 0.299R' + 0.587G' + 0.114B'
I' = 0.596R' - 0.275G' - 0.322B'
Q' =0.211R' - 0.523G' + 0.312B'
4.27. Chrominance transmission signals in terms of colour difference
signals
I' = 0.736(R' - Y') - 0.268(B' - Y')
Q' = 0.478(R' - Y') 0.413(B' - Y')
0 = 0.299(R' - Y') 0.587(G' - Y') 0.114(B' - Y')
(R' - Y') = - 1.963(G' - Y') - 0.381(B' - Y')
(G' - Y') = - 0.509(R' - Y') - 0.194(B' - Y')
(B' - Y')=- 2.623(R' - Y') - 5.144(G' - Y')
4.28. Colour difference signals in terms of I' and Q'
(R' - Y') = 0.9561' 0.621Q'
(G' - Y') = - 0.2721' + 0.647Q'
(B' - Y') = - 1.106/' 1.703Q'
4.29. Colour difference signals in terms of camera signals
(R' - Y') = 0.701R' - 0.587G' - 0.114B'
(G' - Y') = - 0.299R' + 0.413G' - 0.114B'
(B' - Y')=- 0.299R' - 0.587G' + 0.886B'
(R1 - Y')
S'R = 1.14 = 0.615R' -0.515G' -0.100B'
5"13 =
(8 '
2.03Y
')
- 0 147R' - 0.289G' + 0.436B'
4.30. N.T.S.C. tristimulus values in terms of C.I.E. values
The camera outputs, R', G', B' are sufficient to determine the
C.I.E. co-ordinates of the transmitted colour and vice versa, and
these relations are, (see Section 2.25)
R = 1.91(X) - 0.532( Y) - 0.288(Z)

G = - 0.985(X) 1.999( Y) - 0.028(Z)
B = 0.058(X) - 0.118( Y) 0.898(Z)
for a reference white of 0.981(X){ -(Y)+ 1.182(Z) or unit lumi-

nance of illuminant C.
(X) = 0.607R + 0.174G + 0.200B
(X) = 0.299R + 0.587G + 0.114B
(Z) = 0.000R + 0.066G + 1.116B
Notice that these equations refer to the camera signals before
they have been gamma corrected.
In all of the above equations the numerical constants have been
given to three decimal places in order that the reader, when calcula-
ting signal values, might be able to rely on the second decimal
place in the final answer. If only two decimal places are used in
the calculations, the answer may differ significantly from published
values. For most engineering purposes an accuracy of two decimal
places is sufficient.
It is only permissible to use colour difference signals such as
(R' - Y') at the receiver for modulation frequencies which lie
within the Q bandwidth, i.e., for large areas of colour.
4.31. The colour burst signal
In order that the receiver shall be able to decode the colour
difference information it must be able to recognize the phase of the
sub -carrier signal and must therefore have a phase reference with
which to compare it. Only then can the receiver tell I' phase from
Q' phase, or, to look at it another way, recognize the amount of
phase modulation on the sub -carrier.
The transmitter provides a reference phase by generating a con-
tinuous sine wave of sub -carrier frequency at a phase standardized
as 180°, that is to say, it is opposite in phase to the positive direction
of the (B' - Y') axis. In Chapter 5 it is shown that of the four
convenient phases of ± (R' - Y') and + (B' - Y') the reference
signal has the least visibility on the receiver display for the
- (B' - Y') phase.
A section of this continuous sine wave (Fig. 4.14) is transmitted
on the back porch of the line blanking period by gating out 9 cycles
of the reference signal every line (Fig. 4.15). The phase of this
CONTINUOUS REFERENCE SINE WAVE
N1 111NA INN MN NNN [v ivy A

THIS INTERVAL
NOT TO SCALE
(b)
I LINE NMI N
BURST SIGNAL GATED OUT
CHROMINANCE
SIGNAL PHASE
(c)
-IMAM YELLOW
(d)
qui[vy.[I' BLUE
(e) GREEN
Fig. 4.14. Formation of colour burst signal
burst of reference signal is constant with respect to time but due to

the frequency relationship between sub -carrier and line frequencies
the phase of the burst with respect to the line sync pulse reverses on
successively transmitted lines, that is to say between lines 1 and 3,
or 6 and 8, etc.
The peak -to -peak amplitude of this colour burst signal is the
same as the height of the sync pulse signal, i.e. 0.3 of the peak white
transmitted signal.
4.32. Frequency relationships
The approximate choice of frequency for the sub -carrier is deter-
mined by the bandwidth required for the Q' channel. On mono-
chrome receivers the sub -carrier appears as an interfering C.W. signal
and produces a beat pattern of bright and dark dots along the scan-
ning lines, the dots falling into the familiar pattern of interference
stripes across the picture. The higher the sub -carrier frequency,
the finer and less obvious the texture of the beat pattern becomes.
If the sub -carrier frequency is too close to the upper frequency
limit of the luminance channel there is insufficient room for the
upper sidebands of the Q' sub -carrier (Fig. 4.12). To further reduce
the visibility of the dot pattern, the exact sub -carrier frequency in the
required approximate range is chosen so that the dots interlace;
this happens if the sub -carrier frequency fs, is any odd multiple of
half the line scanning frequency fr,
fac== (2n -p 0E11
where n is any whole number. The sub -carrier peaks, corresponding

to white dots, are then displaced on successive scanning lines since
there is always half a cycle of sub -carrier left over at the end of each
line. On any one field, therefore, the dots are neatly staggered.
On the next field the dots are staggered with respect to the dots
on the next upper line of the previous field, since there are, for
example, 2021 lines in a field so that the first half line of the second
field is really a continuation of the last half line of the first field, the
203rd. This is illustrated for a simple system with very quick
flybacks in Fig. 4.16.
The succeeding 3rd field rescans the odd lines again and the sub-
carrier peaks are now displaced relative to the dots on the first
PEAK
16 5 t0.5/es
WHITE
1.5 ± 0.10
9 ±Ips
00
o U BLACK LEVEL
LIFT
..---"---COLOUR BURST
9 ± I C/S OF REFERENCE
FREQUENCY
SYNC. LEVEL
Fig. 4.15. The colour burst reference signal. The burst is not transmitted
during field sync pulses
(b)
2 2
5
4 4
5 5
6 6
7 7
8 8
9 9
10 10
(C)
(I)
Fig. 4.16. Dot interlacing. (a)-picture I, field 1; (b)-picture 2,
field 3; (c)-picture 1, fields 1 and 2; (d)-picture 2, fields 3 and 4
field. Similarly on the fourth field the dots fall between those
illuminated on the even lines of the second field.
The net result is that over two complete pictures, i.e., - s,
12 .5
1
the peaks of the sub -carrier on one picture fall into the troughs of the
sub -carrier on the next picture. This behaviour arises because the
line frequency is an odd multiple of the picture frequency, fp
fL = 405fp
so that the sub -carrier frequency is an odd multiple of half the picture
frequency as well as of half the line frequency,
fsc = 4(2n 1)fr,

f(2n 1)405fp
and as 405 and (2n + 1) are both odd, their product is odd.
Although the eye does not store the image very well over this
relatively long period, the net subjective effect is a marked decrease
in the annoyance value of the beat pattern, equivalent to a decrease
in sub -carrier amplitude of about 20dB. This principle is used in
monochrome television to reduce the interference between stations
sharing the same channel.
525cir-
For 405 lines a convenient odd multiple of half the line frequency,
which has low value factors to make the transmitter dividing
cuits simple, is -y-fL, (525 = 7 x 5 x 5 x 3).
4.33. Asynchronous operation

To ease the design of the reference circuits in the receiver it is
usual to unlock the field frequency from the mains and to make fF
exactly 50 c/s however the mains frequency is varying, so that fr, is
then exactly
405
-2 fF = 10,125 c/s
and the sub -carrier frequency is constant at

525
-2 fr, = 2.6578125 Mc/s
4.34. Vision -sound carrier spacing

To reduce the visibility of any beat patterns produced between
the sub -carrier and the sound carrier, this frequency interval is also
made an odd multiple of half the line frequency and the vision
carrier -sound carrier spacing becomes a multiple of fL
f vision =f sound ± 350fL
4
-f sound + ijsc
This means that the vision -sound carrier spacing cannot be
maintained at exactly 3.5 Mc/s, but must now be 3.54375 Mc/s.
In practice the receiver tuning of the sound carrier is more critical
than the vision carrier and for the benefit of the monochrome
receivers already in use the sound carrier frequency of the station
is left unaltered and the vision carrier is increased slightly in fre-
quency.
In the 525 -line system, although F.M. sound is used and the
vision carrier -sound carrier spacing cannot therefore be locked,
nevertheless the visibility of the beat between the sub -carrier and
the sound carrier is reduced, on average, by making the vision
carrier-undeviated sound carrier spacing a multiple of the line
frequency. The nominal line frequency is 15.75 kc/s and nearest
harmonic to 4.5 Mc/s is the 286th at 4.5045 Mc/s. However, the
change to 4.5045 Mc/s might upset the intercarrier operation of
existing monochrome receivers, so the carrier spacing is kept at
exactly 4.5 Mc/s by reducing the line frequency from 15.75 kc/s
to 15.73426 kc/s.
That is
Line frequency fr, = 4.5286106 - 15,734.26 c/s
(monochrome = 15,750 c/s)
Field frequency fF = 525= 59.94 c/s

(monochrome = 60 c/s)
As far as the sub -carrier frequency is concerned, a Q' bandwidth
of 0.5 Mc/s must be accommodated in the video band of 4.0 Mc/s,
so that the sub -carrier frequency must be near 3.5 Mc/s. A con-
venient multiple is 455, so that
455
sub -carrier freuencyL
q c= -2 fr, = 3,579,545 c/s
where 455 = 13 x 7 x 5
Similarly, in the 625 -line system the 5.5 Mc/s vision carrier -
sound carrier spacing is made a multiple of the line frequency. In
this case the 352nd harmonic of the nominal line frequency of
15.625 kc/s is exactly 5.5 Mc/s, so that the monochrome scanning
frequencies are acceptable.
5.5
Line frequency fr, = 35x 2106 - 15,625 c/s
(monochrome = 15,625 c/s)
Field frequency fF = -2 ft, = 50 c/s

625
(monochrome = 50 c/s)
For a Q' bandwidth of 0.6 Mc/s and a video band of 5 Mc/s,
a sub -carrier near 4.4 Mc/s is indicated. A convenient multiple
is 567, so that
567
sub -carrier frequencyfm= fr, = 4,429,687 c/s
567 = 7x3x3x3x3
The same sub -carrier frequency will probably be used in the
Russian 625 -line system where the vision -sound carrier spacing is
6.5 Mc/s.
4.35. Frequency interleaving and the energy spectrum
In Chapter 1 it was pointed out that the periodic nature of the
luminance signal gives a discrete line structure to the energy spec-
trum of the television video signal (Fig. 1.10). For similar reasons
the sub -carrier signal and its sidebands also exhibit a discrete line
structure and due to the odd multiple of half line frequency relation-
ship adopted, the clumps of energy in the sub -carrier spectrum fall
neatly between the clumps in the luminance signal spectrum (Fig.
4.17). The two signals can therefore coexist in the same frequency
CR CIMIN ANC E
SPECTRUM
LUMINANCE
CLUMPS
SPECTRUM
CLUMPS
(b)
.--
iokc/s
LUMINANCE
C ROMIN A NC E
SPECTRUM
ENVELOPES
FREQUENCY (M C/S)
(a)
Fig. 4.17. (a) is the frequency spectrum for a typical 405 -

line signal. (b) shows an expansion of the frequency scale
band as separate identities. Despite this it is not practical to extract

one from the other as filters with the necessary comb -teeth properties
of changing from pass to reject every 5 kc/s, and ideally every 25
c/s, are not yet manufactured. However, the subjective appearance
of the composite signal on a monochrome picture is improved by
this interleaving technique.
4.36. Delay characteristics
The design of the colour receiver I.F. circuits is rather more
difficult than is the case for monochrome, since the amplitude -
frequency response must be maintained out to well past the sub -
carrier frequency. At the same time, the sound rejection required
is at least as much as for monochrome reception and this implies a
sharp onset of attenuation at the end of the luminance band. With
the minimum -phase type of circuits used in monochrome reception,
such an abrupt change in the amplitude response is reflected in a
lack of linearity in the accompanying phase characteristic. In the
colour receivers so far produced, including those using non -minimum
phase sound rejector circuits, it has not been economic to combine
05 , + 0.5
AVERAGE COLOUR TRANSMITTER CORRECTION
RECE
o 0
0
0.1
1.1.1-/
X 1.1 0.5
-0.5 I5 25 5
.5 I I IS 2 25 3 .5 I 2
FREQUENCY (MC/5) FREQUENCY (M C/S)

(a) (b)
Fig. 4.18. Envelope delay characteristics
an ideal flat amplitude response with an ideal linear phase response.

The phase responses of several types of colour receiver were
measured and computed and an average delay curve drawn. The
transmitter applies pre -correction to balance out this average phase
error in colour receivers, and Fig. 4.18 shows these two envelope
delay characteristics. The transmitter correction is only for fre-
quencies above 11 Mc/s.
4.37. Transmitter tolerances
The transmitter line and field frequencies are counted down from
the sub -carrier frequency, which has a tolerance of ± 8 c/s and a
maximum rate of change of 0.1 c/s2. The burst is 9 cycles of sub -
carrier ± 1 cycle, and its position is defined to within ± 0.1
relative to the line pulse. The tolerances on the sub -carrier signals
are approximately ± 2dB in amplitude (see Chapter 5) and ± 10°
in phase for the saturated primary colours at 75% of peak signal.
The sub -carrier phase accuracy is normally better than this for flesh
tones. The Y', I' and Q' signals should match each other in time
to 0.07 ps. The peak white signal of the transmitter is usually
reduced to 90 % of its monochrome value to allow partially for the
sub -carrier excursions. In other respects the colour transmission
conforms to standard monochrome tolerances.
4.38. The colour test stripe signal
For some time after the inauguration of a colour television service
it is expected that the majority of the programmes will continue to
(Above) luminance and chrominance components displayed
separately; (below) complete lack of l' signal on left of picture
and complete lack of signal on right of picture
The top photograph shows the Marconi next channel direct switching vision
mixer, specially designed to handle colour television signals. To the right of
the mixer, is a special effects unit for producing horizontal and vertical wipes
and corner inserts. The lower picture shows the camera control compartment
of the Marconi two camera channel Mobile Colour Television Unit.
(Courtesy Marconi Ltd.)
be in monochrome. To enable the colour performance of any
particular receiver installation to be assessed whilst monochrome
programmes are being radiated, it is possible for stations to transmit
a special signal. The signal which has been used in America consists
of a burst of sub -carrier frequency at the beginning and end of each
line (Fig. 4.19). The two bursts have the same phase but their fre-
quency does not need to be locked to the line scanning frequency.
On a monochrome receiver these bursts produce two narrow verti-
cal stripes of beat pattern at the extreme edges of the picture which
are often hidden by the picture mask.
On a colour receiver with an automatic colour killer (see Section
7.7) the stripes will be hardly visible since the luminance channel
is designed to reject sub -carrier frequency. To make the stripes
visible, the burst gate of the receiver must be delayed until it accepts
the nearest stripe: this will switch on the chrominance channel and
pull the reference oscillator into phase lock. The second stripe on
the tube face should then appear as a green band along the edge of
the monochrome picture. If the aerial or receiver has inadequate
gain at sub -carrier frequency, the green stripe will not appear. If
BURSTS OF SUBCARRIER
FREQUENCY
Fig. 4.19. Colour test

stripe signal. Test bursts II
have the same amplitude
and duration as colour
burst, but need not be
locked to line frequency 1
VIDEO VIDEO
SIGNAL SIGNAL
LINE SYNC. PULSE
the hue control is incorrectly set, the stripe will have the wrong
colour. The burst gate of the receiver may be delayed either by
adjusting the line hold control or by adding capacity to the sync
separator anode (see Section 10.4).
4.39 Summary
The colour television camera analyses the picture into three
voltages R', G' and B', which represent the amounts of red, green
and blue lights which the receiver must generate to produce an
9
impression of the colour in the studio scene. A luminance video
signal is formed by combining fractions of the primary voltages,
Y' = 0.30R' + 0.59G' + 0.11B'
The rest of the colour information is transmitted by sending band-
width limited I' and Q' signals on a sub -carrier placed in the high
frequency part of the video signal. The I' and Q' signals are derived
from the colour difference signals (R' - Y') and (B' - Y') thus
I' = 0.74(R' - Y') - 0.27(B' - Y')
Q' = 0.48(R' - Y') 0.41(B' - Y')
The resulting colour sub -carrier vanishes for grey tones, I' =Q' =0,
and its amplitude increases with both the purity and the luminance
of the colour being transmitted. The phase of the sub -carrier signal
with respect to the colour burst signal describes approximately the
hue of the transmitted colour.
The luminance signal produces an acceptable black and white
picture on normal monochrome receivers, the sub -carrier producing
an interference pattern which is not obvious; the field frequency is
not locked to the 50 c/s mains.
CHAPTER 5
Colour Specification in
N.T.S.C. Systems
5.1. Introduction
In Chapter 4 the transmitted signal has been discussed in electrical
terms using as units the voltages produced by the cameras on peak
white. Different colours produce different electrical signals and
it is possible to plot the chromaticities of colours on a C.I.E. diagram
(see Section 2.25), and to indicate on the same diagram the corres-
ponding signal voltages. Such diagrams can indicate how the
received picture will appear to the eye as the signal voltages change.
For convenience in measuring and recording colours the C.I.E.
system does not use the red, green and blue primaries but its own
set of theoretical primaries called X*, Y* and Z*. Similarly, the
television transmitter does not use amounts of the red, green and
blue primaries to describe the colour being transmitted; instead it
radiates the amounts of Y', I' and Q', and the behaviour of the
system can be described in terms of these quantities.
It is unfortunate that the letter Y has been adopted for both the
C.I.E. fictitious green primary and for the N.T.S.C. luminance
signal and its corresponding white primary. Since this practice
is widespread, the letters have not been changed since the reader
can easily distinguish which is meant in any particular context.
Although both Y's have the same numerical value, the chromaticity
of the C.I.E. Y* is x = 0 and y = 1, while the chromaticity of the
N.T.S.C. Y' is x = 0.310 and y 0.316.
5.2. System white and primary colours

The choice of white for the colour television system theoretically
affects the size of the units in which the red, green and blue com-
ponents of the picture are measured. The usual choice is C.I.E.
Illuminant C (see Chapters 2 and 4). If a warmer white such as
Illuminant A is chosen, more red light is needed to produce the
system white and the size of the red unit is larger. For practical
117
reasons, e.g., tungsten lighting, the reference or normalizing white
used at the studio is not usually Illuminant C, but the colour receiver
should always be adjusted to reproduce Illuminant C when the
chrominance signals are zero.
In discussing the colour performance of the N.T.S.C. system, the
electrical signals will be related to the chromaticity of the colour
reproduced at the receiver, on the assumption that the receiver is
using the N.T.S.C. red, green and blue primaries (Fig. 4.5) and that
Illuminant C is the reference white.
5.3. Y' luminance signal
The luminance signal is formed after the camera signals have been
gamma corrected (see Section 4.14), thus
1 1 1
Y' = 0.30Rv 0.59GY 0.11/3v

or
Y' = 0.30R' + 0.59G' + 0.11B'
where 0.30 is the luminosity coefficient of the N.T.S.C. red primary,
etc. If different red, green and blue primaries are chosen for the
system, the proportions of R', G' and B' used for the luminance
signal will also be different.
Because of the need for gamma correction, parts of the system
are not linear; a change of 0.1 in a signal such as the R' signal
produces a bigger absolute increase in the light output from the
receiver tube if it is a change from 0.8 to 0.9, than if it is a change
from 0.1 to 0-2.
Ideally, the luminance signal should be formed before gamma
correction is carried out, while the camera signals are directly pro-
portional to the light intensity of the scene
Y = 0-30R ± 0.59G + 0.11B
where R, G and B, without primes, indicate voltages which have
not been gamma corrected. The gamma corrected luminance
signal would then be
Yv = (0.30R + 0.59G 0.11B)Y
which is not the same as the luminance signal actually used
Y' = 0.3ORY + 0 59GY + 0.11Bv

COLOUR SPECIFICATION IN N.T.S.C. SYSTEMS 119
unless R = G = B when the picture is a black and white colourless
scene. For a saturated blue colour
YY = (0.11 x 1)2.2 = 0.37

and
Y" = 0'11(1\22
) = 0'11
so that the Y' signal is smaller than it should be. This result is
true in general for all the saturated and near saturated colours
-the luminance signal is too small. For greys the luminance
signal is correct and for desaturated colours the error is small.
The luminance reproduced by the signal actually used is
( Y')Y = (0.30R' ± 0.59G' + 0.11 B')Y
and for the saturated primaries and their complementaries at full
brightness, i.e., R = 1 or 0, G = 1 or 0 and B = 1 or O. R = R',
G = G', B = B' and (Y')z' ( Y)Y which is not Y, of course, and
is therefore incorrect.
The luminance reproduced by the theoretically correct signal is
(p)y = y
If the reproducing cathode-ray tube has a gamma higher than 2.2,
as may happen for monochrome receivers, then the reproduced
luminance is actually
(y)y2 ==.(y)v2
for the saturated primaries at full amplitude, and for the second
signal would be
1 Y2
( ( Y)Yi
where yi = 2.2 and y2 may be 2.7.

Values for these signals are given in Table 5.1 for the saturated
primaries and their complementaries at maximum brightness, and
at three-quarters of their maximum brightness.
Since the transmitter generates in the first place the primary
R', G' and B' voltages, and these are recovered at the receiver,
the total luminance produced by the colour cathode-ray tube is
correct. It will be the sum of the luminances produced by (R')'Y
Table 5.1
LUMINANCE CARRIED BY MONOCHROME COMPATIBLE SIGNAL
Theoretic- Luminance reproduced Luminance reproduced

Correct Transmitted ally correct on tube with y = 2.2, on tube with y = 2.7, Correct
Colour luminance signal signal i.e. colour tube i.e. monochrome tube luminance
Y Y' I. Y
Yv 1 1
( re 2 ( yyr 2 (n2.7 ( YY)"
White 1 1 1 1 1 1 1 1
'
Yellow 0.886 0.886 0.946 0.77 0.89 0.72 0.86 0.89

io Cyan 0.701 0.701 0.851 0.46 0.70 0.38 0.65 0.70
"5
u 0.59
Green 0.587 0.587 0.785 0.31 0.59 0.24 0.52
Magenta 0.413 0.413 0.670 0.14 0.41 0.09 0.34 0.41

i : t . ,I
Red 0.299 0.299 0.578 0.07 0.30 0.04 0.23 0.30

0''
o,,
Blue 0.114 0.114 0.373 0.01 0.11 0.003 0.07 0.11
Yellow 0.664 0.777 0.830 0.57 0.66 0.51 0.61 0.66

r.
g Cyan 0.526 0.615 0.747 0.34 0.53 0.27 0.45 0.53
"O
u 0.440 0.515 0.689 0.23 0.44 0.17 0.37 0.44
,, Green
E.':
]
N 0.310 0.362 0.587 0.11 0.31 0.06 0.24 0.31

'
Magenta
t14
0.224 0.262 0.507 0.05 0.22 0.03 0.16 0.22

o Red
.-1 Blue 0.086 0.100 0.327 0.01 0.09 0.002 0.05 0.09
of red light, (G')y of green light and (B')" of blue light, the input-
output characteristic of the cathode-ray tube effectively raising the
signals to the power y, or 2.2. Since R' RY, the total luminance
is the sum of the luminance produced by R, G and B,
i.e.
1R mG nB = 0.30R + 0.59G + 0.11B
where 1, m and n are the luminosity coefficients of R, G and B.
This total luminance is proportional to the brightness of the
original scene.
As the final luminance on the colour receiver tube is correct and
the transmitted Y' signal is too small, the rest of the luminance
must be carried by the chrominance signals, such as (R' - Y')
and (B' - Y') or and Q'. The constant luminance principle
is thus only approximately followed for strongly saturated colours
(see Sections 4.12 and 4.14). Since the Y' signal is not conveying
all the luminance it would perhaps be better to call it the mono-
chrome signal, but common usage is to refer to it as the luminance
signal.
If the transmitter radiated the true luminance signal
1 1
YY (0.30R ± 0.59G + 0.11B)Y

the receiver would need complicated circuits to derive from YY,
1 1
(R' - YY) and (B' - YY) the required G' value. By using Y'
the receiver needs only simple linear circuits which add and subtract
fractions of the transmitted signals.
5.4. Choice of burst phase
Since the chrominance signal carries some of the luminance
information, the colour reference burst may produce a coloured
vertical bar on the display tube if the line blanking is not adequate.
The visibility of this bar is minimized by the choice of phase of the
burst as -180°. Thus if the burst amplitude is S and it has a
phase 0 with respect to the
(B '2.03Y)
- axis, then it will have a com-
ponent along this axis of S cos 0 which the receiver will interpret
as a signal
(B' - Y') = 2.03S cos 0
Similarly, the burst will produce a red difference signal of
(R' - Y') = 1.14S sin
These two signals correspond to a green difference signal of
- 0.30 0.11
(G' - Y') = (R - Y') -0.59(B - Y')
see Equation 4.5
- 0.30
0.590 59(1.14 S sin 0) -
0.11
(2.03 S cos 0)
For the colour burst the transmitted Y' signal is zero, so that the
red, green, and blue signals to the tube are
R' = 1.14S sin 0
0. 0-11
G' 0.590 (1.14 S sin 0) -^ S cos 0)
o0.59(2.03
B' = 2.03S cos
and these produce a total luminance of
L8 = 1(107 + m(G')Y + n(B)7
If R', G', or B' is negative, then it makes no contribution to the
total luminance. The four most convenient phases for S are
Table 5.2
LUMINANCE PRODUCED BY COLOUR BURST
0 0° 90° 180° 270°
Red 0 0.45" 0 0
Green 0 0 0.07SY 0.18Sv
Blue 0.52S" 0 0 0
= 0, 90°, 180° and 270°, and Table 5.2 gives the luminance contri-
bution of the resulting red, green and blue signals for these four
phases.
Of the four most convenient phases for the burst, 0 = 180°
gives the least luminance on the picture, and in this case the output
is green in colour. The optimum phase angle is on the +I' axis
side of the -(B' - Y') phase, but 180° is a more useful compromise.
5.5. Colour bar signal
Fig. 4.8 shows a diagrammatic picture of chrominance signals,
the dots marking the ends of the phasors which represent the sub -
carrier signal for each of the three primaries and their comple-
mentary colours. In considering the performance of colour
television systems and apparatus these six colours form convenient
standard points to which to refer, giving rise to signals which are
roughly equally spaced around the sub -carrier diagram. One of
the standard colour television test signals consists of a picture
containing eight equal -width vertical bars, as in Fig. 5.1; each bar
W . z< zw , ..., z
I-
13, ..1
_a ,-- ..,
e.
Z
u.k
. =.
0
t..e ...,
t.)
..:(
_., Fig. 5.1. Colour bar test picture
-
Lo <-,
0
I
g.,
.4 CO
is coloured with either one of the primary colours, a complementary,

black or white. The colours are usually arranged in order of
luminance with the brightest colours on the viewer's left hand side.
Other arrangements are sometimes used for special purposes.
The video waveform corresponding to such a saturated colour
bar picture is shown in Fig. 5.2(a) and (b). The sub -carrier peaks
extend up to 1.33, or above the peak white signal, and down to
-0.33, into the synchronizing region. The weighting factors of
1
1.14
for the (R' - Y') signal and03 for the (B' - Y') signal
ensure that these excursions are not larger. Without such weight-
ing factors the sub -carrier would need to extend down below the
zero carrier level, as shown in Fig. 5.3, which is not possible. If
the weighting factors are chosen to be small, so that the chromin-
ance signals are small also, then the signal-to-noise ratio at the
edge of the reception area is poor. It is desirable to use the maxi-
mum amplitude of chrominance signal which can be radiated without
overloading the transmitter. Highly saturated colours at high
luminance occur for only 1% of the time in average television
programmes and it has been found acceptable to limit the sub -
carrier excursions to within the range 1.33 to -0.33. Between
the range to 1.33 the transmitter is nominally overloaded but
1
if this is only for brief periods the effects are not serious. In the
range 0 to -0.33 the sub -carrier is extending into the synchronizing
region, but here again the practical effect on receivers turns out to
be small.
If the sub -carrier were modulated with the full colour difference
signals (R' - Y') and (B' - Y'), which will be written as RD and
WHITE 1-0 0 01789
0.8 170
0-6 1.59
41
0-4 130 1.00
0.2 1
.11
I0- 1.89
0.8 - 70
BLACK 0-
06- I.59
-08 <
1- 08 1.41
O 2 Y 0-4- 1
30
-1 2 1.4.1
U.1 1.0
< 1, 0 GA
LIJ ,- 12 < {,./ _J -J 0.2-
,.- 00X CO CO CO
-0.54 0
-02
08 -0 -4 -043
06 -0-6
04 a
LL1 tal
0.2
=
. . .
--a
1.4.1
..:(
.-
11
CIL
0 0
L..,
0 0<
0 Y 0 0 5
2 2 CO CO
-02
-0.4 1.4 1-33 133
-1-2
-06
1.0
0-8-
I4 1.33 1-33 6-
\\7I18
1.2 -
0.4-
1.0 - 7 I 00 93 -21
0-2-
0. 8 -
06- 5
=21
0.4 -
02- 5 15
o- 07--
-08 .:(\,///' (b)
-.31 -.33 -33 31
4'6'4-54 -54
(a)
Fig. 5.2. Saturated colour bars at full amplitude. (a) with set-up: set-up is 5% of peak white signal
including sync. (b) without set-up
BD, the subcarrier amplitude would be V'RD2 BD2. The peak
excursions of the composite signal waveform would then be
Y' VRD2 BD2
and the values of this for the primary colours are shown in Fig.
5.3 and in Table 5.3.
Q
tJ ztal Z %e
2 411 41
aw
c..1
I8-
1-
0 az
cc w
cc
...,
co co
I6 -
1.4
1.2 -
1.0 - Fig. 5.3. Theoretical colour
08- bar waveform with full
0.6-
colour difference signals
0.4-
02-
0 -
- 0.2-
- -
- 06-
- 0-8 -
Fig. 5.4. Colour bar

waveform with single
weighting of the differ-
ence signals
If the colour difference signals are reduced by a single weighting

factor 8, thus
Y' BD2 BD2
8 can be adjusted so that on the maximum brightness of a saturated

yellow
Y' sy D2 + BD2
and the luminance signal is
Y' = IR' mG' nB'
.where 1, m and n are the luminosity coefficients of the N.T.S.C.
Table 5.3
MAXIMUM AND MINIMUM EXCURSIONS OF THE COMPOSITE VIDEO SIGNAL
`1-,
FI4
cl
CCI
cg
,= -, +
cl gx1
0tn ,--,a +
,-, MI °A.
1:4 CI
co 6 + li
cG
cG
± II .
s. -I-
s Iq
= V-, t!--. c4
e c-4-4 -H -H
Yellow 0.886 0114 0.0130 -0.886 0.7850 0.7980 0.8933 0.447 1.78 -0.01 1.33 044
Cyan 0.701 -0.701 0.4914 0.299 0.0894 0.5808 0.7621 0.381 1.46 -0.06 1.08 0.32
Green 0.587 -0.587 0.3446 -0.587 0.3446 0.6892 0.8302 0.415 1.42 -0.24 1.00 0.17
Magenta 0.413 0.587 0.3446 0.587 0.3446 0.6892 0.8302 0.415 1.24 -0.42 0.83 0.00
Red 0299 0.701 0.4914 -0.299 0.0894 0.5808 0.7621 0.381 1.06 -046 0.68 -0.08
Blue 0.114 -0.114 0.0130 0.886 0.7850 0.7980 0.8933 0.447 1.01 -0.78 0.56 -0.333
primaries. Their values of 0.299, 0.587 and 0.114 are preferably
not substituted until the end of the calculation to avoid errors in
the second decimal place. For yellow, B' = 0, so that
Y' = 11V+mG'=I+m=1-n
and
RD = 1 - Y' n
RD = n - 1
1k= 1 -n SVn2 + (n - 1)2
=1-n+81/n2+0-2n+ 1
= 1 -n + 8V2n2 - 2n +1
1-1 - 1 + n
8-
V2n2 - 2n + 1
n
1/2n2 - 2n + 1
By substituting 0.114 for n, 8 = 0.50, and the values of
Y' 81/RD2 + BD2
for the primary colours are given in Table 5.3.

It will be seen that the total excursions of the signal are now
limited to the range 11 to and that these maximum values are
reached by only the yellow and blue signals. It is not necessary
to limit the cyan and red signals to less than this and they also can
be allowed to reach 1i and -I if two weighting signals are used
instead of one, thus
Y' 1/ceRD2 + PBD2
For yellow
Y' = 1 - n, RD = n, BD = n - 1
1-1 = 1 -n + 'VoL2n2 + P2(n - 1)2
n+ = V a2n2 132(n - 1)2 [5.1]
For cyan
Y'=m±n=1-1, RD =1-land BD = 1
= 1 - 1 + Voy 1)2+P212
IR +1 ___. / 4_ 1)2 + p212 [5.2]
The same two equations are obtained from the negative excur-
sions of the signal
Y' - VaRD2 PBD2
for red and blue. This arises from the symmetrical properties of the
primary signals and their complementaries.
From Equation 5.1
(3n + 1)2 = 9[0/2n2 + /32(n - 1)2] [5.3]
From Equation 5.2
(31 + 1)2 = 9[0(/ - 1)2 + 132/2] [5.4]
Multiply Equation 5.3 by /2
12(3n + 1)2 9[0,212n2 p212(n 1)2] [5.5]
Multiply Equation 5.4 by (n - 1)2
(n - 1)2(31 + 1)2 = 9[0(n - 1)2(1 - 1)2 + /32(n 1)2/2] [5.6]
Subtracting Equation 5.6 from Equation 5.5

12(3n + 1)2(n 1)2(31 + 1)2 = 90,2u2n2 (n - 1)2(1 - 1)2]
since a2 - b2 = (a - b)(a b), then
(31n ± 1 - 31n + 31 -n 1)(31n ± 1 ± 31n - 31 ± n - 1)
= 9a2(In - nl ± 1 n - 1)(1n ln - 1 -n +1)
(41 -n 1)(61n - 21 ± n - 1) = 90(1 n -1)(21n - 1 -n 1)
Further, since 1 + n=1
(4/ -n 1)(61n - 21 ± n - 1)
-m(21n m)
(41 - n + 1)(1 - 61n ± 21 -n)
[5.7]
m(21n m)
Substituting 1= 0.299, m = 0.587 and n = 0.114
1
= 1.14
Similarly, multiplying Equation 5.3 by (1 - 1)2 and Equation 5.4
by n2 and subtracting gives
9p2 (1 + 2n - 6nI - 1)(1 + 4n - 1)

[5.8]
m(21n m)
and hence
1
2.03
1
Using these two weighting factors of
1.14
and03' both yellow
and cyan signals go up to the maximum value of 11 and both blue
and red go down to the minimum value of -I; in this way the
best signal-to-noise ratio is obtained, at the expense of a more
noticeable dot pattern on monochrome receivers.
The transmission of the colour bar signal of Fig. 5.2 provides a
much more onerous condition for the transmitter than does the
average television signal. Such colour bar signals can be reduced
to 75 % of their peak value, when they still represent 100% purity
colours but at only 75% of their maximum luminance. Trans-
mitter overload can also be reduced by transmitting the waveform
corresponding to colours of full luminance and 75% saturation,
but this is not such a convenient signal for receiver testing.
Fig. 5.5 shows the sub -carrier signals for the colour bar wave-
form at 100% and 75% of maximum amplitude, and the corres-
ponding luminance amplitudes are shown in Table 5.4. The
transmitter tolerances are +10° in phase and 20% (approx. 2dB)
in ratio of sub -carrier to luminance amplitude. Fig. 5.5 corres-
ponds to Fig. 4.8 but the background scale has been changed from
the rectangular co-ordinates which are convenient for plotting
(R' - Y') (B' - Y')
and , to the polar co-ordinates which are con-
1.14 2.03
venient for measuring sub -carrier amplitude and phase; a full list of
values is given in Table 5.5. The chromaticities produced by these
sub -carrier signals are those given in Fig. 4.4.
The locus of the maximum excursions of the sub -carrier signal is
the irregular hexagon formed by joining the ends of the sub -carrier
vectors for the primary signals, as in Fig. 5.6. Our original R', G'
and B' signals can be considered in colour space as being limited by
Table 5.4
LUMINANCE AMPLITUDES FOR 100% AND 75% OF
MAXIMUM GAMMA CORRECTED VALUES
Primary 100% 75%

Magenta 0.41 0.31
Red 0.30 0.23
Yellow 0.89 0.67
Green 0.59 0.44
Cyan 0.70 0.53
Blue 0.11 0.08
Table 5.5
SIGNAL VALUES OF LARGE AREA PRIMARY COLOURS AT FULL LUMINANCE
Colour (R' -Y') (B' -Y') (R' -Y') (B' -Y')

R' G' B' Y' (K -Y') (G' -Y') (B' -Y') 1.14 2.03 1.14 Y' 2.03 Y' Q' 1' V (Q')2+ 8° 4°
White 1 I 0 0 0 0 0 0 0 0 0 0 - -
Yellow 1 1
1H
0 0.886 0.114 0.114 -0.886 0.100
-0.436 0.113 -0.493 -0.312 0.322 0.448 13 167
Cyan 0 1 1 0.701 -0.701 0.299 0.299
-0.615 -0.147 -0.877 0.210 -0.211 -0.596 0.632 2565 283.5 -76.5
Green 0 1 0 0.587 -0.587 0.413 -0-587 -0.515 -0.289 -0.877 -0493 -0523 -0.274 0.591 299.3 240.7 --119.3
Magenta I 0 1 0.413 0.587 -0.413 0.587 0.515 0.289 1.247 0.700 0.523 0.274 0.591 119.3 60.7
Red 1 0 0 0.299 0.701 -0.299 -0.299 0.615 -0.147 2.057 -0.493 0.211 0.596 0.632 76.5 103.5
Blue 0 0 1 0.114 -0-114 -0.114 0886 -0.100 0.436 -0.877 3.829 0.312 -0.322 0.448 193 347 -13
Black 0 0 0 0 0 0 0 0 0 0 0 0 - -
4, is phase with respect to (E'H-2,0E3'v) axis.
8 is delay with respect to burst phase.
A Marconi large screen colour television projector. Picture sizes up to
12 ft x 9 ft are obtainable with a highlight brightness of 5 foot lamberts.
The three images are projected from three in -line Schmidt optical systems
with electronic correction provided to eliminate distortion due to the
lateral displacement of the two wing tubes. (Courtesy Marconi Ltd.)
V aull-5017 -,EN .3.s adnidid padnpoadad uo ti atopnqg 'ism aqui
100° 900 80°
110° o°
120° 0°
130° so°
140°
150°1jD +40° 0°
160°
170°
YELLOW
/67°
44 101
0°
0°
00r7
2 ? .3 0 45E1
leo° --REF. BURST 0216 4/Ik

190° 350°
8 2.23
Fla
' BLUE
200°
lir 340°
&G -E-v
c
210°
217°
220°
I 230° ceW
240° 300°
310°
320°
330°
250°
260° 270° 280
DELAYING BURST WITH RESPECT TO SIGNAL BY INSERTING CABLE IN PATH OF REFERENCE

ROTATES DEMODULATION AXES ON THIS DIAGRAM CLOCKWISE. ORANGE FLESH TONES ARE
THEN REPRODUCED TOO YELLOW -GREEN.
Primary sub -carrier signals for 100';,', and of maximum gamma cor-
rected values.
Transmitter tolerances are j.10 in phase and 20'.';, (approximately
2d6) in ratio of sub -carrier to luminance amplitude.
Fig. 5.5. Sub -carrier phase and amplitude coefficients Jor large areas of
saturated primaries
To
UNIT CUBE
WHITE
GREEN RED
YELLOW YELLOW
CYAN
RED
GREEN
MAGENTA
WHITE
BLUE
BLACK
MAGENTA
BLUE
(a) (b)
PROJECTION ON
SUB -CARRIER PLANE
AFTER LINEAR
TRANSFORMATION
B=Y'
2.03
(c)
Weighting factors reduce maximum levels for R' to 11632, G' to

0.593, B' to 0447. Without weighting factors, Q' axis is 2110"
anticlockwise from (B' - Y') axis with scale factor of 1.81Q'. I'
axis is 1391° anticlockwise from (B' Y') axis with scale factor of
I '461'.
Fig. 5.6. Maximum excursions of sub -carrier signals

a unit cube, and the transmitted signal excursions are derived from
this cube by linear transformation so that straight lines remain as
straight lines, although the angles between them, and also their
lengths, alter. The hexagon in the sub -carrier plane is then the
projection of the transformed cube as in Fig. 5.6.
5.6. Chrominance signals and colour triangle for unity gamma
In Chapter 4 the N.T.S.C. transmitted signals Y', I' and Q'
were derived from the R', G' and B' camera signals by a series of
simple calculations which are described mathematically as linear
transformations. For any colour whose tristimulus values are
R, G and B, it is possible to calculate both the chromaticity co-
ordinates r, g and b (see Section 2.11), and the chrominance signals
(R` - Y') and (B' - Y'), or I' and Q'. The chromaticity co-
ordinates r, g and b, are independent of the luminance of the colour.
The chrominance signals increase linearly with luminance, but
(R' - Y') (B' - Y') I' Q'
their values for unit luminance, - and yl
A
are constant for a fixed chromaticity. It is thus possible to plot

on a chromaticity diagram the corresponding values of chromin-
ance signals for unit luminance.
Fig. 5.7 shows a chromaticity diagram on which the lines of
constant r and constant b have been marked. The red primary
R* chromaticity is given by b = g = 0 and r = 1. Notice that a
line from the blue primary, such as BD, represents a constant ratio of
r R
g
= - i.e. all the chromaticities along the line BD are mixtures of
G'
0.7
the blue primary with a chromaticity D which contains - times as
0.3
much R* as G*. All the lines which represent a constant value of -G
pass through B*. When the chrominance signals are zero, the
colour reproduced by the receiver is the system reference white,
and this is produced by one unit of R*, one unit of G* and one
unit of B*, so that the chromaticity of the reference white is at the
centroid of the triangle. This point can be considered as the Y*
primary, since it is the chromaticity reproduced when the other
two transmitted parameters are zero.
The line marking the locus of a constant value of
R - Y
K
say, can be found as follows:
since Y = 1R mG nB
R*
I.0
0.8 LINES OF CONSTANT r
0.6
LINES OF CONSTANT b
0.4
04
0.2 04 0.6 08 I0
LINES OF CONSTANT -
G
ig. 5.7. Colour triangles: (a) with lines of constant r and b,

and (b) with lines of constant RIG
where 1, m and n are the luminosity coefficients of R*, G* and B*,
(1 - - mG - nB
-K
IR + mG + nB
(1 - - mG - nB =KIR + KmG + KnB
i.e., [1 - /(1 K)]R - m(1 K)G - n(1 K)B = 0
N
N R*
Fig. 5.8. Colour triangle with lines of constant
(R - Y)/Y
0.6
Amy
0.4
y
0.2
R -Y
Y I
G* 0 02 0.4 06 08 1.0 12 (8)

b
dividing throughout by R + G B, (assuming R + G + B 0)

[1 - 1(1 + K)]r - m(1 + K)g - n(1 + K)b = 0
[1 - 1(1 ± K)]r - m(1 + K)(1 - r - b) - n(1 K)b = 0
[1 - 1(1 + K) m(1 K)]r - [n(1 K) - m(1 + K)]b
= m(I K)
This is a straight line which passes through the points
b = 0, r - 1 -(1m(1-m)(1K)+ K) [5.9]
and
r = 0, b - m -n [5.10]
This latter point is independent of the value of K, so that all
(R Y)
lines of constant pass through it. It can be considered as
the (B - Y)* primary and like the C.I.E. primaries does not
represent a real colour. By inserting the values of 1, m and n,
the co-ordinates of (B - Y)* are found to be r = 0, b = 1.24.
This value of b =1.24 places the chromaticity of (B - Y)* well
outside the RGB triangle in a super -purity position which is out-
side the range of physically realizable colours. By inserting specific
values of K in Equation 5.9, the intercepts which various lines
make on the r axis are easily found (see Fig. 5.8). Similarly, the
(R - Y)* primary can be located at b = 0, r = 2.04 and the lines
of constant
(B -
Y)can
be found and are plotted in Fig. 5.9.
From Figs. 5.8 and 5.9 the values of

(R
Y) and
(B - Y)
can be
read for any given chromaticity of the transmitted colour.
(R Y)
The line for = 0 passes through the Y* primary and can
be considered as the
(B - Y)
axis; similarly
(B - Y)
-0 is the
Y Y
line passing through the Y* primary which is the

(R - 1') axis.
Y
(R Y)
Instead of calculating the positions of - constant, the
graphical construction mentioned in Chapter 2 can be used. The
line joining (B - Y)* and (R - Y)* also passes through (G - Y)*
and is the line of zero luminance, the alychne (see Section 2.19).
If we had known the position of this line we could have located
- Y)* and (B - Y)* by its intersection with the sides of the
colour triangle.Any line parallel to the alychne such as UV in
Fig. 5.10, would have enabled the positions of the lines of
(R -Y) The line from (R - Y)* to Y* is
= constant to be drawn.
the line
(B - Y)
- 0. The line from (R - Y)* to G* is the line
(B - Y)
= -1. The intercept " e " then defines the interval for
(B - Y)
varying by steps of 1, and the scale along UV is linear.
22
20
Fig. 5.9. Colour triangle with lines of constant
(B - Y)/Y
18
1.6
1.4
1.2
0.2
02 04 0.6 0.8 B 1.2

Fig. 5.10. Graphical construction for the locus
of normalized colour difference signals
\t,e,
\x4
09
0.8
0.7
0.6
SOO
b
495
0.3 490
620
6S'
'SO
-
700
0.2 - 485
'P
480
!e ar
01
B
a
470
I I I I
I
0 01 400 0.2 0.3 04 0.5 0.6 0.7 0.8

CO -EFFICIENT OF X
Fig. 5.11. Lines of constant r, g, b on the C.J.E. chromaticity diagram
The alychne could have been constructed from the equiluminous

point, which is r = 0.242 and b = 0.635, for the N.T.S.C. primaries
and Illuminant C. This may be calculated from the condition that
IR = mG = nB, and I = 0.299, m = 0.587 and n = 0.114.
Similar calculations can be made for I* and Q*, but it is advanta-
geous to use the C.I.E. chromaticity diagrams.
5.7. Chrominance signals on the C.I.E. chromaticity diagram

5.7.1. COLOUR DIFFERENCE SIGNALS, Y = I
Using the formulae in Section 4.30 the chromaticity of a colour

RR* + GG* + BB* can be converted to x and y co-ordinates,
and Fig. 5.11 shows lines of constant r, g and b. This diagram
09 1 i I I /
, R -Y /
320 t
-= -1
515
y /
530 1.0
I -05 0/ 0.5
_1 _ _ i___ _ t
0.8 _1_ _ _ _ _ ___+. ___
CONSTRUCTION LINE540
/ LINEAR SYSTEM
/
550
0.7 RY = 0
560
0.6
570
-
0.5 560
590
0 600
610
.3 630
0.2 NIr0
Nee,
Nye
Ne
0.I
Y
470 coNsTor
I I I I I I I
1 ."-.
01 k 02 0.3 04 0.5 0.6 01 0.8 0.9 10 (R_Y)*
(B -Y)* CO -EFFICIENT OF X
Fig. 5.12. Construction for lines of constant normalized colour difference signals on C.I.E. diagram
09
0.8
0.7
0.6
LL
0.5
0
0.4
620
.3 630 --
650
700
02
0I
0 t 4000.2
0'1 03 0.4 0.5 0.6 0/ 0.8
(B-Y)* CO -EFFICIENT OF X
Fig. 5.13. Lines of constant normalized colour difference signals, unity

gamma, on the C.I.E. chromaticity diagram
can also be constructed graphically, as described in Section 2.14.

The transmission primary points Y*, (R - Y)* and (B - Y)* can
then be found (see Section 5.6) and here also a graphical con-
struction saves much tedious calculation (Fig. 5.12). Since the
chrominance primaries do not contribute to the reproduced lumin-
ance (as y = 1) they must lie on the Y = y = 0 axis. Also, as
(B - Y),
the line joining G* to R* is a line of constant it must pass
through (R - Y)*. The intersection of y = 0 and the line G*R*
is therefore the (R - Y)* primary and is the point x = 1.070 and
y = 0. (B - Y)* is at the intersection of y = 0 and the line
through G*B* at x = 0.131 and y = 0. The family of straight
0.8
0.7
06
>-
0.5
Ls -
0.4
O
0.3
0.2
,- -, /
...- s
..- 0.1
/
,- /
..
BLUE
/
470
I /
, I
1 I 1 I
-4* -t-
. I I I r I
4 I
-0.3 -0 2 -01 0 0.1 02 a* 03 04 01 0.6 07 013
CO -EFFICIENT OF X
Fig. 5.14. Construction of 1* and Q*, unity gamma, on the C.I.E. chromaticity diagram
lines through each chrominance primary makes linear intercepts
on a line parallel to y = 0. The scale of these intercepts is
determined from lines through the white point Y*, which give
(B - Y)
-0 -(R - Y), and from the lines through R*G* and
(R (B
B*G* which give 1= 17). Fig. 5.13 gives a plot
R
of such lines. Notice that these are also plots of -(R 17) +1
B (B Y)
-I- 1. A similar construction will give y.Y
and
5.7.2. 1* AND Q* SIGNALS WHEN Y = 1

The chrominance signals I and Q, normalized to unit lumin-
ance, -/ and -, can be similarly plotted. The I* and Q* primaries
are also zero luminance primaries and lie on y = 0. A simple
graphical construction for the Q* primary is to consider the line
through Y* and Q*, i.e. the line for which -/ = 0 and hence
I = 0.74(R - Y) - 0.27(B -Y) = 0

Along this line, which is the Q axis,
- Y)= 0.365(B - Y)
and
(R Y) (B Y)
= 0.365
It must cut the side G*R* of the receiver primary triangle at the
(B - Y)
value for which - -1, i.e. at
(R Y)
- 0 365. This
point is readily located from Fig. 5.13. The line from this point
through Y* cuts the y = 0 axis at Q*, x = 0.245 and y = 0 (Fig.
5.14). A similar construction shows that I* lies on the negative
side of the y axis, at x = -0.333, y = 0. Lines of constant -Q and
-Q can then be constructed, as in Fig. 5.15, using the lines through

09
550
08
40
VE7
O 550
0.7
ti
o
560
0-6
Pk
000 tl
0.5
04
0.3 490
Q
Y
-
,).
:l.
il .t,
0111.1*
630
650
700_
0
0.2 405 Oa
0.6
440 piVO
° 1
0I v0 Z% - ::?
v7 rte.5. oc's
B e
la400 I 1 I I I
0I 400 0.2 01 04 01 06 07 0.8

CO -EFFICIENT OF X
Fig. 5.15. Lines of constant normalized chrominance transmission signals,

unity gamma, on the C.I.E. chromaticity diagram
Y* for the -/ 0 = 5Q, lines and noting that equals 2 for the red
primary and -Q equals 2.74 for the blue primary.

As I = [0.74(R - Y) - 0.27(B - Y)] increases in value, the
chromaticity point moves away from the I* primary and not towards
it, and it is therefore referred to as the -I* primary.
5.7.3. SUB -CARRIER AMPLITUDE AND PHASE WHEN y = I
The sub -carrier phase D is related to the values of -/ and Q, thus

I
I 17
o = 33° + tan-' = 33° + tan -I- - -
Q
Y
(R - Y)
2.03 (R - Y) Y
= tan -1
1.14 (B - Y) - tan -1 1.78 (B - Y)
Y
The lines of constant sub -carrier phase are lines representing a
fixed ratio of -/ to -Q and are straight lines through the Y* primary,
Y
as shown in Fig. 5.16. The locus of constant sub -carrier amplitude
for unit luminance
'V/2 + Q2 4/ /2 Q2
Y yz -I-- yz
is an ellipse, and curves for various normalized sub -carrier amplitude

values have been plotted in Fig. 5.16. These curves can be plotted
from the values of -/ and -Q for any given chromaticity, derived from
Y
Fig. 5.15. For any given sub -carrier phase the variations in lumin-
ance and sub -carrier amplitude make the reproduced chromaticity
move along a straight line from the white point, so that a given
sub -carrier phase represents a particular dominant wavelength.
On the other hand, for a given value of sub -carrier amplitude per
unit luminance, variations in sub -carrier phase produce mainly
variations in dominant wavelength, although it is clear that these
are always accompanied by small changes in purity. The colour
change resulting from any change in sub -carrier angle or amplitude
can be deduced at once from this diagram if the luminance is known.
When interpreting the visual appearance of the chromaticity change
it must be remembered that the eye is not equally sensitive all over
the diagram, and reference should be made to Fig. 2.18.
5.7.4. CHROMATICITY MAPS WHEN y 0 1
So far we have assumed that the receiver cathode-ray tube has a
linear relationship between voltage input and light output. Practical
colour tubes are not linear and the N.T.S.C. system predistorts
the signals to correct for this, using in general a value of gamma
equals 2.2. For a colour with tristimulus values R, G and B,
I
the signal voltages are proportional to R2'2, G2'2 and B2'2. The
luminance signal is not Y, but
1 1
1R2.2 nV"2.2 + nB2* 2
The diagrams in Figs. 5.13, 5.15 and 5.16 must be revised for
all cases where y 1. For purities up to about 30 % the errors are
small, but become marked for saturated colours near the com-
plementary primaries. The new curves may be calculated from
the equations of Sections 4.25 to 4.30, but a graphical approach
may be more helpful to the reader.
09
0.8
0.7
0.6
0.5
0.4
10
3 610
600
700
0.2
0I
470
I
I y 1 1 1 I I
01 f 400 0.2 01 0'4 05 Ot 0.7 0.8
(B -Y)# CO -EFFICIENT OF X
Fig. 5.16. Lines of constant sub -carrier phase and normalized amplitude,
unity gamma, on the C.I.E. chromaticity diagram
09
520
0.8
$40
G
0.7
-1.0 "1
-0-71V/tpsf, RE:(i
-0-6
-0.5
-0.4 11:1
0.6
444
1k 570
t414 44-
.4fb
SOO 0.2
04ti
0
0.2
***
°1114** ,
.3
0.3
0.4
05
44.1,04.
lIftlytti:x,5 PAA CIO° p35
700
0.6
0.8 /11#1.41X4rAv--4111.2 35
0.2 I-0
Stral1474*-A5C-1?'.-1Z-FX5
114/
5
0I
7.77 313/0.8
B
470
0I ain 0.2 0.3 0.4 0.5 06 0.7 08

CO -EFFICIENT OF X
Fig. 5.17. Lines of constant normalized colour difference signals, y -2.2,
on the C.I.E. chromaticity diagram. Intersections along B*, R4 line
shown as
- Y')/Y' value
(R' - Y')/Y' value
For the gamma corrected case

(R' - Y') R'
- 1
1
R2.2
= 1
Now Y' is not Y2'2 but 1R' mG' nB' so that

Y2.2 iR
Y' Y
-1
II
1
2. 2
Curves of constant(R- can be drawn upon the chromaticity

Y
diagram using the construction of Fig. 5.12, but substituting for
1
the linear scale cd on the construction line a scale of a". If the

transmitted luminance signal were a true luminance signal so that
1
R( ' Y)
Y2'2 Y' the lines of would still be linear. In fact, a
(R' Y'
plot of is) shown in Fig. 5.17. Diagrams for the normalized
Y'
09
520
5$0
08
540
0.:
510
G 0 550
0/ -089/
360
0.6
>-
500
u- _0
0'5
-0-5
LL
t5 04-
C,
-04
011
03 .,190
-01
-0-2 700
-0-1
0.2 0
0-1
480
0I
2781
470 -21
0 0.1 400 0.2 01 04 05 0.6 01 OB

CO -EFFICIENT OF X
Fig. 5.18. Lines of constant normalized chrominance transmission signals,

y = 2.2, on the C.I.E. chromaticity diagram. (Reproduced from "Television
Engineering Handbook" (Editor D. G. Fink), by Bailey and Loughlin. Copy-
right 1957, McGraw Hill Book Company, Inc.)
09
520
515 530
V( IV (QT
0.8
540
VALUES OF
Y' 0
510 0 = 33° + tan' Q
* 40 300
07
20°
505
/ i 60
/ 0.6
06 4
..,o
(3.8,
'
/ /
500
260° / / . I0° /.."'
A.
/ /
11 / 160° no
I / ,
1 , / /
1
I ,/ csv1 140°
495 \
I:11
// / i' - I
1
,
I
/ / , / 120°
280° I ' I / -C
0-I 0.2 0-3 '4 R
.3 490
CYAN .1,
: 1 1
i t
I
f I -1-
' /, / , - / , scxf 620
630
650
-
I" I 80°
700
02
300°//
--'....
/
465 11
1
-, 60°
-- -
480 320°///o°
a
MAGENTA
0-I 340° 20°
470
1443 0 I I I
400 0.2 0.3 04 0.5 0.6 07 0.8

CO -EFFICIENT OF X
Fig. 5.19. Lines of constant sub -carrier phase and normalized amplitude,
y = 2.2, on the C.I.E. chromaticity diagram
chrominance signals -/'

Y'
and -Q', and for the sub -carrier amplitude and
Y'
phase are given in Figs. 5.18 and 5.19.
On Fig. 5.19, the lines of constant phase angle are curved except
for the special cases of the primary hues.
Fig. 5.19 can be used to determine the changes in chromaticity
of the reproduced picture which occur when the sub -carrier changes,
or when errors in the sub -carrier amplitude or phase are introduced
into the receiver.
Although a small change in sub -carrier phase makes a dispro-
portionate change in chromaticity in the green and magenta regions
09
520
515 530
0.8 -
540
550
0.7
560
570
>-
0 05 5110
590
04-
600
610
620
630
650
-
700
0.1
I I I I I
03 04 01 06 0.7 08
CO -EFFICIENT OF X
Fig. 5.20. Chrominance signal axes on the C.I.E. chromaticity diagram
compared with a similar phase change in the blue region, the net
subjective effect on the eye is fairly constant all round the diagram
(see Fig. 2.18). The -/' and
axes, corresponding to phases of
Y'
-Q'
Y'
123° and 33°, can be plotted on the chromaticity diagram and are
shown in Fig. 5.20, as are the colour difference axes.
5.8. Chromaticity diagram on normalized colour difference axes
Fig. 5.21 shows an interesting chromaticity diagram on which the
axes are
(R' - (B' -Y ').
and The RGB colour triangle is
1. 14 Y' 2.03 Y'
shown and it may be demonstrated that the and -Q' axes make
Y' Y'
their correct 123° and_33° angles with the normalized colour differ -
R( ' - r) (B' - Y')
ence axes, just at I' and Q' axes do to the and
1.14
2.03
axes on the sub -carrier signal diagram, in Fig. 4.8. Lines of constant
C.I.E. chromaticity co-ordinates x and y are shown for comparison.
In general, plots of the chrominance signal axes on a colour triangle
do not give the same phase angle relationship as these axes have
on the sub -carrier phasor diagram (see Fig. 5.5). An interesting
exception is shown in Fig. 5.22(a). The (R - Y) and (B - Y)
axes, plotted on a simple RGB colour triangle without weighting
factors and for unity gamma, are at approximately 33° to the corres-
ponding I and Q axes. This is fortuitous and does not, for example,
hold for the triangle of Fig. 5.22(b).
5.9. Chromaticity transitions at colour changes
Since the Y', I' and Q' signals are not all transmitted with the
full system bandwidth, the preceding sections have assumed signals
which are constant or changing slowly with time. For very abrupt
changes in colour, the I' and Q' signals take time to follow the
changes in chromaticity. During a succession of very rapid changes
FILY'
RED
1.140
2
0'
-YT
AGENTA
C.I.E y CO-ORDINATE
C.I.E X CO-ORDINATE
BLY.
YELLOW 2.03
0-
-I -
B UE
GREEN CYAN
0 I 2 3 4
Fig. 5.21. Chromaticity diagram in terms of (R' - Y')//.14Y' and

(B' - r)12.03Y'
1-0
R -Y
0608
04
--
0
3°
TO BG°
02
B -Y
8°0 02 04 06 08 10
(a)
1-0R*
it\0.8
0-6
0.4
T 3°
TO G'B'
0.2
B -Y
0
G 0 02 04 0.6 \ 0.8
B
(b)
Fig. 5.22. Chrominance axes on simple colour triangle with no gamma

correction; (a) with B* at origin; (b) with G* at origin
the and Q' signals may not be able to follow fast enough and will
adopt mean values. With frequencies above 0.34 Mc/s in the 405 -
line system, the Q' signal will only reproduce a mean value, and for
frequencies above 1 Mc/s the signal also will only follow the
mean value. At such frequencies the luminance signal is the only
signal which changes. This does not mean that the fine detail in
09
0.8
0.7
0.6
PURITY CHANGE
AS Y' ALTERS
05
0.4
620
630
650
700
1464
0.1 400 0.2 0'3 0.4 05 0'6 0.7 0.8
CO -EFFICIENT OF X
Fig. 5.23. Reproduced chromaticity change at a colour transition
the picture can only be reproduced in black and white. The average
colour of the fine detail will be reproduced and this will change in
brightness with the detail information. Similarly, picture detail
which gives rise to frequencies in the range 0.34 to 1 Mc/s will not
be reproduced only in orange -cyan colours, although only the
orange -cyan component in the colour will vary. These considera-
tions only apply along the scanning line; in the vertical direction
the full system resolution is obtained in all colours.
Because of the bandwidth limitation the chromaticity reproduced
by the receiver during a colour transition may not follow the chro-
maticity change at the transmitter, although the final reproduced
chromaticity will be correct. The rates of change of the three
signals Y', I' and Q' are different. The first signal to vary at a
transition is the Q' signal, so that the reproduced chromaticity moves
/'
along a line of constant -, until the I' channel signal begins to change
(Fig. 5.23) when the chromaticity moves along a path determined
by both the -; and Q' signals. When the luminance transition occurs
formsY' /'con-
the chromaticity makes an abrupt change in purity, and then
tinues to trace out the path determined by the - and -Q' wave-
Y'
until the 1' waveform transition finishes, when it follows a
'
line of constant to its final chromaticity. The transition thus
Y'
has five distinct parts during each of which the chromaticity is
moving along a different path. Such transitions are further dis-
cussed in Section 16.8.
5.10. Luminance transitions at colour changes

If the constant luminance principle were obeyed exactly, the
irregular and slow chromaticity transitions would be masked by
the sharp luminance change produced by the Y' waveform. As the
Y' signal is not the true gamma corrected luminance, the chromin-
ance sub -carrier is carrying part of the luminance, particularly for
colours of high purity.
That part of the luminance carried by the chrominance signal is
bandwidth limited and can only change slowly, with the result
that the reproduced luminance also shows five distinct changes
through a colour transient.
As the sub -carrier goes near zero on a slow colour change, the
contribution to luminance which it carries exchanges exactly with
the luminance signal. On a fast colour change, the restricted band-
width of the chrominance channels makes the chromaticity change
slow and the change in the luminance carried by the sub -carrier is
correspondingly slow. This luminance change is therefore relatively
broad and does not match up with the fast luminance change carried
by the Y' signal. These effects are discussed in more detail in

Chapter 16.
5.11. Summary
A graph of lines of constant sub -carrier phase and amplitude on a
C.I.E. chromaticity diagram show that the sub -carrier phase deter-
mines mainly the dominant wavelength of the reproduced colour,
whilst the sub -carrier amplitude for unit luminance determines mainly
the reproduced purity. This is exactly true for the primary colours
and nearly true for pastel colours, but becomes more incorrect as the
purity of the non -primary colour increases.
During a colour transition the reproduced chromaticity can
change along several incorrect paths before assuming its final correct
value. The associated luminance change can also have a slow,
incorrect variation superimposed upon it. These errors are usually
only obvious on special test patterns, such as the colour bars, which
involve changes between saturated colours.
CHAPTER 6
Transmitter Coding Circuits

6.1. Introduction
The processing of the R, G and B signals to form a composite
N.T.S.C. video signal was described in Chapter 4, and it is the
purpose of this chapter to examine the circuits which can be used
for performing the various operations required.
The circuits described here have been used by the authors to
obtain a fully encoded N.T.S.C. signal on the 405 -line standard for
colour receiver test purposes, but it should be pointed out that these
circuits are not the only possible arrangements which can be used.
For this reason, a block diagram of each section of the apparatus
will be described and then a suitable circuit will be given for it.
Should the reader wish to construct a transmitter, he may apply
to the construction those circuits with which he is familiar, provided
they carry out the operations indicated in the block diagram.
6.2. Complete block diagram of transmitter

The operations required to form a complete R.F. signal from an
R, G and B input signal are shown in Fig. 6.1. It is assumed
that each of the three video signals is about 1 volt peak -to -peak
amplitude and is line and field blanked, but without sync pulses.
Note that if the R, G and B input signals are obtained from a picture
source, they must be individually gamma corrected before entering
the unit of Fig. 6.1. On the other hand, gamma correction is not
necessary if a colour bar signal (see Section 6.8) is used, since in this
case each signal has only two possible values, namely, " on " or
" off ".
The encoder unit is fed with the R, G and B video signals, a C.W.
sub -carrier signal and mixed sync, and its output is a composite
N.T.S.C. video signal. The composite video signal is then mod-
ulated on to a suitable R.F. carrier. At the same time the sound
carrier must be generated and modulated with the audio signal and
then both sound and vision carriers are added together. To keep
the special frequency relationships which are called for in N.T.S.C.
operation, the sub -carrier oscillator is used to provide both the
156
OUTPUT
R t
COMPOSITE VIDEO VISION SOUND
ENCODER ADDER A. F. INPUT
MODULATOR MODULATOR
4145625 MO
SUBCARRIER
OSCILLATOR
45 MO IXER
OSCILLATOR
2 6578125 MC/s
r -1
X4
0 8859375 MC /S 3.54575 MC/s
4 0. :
5
03796875 MC/S 151875 MC/S 0 30375 MC/S 010125 MC/S I1 I 20'25 kc s
L DIVIDER CHAIN
50 C/s FIELD
: 405
I(
PULSE FREQUENCY
41(
MIXER
MIXED SYNC. 10125 C/S
:2
LINE*
FREQUENCY
m
CLAMPING PULSE OUTPUT I CONVENTIONAL SYNC. GENERATOR
_
Fig. 6.1. Complete block diagram of' an N. T.S.C. transmitter
ammm-aalmass,...

mixed sync timing and the difference in frequency between the
sound and vision carriers.
In order to achieve the correct frequency relationship between
the sub -carrier and the scanning frequencies, the twice line frequency
oscillator of a conventional sync generator must be replaced by a
signal derived from the sub -carrier frequency. This signal can be
obtained by dividing the sub -carrier frequency by 525 (i.e. 7 x 5
x 5 x 3) and multiplying by 4.
Again, in order to reduce the visibility of the R.F. sound carrier/
R.F. sub -carrier beat, this beat frequency is made an odd multiple
of half the line frequency. Since the R.F. vision carrier/R.F. sub -
carrier beat is also an odd multiple of half the line frequency, it
follows that the R.F. vision carrier/R.F. sound carrier beat is a
whole multiple of the line frequency. The multiple which is most
convenient and which gives a beat nearest to the conventional beat
frequency of 3.5 Mc/s is 350. The R.F. vision/R.F. sound beat
then becomes 350 x 10,125 c/s, or 3.54375 Mc/s. Since the
525
sub -carrier frequency is -2 times the line frequency, and the R.F.
vision/R.F. sound beat is 350 times the line frequency, it follows
that the sub -carrier frequency is x2350 or I of the R.F. vision/R.F.
2
sound beat. Hence, the appropriate relationship between the vision
and sound carriers can be achieved by first deriving the beat fre-
quency of 3.54375 Mc/s by taking of the sub -carrier frequency, and
then mixing this with either the R.F. vision carrier or the R.F.
sound carrier. Strictly, the beat should be mixed with the con-
ventional sound carrier of 41.5 Mc/s so that the sound traps of
existing monochrome receivers are correctly tuned, but in the
equipment used by the authors a crystal controlled 45 Mc/s source
was already available so that for convenience the sound carrier was
derived by mixing the beat with the vision carrier.
It should be noted that while a public transmitter would be
required to conform to the various frequency locking arrangements
described above, for most test purposes they are not essential, and an
R.F. vision signal of composite video in which the sub -carrier and
scanning frequencies are not locked can prove very useful.
The individual blocks of Fig. 6.1 will now be described in detail.
6.3. Encoder
The encoder accepts the R, G and B blanked video signals, a
C.W. sub -carrier signal and the mixed sync signal, and provides a
Y DELAY
I MC/S PHASE
I DELAY BALANCED
L. P. F. SPLITTER
INPUT MODULATOR
COMPOSITE
G VIDEO
INPUT OUTPUT
MATRIX -0 ADDER
VARIABLE C. W.
90° CLAMP
"-MN- PHASE SUBCARRIER
INPUT SHIFTER GENERATOR
DRIVE
0.34 M CiS PHASE 9

BALANCED
L . P. F. SPLITTER
MODULATOR
I.
OUTPUT TO
MIXED SYNC. GATE SUBCARRIER
HMODULATOR DIVIDER
INPUT GENERATOR OSCILLATOR
CHAIN
FIELD MIXED SYNC .+ BURST

BURST ADDER
REMOVAL GATE
INSERTION
UNIT
Fig. 6.2. Encoder block diagram

composite video output. A block diagram of the arrangement
is shown in Fig: 6.2, while a possible circuit configuration is shown
in Figs. 6.3(a) and (b). Referring to Figs. 6.2 and 6.3(a), the R,
G and B video signals are first fed into a matrix network which forms
the transmission parameters Y, I and Q according to the equations
Y= 0.30R + 0.59G + 0.11B
-I = -0.60R + 0.28G + 0.32B
Q= 0-21R - 0-52G + 0.31B
The appropriate fractions of R, G and B are formed by resistor
networks, while a phase inverting stage is used to obtain the negative
red signal for the -I parameter, and similarly for the negative
green signal for the Q parameter. Note that by deriving -I
rather than +I, only one phase inversion for red is required instead
of two inversions, one for green and one for blue. The -I signal
can be readily phase inverted in the modulator stage by appropriate
connection of the sub -carrier drive transformer.
The Y signal is delayed by approximately 0.7 ps (see Section 4.22)
by means of a ferrite loaded delay line which has a constant resist-
ance termination to prevent mismatch of the delay cable by the
input capacity of the following valve.
The I signal is delayed by approximately 0.5 tis also by means of a
ferrite loaded delay line with a constant resistance termination, and
this feeds the grid of a valve whose anode load is a low pass filter
having a cut-off frequency characteristic which is about 1dB down
at 1 Mc/s and 20dB down at 1.8 Mc/s. The low pass filter output
is then fed to a phase splitting stage which provides a push-pull
output to the I balanced modulator. The balanced modulator
consists of two 6F33 valves whose anodes are strapped and whose
control grids are fed push-pull by the output of the phase splitter,
while the suppressor grids are fed push-pull by a C.W. sub -carrier
signal of the appropriate phase. The modulator is really doubly
balanced in that no output is obtained if either the I signal drive or
the C.W. sub -carrier drive is removed. The I modulator output is a
sub -carrier sine wave whose amplitude is proportional to the
incoming I signal and whose phase is always lagging 57° on the burst
phase for a positive I signal, and always leading 123° on the burst
phase for a negative I signal. For a zero I signal (which must occur
during the line and field blanking periods) no output is obtained.
The Q signal path is identical to that of the I signal with three
exceptions. Firstly no delay is required. Secondly the low pass
filter cut-off frequency characteristic is about 1dB down at 0.34
+250V
'
520.2k11
LINE SYNC
INPUT
Tig. 6.3 (a). Encoder circuit diagram
(FOR CLAMPING) (matrix and balanced modulator)
O'IALF
1000.
62on
MIXED SYNC 6BOn
+ BURST INPUT
C.W.SUBCARRIER
INPUT
4'7pF
75n
3.3141
-A
432pF
Io pH 368pH
I CHANNEL
COMPOSITE
VIDEO
OUTPUT
7511
101d).
Q CHANNEL -4-
7511
MATRIX DELAYS LOW-PASS CATHODE

PHASE SPL IT TERS CLAMPING C.W. SUBCARRIER BALANCED ADDER FOLLOWER
F ILTERS DIODES DRIVER MODULATORS OUTPUT
+2soV
47kfl 6.13kfl :00/n 150kn
100/2H
C.W. SUBCARRIER 10PF

47kfl
OUTPUT TO MODULATORS -4-1
E. DIVIDER CHAIN
V4
15pF
330k11 100pFT 220E1

2200.
j17,000pF
int IMfq
470n
01yE- I0 1,11
220k
'50pF 22011
POSITIVE MIXED SYNC
+ BURST OUTPUT
613011 221,11 CrIALF SYNC AMPLITUDE NEGATIVE MIXED SYNC

ADJUSTER +BURST OUTPUT
SUBCARRIER FREQUENCY CRYSTAL
(26578I25MO5) POSITIVE MIXED
SYNC INPUT
10k11 3.3k11,
470pF, 5000 PET SECTION 8i/F 8/F
I kfl 3.3kfi
33AE1
10611
VIO VII VI2

0 LciF
1.000pF 3 OpF I,000pF
11
27 27 277
I Mfl IMn
100k11 47611
son
1,000pF
Fig. 6.3 (b). Encoder circuit diagram (burst insertion unit and sub -carrier oscillator)
TRANSMITTER CODING CIRCUITS 163
Mc/s and 10dB down at 0.5 Mc/s, and thirdly the phase of the C.W.
sub -carrier push-pull drive to the Q balanced modulator lags 90°
on the drive to the I balanced modulator. The Q modulator
output is a sub -carrier sine wave whose amplitude is proportional
to the incoming Q signal, and whose phase is always lagging the
burst phase by 147° for a positive Q signal, and always leading the
burst phase by 33° for a negative Q signal.
It is necessary to preserve the D.C. components of the Y, I and
Q signals but it is inconvenient to provide D.C. coupling throughout.
Hence, A.C. coupling is employed, together with a clamping circuit.
Clamping of the I and Q signals can be conveniently carried out
at the grids of the I and Q modulators, while Y signal clamping is
usually arranged on the composite video signal input to the R.F.
modulator. If a video signal only is required, the clamping can
be made on the composite video fed to the grid of the output
cathode follower.
The principle of the clamping method used here is to arrange
to return the point to be clamped (which is usually the grid of a
valve A.C. coupled to the previous stage) to a certain D.C. potential
at the start of each line scan period. To do this, line frequency
pulses are fed to a clamp generator which is essentially a phase
inverter providing a push-pull output of equal amplitude but
oppositely phased pulses. These outputs are connected to two
diodes in series, so that conduction occurs during the pulse period
and the anode/cathode junction of the diodes assumes the D.C.
potential to which the diode leak resistors are connected. Hence,
by connecting the point to be clamped to the junction of the anode
and cathode of the diodes, this point will assume the D.C. potential
of the diode leak resistors during the pulse time. Note that during
the period between pulses, i.e. during the active line scan, the diodes
are biased off and do not affect the signal on the point where
clamping is applied. The clamping pulses themselves should be
long enough to ensure discharge of the coupling capacitors, yet
short enough not to interfere during the active line period. As
far as the balanced modulators are concerned, the clamping pulses
may be as long as the line blanking period, but for clamping the
composite video when sync pulses are present, they must obviously
not be longer than the line sync period.
For most applications, continuous line sync pulses or mixed
sync pulses provide a suitable clamp source, but in refined equip-
ment a pulse generator which develops line frequency repetition
pulses having a smaller width than line sync pulses and centred
within them in time position, and with relatively poor rise and
decay times, is recommended. Then spurious differentiated pulses
due to stray capacity effects are avoided in the waveform being
clamped.
The purpose of the burst insertion unit, shown in the lower half
of Fig. 6.2 and in circuit form in Fig. 6.3(b), is to add the 9 cycles
of colour synchronizing burst to the mixed sync waveform. This
is done by first forming a suitable gating pulse waveform which
(a)
(b)
(q)
(h)
(i)
(J)
Fig. 6.4. Burst insertion waveforms. Shaded areas

represent peak -to -peak sub -carrier sine wave
consists of rectangular pulses of line frequency repetition rate, having

a leading edge lagging 1.5 /Is on the trailing edge of the line sync
waveform, and having a duration of -2.66 that is, 3.4 t.t.s. Since
the burst signal should be omitted during the field sync period, a
field repetition pulse which has a duration of 4 lines is applied to
the gating waveform so that the line frequency pulses are removed
during the field sync period. The field removal gate can be taken
from the conventional sync generator, but if this is not readily
accessible it can be generated from the mixed sync waveform by
means of an integrating and shaping circuit; see Fig. 6.3(b).
A convenient method of generating the burst insertion gating pulses
is by way of a lumped constant delay line. The principle of opera-
tion is indicated in Fig. 6.4. Positive going line sync or mixed
sync pulses are fed to the grid of a pentode whose anode is con-
nected to a point in the delay line, which is short circuited at one end
and characteristically terminated at the other.
Hence, the line sync pulses applied to the grid of the valve
(Fig. 6.4(a) ), are inverted and applied to the delay line (Fig. 6.4(b) ).
Having travelled down the delay line, the pulses are reflected and
inverted at the short circuit so that they arrive back at the anode
connection as delayed pulses (Fig. 6.4(c) ). The anode waveform is
therefore the sum of those in Fig. 6.4(b) and (c) and is drawn in
Fig. 6.4(d). This waveform travels towards the characteristically
terminated end of the line, being delayed in the process, and is not
reflected from the characteristic termination. The output is there-
fore as indicated in Fig. 6.4(e).
This method has the advantage that there is an independent
control of the pulse width (moving the short circuit towards the
anode tap reduces the pulse width) and for the time position of the
leading edge of the pulses; see Fig. 6.4(e). (Moving the output tap
towards the characteristic termination increases the delay.)
The pulse waveform (Fig. 6.4(e) ) can now be clipped by a grid
current biased valve (a " sync separator " stage) to yield an
output (Fig. 6.4(f) ), and this is finally inverted as shown in
Fig. 6.4(g). The field removal gate can be applied to the clipping
stage to remove the line gate during the field sync period.
The burst insertion gating pulses are fed to a modulator stage
which is also fed with a C.W. sub -carrier signal. In effect, this
modulator is an amplifying stage which is switched on only during
the gating pulse period and its output consists of the required burst
signal sitting on a negative pedestal (Fig. 6.4(h) ). Differentiation
of this waveform removes the pedestal to give the burst waveform
(Fig. 6.4(i) ). The differentiation produces " pips " corresponding
to the leading and trailing edges of the pedestal but if these edges
are not sharp the " pips " are only small in amplitude. Finally,
the mixed sync is added to give the mixed sync plus burst waveform
as shown in Fig. 6.4( j).
The complete composite video waveform is obtained by adding
together the Y signal, the modulated I and Q signals, and the mixed
12
sync plus burst waveform. One way of carrying out the addition
is by means of four pentode stages whose anodes are strapped, as
shown in Fig. 6.3(a).
The burst phase should lead the I phase by 57° and in order to
set this correctly a variable phase shifter is included in the C.W.
sub -carrier feed to the sub -carrier driver which operates the I and Q
balanced modulators. The I phase should, of course, lead the Q
phase by 90°.
Several adjustments are necessary to ensure the correct operation
of the circuits shown in Fig. 6.3(a).
Thus, in the matrix circuit, the inverters V1 and V2 must have the
correct gain settings so that the matrix equations
-I = -0.60R + 0.28G -I- 0.32B
and
Q = 0.21R - 0.52G + 0.31E
are satisfied. This may be done by connecting a video oscillator
operating at about 100 kc/s to the blue input and adjusting the
voltage output until a 1.0V reading is obtained on a valve volt-
meter connected to the grid of V3. Next, transfer the oscillator
output to the green input and adjust P1 until 1.68V is obtained at
the grid of V3. This ensures that the green to blue ratio is 0.52
to 0.31, or 1.68, in magnitude.
Similarly, with the oscillator still connected to the green input,
transfer the valve voltmeter to the grid of V4 and adjust the oscil-
lator output to give a reading of 1-0V. Next, connect the oscillator
to the red input, and adjust P2 until 2.14V is obtained at the grid
of V4. This ensures that the red to green ratio is 0.6 to 0.28, or
2.14, in magnitude. (See Fig. 6.3(a), odd numbered valves up to
17 refer to the Q channel, and even numbered valves up to 18 refer
to the I channel.)
The operating conditions of the balanced modulators may be set
up with the aid of a colour bar generator, as described in Section
6.8. This generator provides equal peak to peak values of square
waves of R, G and B which simulate a video signal consisting of a
vertical full white stripe immediately following the line blanking,
adjacent to which is a full yellow, then cyan, green, magenta, red,
blue and black. Thus, full white corresponds to R = 1, G = 1
and B = 1, while full yellow corresponds to R = 1 and G = 1
and so on.
With the colour bar output signals connected to the appropriate
R, G and B input sockets and with the mixed sync plus burst and
line clamping inputs connected, together with the C.W. sub -carrier
input, connect a D.C. voltmeter between the slider of P3, P6 and P7
and earth in turn, and adjust each one to give a reading of 2.5V.
P3 sets the reference voltage to which the black level of the video
signal is restored and controls one valve of each modulator pair
(V14 and V15). The second valve of each pair (V13 and V16) may
be set up as follows:-
With an oscilloscope which is capable of displaying the sub -
carrier sine wave and synchronized to the line frequency, observe
the waveform at the grid of V17. Adjust P5 to balance out the
sub -carrier during the line blanking period and white bar period,
using the trimmers between anodes and suppressor grids of V
and V15 for final adjustment.
Exactly the same procedure may be adopted for balancing the
I modulator by connecting the oscilloscope to the grid of V18 and
adjusting P4 and the trimmers connected to V14 and V16.
The next step is to set the correct relative levels of the Y, I and
Q components. With the composite video output displayed on
an oscilloscope synchronized to line frequency, remove the I signal
by earthing the grid of V18 and adjust P9 in the Y channel, so that the
amplitude of the white bar above black level is
1
-0 96 times
2 x 0.52
as great as the peak -to -peak amplitude of sub -carrier on the green
or magenta bars. This ensures that the magnitude of the Q sub -
carrier signal for full green is 0.52 of the black to peak white ampli-
tude.
Similarly, the correct I sub -carrier amplitude is set up by earthing
the grid of V17 to remove the Q signal and adjusting P8 in the I
channel so that the peak -to -peak amplitude of sub -carrier on the
red or cyan bars is
2 x 0.6 = 1.2 times
as great as the amplitude of the white bar above black level.
Notice that the peak sub -carrier amplitude, not the peak -to -peak
subcarrier amplitude, is equal to the I or Q signal. However, the
peak -to -peak sub -carrier is more convenient to measure on an
oscilloscope and this accounts for the factor of 2 in the above
relationships.
The final step in setting up the correct amplitude relations con-
sists of adjustment of the sync. amplitude relative to the burst
amplitude, and the amplitude of the burst plus sync waveform
relative to Y.
Observe the mixed sync plus burst output on an oscilloscope
(see Fig. 6.3(b) ), and adjust the " sync amplitude adjuster " so that
the peak -to -peak sync waveform is equal to the peak -to -peak burst
waveform. Next, observe the composite video output (Fig. 6.3(a) )
and adjust P10 so that the sync height is 0.43 of the black to white
amplitude. This figure of 043 is derived as follows:-
Since 100% carrier corresponds to peak white and 30 % carrier
corresponds to black, the black to peak white range is 70 % carrier.
30
Hence the sync height is - or 043 of the black to peak white
70
range.
The above adjustments ensure the correct relative amplitudes of
burst, sync, Y, I and Q signals but in addition the correct phase
relation must obtain between the I and Q sub -carrier signals and
the burst. Thus, I must lead Q by 90° and the burst phase must
lead the I phase by 57°.
Measurement of sub -carrier phase angles can be readily carried
out by means of a calibrated phase shifter and an oscilloscope
whose X amplifier can be fed with an external signal. A suitable
phase shifter is described in Section 12.5, while most oscilloscopes
are fitted with an X input terminal, though unfortunately relatively
few have an X amplifier with a cut-off frequency above the sub -
carrier frequency of 2.7 Mc/s. It has been found that oscilloscopes
having X amplifier bandwidths of only some 300 kc/s can be used
at sub -carrier frequency provided that sufficient amplitude of signal
is available.
The principle of phase measurement is to apply one signal (pre-
ferably the smaller) to the Y amplifier of the oscilloscope, while the
larger signal is applied via the phase shifter to the X amplifier.
Since both signals are at the same (sub -carrier) frequency, the dis-
play is an ellipse. By adjustment of the phase shifter the ellipse
can be collapsed into a straight line, in which case the two signals
on the oscilloscope X and Y plates are either in phase (straight line
with positive slope) or in anti -phase (straight line with negative
slope).
Thus, to set the phase between the / and Q sub -carrier signals,
apply the composite video output, see Fig. 6.3(a), to the Y amplifier
of an oscilloscope and a reference C.W. sub -carrier signal from the
sub -carrier oscillator to the X amplifier via the calibrated phase
shifter. With a suitable signal (e.g. the green output of the colour
bar generator) connected to the grid of V,, adjust the phase shifter
PHASE ANGLES
PEAK TO PEAK RELATIVE TO BURST
VALUES OF
SUBCARRIER
SINE WAVE
-13° +103°
c) -88:M26:
N'SJ
LUMINANCE +61°
LEVELS 1-18
-119°
1007. CARRIER ---1.0
0.89
0.7
059
0 41
BURST
-+ 0-215
30/. CARRIER
--0215
CARRIER 0 43
WHITE YELLOW CYAN GREEN MAGENTA RED BLUE BLACK
Fig. 6.5. Colour bar composite video waveform
until a straight line display is obtained. Note the phase reading,

which gives the phase of the modulated / signal. Switch the phase
shifter to a reading 90° lagging on this, connect the green output of
the colour bar generator to the grid of V3 and adjust the trimming
capacitors connected to the driver transformer (Fig. 6.3(a)) so that a
straight line display is again observed on the oscilloscope. If this
straight line has the opposite slope to that obtained previously,
reverse the connections between the driver transformer and either
the I or the Q modulator. Re -check the phase of the modulated
I signal and repeat the whole process until the I signal leads the Q
signal by 90°.
Next, the I phase must be adjusted to lag the burst phase by 57°.
With the composite video applied to the Y amplifier of the oscil-
loscope and a reference signal applied via the calibrated phase
shifter to the X amplifier, adjust the calibrated phase shifter so that
a straight line is obtained for the burst signal. With 57° added
to the reading of the phase shifter and with the green output of the
colour bar generator connected to the grid of V4, adjust the phase
shifter P11, which feeds the C.W. sub -carrier drive to the modulators
(Fig. 6.3(a) ) until a straight line display is obtained for the green
bar period.
For a 100% saturated colour bar signal, the composite video
output should appear as shown in Fig. 6.5 and the phases of the
sub -carrier signals relative to the burst should be as shown.
It should be noted that, in the authors' opinion, the lengths of
delay line in the Y and I channels should be such that, when the
composite video is modulated onto an R.F. carrier and displayed
on an average monochrome receiver, the sub -carrier dot structure
due to a coloured portion of the picture is symmetrically placed
about that coloured portion. This may be checked, for example,
by applying the blue output of the colour bar generator to the grid
of V21 (Fig. 6.3(a) ), and also to the grid of V3, having first discon-
nected the V3 grid from the V1 anode circuit and having earthed the
grid of V4. The resulting composite video, having been modulated
onto an R.F. carrier, should give a display on a monochrome
receiver which has equal widths of dot structure along each side
of each bar. Adjustment of the Y delay line length should be
made, if necessary, until this condition is obtained: Note that the
BBC delay specification differs from this (see Section 4.22).
Similarly, the I delay line may be checked by feeding the blue
bar output to V21 as before, and also to the input of the I delay
line with the grid of V3 earthed. Again, adjustment of the I delay
line length should be made to give a symmetrical dot pattern.
The encoder shown in Fig. 6.3(a) can be readily converted to
equiband I and Q working. This can be done by replacing the
I delay line with a conventional lead, removing the I and Q low
pass filters in the anodes of V4 and V3 and replacing these by simple
resistive loads shunted by a capacitance. Thus, 1.5k0 resistor loads
shunted by 150pF will give a 3dB response at 700 kc/s. The Y
delay line must be altered so that a symmetrical dot pattern is
observed on a monochrome receiver, as described above.
6.4. Vision:modulator and carrier oscillator
The vision modulator and carrier oscillator unit accepts a com-
posite video signal and a suitable clamping waveform (such as
continuous line sync) and supplies a modulated carrier output
and also a C.W. vision carrier output from which the sound carrier
is derived, as described in the next section.
A suitable circuit arrangement is shown in Fig. 6.6. The com-

posite video (sync negative going) is applied to the grid of the
pentode V4 and clamping is applied from the amplifier V, and the
diodes V2 and V3, so that the V4 grid video signal is instantaneously
earthed during each line sync period.
The operation of the valves V4 and V5, which comprise the
modulator stage, may be explained as follows :-
Suppose that, in the first instance, no signal is applied to the
grid of V,.Then part of the V4 anode current will flow through
45Mc/s MODULATED OUTPUT
45Mc/5 C.W. OUTPUT
+250V
5.6kriti500pF
-3.31,11
MODULATION
DEPTH
ADJUSTMENT -----
8201.1 471,11 I5Mc/s

CRYSTAL
0IAn 100k11
681.11 ,500 pF
LINE VI
10 II.
SYNC
INPUT
-44-11
0.1ALE
lOokEt
1-7
,500 pF
15011
oekn F.1,500pF
150n 60p FL- 15011 1001,11
V3 =
0.01pq
VIDEO
INPUT
CLAMP DIODE MODULATOR BUFFER I Mc/s OSCILLATOR

AMPLIFIER CLAMP AMPLIFIER & x3 MULTIPLIER
Fig. 6.6. Vision modulator and carrier oscillator
the resistor R, and the remainder will flow through V5 and V,.
Since the grid of V5 is held at a fixed D.C. potential, the cathodes
of V5 and V, will likewise be held at a fixed potential so that the
current through the resistor R will be constant and independent of
the V4 grid drive. Hence, by a suitable choice of parameters, it
is possible to make the current through R equal to the anode current
of V4 when the V4 grid is earthed. Then zero V4 grid voltage gives
zero current through V5 and V,, the anode current of V4 flowing
wholly through the resistor R. Now when the V4 grid is driven
positive, the current through R remains the same and the increase
in V4 anode current therefore passes through V5 and V,. Thus,
the V5 current bears a linear relation to the V4 grid voltage down to
the point where this is zero; that is, the operating point on the
mutual characteristic of V4 is lifted by the current bias through R.
When the carrier signal is injected into V5 by V,, it is modulated
linearly by the V4 signal and the modulated output is obtained from
the V5 anode load. The V5 grid potential should be adjusted so
0.1pF
8pF
22pF
opH
35437581cts
- INPUT
,500pF
°Old).
Fig. 6.7. Mixer for generating sound carrier
sync signals correspond to zero

output. This can be done by applying the modulated output to a
simple crystal detector and observing the demodulated video wave-
form on an oscilloscope. If the oscilloscope input is now shorted
out intermittently, a line will be observed corresponding to zero
input and by adjusting the potentiometer PI the negative tips of the
line sync can be made coincident with this zero line.
The carrier frequency of 45 Mc/s is generated from a 15 Mc/s
crystal oscillator which has a load tuned to 45 Mc/s. The buffer
stage V, provides additional filtering to remove the 15 Mc/s com-
ponent and it also provides a C.W. 45 Mc/s signal from which the
sound carrier may be derived.
The anode loads of V5 and V, consist of 45 Mc/s tuned circuits
with loosely coupled secondaries. Each may be about 8 turns, with
a 1 turn secondary, on a i in. former with dust core tuning.
The modulated vision carrier should finally be passed through an
asymmetric sideband filter and a phase correcting circuit, but for
ordinary test purposes these refinements can be omitted.
6.5. Sound carrier mixer
Fig. 6.7 shows a suitable circuit for generating the sound carrier
frequency from the 45 Mc/s vision carrier and the 3.54 Mc/s derived
by taking of the sub -carrier frequency. The 45 Mc/s input is fed

to V1, whose anode load is tuned to 45 Mc/s, and thence to the mixer
V2. Similarly the 3.54 Mc/s input is fed to V,, whose anode load
is tuned to 3.54 Mc/s, and thence to the suppressor grid of V2.
The V2 anode load is tuned to 41.46 Mc/s.
The anode loads of V1 and V2 may be typically some 8 turns on
a j in. former with a dust core, the secondary windings being 1 turn.
The circuit (Fig. 6.7) can be used to derive a 45.04 Mc/s vision
carrier from a 41.5 Mc/s sound carrier, by tuning the anode loads
of V1 and V2 appropriately. This is the correct specification but as
mentioned earlier it may prove more economical to derive the sound
carrier since the vision carrier frequency is probably already avail-
able from a monochrome picture source.
6.6. Sound modulator
The sound modulator circuit can be similar to the vision modula-
tor shown in Fig. 6.6, with minor modifications.
Thus (Fig. 6.6), the sound carrier is fed to the grid of V8 and the
audio input is fed via a potentiometer to the grid of V4. The
valves V1, V2 and V3 must be omitted, of course, and also V, and
÷3 REGENERATIVE DIVIDER x4 MULTIPLIER
GEX54's
C.W. SUBCARRIER "IMF 22PF OUTPUT TO

INPUT -3.--i I-I...SOUND CARRIER
f = 2-657BIZ5Mcis MIXER
-r 314375Mth
4.7pF
!Son
CH CI( POINT CR CK POINT
= 011E159375MR/5 3.54375MOR
Fig. 6.8. 413 multiplier for deriving sound/vision beat

from sub -carrier
V8. If a sound carrier of 41.5 Mc/s and a vision carrier of 45.04

Mc/s are used, as described in Section 6.5, then V, and V8 can be
tuned to 41.5 Mc/s (with a crystal frequency of 13.833 Mc/s) in the
sound modulator, in which case V, and V8 can be omitted in the
vision modulator. Since the modulating frequency is very much
lower in the sound modulator, economy of H.T. current can be
effected by replacing V4 by a Z77 and increasing R to 22ka In this
case an additional fixed resistor of 33142 must be included in the
top leg of P1 so that the correct bias can be set for V5.
6.7. Divider chain
A suitable circuit for deriving the sound/vision carrier beat from
the sub -carrier frequency is shown in Fig. 6.8.
The C.W. sub -carrier frequency is divided by 3 in the regenerative
divider V1 and then multiplied by 4 in the multiplier V2. The two
coils forming the transformer anode load of V1 may be wound on a
in. former with a dust core, and sufficient coupling between primary
and secondary is obtained if the two coils are wound adjacent to
each other on the former. The primary is tuned to of the input
cis
1(l'IN-I)fcts=
- I) TIMES
MULTIPLIER
INPUT
OUTPUT
FREQUENCY MIXER
FREQUENCY
fC/S LC/S
N
LC/S
Fig. 6.9. Principle of regenerative divider
frequency. The anode load of V2 is tuned to of the input fre-

quency and therefore amplifies the fourth harmonic of the V1 output.
The principle of the regenerative divider is shown in Fig. 6.9.
The frequency f to be divided is fed into a mixer stage whose output
load is tuned ton of the input frequency, where n is the required
division factor. Suppose that an output fn exists, then when it is
multiplied by the (n - 1) times multiplier an output frequency f -

results. When it is mixed with the input frequency f, an output In
is obtained. Hence, positive feedback at the frequency -n maintains

continuous operation of the circuit.
In the circuit of Fig. 6.8, the multiplication and mixing processes
are carried out in the diode ring in the grid of V1. Thus, the anode
+7 DIVIDER 84 MULTIPLIER BUFFER +5 DIVIDER SQUARER & CATHODE FOLLOWER OUTPUT
+250V
4714i1. 470. 4.7krt loft 4.7th F4-7 kh.

osi,ur 0.1ALF 0 pF 2P F 2/2
.01pF
1-11 1.2MH 1-1 1 -II 1-11-
225mH
-IN
.150,01 .150mHi
2opF T68pF 68pF
90pH 150 pH) 4:1 214
Tio"rookf,
C.W. SUEICARRIER VI V4 VS
INPUT
f = 2667Bi25Mcfs 22pF 10 pF 22pF 22pF
030375 MC/5
1 --41 SQUARE -WAVE
Z 7
0.0ipF OUTPUT
47pF 4.7pF
110015
613011' 101,n.
I() 10
45754's GEX54's 100kn
TO1/2F 0IpF 010

f ir T
150fl 41kn 1015 41'0Q soft Ion 4.4n ion 2,1F 156a
CHECK POINT CHECK POINT CHECK PO NT

4f
7 . 0.3796B SMc!s 44 - 151875Mcfs -33= 03037 Mcfs
Fig. 6.10. 4/35 section of sub -carrier to twice line frequency divider chain
load of V1 is tuned to of the input frequency and the non -linearity
of the diode circuit produces a fourth harmonic of this which mixes
with the input frequency to produce a frequency component which,
being fed to the V1 grid, maintains continuous operation.
It should be noted that the phase of the transformer connections
of the V1 load should be made so that no output is obtained when the
input is removed. This ensures that the divider is operating
regeneratively and not as a synchronized oscillator.
For setting up the correct tuning of the divider circuits, it is
recommended that the input signal be displayed on one beam of a
double beam oscilloscope, while the output signal from each check
point in turn is displayed on the second beam. The correct fre-
quency relationship can then easily be observed.
A suitable circuit for multiplying the sub -carrier frequency by
as part of the division process to obtain a locked twice line
frequency, is shown in Fig. 6.10. Here, regenerative dividers are
used, the circuits being similar to those described above for the 4.
process except that the frequencies are different.
The output of the final divider of Fig. 6.10 is squared up to
4f
produce a square wave of the -35 frequency, so that an Eccles -Jordan
type of divider may be used for the final division by 15. The
regenerative type of divider tends to be unsuitable for low frequency
operation because of the low frequency but high Q tuned circuits
which would be required.
A suitable Eccles -Jordan circuit for carrying out the final division
by 15 is shown in Fig. 6.11. The division is made in two steps of
3 and 5, respectively.
The principle of the Eccles -Jordan circuit may be described as
follows:-
Considering the 3 to 1 divider as a simple example, suppose that
V1 is taking current. Then examination of the D.C. coupling
from the anode of V1 to the grids of V2 and V3 shows that V2 and
V3 will be held cut off. Now suppose a positive pulse is fed to the
cathode of V1, then a positive pulse will appear at the V1 anode and
this will cause V2 to conduct. As soon as V2 conducts, it biases
off V1 and V3 and equilibrium again occurs until another positive
cathode pulse is applied, when V3 will conduct. Thus, as the
input pulses are applied, so each valve will conduct in turn and the
output from any one valve will have a frequency equal to the input
frequency divided by the number of valves in the circuit.
The output from the circuit shown in Fig. 6.11 is A, of the sub -
carrier frequency, or 20.25 kc/s. This output can now be fed as the
+3 DIVIDER .÷5 DIVIDER
ACHIOOkr1
220kn-
IUMMIIIIIIIMUMMIIIIMINNIMII
INIMEMMEINRIIIIMIIIIMMEMIMII
''''
Hilininiaiirmatfelli
fill in 1111 111-4"ockHn
vo
220pF 220pF
rei ei e. grol ,
INPUT OUTPUT
0.30375 MC/s 2015keis
13ka IMfI 221cD.
it 113kft
221a
CHECK POINT
0.10125MC/A
Fig. 6.11. Eccles -Jordan = 15 circuit diagram. Note all screen grids are connected to H.T. line
" master oscillator " signal to a conventional mixed sync generator,
as shown in Fig. 6.1.
6.8. Colour bar generator
Suitable video sources of R, G and B inputs to the encoder may
be obtained from either a colour camera or a colour flying spot
scanner. However, since both of these items are costly, it is con-
venient to have an inexpensive apparatus for artificially generating
TRAILING EDGE
LEADING EDGE
OF LINE BLANKING
OF LINE BLANKING
1.4 ACTIVE LINE PERIOD >
ON
FRED AND BLUE
TRIGGER EDGE
GREEN
BLUE ON
.4 -TRIGGER TRIGGER
EDGE EDGE
RED
I._BLUE OFF
ON
BLUE
OFF
G+R+B G+R G+B R+ B
G R B 0+0+0
=WHITE =YELLOW =CYAN =GREEN =MAGENTA =RED =BLUE =BLACK
Fig. 6.12. Colour bar generator waveforms
suitable video signals which simulate fully saturated primaries and

their complementaries. Such a device is called a colour bar generator
and one form of it provides an R, G and B video signal equivalent
to a colour picture consisting of vertical colour stripes with full
amplitude white on the left hand side, then full yellow, cyan, green,
magenta, red, blue and black. This arrangement gives the full
amplitude colours in order of decreasing luminance from white on
the left to black on the right.
The three waveforms required to produce this picture are shown
in Fig. 6.12. All waveforms have equal peak -to -peak values and
each waveform is a square wave, but the repetition frequency of
the red signal is twice that of green and the frequency of the blue
signal is twice that of red. If the colour bars are to be of equal
widths, the " on " and " off " periods of the B waveform should
each be of the active line period (i.e. the time of one line scan
less the line blanking period), while the " on " and " off " periods
of the red and green signals should be I and 4-, respectively, of the
active line period.
Thus, during the first of the active line period, R, G and B are
each of full amplitude to give white. During the next of the active
line period, R and G are still " on " but B is " off " so that yellow
is obtained. Following this process through from left to right,
it can be seen that white, yellow, cyan, green, magenta, red, blue and
black will result.
A suitable circuit for generating the waveforms shown in Fig.
6.12 is drawn in Fig. 6.13.
Each colour channel consists of a monostable multivibrator
triggered from the line blanking input, followed by a field blanking
insertion stage, a phase inverter, and a cathode follower output.
An amplitude adjustment is provided in each output so that the
peak -to -peak values can be made equal.
The operation of the circuit may be described as follows:-
Positive going line blanking pulses are inverted by V6 and differ-
entiated so that positive " pips " are obtained corresponding to the
trailing edges of the line blanking pulses. Now referring to the
green channel, V1 and V2 are arranged as a monostable multivibrator
in which, with no signal input, V2 is normally conducting. The
positive trailing edges of the line blanking cause V1 to conduct
and this renders V2 non -conducting but since the V1 anode is capacity
coupled to V2, the non -conducting periods of V2 will last only for a
time determined by the time constant of the RC circuit between the
anode of V1 and the grid of V2, and this time can be adjusted by the
potentiometer in the grid of V2. Thus, positive going pulses are
obtained at the V2 anode whose lengths can be adjusted and whose
leading edges coincide with the trailing edges of the line blanking.
Note that corresponding negative going pulses are obtained at the
V2 cathode.
The V2 output is applied to V3 which is switched off during the
field blanking period by the negative going field blanking pulses
from the anode of V,. Thus, a negative going and field blanked
G waveform is obtained at the V3 anode and this is inverted by V,
and fed to the cathode follower output V5.
The red channel operates in a similar manner to the green, but
here the monostable multivibrator time constant is shorter. Also
the negative going green waveform at the V2 cathode is differentiated
to produce positive " pips " at the times of the trailing edges of the
green waveform. These trigger the red and blue multivibrators
+2s0V
WW1 IOkR
sin EWE
4.7kR 4.70. 0.1 F
LINE
BLANKING V6 V7
INPUT 0.1,2F
0.1pF
33pr'
IMEL
Mn
-161411
47 R
FRAME
LARKING +250
4.7 rt 220kri 47k
014
100k T1 I'81,11
VI
150kT1 220kR GREEN

OUTPUT
82p1
150 P2kR
12 An
mais ovF
+250V
4.7 n 270k11
30pF
- 1000
Vio
9
22.F ovi
001(1 100811. 10k(1
I N
2TpF
r20 IMn
47orl
471R olgr
6e,F +250V
BLUE
OUTPUT
6 Bk
Fig. 6.13. Colour bar generator circuit

which are also triggered by the trailing edges of the line blanking
pulses.
Similarly, the blue channel multivibrator has a shorter time
constant than the red. It is additionally triggered by the trailing
edges of the red waveform, as well as by the trailing edges of the
green and the line blanking waveforms.
The various triggering functions are illustrated in Fig. 6.12, and
the amplitudes and on -off ratios of the three waveforms may be
easily set up by observing each waveform in turn on an oscilloscope
synchronized to the line frequency. It must be remembered that
if equal bar widths are required, it is the active line period and not
the whole line period, which must be divided by f by the green
waveform, and by 1- and I by the red and blue waveforms, res-
pectively.
6.9. Summary
In this chapter the operations required for generating a complete
R.F. signal encoded in accordance with the British adapted version
of the N.T.S.C. system have been described in the form of block
diagrams. In addition, actual circuits have been given for each
block but, of course, any circuits which carry out the operations
indicated would be suitable.
A colour bar generator has also been described, since this provides
a useful and inexpensive source of R, G and B signals.
A considerable amount of the circuit complication in an N.T.S.C.
transmitter is due to the requirements of frequency locking between
the scanning frequencies, the sub -carrier frequency, and the R.F.
vision/R.F. sound carrier beat. Again, the encoder unit which
provides a composite video output from an R, G and B input is
complicated by the differential bandwidth limitation of the I and Q
signals. It should therefore be pointed out that an equipment
which does not provide locking between the various frequencies,
and which employs an equiband encoder, does prove adequate for
most testing procedures, especially when a colour bar signal is
available.
Thus, a simplified but extremely useful test set up would consist
of an equiband encoder, a colour bar generator, and a conventional
mixed sync generator all for providing a composite video signal,
together with a vision modulator for providing the corresponding
R.F. signal.
CHAPTER 7
Introduction to Colour Receiver

Design
7.1. Introduction
The greater part of a colour television receiver is basically similar
in operation to a monochrome set. The radio frequency signal is
selected, amplified and converted to the standard I.F. by a normal
type turret or tuner unit. The I.F. stages are similar to monochrome
practice; marked differences only appear at video frequencies.
The time bases and power supplies are like their monochrome
counterparts but in general are required to deliver higher currents.
For three -gun types of colour tubes convergence circuits are needed,
while single -gun tubes need special gating circuits for selecting the
appropriate modulation at the right time.
The greatest divergences from monochrome usage occur in the
circuits for demodulating the colour difference information carried
by the sub -carrier. Since the sub -carrier has two distinct kinds of
modulation, at least two demodulators are required. Each demodu-
lator must pick out the modulation which is occurring at one
specific phase of the sub -carrier, say 0°, and reject the modulation
which the sub -carrier has at phase 0° + 90°. Demodulators which
can do this are called synchronous demodulators as they must be
synchronized in phase with the sub -carrier.
Synchronous demodulators have to be supplied with a continuous
reference sine wave signal which they can use to judge the phase
of the incoming sub -carrier signal. This reference sine wave must
have the same frequency and phase as the reference signal used at
the transmitter when the signal is coded. The circuits which
derive this receiver reference signal from the colour burst are called
reference generators. The chrominance demodulators and the
reference generators are narrow band devices and service engineers
who are familiar with black and white receivers will experience little
difficulty in handling these new circuits.
The type of colour cathode-ray tube which the receiver uses has a
decisive effect on the exact method of signal decoding. Receivers
182
INTRODUCTION TO COLOUR RECEIVER DESIGN 183
which use three -gun tubes or which combine optically the pictures
from three separate tubes, require simultaneously, at the electron
guns, voltages proportional to the three tristimulus values R, G
and B. Other types of display require the information in a different
form which can always be derived from the tristimulus values
although it is usually more economical to tailor the receiver to suit
the display tube it is using.
Most receivers use three -gun shadow mask tubes and this is the
type of receiver which will be discussed here in greater detail.
7.2. Block diagram of typical receiver

A general schematic of the receiver sections needed to perform
the functions described in Section 7.1 is shown in Fig. 7.1.
The video signal at the detector consists of the normal luminance
waveform with mixed sync, together with the chrominance sub -
carrier, its sidebands and the colour burst signal. It is possible to
amplify these signals together but certain difficulties, to be discussed
later, can be avoided if the chrominance and luminance signals are
separated out at the detector.
The luminance signal is delayed with respect to the chrominance
signals by a delay line which is usually a piece of coaxial cable specially
made to have a low transmission velocity. The inner of the cable
is wound in a spiral and the polythene medium is loaded with ferrite
dust. This delay is needed because the chrominance signals go
through narrower band circuits than the luminance signal and have
different time delays. The centres of chrominance transitions thus
occur later than the centre of the corresponding luminance transition
unless the luminance signal is delayed to compensate (see Fig. 4.13
and Section 4.22). The luminance amplifier corresponds to the
normal video amplifier, except that the final output signal required
by the shadow mask tube is some 80-130V peak -to -peak video signal,
depending on the screen potentials. The chrominance sub -carrier
is usually rejected by this amplifier before the signal reaches the tube.
The chrominance signal is extracted from the composite encoded
video by giving the chrominance amplifier a suitable band-pass
response which accepts the sub -carrier and its sidebands but rejects
the unwanted lower -frequency luminance signals (see Fig. 7.2). Of
necessity, the higher -frequency luminance signals are also accepted
by the chrominance amplifier but there is no practical way of
discarding these luminance components, despite the frequency inter-
leaving (see Section 4.35), whilst still retaining the chrominance
components. The undesired luminance components cause some
SYNCHRONIZING
TURRET PULSE
SEPARATOR
REFERENCE GENERATOR
REACTANCE
-4. CONTROL
.A. G. C. BURST PHASE
COMMON
GATE DETECTOR 't
I. F.
OSCILLATOR
LINE
FI LD CONVERGENCE TIME
TIME BASE
SOUND, VISION BASE
I. F. I. F. E. H. T.
COLOUR
DECODER
-40.- t
SYNCHRONOUS
1'
DETECTOR
CHROMINANCE DELAY
DET. DET
AMPLIFIER
SYNCHRONOUS
0
DETECTOR MATRIX
-b.
Y' Ncs.
AUDIO LUMINANCE PURITY
LUMINANCE DELAY H MAGNETS
AMP AMPLIFIER
BLUE
LATERAL
SHIFT FIELD
MAGNET
EQUALIZER
TUBE MAGNETS
POWER PACK
POTENTIALS
L
POWER SUPPLY FOR RECEIVER
Fig. 7.1. Basic functions for a shadow mask tube receiver

spurious colour effects which are not very noticeable on colour
pictures.
The complete chrominance signal is passed to each of the syn-
chronous detectors. At least two synchronous detectors are required
and in some arrangements it is advantageous to use three. Several
types of synchronous detectors are described in Chapter 9. One
type may be thought of as a gating valve which is opened for a short
SUBCARRIER
FREQUENCY
UNWANTED WANTED
LUMINANCE CHROMINANCE
FREQUENCIES SIDEBANDS
, r r
0 I 1 , 3
FREQUENCY (MCA)
Fig. 7.2. Chrominance band-pass response
time at the same phase on every cycle of the reference voltage and
accepts the magnitude of the chrominance signal at these instants
only.
The outputs of the synchronous detectors depend on the phase
at which they are demodulating, but are ideally I' from one detector
and Q' from the other. The voltage level at which the synchronous
demodulators operate may vary from a few volts to the hundred or
so required for application to the display tube.
In general the low frequency demodulated chrominance signals
are time matched if they have different pass -bands from each other
and are then passed to the matrix circuits where, together with the
Y' luminance signal, they are recombined to form R', G' and B'.
The matrix circuits, in effect, solve the three equations
R' = Y' 0.961' 0.62Q'
G' = Y' - 0.27F - 0.65Q'
B' = Y' - 1.11I' + 1-70Q'
and derive their name from the matrix notation used in algebra as a
form of shorthand for writing and solving such equations.
R' 1 0.96 0.62 1
G' = 1 - 0.27 - 0.65 1'
B' 1 - 1.11 1.70 Q'

The matrix circuits are commonly a few resistors connected so
that appropriate fractions of the three incoming signals are added
together. Valve converters are normally used to get the right sign
of signal for the matrix equations. The three outputs from the
matrix are used to modulate the cathode-ray tube, after further
amplification if necessary.
The reference generator which provides the reference frequency
to the synchronous detectors may take several forms but the basic
functions shown in Fig. 7.1 are representative. The stable oscillator
is tuned to the sub -carrier frequency but requires to be locked very
accurately in phase as well as in frequency with the transmitter's
reference sine wave. The colour burst signal is gated out from the
composite chrominance signal and the phase of the reference oscilla-
tor is compared with the phase of the colour burst signal. Any
difference in frequency or phase between the burst and the reference
oscillator produces a voltage which is used to change the frequency
of the oscillator slightly until it pulls into phase lock.
The line time base provides about 1+ A of scanning current to the
deflection yoke and also generates the 22-25 kV required for the
final anode of the shadow mask tube. The E.H.T. must be stabilized
up to currents of 1 mA or so, since the convergence varies with the
E.H.T. voltage. The electron guns of the shadow mask tube are
electrostatically focused and the line time base also provides the
5kV focus voltage at negligible current. The field time base
generates some 500 mA of 50 c/s or 60 c/s sawtooth current for the
field deflection coils.
Voltage waveforms from both time bases are processed in the
convergence circuits to provide parabolic current waveforms
through the dynamic convergence coils of the shadow mask tube.
All these circuit functions will be described in more detail in the
next four chapters. In the rest of this chapter we shall discuss a few
general topics which arise in the design of colour receivers.
7.3. Asynchronous operation
For normal monochrome transmission in the United Kingdom the
field frequency is kept in synchronism with the mains frequency.
As the national grid frequency varies so the television field fre-
quency is varied. By this means any effects on the television picture
due to mains hum, such as brightness modulation or displacement
of the scanning lines, are kept stationary with respect to the picture,
or at least very slow moving.
The hum effects are then much less noticeable than if they are
moving up or down the picture. The most objectionable rate of
movement is about 2 seconds for a complete sweep from top to
bottom of the picture.
Such mains -locking enables appreciable economies in smoothing
to be made in every receiver. However, in colour transmission it
is desirable to keep the sub -carrier frequency constant to ease the
design of the reference generator. Since
525, 405 525,
j 8C = 11' = -2- - x -2-- 1 F
a constant sub -carrier frequency implies a constant field frequency

and it is not then practical to lock to the mains frequency.
Various suggestions have been made for overcoming this difficulty,
but the present practice is to operate asynchronously (see Section
15.2).
7.4. Crosstalk between chrominance signals

Since the I' signal is transmitted in a vestigial sideband fashion,
it produces a spurious signal 90° out of phase with itself. A simple
outline of the effect is given in Fig. 7.3.
An unmodulated carrier can be represented in amplitude and phase
by a phasor, such as OC shown in Fig. 7.3 (a). If the carrier is
modulated then its amplitude will vary so that some time later than
Fig. 7.3 (a) it is represented by the phasor OC' in Fig. 7.3 (b). A
convenient way of representing such a modulated carrier is to
consider it to be composed of the original unmodulated carrier OC,
together with its sidebands. If the modulation is a simple sine
wave, the sidebands can be represented by a pair of additional phasors
C'U and C'L, the upper and lower sidebands respectively, rotating
in opposite directions symmetrically around the end of the carrier
OC. These sideband phasors have an amplitude dependent on the
depth of modulation and rotate with an angular frequency deter-
mined by the frequency of the modulating sine wave. The resultant
of the carrier and sidebands is the phasor OD shown in Fig. 7.3 (e),
since the sideband phasors can be resolved into components along
the carrier OC and perpendicular to OC, as in Fig. 7.3 (d). The two
components perpendicular to the carrier cancel out for all positions
of the sideband phasors as they rotate about C, leaving only a
0
(0)
(e)
C F
(f) 0 (h)
Fig. 7.3. Quadrature signal clue to vestigial sideband transmission

component CD which adds to or subtracts from the main carrier
phasor.
However, if one sideband is missing, as happens with the higher
video frequencies in vestigial or asymmetric sideband transmissions,
the two components perpendicular to the carrier no longer cancel,
see Fig. 7.3 (g), and the resultant, see Fig. 7.3 (h), consists of the
carrier with only half the amplitude of the original wanted modula-
tion, CE, together with an unwanted component CF in quadrature
with it.
Since there is not room between the sub -carrier frequency and the
end of the video band to send the I' signal double sideband, the
4I.
o°
0°
If iI
203 2.01
(b) _iy=t
1.14
0.70
Fig. 7.4. Demodulator axes
higher I' frequencies, from 0.34 Mc/s to 1.0 Mc/s, are sent single
sideband. These I' modulation frequencies therefore give rise to a
quadrature signal, equal in amplitude to the remaining wanted
modulation but 90° out of phase; i.e., in the Q' phase.
The Q' signal is double sideband throughout its modulation range
of 0 to 0.34 Mc/s and does not produce a spurious quadrature signal
in the I' channel. As the I' signal is only single sideband from
0.34 Mc/s to 1 Mc/s, the spurious quadrature signal in the Q'
channel falls outside the Q' acceptance bandwidth and is therefore
rejected. In the range 0 to 0.34 Mc/s the I' signal is double side -
band and does not produce any crosstalk into the Q' channel.
If crosstalk from I' to Q' is to be avoided the Q' channel must
not accept frequencies outside its specified bandwidth of + 0.34
Mc/s.
7.5. Types of chrominance channel response
If the full chrominance bandwidth of the system is to be used, one
colour detector must demodulate along the I' axis (see Fig. 7.4), at a
phase of 123°. The chrominance frequency response for the 1'
channel should then be 1 Mc/s. The sub -carrier modulation is
vestigial sideband (Fig. 7.2) and amplitude correction should be
applied for the missing sidebands either before or after demodula-
tion. Thus in Fig. 7.5 (a), if the sub -carrier chrominance response is
flat, then the chrominance video response must incorporate a 6dB
amplitude boost to compensate for the missing sidebands. Alter-
natively the response can be shaped as in Fig. 7.5 (b) so that the
CHROMINANCE RESPONSE AT
SUBCARRIER FREQUENCIES CHROMINANCE
VIDEO RESPONSE
u.1
=
o-
20-5
x
"c 0
10 15 20 25 30
0 05 15
I
MC/S M C/S
(a)
CC
VIDEO RESPONSE
CC
CC
(-)
= o ,
ti -
2 05 EZ 0 5
0 0
10 I5 20 25 30 0'5 I 15
MC/S MC/S
(b)
Fig. 7.5. Amplitude correction for I' vestigial sideband at (a) video and
(b) sub -carrier
sub -carrier is 6dB down relative to the peak of the chrominance

response. In this case the following video stages can have a flat
amplitude response. The important thing is the resultant overall
response up to the display, and a boost in one part of the channel
can be compensated for in another stage, within limits. The
asymmetric sideband correction is often omitted and the sharper
luminance waveform is relied upon to mask the resulting chrominance
smear.
For accurate waveform reproduction the overall phase response
is required to be as linear as possible. The exact shape of the phase
response is not critical as far as the accurate reproduction of hue
in large areas of the picture is concerned.
Receivers using the full I' bandwidth of 1 Mc/s are called wideband
chrominance receivers. The majority of colour receivers do not
use the full I' bandwidth but have both chrominance channels
double sideband with the bandwidth of the Q' channel, I Mc/s.
Such receivers are called narrowband chrominance receivers or
equiband receivers. Although narrowband receivers have relatively
poor colour definition, they have certain design advantages which
tend to offset the loss in chrominance resolution. One advantage is
that the phase angles at which the colour detectors demodulate can
be chosen without regard to any unwanted interaction between the
chrominance signals and it is not essential to use I', Q' axes. If the
decoding axes are not I' and Q' it is still possible to use wider
bandwidths for both channels than the Q' bandwidth, but at the cost
of some crosstalk between the two chrominance signals. This is
called wideband equiband operation.
7.6. Choice of demodulation axes
In equiband receivers where both chrominance channels are only
± 0.34 Mc/s in width, there is no possibility of the I' signal cross -
talking into the Q' channel due to vestigial sideband distortion.
Synchronous demodulation can therefore take place along any
convenient axes. An obvious choice is (R' Y') and (B' - Y')
(R' - Y') (B' - Y')
axes, Fig. 7.4. These are strictly the and axes,
1.14 2 .03
and the gain factors must be remembered in designing the matrix
circuits. Having obtained the (R' - Y') and (B' - Y') signals
(G' - Y') is easily obtained, since
Y' = 0.30R' + 0.59G' + 0.11B'
and 0 = 0.30(R' - Y') 0.59(G' - Y') 0.11(B' - Y')
(G' - Y') = - 0.30 - Y')- 0.11
c(B' - Y')
The simple addition of the Y' signal to the three colour difference
signals (R' - Y'), (G' - Y') and (B' - Y') then gives the three
tristimulus values R', G' and B'. The display tube itself can be
used to carry out this addition by modulating the three cathodes with
the negative luminance signal and the three grids with the colour
difference signals. It is preferable not to modulate the cathodes
with the colour difference signals (see Section 118).
The (R' - Y') and (G' - Y') axes can also be used for demodula-
tion and this has the advantage that the (G' - Y') signal is received
at a relatively higher level than the (B' - Y') signal. The angle
and gain factor of the (G' - Y') axis, or indeed of any other
demodulation axis, is easily found.
The signal along the (G' - Y') axis is the resultant of the pro-
jections on this axis of the I' and Q' signals, or, which is the same
thing for narrowband signals, the projections of the (R' - Y')
and (B' - Y') signals. Let the (G' - Y') axis make an angle a
with the burst phase, Fig. 7.4, and let the magnitude of the (G' - Y')
signal along this axis be
(G' Y')-
then
(G' - Y') (R' - Y') (B' - Y')
sin a cos a
C 1.14 2.03
or
C sin a C cos a
(G' - Y')= 1.14
(R' Y') (B' - Y')
2.03
but also
(G' - Y') - -0.59 (R' - Y') -059(B'
. 0.11
- Y')
0 30. 1.14
C sin a -
0.59
and
0.11. 2.03
C COS a =
0.59
so that C = 0.70 and a = 57°.
For stability it is sometimes desirable to use three demodulators,
and various axes have been suggested for such arrangements and
will be discussed further in Chapter 9.
7.7. Cross -colour and parc
As well as the vestigial sidebands of I' interfering with the Q'
channel, various other forms of crosstalk can occur between the
components of the multiplexed colour television signal. The
subcarrier itself appears in the luminance channel and causes an
unwanted beat pattern on the picture (Section 15.4). The lumi-
nance signal components in the high frequency part of the video
band are accepted by the chrominance channel and demodulated as
if they were colour signals, when they give rise to spurious colour
effects which are called cross -colour. These effects are only notice-
able in the picture areas where there is fine detail corresponding to
video frequencies close to the sub -carrier frequency. The 21 Mc/s
block of resolution bars in Test Card C normally gives rise to such
cross -colour effects. A luminance frequency of 2.5 Mc/s is treated
by the chrominance demodulators as a chrominance frequency of
2.66 - 2.5 Mc/s = 0.11 Mc/s
so that a high frequency luminance signal causes low frequency
chrominance crosstalk. Cross -colour is dealt with further in
Section 16.10.
Similarly, any random noise in the high frequency part of the
luminance spectrum, which is normally relatively unimportant
because of its fine structure, is demodulated by the synchronous
detectors into low frequency noise in the chrominance channels,
where it is more obvious because of its coarser structure despite
the protection afforded by the approximately constant luminance
operation of the system. This heterodyning of random noise
frequencies produces annoying low frequency colour effects which
the Americans have called parc.
To remove such spurious colour effects from the monochrome
picture which the colour receiver should produce during mono-
chrome transmissions, a circuit called the colour killer reduces the
gain of the chrominance channel to negligible proportions during
black and white programmes. The colour killer does this auto-
matically, depending on whether or not there is a colour burst
present in the back porch of the line sync pulse.
7.8. Sound -sub -carrier beat and chrominance buzz
Beat effects between the vision and sound carriers, resulting in a
3.5 Mc/s pattern on the picture, are familiar enough in monochrome
reception. In colour reception there is the further possibility of the
chrominance sub -carrier beating with the sound carrier to produce
a black and white pattern at the difference frequency of
3.5 - 2.66 = 0.84 Mc/s
which is a much coarser beat pattern than the vision carrier -sound
carrier beat itself.
It is also possible for enough of the 3.5 Mc/s beat frequency to
reach the synchronous detectors, where it is demodulated to produce
a coloured 840 kc/s beat pattern. Both these 840 kc/s patterns must
be guarded against in receiver design.
The vision signal may produce a buzz in the sound channel if it
is not adequately rejected, due to the low frequency picture and
field modulation carried by the vision signal. The chrominance
sub -carrier is also interrupted at field frequency and is much closer
to the sound carrier. Receivers which have barely adequate vision
rejection in the sound channel may reproduce the familiar vision
buzz on colour transmission but not on monochrome transmission,
due to insufficient chrominance sub -carrier rejection in the sound
channel.
7.9. Colour controls
Colour television receivers have the same customer controls as
most monochrome receivers, with one or two additional knobs.
Channel selection and tuning, contrast, brightness and volume
(sound) are as on monochrome receivers. In addition there is
usually a colour control knob marked " saturation ". This controls
the gain of the chrominance channel and thus varies the purity of
the reproduced colours, the colours becoming more pure as the gain
is increased, whilst decreasing the chrominance gain eventually
results in a black and white picture.
A second colour control is also incorporated in receivers but
may only be accessible to the service engineer. The hue control
knob varies the phase of the reference generator output in the
receiver and thus varies the colour which a given phase of sub -carrier
signal will produce. For example, rotation of the hue control may
change the appearance of flesh tone from a greenish -yellow through
a normal facial tone to a purplish colour.
Other controls are available for the service engineer who instals
the colour receiver. For three -gun type tubes there are usually
three bias or background controls, one for each electron gun, with
an overriding master brightness control for the viewer. There
may also be three screen grid controls to vary the slope of the
characteristic of each gun. Occasionally there are also separate
gain controls in each R, G and B channel. The level at which' the
colour killer decides that there is no burst present in the signal may
be adjustable, as may the gate which is picking out the burst from
the composite colour signal.
Different types of display tube also call for their own setting -up
procedure, but these matters are more fully dealt with in Chapters
11 and 13, for the case of the shadow mask tube.
7.10. Power supplies and heat dissipation
The power supplies may be conventional in colour receivers, but
are required to deliver roughly twice the power output of their black
and white counterparts. At the same time, three -gun colour tubes
are very susceptible to unwanted stray magnetic fields. Trans-
formers and chokes should be placed as far away from the tube as
possible, at the back of the receiver. The fields from the trans-
formers should be kept from the tube by means of magnetic shields,
copper shorted -turns around the transformer, opposing the fields
from two transformers, and such -like techniques.
With such increased power dissipation in a 21 in. tube cabinet, the
problem of cooling is greater than in a monochrome receiver. The
components controlling colour balance and convergence are parti-
cularly sensitive parts of the receiver and due attention must be paid
to keeping the effects of heat away from these circuits.
7.11. Summary
As well as its normal monochrome functions of reproducing the
sound, scanning the display tube and amplifying and detecting the
luminance or Y' signal, a colour receiver has three new basic func-
tions to perform.
It must generate a continuous reference sine wave locked in
frequency and phase to the transmitted colour burst. It must use
this reference voltage to demodulate at least two chrominance signals
from the sub -carrier signal and it must then matrix the two chromi-
nance signals with the luminance signal to produce the final red,
green and blue modulating voltages which the display tube needs.
Certain types of display tube need additional circuits for their
operation but these vary from tube to tube.
The multiplex technique used in colour transmission admits the
possibility of various forms of crosstalk between the five distinct
signals which the composite television signal contains, namely: the
luminance, the sound, the synchronizing signals and the two chromi-
nance signals.
To avoid crosstalk from I' to Q', chrominance demodulation
must be either along the I' and Q' axes or both chrominance channels
must be restricted to the narrow Q' bandwidth.
CHAPTER 8
Colour Receiver Amplifiers

8.1. Introduction
The incoming R.F. signal is received and amplified by the colour
receiver very much as in a monochrome receiver. The vision
carrier and its colour sub -carrier are treated as one signal throughout
the R.F. and I.F. stages, whilst the sound carrier is separated out
in the I.F. amplifier and handled exactly as in black and white
television practice. The chrominance sub -carrier signal is filtered
from the vision signal either just before, at or soon after the lumin-
ance I.F. detector, and is further amplified before being synchron-
ously detected to provide the low frequency colour -difference
information. The luminance amplifier is similar to the video
amplifier in conventional receivers.
8.2. R.F. tuners
Any well designed monochrome tuner unit will function satis-
factorily on colour transmissions.
All the extra colour information
is carried in the same bandwidth as the normal monochrome
transmission and no extra circuits or considerations are involved.
However, whereas liberties can often be taken with the amplitude
and phase response at the higher video frequencies without producing
very noticeable effects on monochrome pictures, such distortions
in colour reception will affect the colour sub -carrier and its chromin-
ance signals. After the sub -carrier has been demodulated these
distortions become low frequency chrominance distortions rather
than high frequency luminance distortions. The lower frequency
makes the chrominance waveform distortion more obvious on the
reproduced picture but this is countered by the masking effect of the
superimposed luminance waveform. Since the relative amplitude
of the luminance carrier to the sub -carrier is conveying colour
information about the purity of the transmitted colour, the ampli-
tude response should avoid peaks around the sub -carrier frequency.
Most monochrome receiver turrets are sufficiently flat and broad
in their responses to be quite acceptable for colour.
The noise factor requirements of the tuner are approximately
the same as for monochrome reception since the system operates
196
COLOUR RECEIVER AMPLIFIERS 197
at nominally constant luminance. A good noise factor is still
required as even in noisy fringe areas most observers feel that the
chrominance signal adds something worthwhile to the television
picture, providing they are viewing from more than four times
the picture height.
8.3. I.F. stages
The design of I.F. stages for colour receivers tends to be rather
more critical than for black and white receivers, although the same
principles apply.
The basic requirements are that the I.F. amplifier and video
stages together shall have a flat amplitude response over the pass -
band of 0 to 3 Mc/s, with a linear phase response. Since the trans-
mission is vestigial sideband this implies that either the I.F. or video
stages must attenuate the low frequencies, as in monochrome
receivers. The vision I.F. stages must reject the sound signal, and
the sound I.F. stages must reject both the vision and chrominance
signals. Protection against the signals on adjacent channels must
also be provided. There must be no cross -modulation between the
vision, sound and chrominance signals, and the input-output
characteristics must be linear. With the types of " minimum phase "
circuits which can economically be used in mass-produced receivers,
compromises are necessary between these requirements.
The necessity of providing adequate rejection at sound and
adjacent sound frequencies makes it difficult to keep the vision
phase response linear at the edges of the pass -band. The situation
at the vision carrier position is the same as in monochrome recep-
tion but it is more critical at the sub -carrier end of the band. The
transmitted waveform is pre -distorted to correct for the phase
response of the average receiver around the sub -carrier and this
eases the receiver design.
If the overall amplitude response is flat and the overall phase
response is linear, then two advantages accrue.
Firstly, there is no distortion of the Y', I' or Q' waveforms, and
hence the colour transitions in the picture are not distorted. Sec-
ondly, there is no crosstalk between I' and Q' signals which could
cause an I' signal to give rise to a spurious signal in the Q' channel
and produce the wrong colour at the edges of coloured picture
areas.
In general, phase distortion at the sub -carrier frequency will not
cause incorrect hue reproduction in large areas of the picture.
It is important to keep clear the distinction between phase angles
in the phase -frequency characteristic of the receiver, and the phase
14
angle between the reference burst and the chrominance signal
carried by the sub -carrier. It is the latter which determines the
hue of the reproduced colour. The colour in large areas is not
affected by phase distortion in the receiver response at 2.66 Mc/s,
i.e. by the time delay between the sub -carrier and the vision carrier.
For large areas of the same colour the chrominance signal is at
the same frequency as the reference burst and any phase distortion
in the receiver applies to both equally and does not affect the differ-
ence in phase between them, providing both burst and chrominance
traverse the same paths through the receiver circuits. For example,
if an all -red picture is being transmitted a colour receiver will dis-
play red no matter how distorted the phase response of the signal
channel, because both the reference burst and the chrominance
operate at the same point on the receiver's phase -frequency character-
istic. Even when the burst and signal follow different paths, the
probability is that the phase difference between them will be con-
stant and can be taken up on the hue control.
However, phase distortion in the receiver can give rise to I'Q'
crosstalk. For simplicity, consider a chrominance sub -carrier E
which has the phase of the I' component.
E = Eo cos (cot + 33°)
I, I. D.S.B.
EuVi.EL
II II I
cui I
EL //6
EL
I I 0 D.S.B.
---J-
KIEL
D.S.B. S.S.B.
ED Eo a
(a)
I
2 Eu
AMPLITUDE DISTORTION PHASE DISTORTION

(b) (c) (d) (e)
Fig. 8.1. I'Q' crosstalk due to amplitude or phase distortion. The
chrominance sub -carrier vector is E = Eo cos (cot + 33°) with an upper
sideband vector Eo cos [(w + a)r)t + 33°] and a lower sideband vector
EL cos [(w - wr)t + 33°]
and upper and lower sidebands
Eu cos [(to + wr)t + 33°]
and
EL cos [(w - (.0r)t + 33°]
representing amplitude modulation of E at a modulation frequency
of wr as in Fig. 8.1. It is clear that both the sideband vectors will
,,,,,e5UBCARRIER
(a)
Fig. 8.2. Enhancement of the

luminance amplitude due to over -
amplification of the sub -carrier
ZERO
CARRIER
(b)
have components along the Q' axis but that these will always be
equal and opposite so that there is no crosstalk into the Q' channel,
over the frequency range which the Q' channel accepts and for which
I' is double sideband. Any double sideband signal which does not
have the sidebands equal and symmetrically disposed about the
demodulation axis will give rise to crosstalk; the crosstalk wave-
form will differ in shape from the original waveform since all the
components suffer a 90° phase change irrespective of their fre-
quency. The sufficient condition for no crosstalk is that the receiver
amplitude response characteristic should be symmetrical about the
sub -carrier frequency while the phase characteristic must be skew -
symmetrical. For the I' and Q' waveforms to be undistorted it is
also necessary, of course, for the amplitude response to be flat and
the phase response linear.
The preceding remarks apply to the overall response of the I.F.
and chrominance channel up to the synchronous detectors. It is
possible for the I.F. response to depart from the ideal characteristics
and to compensate for this in the chrominance amplifier. There is
some advantage in reducing the level of sub -carrier before the lumin-
ance detector and incidentally rejecting the sound carrier more
effectively, whilst correcting the response later in the receiver.
Reducing the sub -carrier level effectively reduces the depth of
modulation and hence the single sideband distortion. It is important
that the sub -carrier amplitude should not be emphasized in the I.F.
response. An increase in sub -carrier level may well carry the video
waveform on the saturated low luminance colours, such as red and
blue, down to the zero carrier level as shown in Fig. 8.2, with con-
sequent rectification of the sub -carrier and enhancement of the
luminance level.
The cross -modulation requirements are similar to those for
monochrome receivers but can give rise to further intermodulation
effects such as sub -carrier buzz on the sound channel. Gain control
is similar to black and white practice but it is important that it
shall not change the response shape and upset the carrier to sub-
carrier ratio (see also Section 8.6).
8.4. Sound rejection
The sound rejection in the vision I.F. must be sufficient for the
840 kc/s difference frequency beat between the sub -carrier and sound
carrier, which can be produced at the luminance detector, to be
imperceptible on the picture. The worst case, when the 840 kc/s
beat is strongest relative to the luminance signal, is when the sound
carrier is modulated 100 % and the picture is displaying a saturated
red at maximum amplitude. Experiments indicate that a minimum
of 37 dB of sound rejection is required, which is rather less than most
monochrome receivers incorporate anyway. This is another
instance where the colour transmission is compatible with mono-
chrome reception. For systems using F.M. sound where the sound
carrier is not locked in frequency and hence the sound sub -carrier
beat is not an odd multiple of half the line and field frequencies,
some 20 dB more sound rejection is required.
Normal type monochrome sound rejection circuits have been
used satisfactorily in colour television receivers, as in the circuit
of Fig. 8.3. R.C.A. have developed an interesting sound rejector
circuit called the bifilar-T trap and this has been used in American
and British receivers. The bifilar-T circuit is illustrated in Fig. 8.4.
L1C1 is the trap circuit proper and determines the frequency of
rejection whilst the series bifilar elements resonate at a midband
frequency. Theoretically the circuit is a non -minimum phase
type of circuit and the phase response is rather better than a normal
+H.T.
200V
2.2kn.
oooltcF
SOUND
TAKE OFF TO LUMINANCE
DETECTOR
47ka
2.2 pF
F-40.
TO CHROM !NANCE
DETECTOR
X0000
23pF
FROM
TURRET
80pg,
ql
0.00IpF 0.001,uF
ITT kn
oolp.F 0.001µF 0 00IpF 3001/..LF

0.00IyF
270n
I 4 A.G.C.
Fig. 8.3. I.F. circuits of the G.E.C. colour receiver TTIV

4L
TUNED TO
REJECTION
/FREQUENCY
(a) ( b)
Fig. 8.4. Bifilar-T trap (a) and equivalent circuit (b)
I VISION I.F. 2^4 VISION I.F. 3,11 VISION LUMINANCE
+Fit
DETECTOR
ty. SEX 34
L
TURRET 4
TUNER
GL
AMPL IF I EN. +.4 00000

OSCILLATOR 6 I
6 iff 11
N MIXER)
DELAY
NE
T
SOUND
REJECT ION
TRAP
ADJACENT -CHANNEL TO CHROMINANCE

SOUND REJECT ION TRAPS AMPLIFIER
A G.C. LINE
H.T.
L.S.
A.G.C. DELAY & SOUND SOUND

I It SOUND I.F. Og SOUND I.F. NOISE LIMITER AMPLIFIER OUTPUT
N.5. I.F. and sound circuits of the Murphy colour receiver

type circuit, but the inevitable losses in the circuit and the fact
that oscillator drift makes it difficult to keep the sound I.F. exactly
at the peak rejection, tends to nullify some of its advantages. A
further trap can be incorporated in the same circuit for rejection
of the adjacent sound frequency. Fig. 8.5 shows how they are used
in a Murphy receiver.
8.5. Vision I.F. detectors
A single diode can be used for demodulating the multiplex I.F.
vision signal, as in Fig. 8.5, and the design of this stage can be quite
normal providing care is taken to keep the overall video response
INPUT OUTPUT
ow 10- t. 10
<0.5 < 05
0 'it - 0
TIME TIME
"" I.0 g 1.0-

0_
E
< 0.5 .4 0.5
1.1.1
I=
TIME
- 0
TIME
CONSTANT AMPLITUDE LUMINANCE CI 1.0

STEPS AND CONSTANT AMPLITUDE
SUBCARRIER
LEVELS BEFORE
C 05 FILTERING
d
TIME
AFTER FILTERING
SUBCARRIER HARMONICS
Fig. 8.6. DiJjerential amplitude distortion
flat to 3 Mc/s. The input impedance of diode detector circuits

tends to increase over the single sideband region of the signal.
This can boost the sub -carrier with respect to the carrier and may
cause the sub -carrier peaks on saturated low luminance colours to
extend down to the zero carrier level, and in extreme cases may cause
enough distortion to enhance the luminance.
Amplification of the complete video signal as one entity raises
difficulties in the design of the video amplifiers and there is some
FUNDAMENTAL HAS PHASE SHIFT
RELATIVE TO ORIGINAL SINE WAVE
(o) TIME (b)
Fig. 8.7. Differential phase distortion with diagonal clipping

shown at (b)
VIDEO AMPLIFIER 2nd VIDEO AMPLIFIER
+ 30 0 V
+385V
B2pF
2.7kn.
TO SYNC 1-* -TO CHROMINANCE
A.G.C. AMPLIFIER
120/2H1
I kfl VISION -SOUND
CARRIER
680pF BEAT TRAP
545pF
suon
VI
6AWI3
9 I Oil
680pF
3 3 fl
22kil 180011 56kfl
2/2F
V2
I2BY7A 0.01/2F
8.2kfl
DELAY LINE 0.1µF
120kfl
8 11
140/2H
6130k11 VERTICAL
22pF BLANKING
I 5 kfl
IOOpF
{Iskn I 8k
250k11
470kfl 0GIALF
100n
10.01/a I T330pF
T'200PF
-40V -I5V
Fig. 8.8. Video circuits of the R.C.A. colour receiver CTC 5N

advantage in using two detectors, one accepting the full luminance
bandwidth and the other the chrominance bandwidth only. The
use of two detectors also overcomes difficulties due to reflections in
the luminance delay cable affecting the chrominance response. An
example of this kind of technique is shown in Fig. 8.3.
8.6. Luminance amplifiers

If there is only one vision detector, and the chrominance and
luminance signals are amplified together in one or more video stages,
then the input-output linearity of the stages becomes much more
important than in black and white reception. The sub -carrier signal
rides up and down on the luminance signal as the latter varies in
amplitude (see Fig. 4.2) and it is important that neither the normalized
amplitude nor the phase of the sub -carrier signal shall depend on the
luminance level at which the signal passes through the video stage.
If this sort of distortion does occur then, for example, the colour of
actors' clothes will vary as they emerge from the shadows into the
brightly lit parts of the scene and will also change with the setting
of the contrast control.
When the normalized amplitude of the subcarrier varies with the
luminance amplitude it is called differential gain distortion. Of
course, for a given purity of colour the sub -carrier amplitude should
change with luminance level but the ratio of sub -carrier amplitude
to luminance amplitude should be constant for any particular
colour however much light is falling upon it. Fig. 8.6 illustrates
what happens. The sub -carrier signal is stretched non -uniformly
from the positive peaks to the negative peaks. The mean level
of the resultant signal is then different from the true luminance
level, and since the harmonics of the sub -carrier are rejected by the
video pass -band the final peak -to -peak fundamental amplitude of the
sub -carrier may be different from the true value. Both these effects
can change the purity of the reproduced picture.
If the phase of the sub -carrier varies with the luminance level
then the effect is termed differential phase distortion. Such effects
can occur due to diagonal clipping as in Fig. 8.7, or in the I.F.
amplifiers due to variation in bias altering the amount of feedback
in the circuit. Similarly, a variable cathode resistance used as a
control in the luminance amplifier can cause differential phase
effects due to the shunt capacitance to ground across the cathode
resistor.
Fig. 8.8 illustrates the first video amplifier in an R.C.A. colour
receiver. The first valve is used to split the chrominance signal
ISL LUMINANCE LUMINANCE
STAGE STAGE
+400V
+230V
DELAY LINE
CATHODES
J00 OF C.R.T.
SYNC
SEP
SPOT
LIMITER
Fig. 8.9. Video circuits of the Murphy colour receiver
from the luminance signal, and the output from the detector is
applied between grid and cathode of the valve. In this way the
amplification of the valve is divided between the chrominance and
luminance channels. The Murphy receiver of Fig. 8.9 separates
the signals immediately after the detector, whilst a G.E.C. receiver
uses separate detectors, Fig. 8.10.
Once the chrominance has been separated out the luminance
signal is delayed so that its waveforms will be in time coincidence
with the waveforms from the narrow band chrominance circuits
(see Section 4.25). The delay is usually accomplished by delay
cable but lumped low-pass filter circuits or all -pass bridged -T
phase -equalizing circuits may be used. The delay required depends
upon the chrominance bandwidth and shape but is typically about
0.6 to 0.8 /is and requires about 1 ft of delay cable.
The termination of the delay cable is important if reflections and
the production of spurious echoes on the picture are to be prevented.
The constant resistance network termination of Fig. 8.10 is useful
in this respect.
The luminance signal is carrying the 2.66 Mc/s sub -carrier which
is an unwanted interfering signal in the luminance channel. Al-
though the choice of sub -carrier frequency as an odd multiple of
half the line and field frequencies helps to reduce the visibility of
the dot pattern, it can be disturbing and can produce an objectionable
beat pattern with the dot structure of the shadow mask tube screen.
Further, because the input-output characteristic of the display tube
is curved the sub -carrier gets partially rectified by the tube, as shown
in Fig. 8.11, and the resulting D.C. component of the signal decreases
the saturation of the colours. On monochrome receivers this
can be an advantage (see Chapter 15) but it is to be avoided in
colour reception. The luminance amplifier therefore includes a
sub -carrier rejector circuit, which is adequate to remove the dots
from the large areas of the picture without unduly upsetting the
luminance transient response. On Test Card C the sub -carrier
rejector circuit reduces the contrast of the 2.5 Mc/s resolution bars
but hardly affects the 2 and 3 Mc/s bars. Since the sidebands
of the sub -carrier are similarly accepted, dots still appear in small
coloured areas and at the edges of larger ones.
Even though it is not carrying chrominance, the luminance
linearity is still important. Any curvature of the input-output
+400V
+ 2oo V
2.2 kIL
)TO SYNC
& A.G.C.
R.F.C. 10 pF
41011
C.R.T.
CATHODES
410.0.
GE% 35
LZ 319
R.F.C. 33011
411dt
6.8pF
1..32000 N 709
15011 .1'00 pF
R.F. C. 27k11
-"\AAAA.e.-4-
2.2 pF gro DELAY CABLE FRAME
BLANKING
+200V
10pF
10811
z SPOT.
SEX 54
LIMITER
ISUBCARR I ER
Ikn REJECTOR
:100pH 250#11
LAST R.F.C. =025,0 680 pF
I.F. 42on 2201-1-
BRIGHTNESS
TO CHROM I NANCE
100k11
isoV
Fig. 8.10. Video circuits of the G.E.C. colour receiver 77'1V

characteristic will produce the usual monochrome effects of incorrect
tonal gradation and will cause variation of the saturation values.
The Y' signal after suitable wideband amplification goes to the
matrix circuit. Usually the matrixing may be performed by the
cathode-ray tube itself, in which case a final output voltage of about
120V peak -to -peak video may be needed.
A spot limiter may also be incorporated in the luminance channel
and this will normally be sufficient because of the constant luminance
principle, although it is possible to spot limit each R, G and B
channel separately.
8.7. Chrominance amplifiers
Chrominance amplifiers have no counterpart in monochrome
receivers. They are essentially video bandpass amplifiers accepting
the sub -carrier frequency of 2.66 Mc/s with its sidebands and reject-
ing the lower luminance frequencies. The input to the chrominance
LIGHT
OUTPUT
GRID - CATHODE CORRECT

VOLTS LUMINANCE
LEVEL
SUBCARRIER
SIGNAL
LUMINANCE
FREQUENCY (MC/S1 (b) SIGNAL
(a)
Fig. 8.11. Rectification of the sub -carrier by the display tube.

(a) shows the sub -carrier rejection with a notch bandwidth of 0.4 Mc/s
at 3dB. (b) shows the display tube transfer characteristic
amplifier may be taken either from a separate detector or from a

point in the luminance channel. The later the take -off point the
less gain is needed in the chrominance channel but the greater is the
risk of differential phase and gain effects. A typical chrominance
channel response is shown in Fig. 8.12.
From the point of view of frequency response there are two
alternative designs for the chrominance channel. If full use is to
be made of the I' information then the chrominance bandwidth
must be wide enough to accept not only the ±0.34 Mc/s over which
both I' and Q' are double sideband, but also the 0.66 Mc/s or so
over which I' is single sideband. This means a total bandwidth of
approximately 1.3 Mc/s. The overall response, including the R.F.
SUBCARRIER
I0
0.8
04
02
0
2 3 4
FREQUENCY (MCA)
Fig. 8.12. Typical chrominance response for an equiband

receiver. The amplitude is measured between aerial and
colour tube grids
BURST OUTPUT
+400V
+200V
47 kn.
o TO
o DEMODULATORS
TO LUMINANCE
O
COLOUR
KILLER
22pF GEO 54 1,000 pF ,000 F
Z77 33krl
IMn
100n: 4,700pF T4,700pF
3.5Mcis REJECTOR
A.C.C.
Fig. 8.13. Chrominance amplifiers in the G.E.C. colour receiver TTIV

and I.F. amplifiers, should be flat over the double sideband region
and have a 6 dB boost over the 0.6 Mc/s region of single sideband
operation in order to correct for the missing sideband. Alter-
natively, correction can be applied in the I' channel after synchronous
detection. Some I'-Q' receivers omit this amplitude correction
altogether, partly because it emphasizes the cross -colour from the
+H.T.Z
+Hit
FROM
DETECTOR
A.C.C. &
COLOUR
KILLER
BURST OUTPUT
Fig. 8.14. Chrominance amplifiers in the Murphy colour receiver
lower frequency luminance components which are accepted by the

chrominance channel. The lower frequency limit of the I' channel
is 1.7 Mc/s approximately, and there is a considerable amount of
energy at this frequency in the average luminance signal.
For equiband operation the chrominance acceptance band can
be narrower, +0.34 Mc/s, whilst wideband equiband receivers
may use +0.5 Mc/s, again without boost for the missing sideband.
Figs. 8.13 and 8.14 show two British examples of chrominance
amplifier design. The G.E.C. design is a wideband equiband
channel while the Murphy receiver is for equiband operation.
Wideband I'Q' receivers should incorporate a delay in the I'
channel to ensure registry of the wider band 1' signal with the Q'
signal.
A manual gain control is normally fitted to one or more of the
chrominance amplifier stages to control the reproduced purity of
LY,-111
the colours and this control is usually labelled saturation. Often,

to avoid changing bias conditions, a potentiometer is used to vary
the signal level at the input to the amplifier. The contrast control
operating on the gain of the I.F. stages varies both chrominance
and luminance together so that the final purity of the reproduction
is not changed, only the peak brightness of the picture is altered.
Some form of automatic chrominance control (A.C.C.) is often
fitted, to keep the chrominance gain constant at the predetermined
manual setting as the receiver is switched from one channel to
another, or as it fades in fringe area reception. This A.C.C. voltage
can conveniently be obtained by measuring the burst amplitude, just
as the A.G.C. voltage can be obtained by measuring the sync pulse
amplitude. The derivation of the A.C.C. voltage is discussed in
Chapter 10, as also is the production of the colour killing voltage.
The chrominance amplifier has to feed both or all the syn-
chronous detectors and also the reference generator. If the feed to
the reference generator is after the point at which the colour killing
is applied, then the colour killing voltage must be cancelled during
the back porch so that the colour killing circuits may judge the
presence or absence of the burst.
A completely automatic chrominance gain control voltage,
which sets the chrominance gain to the value required for correct
colour reproduction, can be obtained by comparing the peak -to -peak
amplitude of the burst with the height of the sync pulse. These
two amplitudes should be the same at equivalent gain points in the
receiver. The sound carrier must be firmly rejected in the chromin-
ance amplifier, since a sound signal which is 40 dB down at the
luminance and chrominance detectors will produce sufficient
amplitude of 3.5 Mc/s for it to be amplified by the chrominance
channel and then beat, at the synchronous detectors, with the 2.66
Mc/s reference frequency. The net result is the production of an
840 kc/s difference frequency which causes a coloured beat pattern
on the display tube. It is customary therefore to put a 3.5 Mc/s
rejector circuit in the chrominance amplifier.
The required voltage output of the final chrominance stage
depends upon the type of synchronous demodulator used. For
high level demodulation an output of some 120V may be requiied,
whereas for low level demodulation a few volts may suffice. Low
level demodulators require further video amplifiers to raise the level
of the video signals to the 120 volt level required by the shadow
mask tube. If these video stages are amplifying colour difference
signals then they can have a narrow frequency response. If the
signals are matrixed to R', G' and B' before further amplification,
each video amplifier must have the full video bandwidth and the
gains of the three channels must be accurately stabilized. Any
variation in gain of- a primary signal amplifier produces noticeable
changes in hue, saturation and colour balance. Changes in gain of
a colour difference amplifier produce only changes in hue and
saturation and leave the grey scale balance unaltered (see Chapter 4).
For this reason it is usual to leave the signals in colour difference
form until the last stages of the receiver.
For 100V of luminance drive the colour difference amplifiers
must be able to produce these maximum signal excursions
(R' - Y') = +70V
(G' - Y')= ±41V
(B' - Y')= ±89V
If the display device has phosphors with different efficiencies,
these figures must be suitably modified (see Section 11.8).
8.8. Summary
Colour tuners and I.F. strips are basically similar to good quality
monochrome units and careful design is enough to prevent the
vision carrier, sub -carrier and sound carrier from intermodulating
unduly. Phase distortion at the sub -carrier frequency will not
affect the colour reproduction of large picture areas if the chromin-
ance and burst traverse similar paths in the receiver. For good
colour transients a linear phase and flat amplitude response are
needed throughout the whole video pass -band.
Inadequate sound rejection can produce both a luminance only
or a coloured 840 kc/s beat pattern on the received picture. Inade-
quate vision rejection in the sound channel will allow the sub -
carrier to produce an audible low frequency buzz.
To avoid differential gain and phase effects the video luminance
and video chrominance signals are preferably amplified separately.
Sub -carrier rejection is needed in the luminance channel and 3.5 Mc/s
rejection in the chrominance amplifier. The Y' and I' signals must
be delayed to ensure time coincidence with the Q' signal. If equi-
band working is used, only the luminance signal need be delayed.
Improved colour stability results if the chrominance signals are
left as colour difference signals as long as possible.
A saturation control to vary the gain of the chrominance amplifier
is required, and the gain of the chrominance channel must be reduced
to zero, or killed, when the transmission is in monochrome.
CHAPTER 9
Colour Receiver Decoding Circuits

9.1. Introduction
The purpose of the decoding circuits in a colour receiver is to
derive, from a composite video input, the red, green and blue signals
for driving the display tube.
Normally, the composite video input is obtained from the second
detector which demodulates the I.F. signal. The usual practice is
to feed the detector output into two channels, one being the lumi-
nance channel whose output is the luminance or r signal and the
other being the chrominance channel which accepts the modulated
sub -carrier signal, synchronously detects the sub -carrier, and provides
an output of either I' and Q' signals, or difference signals. The
combination of the Y' signal and the I' and Q', or difference signals,
in a matrix circuit results finally in the required R', G' and B' signals.
Because the Y' channel of a colour receiver is almost identical
with the usual video channel of a monochrome receiver, it has become
common practice to refer to the processing of the modulated sub -
carrier signal to yield chrominance signal outputs as the decoding
process, since this is peculiar to colour receivers. It may therefore
be stated that a colour receiver decoding circuit performs two
functions: synchronous detection of the modulated sub -carrier
signal and simple algebraic manipulation (if necessary) of the
resulting outputs to give the difference signals (R' - Y'), (G' - Y')
and (B' - Y') and ultimately the primary signals R', G' and B'.
This latter process is usually known as matrixing.
9.2. Synchronous detection
The modulated sub -carrier consists of two amplitude modulated
carriers which have the same frequency but a 90° phase difference
between them. The process of synchronous detection enables the
modulation on one of the carriers to be detected without interference
from the modulation on the other carrier and consequently by using
two separate synchronous detectors both sets of modulation may be
recovered independently.
Normally, one carrier is modulated with the I' parameter and the
other with the Q' parameter, though it will be seen later that in some
15 213
(a)
(b)
21:
(c)
Fig. 9.1. Principle of synchronous detection. o, 27r sub -carrier

frequency. (a) shows I' component of sub -carrier, (b) the Q' component
of sub -carrier and (c) the total sub -carrier signal
COLOUR RECEIVER DECODING CIRCUITS 215
cases this arrangement is equivalent to quadrature modulation by
the (R' - Y') and (B' - Y') signals.
As an example of the synchronous detection process, assume that
the I' and Q' signals are simple D.C. values. Then the modulated
sub -carrier will have two components: the first will be a sine wave
of sub -carrier frequency having an amplitude proportional to the I'
signal, and the second will be a similar sine wave having an amplitude
proportional to the Q' signal but with a phase lagging 90° on the I'
component phase. Note that, if viewed on an oscilloscope, the
above modulated sub -carrier would appear as a simple sine wave
because the addition of two equal frequency sine waves 90° apart
in phase is itself a sine wave. Thus, the I' component of the sub -
carrier is shown in Fig. 9.1 (a), the Q' component is shown in Fig.
9.1 (b) and the resulting modulated sub -carrier is shown in Fig. 9.1
(c), for a modulated sub -carrier which has 2 units of I' and 1 unit
of Q'.
Note that since the I' and Q' carriers are 90° apart in phase, at the
instant when either is at a positive or negative maximum, the other
is at zero. Referring to Fig. 0.1 (c), the I' parameter may there-
fore be recovered by measuring the instantaneous value of the
modulated sub -carrier at the particular instant when the I' carrier is
at a maximum and the Q' carrier is zero. This time instant is
labelled " A ". If the subcarrier is measured at instant " B" then
a negative version of the original I' modulation will be obtained.
Similarly, Q' may be detected by measuring the modulated sub -
carrier at the instant " C ", and a negative version of Q' will be
obtained by measuring at the instant " D ".
While a simple case of D.C. values of I' and Q' has been taken
here, it obviously follows that if I' or Q' or both have A.C. com-
ponents, then these can be recovered by the same process of measur-
ing the modulated sub -carrier at the appropriate jnstants in time.
An essential feature of synchronous detection is the measurement
of the instantaneous amplitude of the sub -carrier signal at the
appropriate time instants. This timing information is obtained
from a continuous, unmodulated sine wave which is locked in
frequency and phase to the originating sub -carrier oscillator at the
transmitter.
This continuous sine wave is referred to as the reference signal,
and the colour receiver must generate it from the colour burst or
short 9 cycle sample which is transmitted during the post sync
blanking period. Reference generation is discussed in detail in
Chapter 10, but for the purposes of this chapter it should be assumed
that a reference signal is available at the receiver.
In synchronous detector arrangements, it is usual to specify the
detection axis along which detection takes place. The angular
position of this axis is dependent on the phase of the reference
signal used in the detector. Thus, in the example above, the
parameter is recovered by detection along the + I' axis (which lags
the burst phase by 57°) and the Q' parameter by detection along the
Q' axis (which lags the burst phase by 147°).
However, it will be seen later that in equiband working detection
axes other than I' and Q' may be used, and referring to Fig. 9.2,
(RLY')
114
O DETECTION AXIS Fig. 9.2. Derivation of out-

put of synchronous detector
operating along a particular
detection axis
BURST
PHASE (B' -Y')
203
the output from a detector operating along an axis leading the
-.
(B'2- Y')
axis by 0 is equal to the sum of the projections of the
03
.
(B'2-.03Y') (R'1-.14Y')
and the axes on to this detection axis. Thus,
the output from a detector operating along the 0 axis would be
(B' - Y')
cos 0+
(R' - Y')
sin 0
2 03 1 14
which is the same as
Q' cos (0 - 33°) + I' sin (0 - 33°)
provided the modulation frequencies involved are low enough to be
within the Q bandwidth.
9.3. Diode detection
Several different types of synchronous detector will be described
in this chapter, but by way of illustration, consider the operation of
the simple double diode detector shown in Fig. 9.3.
Suppose a sub -carrier chrominance signal of amplitude S and
phase 0 relative to the Q' axis is fed to a two to one step-up trans-
former, and suppose a reference sine wave of sub -carrier frequency
(derived from the colour burst) of amplitude P, and in the same
phase as Q', is injected as shown. Then the amplitude of the
voltage across Di will be equal to the vector sum of S and P, and
this voltage will appear as a positive voltage between A and C.
Similarly, the vector difference between S and P will appear as a
positive voltage between B and C. The voltage between A and B
will then be the algebraic difference between AC and BC. This
voltage between A and B can be calculated.
The voltage between A and C will equal the peak value of the
vector sum of S and P. Thus, referring to Fig. 9.4 (a), if 9 is the
phase angle between the chrominance input signal and the reference
signal, then A will be positive relative to C by an amount equal to
the magnitude of the vector LN.
Now suppose that P is made very much greater than S, then in the
triangle LMN, LM LN and therefore
V Ac = LN P S cos 0
Similarly, the corresponding conditions for the voltage between
B and C are shown in Fig. 9.4 (b), from which
V Bc LN P -S cos
SUOCARRIER
Fig. 9.3. Double diode -- CHROMINANCE-.-
synchronous detector SIGNAL INPUT
12 2
AI9
vpG
,"!
"(9 /s
t- J
P
(a) (b)
Fig. 9.4. Vector relations for a double diode synchronous detector
Now because the A to C and the B to C terminals are decoupled

by the capacitors, the resultant voltage between A and B will be
equal to the algebraic difference (VAC -V BC), that is
V AB P S cos 0 -P S cos 0 2S cos 0
Now S cos 0 is the component of S which has the same phase as
P, that is, the above detector measures the amplitude of S in the P
phase direction. Hence, by making the reference P phase equal to
the Q' phase, for example, the detector output will be equal to Q'.
Obviously, if the P phase is made equal to that of I' (i.e. leading
the Q' phase by 90°) the detector output will be I'. Consequently,
two identical double diode detectors as described above will demodu-
late the sub -carrier signal if both detectors are fed with the same
chrominance signal but with two differently phased references, one
in the Q' phase and one in the I' phase.
Note that a sign reversal of the detector output can be obtained
by changing the reference phase by 180°. Thus, for a phase dif-
ference of 8 between S and P the output is 2S cos 0, while for a
phase difference of (0 + 180°) the output is
2S cos (0 + 180°) = - 2S cos 0
While the above results for the phase detector are approximate in
that the P signal has been assumed to be very much greater than the
S signal, in practice the approximation is justified since even for the
severe case where P and S are equal, the maximum amplitude error
is only 1dB and the maximum phase error is only 3°, for two detectors
operating in quadrature. Normally P is made at least twice the
largest S value. It should be noted that the diode detector circuit of
Fig. 9.3 will operate satisfactorily if the sub -carrier and reference
inputs are interchanged. This may be seen by rotating the vector
diagram of Fig. 9.4 (b) through 180°, so that S rather than P becomes
the " reference vector " between Fig. 9.4 (a) and Fig. 9.4 (b).
9.4. Matrixing
When the I' and Q' parameters have been obtained from the syn-
chronous detectors, the next step is to derive the difference signals
(R' - Y'), (G' - Y') and (B' - Y') from I' and Q' and then Y'
has merely to be added to these to give the required R', G' and B'
signals. A method for deriving R', G' and B', given the outputs
from an I' and a Q' detector and a Y' signal will be described.
The equations to be satisfied are
R' = (R' - Y') Y' = 0.961' + 0.63Q' Y'
G' = (G' - Y')+ Y' - 0.281 - 0.64Q' Y'
B' = (B' - Y') +- Y' = - 1.111' + 1.72Q' Y'
Notice that the total proportion of 1' required is 2.07, that is, from
- 1.11 to + 0.96. On the other hand, for Q' the total proportion is
from - 0.64 to + 1.72, or 2.36. Since Q' has the greater range,
let us first see how to obtain the various proportions of Q we require,
namely, + 0.63 for red, - 0.64 for green and + 1.72 for blue.
Referring to Fig. 9.5 suppose the points A and B are the output
terminals of the simple double diode detector of Fig. 9.3. Suppose
the total load resistance between A and B is L. Call the total output
A
+79
+1.72
TO Ct' t
SYNCHRONOUS +0.63 b 56
DETECTOR
a+b+c=L
-0.64 C
B4
Fig. 9.5. Q' matrix circuit
+0.29
ro e 2 36
SYNCHRONOUS 0
DETECTOR
2 07 -028
II
d+e+f +g=L
-0133
Fig. 9.6. I' matrix circuit
2.36. First, find a tapping point 0 on the load L which divides L

in the ratio of 1.72 to 0-64. Then, for 2.36Q' across the whole
load, there is +1.72Q' at A and - 0.64Q' at B. 1.72Q' is needed
for blue and - 0.64Q' for green, but + 0.63Q' is also required for
red. So take another tap point between 0 and A which will give
+ 0.63Q'. Hence, three separate resistors a, b and c are required
and the value of these can be found in terms of their sum, which is,
of course, the total load L. Thus
a 1.09 b 0.63 c 0.64
L 2.36' L 2.36' L 2.36
i.e. a = 0.462L, b = 0.267L, c = 0.271L

The three required proportions of Q' are now available, and next
the three values of I' will be derived. The total Q' range was 2.36,
but the total I' range is 2.07. If the two detectors are exactly
similar, the total I' output must be taken as being 2.36 to get the
correct gain relationship between Q' and I', for Q' and I' have been
equally modulated (in amplitude terms) on to their carriers according
to the sub -carrier equation
Q' sin tot + I' cos ad
Thus, in Fig. 9.6 the total I' output is 2.36, of which only 2.07 is
required. A resistor d is therefore needed so that
d 0.29
L= 2.36
For the 2.07 range of I', we can now proceed as for Q'. Thus,
the tap point 0 will divide the 2.07 range in the ratio of 0.96 to
1.11, and the resistor values are given by
e 0.96 _f_0.28 g 0.83
L = 2.36' L 2.36' L 2.36
i.e. e = 0.406L, f = 0.118L, g = 0.351L.
The various proportions of I' and Q' are now available and the
final step is to add the particular I' and Q' values, and Y'.
It should be noted that, in practice, a balanced arrangement is
desirable in order to minimize the effects of stray capacities on the
+1.09
+0.63
TO
SYNCHRONOUS 0 3.44
DETECTOR
-0.64
-1.08
4
Fig. 9.7. Balanced Q' matrix circuit
various I' and Q' outputs. Then the points 0 of Figs. 9.5 and 9.6
would become the centre taps of the detector loads. The Q'
detector load can be balanced by including an additional resistor c'
in series with c, such that
a+b=c c'
c' 1.08
This would make and the total load would have the
proportional value 3.44 instead of 2.36. This is shown in Fig. 9.7.
The I' detector load must now be modified to have the proportional
value 3.44, and this can be done by changing the resistor d so that its
output becomes (3.44 - 2.07), i.e. 1.37. Next, in order to balance
the I' detector load, d may be split into d, and d2 (so that
+0176
+096
TO 1'
SYNCHRONOUS 0
DETECTOR
- 028 3.44
4
II
-013
4
d1fe+f+g+d2= L
-011
46
.H
Fig. 9.8. Balanced I' matrix circuit
d= d2), and d2 can then be placed in series with g. The values

of d, and d2 can be calculated as follows:
3.44
The output from d, must be - 0.96 = 0.76
2
0.76
Hence, d = -344L = 0.221L
3.44
Similarly, the output from d2 must be 1.11 = 0.61
0.61
Hence, d2 0.178L
= 3.44
The balanced I' detector load is then as shown in Fig. 9.8.
The complete circuit for an I'Q' diode decoder is shown in
Fig. 9.9.Note that delay must be included in the Y' channel and
some delay is also necessary in the I' channel. The I' and Q'
detector outputs include low pass filters with cut-off frequencies of
about I Mc/s and 0.34 Mc/s, respectively.
If the synchronous detectors are transformer coupled, as shown,
the tap points 0 of the loads can be connected together and to the
Q' DETECTOR 0.34Mcls LOW-PASS FILTER
SUBCARPIER
INPUT
-*HT
IMcis LOW-PASS FILTER & DELAY
Y'DELAY----I74f7
Fig. 9.9. l', Q' diode synchronous detection and matrixing
Y' signal. The appropriate voltages can now be added in cathode

followers to give the required red, green and blue voltage outputs,
which must be amplified individually before being applied to the
display tube. The correct gain relationship between the Y' and
the I', Q' parameters can be obtained by adjusting the saturation
control which varies the drive applied to the synchronous detectors
relative to the Y' signal.
The above matrix arrangement has been described in detail in
order to illustrate the process of transformation from one set of
parameters, such as I', Q' and Y', to another set, such as R',
G' and B', and equal phosphor efficiencies have been assumed. In
general, the matrix circuit which follows the synchronous detectors
is intimately connected with the type of detector used, and it is
therefore desirable to consider the design as a complete unit. For
this reason, the detectors to be described will include the appro-
priate matrix circuits.
It is sometimes required to derive one of the difference signals
given the other two. This can be arranged by using a simple
matrix and a phase inverting stage. The relationship between the
difference signals can be deduced from the luminance equation
Y' = 0.30R' + 0.59G' + 0.11B'

by writing
0.30R'+0.59G'+0.11B'=1 Y'=(0.30+0.59+0.11)Y'
Hence
0.30R' - 0.30 Y' 0.59G' - 0.59 Y' 0.11B' - 0.11 Y' = 0
i.e.
0.30(R' - Y') 0.59(G' - Y') 0.11(B' - Y') = 0
Thus to derive (G' - Y'), add (R' - Y') and (B' - Y') in the ratio
of 0.30 to 0.11 so that
- 0.59(G' - Y') = 0.30(R' - Y') 0.11(B' - Y')
If this mixture of (R' - Y') and (B' - Y') is applied to a simple
valve amplifier with a gain of the output will be (G' - Y').
This technique is described in Section 9.9.
9.5. Decoding circuits in general

There are several general points in the design of synchronous
detectors and matrix circuits which should be mentioned.
If difference signals and the Y' signal are applied to different
electrodes of the display tube, then the Y' signal should be applied
to the cathode. This is because the difference signal circuits are
relatively narrow band, and if they are included in the tube cathode,
selective feedback may occur and upset the frequency response of
the displayed luminance. It should also be pointed out that a
matrix circuit which accepts the Y' signal and either difference signals
or I' Q' signals must be capable of handling the highest Y' frequency.
Conversely, matrix circuits which do not manipulate the I" signal
need have a frequency response which is only as wide as the widest
chrominance component, i.e. 1 Mc/s for I'Q' working, and about
0.5 Mc/s for equiband working, in the 405 -line system.
In matrix circuits, one sometimes has the choice of deriving
R', G' and B' or deriving difference signals and adding Y' in the
display tube. Generally speaking, the latter is preferable from the
point of view of white balance stability, since for white the difference
signals are zero and the arrangement is therefore independent of
gain variations in the difference signal circuits.
However, in some arrangements it may be possible for the standing
voltage outputs from difference signal detectors to drift differentially
and thereby upset the white balance, in spite of the difference signals
being zero.
Another important consideration in decoding design is the main-
tenance of the D.C. components of all the parameters. D.C.
suppression in the luminance signal will result in the wrong value
of luminance being added to the right value of difference signals, so
that errors in saturation will occur. Again, D.C. suppression in the
I' Q' or difference signals can produce undesirable hue errors in the
colour picture. For example, suppose the whole left hand half of
the picture is a saturated red, while the remaining half of the picture
is grey. Lack of D.C. response in the chrominance signals would
tend to give a cyan (the complementary of red) cast to the grey half
of the picture. The various waveforms for this case are shown in
Fig. 9.10, in which a 3dB D.C. suppression of the difference signals
has been assumed. Notice that although the peak to peak values
of all the signals are unchanged by the D.C. loss, the red signal
becomes slightly desaturated and the grey signal becomes a desatu-
rated cyan.
While the above type of signal would rarely occur in practice, it
often happens that large areas of the picture are predominantly of
one colour so that insufficient D.C. maintenance would tend to
produce a complementary hue cast in the remainder of the picture.
D.C. loss in the difference signals can occur if the D.C. bleeds
to the background controls (the three " brightness controls ") are
of too low an impedance, but usually it is possible to maintain the
D.C. to within approximately 3dB.
The Y' and chrominance signal D.C. components can be preserved
either by D.C. coupling after detection (or synchronous detection)
or by a D.C. clamping technique. The XZ detector described later
(see Section 9.11), is an example of chrominance signal clamping.
In the design of synchronous detectors, it must be emphasized
0.3 03
0 0
07
06
F -Y'
0 0
-01
0.04
0 0
-026
-03 B' -Y'
004
0 0
-0.3 -026
1.0
09
03
02
0 0
0
0
0
0.34
0.3
11' 0.04
0 0
k kIA
PURE PURE SLIGHTLY DESAT.
RED GREY DESAT. CYAN
RED
FULL D.C. 3d B. D.C. LOSS
Fig. 9.10. Hue error resulting from a loss of

3dB in D.C. component of difference signals for
full red to equal luminance grey transition
that when a synchronous detector is operated along one of the
difference signal axes, the output will contain a factor such as
1.14
for the (R' - Y') axis, and i.t:T3. for the (B' - Y') axis. Thus, to
obtain (R' - Y'), for example, a gain of 1.14 must be applied to the
output of a detector operating along the
(R '- )
axis.
1.14Y
These factors 1.14 and 2.03 are introduced at the transmitter to
limit the sub -carrier sine wave plus luminance swing so that it will
not exceed, by more than -I, the black to peak white amplitude range,
as explained in Section 5.5. In decoding circuit design, it is the
1-14 1
ratio between the factors (i.e. -2.03 or 1-) which is significant, since
the absolute value of chrominance gain can be set as required by
the receiver saturation control.
Another factor which is important in decoding circuits is the
relative drive to the three guns of the display, assuming a three -gun
tube is used. Since the red phosphor is the least efficient and the
blue phosphor is the most efficient, it is recommended by the tube
manufacturer that the drives be adjusted accordingly. For the
R.C.A. 21AXP22A tube, for example, the R, G and B drives recom-
mended are in the ratio 1 to 0.8 to 0.6 respectively. Thus, while an
equal phosphor efficiency tube would require (R' - Y'), (G' - Y')
and (B' - Y'), with a - Y' signal to each cathode, the 21AXP22A
tube requires (R' - Y'), 0.8(G' - Y') and 0.6(B' - Y'), with
cathode drives of - Y', - 0.8 Y' and - 0.6 Y' respectively. These
constants must be taken into account when designing the detector
and matrix circuits.
9.6. Equiband working
A decoding process which recovers the I' and Q' parameters
makes use of all the transmitted chrominance information; that is,
(for the 405 -line system) colour detail up to a bandwidth of about
340 kc/s is reproduced in three colours, and detail between 340 kc/s
and 1 Mc/s is reproduced as a mixture of two colours along the I'
axis, these two colours being roughly speaking an orange and a
cyan together with the mean colour of the background. Above
1 Mc/s detail, the colour reproduction is in terms of Illuminant C
white only. A receiver which makes use of all the transmitted
information is forced to use I' and Q' detection axes, so that it may
reject the crosstalking single sideband portion of its I' channel (for
modulation frequencies between 340 kc/s and 1 Mc/s) from its
Q' channel. Hence an I'Q' receiver must include a matrix circuit
arrangement to convert I', Q' and Y' to R', G' and B' for the display
tube, and generally this matrix is fairly complex and therefore
expensive. In addition, delay is required in the I' channel since the
I' bandwidth is greater than that of the Q' channel.
However, below 340 kc/s, there is no differential bandwidth
limiting and any detection axes may be used provided the chromi-
nance bandwidth above 340 kc/s is rejected by the receiver. Thus,
by having three colour reproduction for detail not exceeding 340 kc/s
bandwidth, and for finer detail than this, reproduction in terms of
Illuminant C only, together with the mean background colour,
considerable simplification of the matrix circuits is possible. In
this case, the two synchronous detectors would nominally operate
with equal bandwidths of 340 kc/s each, and such an arrangement
is called equiband detection, for obvious reasons.
It might be argued that, since an equiband receiver rejects the
higher frequency chrominance components, its resulting picture
quality must be inferior to that of an I'Q' receiver. However, it
should be borne in mind that the N.T.S.C. signal employs band-
width sharing, so that I'Q' operation would, on average, produce
more luminance to chrominance crosstalk than equiband operation.
Thus, while an I'Q' receiver would display finer colour detail, it
would also display, as colour, luminance information in the video
range from 1.66 Mc/s to 3 Mc/s, but the corresponding crosstalk
range for narrow equiband working would only be from 2.32 Mc/s
to 3 Mc/s.
A further point is that if the I' channel were to be given a 6dB
single sideband boost between 340 kc/s and 1 Mc/s, the luminance
crosstalk in the video range from 1.66 Mc/s to 2.32 Mc/s would be
emphasized.
It would seem that further experimental work is required to
determine whether l' Q' or equiband operation is to be preferred,
and, in fact, whether I'Q' or equiband modulation should be
employed at the transmitter. However, it has been found that very
good quality pictures can be obtained by using equiband detection
with a slightly greater bandwidth than the Q' bandwidth, and the
majority of American receivers employ this technique. Thus,
equiband operation for modulation frequencies up to about 0.5 Mc/s
(in the 405 -line system) gives very satisfactory results. Strictly
speaking, of course, crosstalk between the two chrominance channels
occurs for modulation frequencies between 0.34 Mc/s and 0.5 Mc/s
(the single sideband region), but it seems that the gain in colour
detail reproduction outweighs the inter-chrominance crosstalk over
this range.
9.7. Choice of axes
With equiband operation, we are not concerned with inter-
chrominance crosstalk and we can use whichever detection axes
are most convenient. This leads to a number of possibilities: for
example, detection along (R' - Y'), (G' - Y') and (B' - Y') in
three separate detectors, or detection along (R' - Y') and (G' - Y'),
with a simple matrix to derive (B' - Y'). There are also XZ detec-
tion and symmetrical detection, which are mentioned later.
The (G' - Y') axis can be deduced from the equation
0.59(G' - Y') =-- - 0.30(R' - Y') - 0.11(B' - Y')
and it is shown in Section 7.6 that detection along an axis which
leads the burst by 57°, and with a gain of 0.7, will give the (G' - Y')
signal (see Fig. 9.11).
The fact that difference signals can be derived without matrixing
in equiband operation reduces the total matrix operation to a simple
(RL.t.')
1.14
BURST (B'Y')
PHASE
2.03
57°
Y')
07
Fig. 9.11. Position of (G' - Y')

detection axis
addition of Y' to the difference signals. Even this simple matrix

can be further economized by detecting at a level which is sufficiently
great to enable the display tube to be driven directly from the
detector output. This technique is called high level detection.
In this application, the three difference signals are fed to the
grids of the appropriate displays, and the three display cathodes are
strapped together and fed with the - Y' signal. (In practice, the
cathodes are normally fed with the appropriate fractions of the Y'
signal to allow for the differing phosphor efficiencies.) The matrix
operation of Y' addition is then carried out in the display tube itself.
9.8. Mixer detection
Another form of synchronous detector which is quite extensively
used is a multiplicative type similar to a frequency changer. In a
+NJ.
LOW-PASS CHROMINANCE
FILTER OUTPUT
SUBCARRIER
CHROMINANCE 1-4-SMARR ER
SIGNAL, REFERENCE
cos tut + Q sin w t P sin (wt + 9)
Fig. 9.12. Mixer detection circuit
conventional mixer, an R.F. signal is fed to one grid of a valve while

a local oscillator signal is fed to a second grid, and the anode
current is proportional to the product of the amplitude of the two
signals and has a frequency component equal to the frequency
difference between the signals.
In a mixer type synchronous detector, the R.F. signal becomes the
sub -carrier chrominance signal, while the local oscillator signal
becomes the reference signal. This is similar to the synchrodyne
process in which the I.F. has zero carrier frequency, and yields the
modulation components directly.
Referring to Fig. 9.12, suppose a reference signal P sin (cot 9)
is applied to one grid of the mixer, while the sub -carrier signal
I' cos wt Q' sin wt
is applied to a second grid, the phase 0 being measured relative to
the Q' axis. Then the anode current is will be proportional to the
product
(I' cos wt Q' sin wt) P sin (wt + 0) = Kia
where K is a constant depending on the valve characteristics.
Now, using the trigonometric identities
sin A cos B = 4 [sin(A B) sin(A - B)]
and
sin A sin B = 4 [cos(A - B) - cos(A B)]
then
Kia= [I' sin(lot + 0) I' sin 0 Q' cos 0 - Q' cos(2wt + 0)1
The double frequency terms will be rejected by the low pass filter
so that
Kia = sin 0 Q' cos 0]
Hence, if the reference phase 0 is made 90°
Kia =2I'
2
while if 0 = 0°
Kia = Q'
2-
It follows that the detection axis of the detector is the same as the
phase of the reference voltage, so that if an (R' - Y') output is
required, for example, then the anode current will be proportional
to the (R' - Y') component if the reference phase is made equal to
the burst phase minus 90°.
For I' Q' operation, two detectors operating 90° apart are required,
while for equiband operation three detectors can be used along the
(R' - Y'), (G' - Y') and (B' - Y') axes. However, there are
several alternative axes which may be used and the particular
choice will depend on the type of matrix arrangement which follows
the detector. Suitable matrix circuits are discussed under Sections
9.10 and 9.11.
In mixer detectors it should be noted that the output is propor-
tional to the amplitude of the reference signal, so that any change in
this during the line scan, for example, would produce a differential
detector gain which would show as a saturation change from left
to right of the picture.
While the use of a valve with two control grids has been discussed
for synchronous detection, it is clear that a triode could be used if
one signal (usually the sub -carrier chrominance signal) is applied to
the grid and the second signal (usually the reference) is applied to
the cathode. This technique is described in Section 9.10 on sym-
metrical detection.
9.9. High level detection

It is possible to amplify the sub -carrier chrominance signal so that
sufficient chrominance output is obtained after detection to drive the
display tube. This method of high level detection can be used
+H.T.
DIFFERENCE
SIGNAL OUTPUT
SUBCARRIER
INPUT
REFERENCE
Fig. 9.13. High level gated
INPUT triode detection
economically only for equiband working, since considerable loss

occurs in l' Q' matrix circuits.
A popular form of high level detector is the gated triode, which is,
in effect, a grid controlled rectifier. The circuit arrangement is
shown in Fig. 9.13.
A sufficiently large amplitude of sub -carrier chrominance signal
is applied between the anode and cathode of the triode and a fixed
phase of reference signal is applied to the triode grid. The grid
time constant is made sufficiently long to bias the triode back by grid
current to class C operation, but the grid resistor is kept low so that
the triode conducts heavily during the positive peaks of the reference
voltage. At the instants of conduction the anode voltage is low,
usually about ± 25V, and is virtually independent of the H.T. line
voltage, the anode load resistor, and the applied signal.
Now suppose a coloured area having a positive D.C. component
of (R' - Y') is being transmitted. Then the (R' - Y') component
of the sub -carrier chrominance waveform will be a sine wave in the
(R' - Y') phase whose amplitude is equal to
1 14 com-
- of the D.C.
ponent, and this sine wave, together with any (B' - Y') sine wave
which may also be present, will be applied between the anode and
cathode of the triode. If the phase of the reference sine wave on the
triode grid is in the - (R' - Y') phase, i.e. 90° ahead of the burst
phase, then the triode anode will be clamped to + 25V at those
instants when the (R' - Y') sine wave is at a negative maximum.
11.-1 COMPONENT OF
SUBCARRIER SIGNAL
1(11.-Y9
D.C. COMPONENT
-----+25V
INSTANTS OF TRIODE COND CTIO
(a)
(B-Y)COMPONENT OF
SUBCARRIER SIGNAL
-+25V
(b)
Fig. 9.14. Triode anode voltage waveforms for positive

(R' - Y') and (B' - Y') components
This is shown in Fig. 9.14 (a) where the dots represent the conduction
times of the triode.
The anode waveform is therefore a sine wave whose negative
peaks are standing on a D.C. level of ± 25V. The RC filter con-
nected to the triode anode removes the sub -carrier sine wave and
leaves a positive D.C. value proportional to the original (R' - Y')
signal together with the + 25V clamping potential.
It is important to notice that any (B' - Y') D.C. components
which are present will appear as a sine wave in the anode wave-
form, but this waveform will be 90° out of phase with the (R' - Y')
sine wave so that at the instants of conduction the (B' - Y') com-
ponent will be zero, as shown in Fig. 9.14(b). Hence, no (B' - Y')
component will appear in the (R' - Y') output.
If the chrominance signal contained a negative (R' - Y') com-
ponent, the detection process would be similar except that the
conduction times would occur at the positive peaks of the
- (R' - Y') sine wave reference so that a negative D.C. voltage
measured from the + 25V clamping level would be obtained. This
case is shown in Fig. 9.15. Notice that the triode anode can go
appreciably negative. For example, if a full saturated cyan is being
transmitted, the normalized value of (R' - Y') is - 0.7, so that if
the display tube requires a drive of 100V, the red difference signal
would be - 70V.
The detection process which has been described for positive and
negative D.C. components applies equally well for any waveform
which may be transmitted. Thus, if the (R' - Y') waveform is
of the form shown in Fig. 9.16(a), the (R' - Y') component of the
sub -carrier chrominance signal at the triode anode will be as shown
in Fig. 9.16(b) which, after filtering, yields the original waveform
of Fig. 9.16(a).
It can be seen that the gated triode detector has a detection axis
which is the negative of the reference phase applied to the grid. In
INSTANTS OF TRIODE CONDUCTION
- - + 25 V
:
(RL '6 D.C.COMPONENT
(R' -Y') COMPONENT OF

SUBCARRIER CHROMINANCE
SIGNAL
Fig. 9.15. (above): Triode anode voltage waveform for

negative (R' - Y') component.
Fig. 9.16 (below): (a) transmitted (R' - Y')

waveform; (b) (R' - Y') component of triode
anode voltage waveform for signal shown in (a)
(a)
the case we have considered, detection along the (R' - Y') axis is
achieved when the reference phase is - (R' - Y'). It follows that
if we have three separate detectors fed with phases - (R' - Y'),
- (G' - Y') and - (B' - Y'), then the required red, green and
+H T.
g
s.
vv nn
(11'-r)
- (G' -Y') (B -Y')
+90'
SUBCARR ER
CHROMINANCE
OUTPUT STAGE
Fig. 9.17. Complete 3 detector high level decoder. Phases of reference signals
are measured from the burst phase
blue difference signals will be obtained. The complete circuit for

such an arrangement is shown in Fig. 9.17 in which the necessary
gains of 1.14, 0.7 and 2.03 for the red, green and blue difference
signals, respectively, are derived from suitable windings on the
output transformer. The reference phases required are - (R' - Y'),
- (G' - Y') and - (B' - Y'), or + 90°, - 123° and 0° measured
from the burst phase. The inconvenient - 123° phase for the
(FILO
114
07
123*
BURST 0B-v)Fig. 9 18 (G' - Y') and - (G'-Y')

PHASE
57 2 o3
detection axes
(GL y')
0.7
green difference detector can be obviated by reversing the green

difference winding and detecting along an axis 57° from burst,
as shown in Fig. 9.18.
The circuit shown in Fig. 9.17 provides a very stable arrangement,
since the triodes are used as on -off " switches " and the various
proportions of the difference signals are obtained from transformer
windings. However, there may be some difficulty in achieving a
sufficient drive for the (B' - Y') detector since the gain required in
this case is 2.03. The drive requirements can be eased by using
(R' - Y') and (G' - Y') detectors and deriving the (B' - Y')
signal by means of a simple matrix.
Since
0.30(R' - Y') 0.59(G' - Y') 0.11(B' - Y') = 0

it follows that
- (B' - Y') = 2.72(R' - Y') 5.37(G' - Y')
Hence, if we add (R' - Y') and (G' - Y') in the ratio of 0.30 to
0.59, the sum will be proportional to - (B' - Y'). If this is
(R'- Y') (G'-
DETECTOR DETECTOR AMPLIFIERcs'-
H.T.
OUTPUT
Fig. 9.19. (B' - Y') matrix and amplifier
applied to a simple amplifier, the required phase inversion will be

obtained, and the gain of the amplifier can be adjusted so that the
correct ratio between the difference signals is achieved.
Fig. 9.19 shows a circuit of a (B' - Y') amplifier which supplies
the (B' - Y') output given (R' - Y') and (G' - Y'). The matrix
equation above is satisfied by adjusting the values of R, and R2
to obtain the correct proportions of (R' - Y') and (G' - Y'), and
the resulting - (B' - Y') signal is inverted by the amplifier to give
the required (B' - Y') output.
In calculating R, and R2, the impedance looking back into the
anodes of the detectors must be taken into account (usually this is
about 51(12), and it is preferable to measure the (B' - Y') output
and adjust R1 and R2 until the correct (B' - Y') detection phase is
obtained, while disregarding, for the moment, the amplitude of the
(B' - Y') output relative to (R' - Y') and (G' - Y'). Experiment-
ally, this adjustment for phase can be made by feeding a constant
amplitude but variable phase sub -carrier sine wave into the chromi-
nance channel, and noting the phase required to give maximum
(B' - Y') output. This phase should be lagging 90° on the phase
reading which makes the (R' - Y') output a maximum, assuming
that the relative (R' - Y') and (G' - Y') phases have been set up
correctly.
The correct relative amplitude of (B' - Y') can be obtained by
adjustment of the anode load of the (B' - Y') amplifier. The
cathode load should be fairly high to give plenty of negative feed-
back, and for a Z77, typical anode and cathode loads would be
251a2 and 3.31d2 respectively, for a 400V H.T. line.
Notice that
the (B' - Y') amplifier is D.C. coupled.
Some improvement in (B' - Y') transient response can be
effected by means of a small capacitor across R2, the larger matrix
resistor.
It should be noted that it is usually necessary to arrange for
different drive ratios of the difference signals to allow for the display
tube phosphor efficiencies, and in the circuit of Fig. 9.17, for example,
the (G' - Y') output required may in fact be 0.8(G' - Y') rather than
(G' - Y'). If a matrix is used to derive (B' - Y'), it must obviously
be designed to include such drive considerations.
A typical high level detection circuit using (R' - Y') and (G' - Y')
detectors, with a (B' - Y') matrix amplifier, is shown in Fig. 9.20.
This circuit is used in the G.E.C. TTIV series of receivers, and is
designed for the 405 -line system for drive ratios of 1 to 0.875 to
0.75 for R to G to B.
In this circuit, small R.F. chokes are included in the anodes of the
(R' - Y') and (G' - Y') detectors to prevent parasitic oscillation
during the heavy positive grid drive period, and sub -carrier filtering
is by way of 1.5mH chokes which resonate at 2.7 Mc/s. The series
3.3k S1 resistors prevent radiation of sub -carrier harmonics which may
otherwise be picked up by the I.F. or R.F. stages of the receiver and
be displayed as a coloured herringbone pattern. It must be
remembered that the detector anodes have sub -carrier signals of the
order of 100V peak applied to them.
The standing D.C. bias for the background controls is injected
from potentiometer arrangements via 270142 resistors, and the D.C.
input resistance to the difference signal outputs is increased by the
150k0 resistors which are heavily decoupled so as to maintain the
R' -Y. DETECTOR DETECTOR fY-Y'AMPLIFIER
+400V
330.
33pF
47kR
33k f1
1.5mH 15mH
V3
112B117
11,
CHROMINANCE +400V +400V

OUTPUT 270kfl
1.5kn.
RED GREEN ), BLUE
BACKGROUND BACKGROUND BACKGROUND
100 kn.
3 7,kn ),
0.1g -150V 0.10 -150V -150V
T.PIAS
0° REFERENCE DRIVE -3? REFERENCE DRIVE

[RELATIVE TO - AXIS] [RELATIVE TO -(11'-Y) AXIS]
Fig. 9.20. High level detector for 405 -line system

frequency response of the outputs. Some difference signal D.C. com-
ponent loss occurs through the background controls, this D.C. loss
0.27
being (0.27 or about 4dB maximum.
+ 0.15
Correct clamping action of the detector triodes is assisted by
driving the grids with a considerable amplitude (at least 50V peak)
of reference signal, and fron.ka low impedance source of 500 S2 or
less.
It should be noted that in high level detectors the reference signal
must not have a " kink " or slight discontinuity during the burst
period, otherwise the clamping action is disturbed and the detector
anodes take several microseconds to recover their correct bottom-
ing potential. This effect shows as a coloured shading on the
extreme left of the picture, and it must not be confused with the
green shading due to the burst which is evident if there is inadequate
line flyback suppression of the display.
Another effect to be borne in mind is that if the reference signal
is removed from the detectors, their anode potentials will rise, but if
a (B' - Y') amplifier is used, its output will not rise proportionately
because the matrix circuit is now incorrect since the anode imped-
ances are greater. The white balance is therefore upset. If three -
Fig. 9.21. Similarity between

120° spaced detection axes and
0.7 colour difference axes
difference detectors are used, the main effect of removing the reference
signals will be to produce a brightness change. Hence, it follows
that reference drive must be applied to the detectors even when the
receiver is operating from a monochrome signal, if bias adjustments
between monochrome and colour signal operation are to be avoided.
9.10. Symmetrical detection
In symmetrical detection the principle is to take advantage of the
fact that on a vector representation of the three colour difference
axes (R' - Y'), (G' - Y') and (B' - Y'), three detection axes may
be drawn, as shown in Fig. 9.21, which are 120° apart in phase and
(d) ( b)
VI
4
REFERENCE
INPUT
Fig. 9.22. 120° reference drive circuit
each of which does not differ by more than 15° from the phase of a
colour difference axis.
Hence, three detectors operating along axes spaced 120° apart
may be arranged to give outputs which are very close to the required
difference signals, and a simple resistive matrix network may be
included to adjust the detector outputs so that the difference signals
are actually obtained.
The reference drive for the detectors may be obtained by means
of the transformer arrangement shown in Fig. 9.22(a). The second-
ary windings AB and CD are loosely coupled and tuned so that the
voltage between B and 0 is 90° out of phase with the voltages
between C and B, and D and B, as shown in Fig. 9.22(b). By
connecting B to the centre tap between C and D, the equal magnitude
and 120° phased outputs V1, V2 and V3 are obtained provided that
(i) VRO = 0.5 V1
(ii) VCR = VDB
(iii) VCR = 0.866 V,
A suitable symmetrical detection circuit for the American system,
designed by the Hazeltine Research Corporation, is shown in Fig.
9.23. The modulated sub -carrier signal from the chrominance
20k11
MATRIX CIRCUIT
4.7pF 47kil
0.047g 0047/2F 0.047,uF

390kr1 390k11 390kii
SUBCARRIER
INPUT
yr G Y'
HMO.
worm
REFERENCE 000. 00_J TO BACKGROUND CONTROLS
INPUT
-I,500pF
!kn. 0.01µF
Fig. 9.23. Symmetrical detector circuit for

525 -line system
amplifier is fed to each grid of the three triode detectors, while the
cathodes of the triodes are returned to earth through the 120° phase
shifting transformer. The 10052 cathode resistors are included to
stabilize against differences in conductance of the triodes.
Each detector operates along the appropriate axis as shown in
Fig. 9.21, and the matrix circuit slightly changes the effective detec-
tion axes so that the correct difference signal outputs are obtained.
The 4 -7pF capacitor across the 47kS2 resistor in the green difference
matrix input helps to maintain the green channel frequency response.
Sub -carrier frequency components appearing at the triode anodes
are attenuated by series tuned traps, and the series inductance in
each output forms a low pass filter in conjunction with stray capaci-
tance and capacitative loading by the traps on one side, and stray
capacitance and capacitative loading by the display tube on the
other side.
The standing D.C. bias for the background adjustment is applied
through 1.5MS2 to each output, and the D.C. input resistance is
increased by means of 390k0 resistors so that the D.C. bias input
voltage required to be applied to the 1.5MS2 resistors is not excessive.
The A.C. response is maintained by the 0.047µF capacitors which
shunt the 390kS2 resistors, but some difference signal D.C. component
loss occurs across the 3901d2 resistors. However, even if the back-
ground control resistance were very small, the D.C. component
1.50
attenuation is only or about 2dB.
+ . 39)
Since the reference drive transformer is arranged to give 120°
phased outputs, the sub -carrier frequency currents due to any two
of the detectors inject an equal but opposite component into the
third detector, that is, the symmetry of the reference drive arrange-
ment prevents crosstalk between the detectors. In practice a small
amount of crosstalk is to be expected since the symmetry will not
be perfect.
Two advantages of the circuit shown in Fig. 9.23 are, firstly the
effective coupling between the detectors is small and secondly the
symmetry of the anode circuits of the detectors means that any
drift in H.T. line voltage will produce equal changes in each of the
difference signal outputs, so that a substantially " brightness only "
change will occur in the picture.
9.11. XZ detection
An interesting and effective decoding method known as XZ
detection has been introduced by R.C.A. The novelty of the circuit
lies not so much in the detectors themselves, which are of the normal,
simple low level mixer type as described in Section 9.8, but rather
in the choice of detection axes and the matrix arrangement which
follows the detectors. The circuit is intended for equiband detec-
tion of the modulated sub -carrier.
In XZ detection, two synchronous detectors are fed with the
modulated sub -carrier signal and each detector is also fed with a
particular phase of reference signal so that one detection axis is
Fig. 9.24. XZ detection axes. The

factor 11178 = 1.1412.03 is the trans-
mitter weighting of (B' - Y') relative
to (R' - Y')
BURST (B- Y)
PHASE 1.78
-x
-X and the other is - Zi as shown in Fig. 9.24. Notice that the

output of each detector will consist of a certain proportion of the
(R' - Y') and (B' - Y') signals, the proportions being different
for the two detectors (e.g. the -Z detector output will contain
relatively more - (B' - Y') signal than will the -X detector
output). The two detector outputs are fed to a three triode matrix
with a common cathode load, as shown in Fig. 9.25. The triode
-X -2
INPUT INPUT
NEGATIVE
I-4-CLAMPING
PULSE INPUT
Fig. 9.25. Basic XZ matrix circuit

arrangement is perfectly symmetrical, the anode loads being the
same for each triode as well as the grid circuit time constants.
The grid of the centre triode is returned to earth, so that its
cathode voltage and therefore the output from its anode will be
proportional to the sum of the signals fed to the triodes 1 and 3.
Now it will be remembered that the green difference signal may be
obtained by adding - (R' - Y') and - (B' - Y') in accordance
with the equation
0.59(G' - Y') = - 0.30(R' - Y') - 0.11(B' - Y')
so that if the sum of the -X and -Z signals is made to contain
1.0 (R -Y1
0.6 (BI -Y')
Fig. 9.26. Vector diagram of the XZ

gg matrix circuit
- (R' - Y') and - (B' - Y') in the ratio of 0.30 to 011, then the
output of triode 2 will be proportional to (G' - Y').
Referring to the vector diagram of Fig. 9.26, the vector sum of
-X and -Z gives the vector 2 after multiplication by a suitable
factor A, which is chosen to give a green difference drive of 0.8 in
this case. The vector 2 is proportional to the output of triode 2,
and of course it should ideally be coincident with the 0.8(G' - Y')
vector shown. However, the red and blue difference signals are
also required and in order to obtain these as well some error in each
difference output must be tolerated since there are not enough
variables to satisfy all the required conditions simultaneously, unless
the symmetry of the circuit is destroyed.
The grid drive to triode 1 consists of the -X signal minus
the cathode voltage - A(X Z), i.e. the grid drive is
-X + A(X Z), so that the triode 1 output is proportional to the
negative of this (because of the phase inversion from grid to anode),
which is X - A(X Z), and is given by the vector 1. Again,
ideally this vector should be coincident with the (R' - Y') axis.
Similarly, the grid drive to triode 3 consists of the -Z signal
minus the cathode voltage - A(X Z), so that the triode 3 output
is proportional to Z - A(X Z) and is given by the vector 3.
The X and Z vectors of Fig. 9.26 are chosen so that the outputs of
the triodes 1, 2 and 3 are proportional to (R' - Y'), 0.8(G' - Y')
and 0.6(B' - Y'), the constants 1 to 0.8 to 0.6 being the phosphor
efficiency drive ratios. Since the exact pure outputs cannot be
obtained, X and Z have been chosen to give least crosstalk in the
difference signals. Thus, output 1 consists of (R' - Y') and a
small amount of -(B ' - Y'), output 2 consists of 0.8(G' - Y') and
a small amount of - (B' - Y'), while output 3 consists of 0.6
(B' - Y') and a small amount of (R' - Y').
More exactly
Output 1 is 1.0 [(R' - Y') - 0.07(B' - Y')]
Output 2 is 0.8 [(G' - Y') - 0.07(B' - Y')]
Output 3 is 0.6 [(B' - Y') 0.07(R' - Y')]
where the 7 % crosstalk terms are the result of choosing X and Z
to make the crosstalk a minimum.
Calculation of the X and Z axes is given in the next section, where
it is shown that crosstalk can be avoided only when the red to blue
drive ratio is 1 to 0.367. However, for a red to blue drive of 1 to
0.6, the crosstalk is only 7 %, and since in practice the vectors 1,
2 and 3 may be rotated by means of the hue control until correct
flesh tone is displayed, the small crosstalk effect is not significant.
If desired, a resistance connecting outputs 2 and 3 may be fitted to
" pull together " vectors 2 and 3 thereby reducing the crosstalk.
Alternatively, crosstalk can be removed by applying a suitable
fraction of the triode 3 input to the grid of triode 2, instead of
earthing the latter. This technique is detailed at the end of
Section 9.12.
It will be noticed that the outputs from the synchronous detectors
are A.C. coupled to the triode matrix circuit. This is necessary since
the anodes of the detectors have a D.C. voltage near that of the
H.T. line, but obviously some means of D.C. restoration is required.
The D.C. components of the chrominance signal are recovered by a
clamping technique in which a large negative pulse occurring during
the line blanking period is applied to the common cathode connec-
tion of the three triodes. This pulse makes the three triodes conduct
during the line blanking period so that they take grid current
which charges the grid coupling capacitors. The bias on the three
triodes is therefore clamped to the same voltage at the end of each
line scan so that any D.C. components present in the chronainance
signal will be applied to the triode grids during the line scan. The
triode anodes are D.C. coupled to the appropriate electrodes of the
display tube. In order to ensure correct clamping action, the sub -
carrier signal fed to the synchronous detectors should be blanked
during the line blanking period to remove the burst and any spurious
information. This can be done, for example, by applying a suitable
positive pulse to the cathode of the sub -carrier chrominance output
stage, so that this becomes inoperative during the clamping period.
This clamping technique has two other advantages besides D.C.
restoration. Thus, since each triode is biased by its own grid
current during the flyback time, any falling off in emission in a
triode will produce less grid current and hence less bias on that
triode. Consequently, the gain of any triode tends to remain
constant as ageing progresses. Since a large negative pulse is
applied to the triode cathodes, the triode anodes are driven very
negative during the flyback time so that all three displays are blacked
out during retrace, thereby providing line flyback blanking.
The XZ detection circuit used by R.C.A. for the CTC 5N series
of receivers for the 525 -line system is shown in Fig. 9.27.
The X and Z synchronous detectors are of the mixer type and
are fed with the sub -carrier signal and the appropriately phased
reference signals. The cathode earth point is via a potentiometer so
that the two detector gains may be equalized, and both anode
circuits are identical. Some inductance correction is included to
maintain the frequency response of the detector outputs to 1.3
Mc/s for 3dB down, so that " wideband equiband " operation is
obtained.
Series chokes in the detector outputs attenuate sub -carrier com-
ponents fed to the matrix and the matrix triode grids are connected
to equal time constant circuits. The negative clamping pulse for
the triode cathodes is obtained from an amplifier which is fed with a
suitable pulse from the line time base output transformer. Since
the clamping pulse causes a negative bias to be applied to the triode
17
+H.T.
300V
10kn.
+I45V
390kfl Ly'
V3
10011 I2BH7A
0.047AF
0.047,0
620uN
X REFERENCE
INPUT
+300V RED
BACKGROUND
390k11
MATRIX
BALANCE,
0047,2F
SUBCARRIER
INPUT
180,1211
GREEN
+300V
BACKGROUND
_fir
390111
Z REFERENCE
INPUT
V2 0047uF
6201.LH
IMI1
+300V
22kfl BLUE
BACKGROUND
V6
712BH7A
1
4/IF
39011
+70V 0.01,0 LINE TIMEBASE
PULSE INPUT BOOST LINE
FROM LINE TIMEBASE
470kf1 ) TO CATHODE OF
CHROMINANCE OUTPUT STAGE
BLANKING AMPLIFIER
Fig. 9.27. XZ detection and matrix circuit for 525 -line system
grids by grid current flow, a suitable positive bias (in this case,
+ 60V) must be applied to the grid leaks of the triodes so that the
correct grid to cathode bias is obtained. The cathode of the
blanking amplifier is connected to the cathode of the sub -carrier
chrominance output stage to remove the burst during the clamping
period.
Note that although the grid of the green difference triode is
" earthy " to A.C. signals, it is connected to the H.T. line through
an R.C. circuit of exactly the same value as in the red and blue
difference triodes. Hence, H.T. line variations will not produce
+VI
(a)
(b)
Fig. 9.28. Three -triode matrix (a) and its equivalent circuit (b)
differential changes between the potentials of the triode anodes and

consequently white balance stability will tend to be independent of
such variations, which will therefore chiefly affect brightness only.
The three triode anode circuits are identical, and variable D.C.
bias for background control is applied through a high impedance
circuit to prevent loss of D.C. chrominance components. The high
value of D.C. supply required as a consequence of this is obtained
from the line time base boost line.
9.12. Calculation of the X and Z axes

In order to calculate the XZ axes mentioned in the previous
section, the operation of the three triode matrix circuit must first be
investigated. This is shown in Fig. 9.28(a), in which the anode
loads RL are equal, and the common cathode load is R.
Assuming that the three triodes are exactly similar, the equivalent
circuit is shown in Fig. 9.28(b) whereµ and Ra are the amplifica-
tion factor and anode resistance of each triode. Note that although
we are interested in the case where v2 is zero, the symmetry of the
circuit and the calculation is preserved if v2 is not put equal to zero
until the final steps of the working.
In the circuit of Fig. 9.28(a), it is required to find il, i2 and i3 in
terms of v1, v2 and v3 and the circuit constants.
Thus the voltage of the equivalent generator of triode 1 is tt(vi - v),
and this must equal the voltage drop across R, RL and Ra.
Hence µ(vi - v (RL Ra)ii
or (RL R011 = µv1 - + 1)v [9.1]
Similarly (RL Ra)i2 = v2 - (Ft. + 1)v [9.2]
and (RL Ra)i3 = ttv, - (IL + 1)v [9.3]
These expressions give the anode currents in terms of the cathode
voltage v. To find v, note that
v
11- 12 + 13 =T?
Hence, adding Equations 9.1, 9.2 and 9.3
(RL Ra).k = P,(v1 + v2 v3) - 3(11. + 1)v

from which
µR(vi + v2 + v3)
v- RL Ra 3R(p. ± 1)
Substituting in Equation 9.1
r + 1) (v1 + v2 +
11 RL Ra vl RL + Ra 3.12(A. + 1) 1
For convenience put
A - RL Ra
+ 1)
3R(µ + 1)
then
11 A (v1 4- V2 + v3)]
RL Ra[
and similarly
i2 [ v2 - A(v1 + v3 + v3)]
RL RL Ra
and
1
3
- RL Ra
[
v3 A(v1 v2 v3)]
In this case, the grid of triode 2 is earthed so that v2 = 0. The

output voltages are as follows:-
triode 1 gives p
AL
R
, -rLIc
a
[
v1 - A(v1 v3)1
[ yid
triode 2 gives A(v1
RL +LRa
[
triode 3 gives v3 - A(vi v3)]
RL
Now it is required to find v1 and v3, each of which contains both red
and blue difference signals, so that r(R' - Y'), g(G' - Y') and
b(B' - Y') are obtained from the anodes of triodes 1, 2 and 3,
respectively, where r, g and b are the phosphor efficiency drive
factors. (Typically, r = 1.0, g = 0.8 and b = 0.6.)
Hence
(RL Ra)
- v1 A(v, + v3) - µRL
rRD [9.4]
(RL Ra)
A(v + v3) = gGD [9.5]
tot&
and
(RL Ra)
- v3 + A(v1 + v3) - µRL
bBD [9.6]
where RD = (R' - Y'), GD = (G' - Y') and BD = (B' - Y')

Now
0.59GD = - 0.30RD - 0.11BD
so that Equation 9.5 may be written as
- A(v1 + v3) - (RL + Ra)

g(pRD + qBD) [9.1
pRL
0.30 0.11
where p = 6:-59 = 0.508 and q = iiT§ = 0.186.
Equation 9.7 indicates that v1 + v3 must contain RD and BD in the

ratio of p to q, that is, 1 to 0.367. But, adding Equations 9.4 and
9.6/
-(v1 + vs) + 2A(vi + v3) = -(1 - 2A) (vi + v3)
(RL + BO
- (rRD + bBD) [9.8]
IL&
which indicates that v1 + v3 must contain RD and BD in the ratio of
the drives r to b. The required outputs cannot therefore be obtained
r 1
unless
b= 0.367
However, suppose the red to blue drive ratio is required to be 1 to
0.6. This can be arranged provided that some crosstalk is tolerated.
Thus, Equation 9.8 would have an RD to BD ratio of 1 to 0.6, while
Equation 9.7 would have an RD to BD ratio of 1 to 0-367. To make
these two ratios equal (since Equations 9.7 and 9.8 must be satisfied
simultaneously), increase the ratio in Equation 9.8 by increasing the
RD content and reducing the BD content by allowing a crosstalk
term /3RD in the BD output of triode 3, and a crosstalk term
in the RD output of triode 1.
-
pBD
Then Equation 9.8 becomes

(1 - 2A) (v1 + v3) = (RL ± Ra) [r(RD - PBD) + b(BD + PRD)]
µ RD
[9.9]
r + bi3.
that is, v1 + v3 has an RD to BD ratio of In addition, let us
b rp
decrease the RD to BD ratio in Equation 9.7. This can be done
either by reducing the RD content or by increasing the BD content.
Since p is greater than q, and since least crosstalk is required, it is
better to increase the BD content. (For example, p = 0.508 and
q = 0.186. If RD were reduced by, say, 0.1, the RD to BD ratio
0.508 0.186408
would be changed from
0r6 or 2.72, to 0.186 or 2.2. But if BD
were increased by 0.1, the RD to BD ratio would be changed from
0.508
2.72 to cTi§ or 1.78. That is, a bigger change can be made by
increasing BD rather than by reducing RD.) Therefore, allowing a
crosstalk term y BD in the triode 2 output, Equation 9.7 becomes
(RL +& Ra) g{
- A(vi + v3) - 0.508RD + (0.186 + y)BD1 [9.10]
IL
0.508
which has an RD to BD ratio of
0.186 -I- y
The RD to BD ratios of Equations 9.9 and 9.10 must be equal, so that
r 113 0.508
b -rp 0.186+y
If the required drive ratios are r = 1, b = 0.6, then
1 + 0.6/3 0.508
0.6 -p 0.186+y
which gives the relation between the crosstalk terms. Obviously,
the crosstalk terms are required to be as small as possible, and this
will occur when p = y = /3 = x, say.
Then
1 + 0.6x 0.508
0.6 -x 0.186 + x
Or
0.6x2 + 1.62x - 0.119 = 0
whence x = -2.77 or 0.07.
Taking the smaller of these two values, it can be seen that the
required drive ratios can be obtained provided that the outputs of
the triodes are accepted as
RD - 0.07BD
instead of RD for triode 1
- 0.8(0.508RD + 0.186BD + 0.07BD) = 0.8(GD - 0.07BD)
instead of 0.8GD for triode 2
0.6(BD + 0.07RD)
instead of 0.6BD for triode 3
Equations 9.4, 9.7 and 9.6 therefore become
(RL + Ra)
- v1 + A(v1 + v3) = (RD - 0.07BD) [9.11]
tiRD
- A(v1
(RD ±
v3) µALRa) 0 8 (0.508RD + 0.256BD) [9.12]
(RL + Ra)
- v3 + A(v1 + v3) = 0 6(BD + 0.07RD) [9.13]
pRD
Substituting Equation 9.12 in Equation 9.11

(RL + Ra)
vi (1.406RD + 0.135BD) [9.14]
Substituting Equation 9.12 in Equation 9.13

(RD + Ra)
V3 = - (0.448RD + 0.805BD) [9.15]
RL
Now it is required to obtain v1 and v3 by synchronous detection

of the sub -carrier. Suppose v1 is derived by detecting with a gain
G1 along an axis O from the - (R' - Y') axis, while v3 is derived
(sloe)
BURST
PHASE
6,
Fig. 9.29. Derivation of the XZ detection axes
by detecting with a gain G3 along an axis 03 from the - (R' - Y')

axis, as shown in Fig. 9.29. Both detection axes are required to be
in the third quadrant, since negative amounts of (R' - Y') and
(B' - Y') are required in each case.
The output of detector 1 is
-G1 cos 01RD-Gi sin 01

BD
1.78
- (RL+ Ra)
tARL
(1 406RD+0.135BD)
sin 0 tan 01 0.135
1.78 cos 01 1.78 - 1.406
tan 01 = 0.175, i.e. 01 = 9.5°
and
1.406(RL + Ra) (RL + Ra)
G1 - - 1 43
ttRL, cos 01 IL&
The output of detector 3 is
BD (RL+
-G3 cos 03RD - G3 sin 03 = IA&Ra)(0 448RD + 0.805BD)
1-78
sin 03 tan 03 0.805
1.78 cos 03 1.78 0.448
tan 03 = 3.2, i.e. 03 = 72.5°
and
G3 - 0.448(RL + Ra)
- 1.49 (RL + Ra)
p-RL cos 03 iiRL
Note that the ratio of the detector gains required is 1.43 to 1.49,
i.e. 0.96.
The detection axes 01 and 03 are defined as the -X and -Z
axes, respectively.
The constant
A - RL ±R(L + 1)
Ra ± 341 + 1)
defines the green drive according to Equation 9.12. It may be
calculated by adding Equations 9.14 and 9.15 and substituting for
v1 + v3 in Equation 9.12. Thus, Equations 9.14 and 9.15 give
vi+ v3 = - (RLaaRL
+ Ra)
(1.854RD + 0.94BD)
Substituting in Equation 9.12

(1.854RD -I- 0.94BD)A - 0.4064RD + 0.2048BD
whence
A - 0.4064RD + 0.2048BD
1.854RD + 0.94BD
- 0.22
For a given valve type and anode load, the required value of A
can be obtained by a suitable choice of the cathode resistor R.
It should be noted that the values of 01 and 03 of 9.5° and 72.5°,
respectively, have been calculated above on the assumption that
equal percentage crosstalk applies in the three difference signal
outputs. An alternative approach is to arrange for the error
angles to be equal, so that in Fig. 9.26, the angles between vector 1
and (R' - Y'), between vector 2 and 0.8(G' - Y'), and between
vector 3 and 0.6(B' - Y'), are all equal. Use of the equation
1 + 0.6/3 0.508
0.6 -p 0.186 + y
enables the equal angular error to be calculated, its value being
approximately 7°. In this case, 01 = 9° and 03 = 69°, approximately.
These values are very close to the equal percentage crosstalk detec-
tion axes of 9.5° and 72.5°.
It is interesting to note that crosstalk can be avoided completely
if a suitable signal is applied to the grid of triode 2. Furthermore,
it is possible to arrange for this v2 signal to be a fraction of the v3
signal.
Thus, allowing for the case when v2 0 0, and assuming drive
ratios of 1.0 to 0.8 to 0.6 for r to g to b, Equations 9.4, 9.5 and 9.6,
after adding the three equations so that v1 + v2 + v3 can be elimin-
ated, become:-
µRD
RD
(1 - 3A)v1 = (1 - 2.405A)(- RD) + 0.45A( - BD) [9.16]
Ra
µRD
RD
(1 - 3A)v2 = (1.81A - 0.405) (- RD) + (0.9A -0.15)
Ra
(-BD) [9.17]
µRD
RD + Ra
(1 - 3A)v3 = 0.595A(-RD) + (0.6 - 1.35A) (-BD)
[9.18]
Now, by a suitable choice of the constant A, it is possible to make

the ratio between RD and BD in Equation 9.17 the same as that in
Equation 9.18.
1.81A - 0.405 0.595A
Solving the equation
0.9A - 0.15 0.6 - 1.35A
it will be found that A = 0.245. The root A = 3 is trivial.
ihRL
Putting C = RL
(1 - 3A) for simplicity, the above equa-
Ra
tions become:-
Cy, = 0.44- RD) + 0.14- BD)
Cv2 = 0.04( - RD) + 0.07( - BD)
Cv,-= 0.146( - RD) + 0.27( - BD)
in which it will be seen that

0.27Cv3 = 0.04( - RD) + 0.07( - BD) = Cv2
Hence, by applying a fraction 0.27 of the signal applied to the

grid of triode 3, to the grid of triode 2, the required triode outputs
of RD, 08G D and 0.6BD are obtained exactly.
For a complete decoder, therefore, two synchronous detectors
are required to obtain v, and v3, and a simple resistive matrix to
derive v2 from v3.
The required detection axes for v, and v3 will be found to be
0, = 25.5° and 03 = 73°, as measured lagging on the - (R' - Y')
Gi
phase. The relative gains required are = 0.9.
9.13. Chrominance signal matrixing

It has been pointed out earlier in this chapter that if I'Q' working
is adopted, the matrix operations after detection of the I' and Q'
signals are fairly complex. However, a decoding method has been
suggested by the Hazeltine Research Corporation which may
simplify the matrix circuits so that difference signal outputs are
obtained directly after detection while maintaining I'Q' operation.
In the usual method of sub -carrier demodulation a single sub -
carrier signal is compared in phase with two separate reference
signals in quadrature. However, the reverse process may be
employed in which two separate sub -carrier signals in quadrature are
compared in phase with a single reference signal. In this case, the
two sub -carrier signals may be treated as " / " and " Q" signals so
that matrixing can be carried out before the synchronous detection
process.
Since the sub -carrier is modulated according to the picture content
and therefore has sideband frequencies, the quadrature sub -carrier
signals cannot be obtained from simple LC or RC circuits, but if
the modulated sub -carrier is applied to a delay line which has a delay
time equal to one quarter of a cycle of the unmodulated sub -carrier
frequency, the output will consist of an undistorted but quadrature
phase shifted carrier version of the input.
The principle of chrominance signal matrixing is shown in Fig.
9.30.
The modulated sub -carrier signal is fed to one band-pass filter
which accepts the sidebands of the I' signal (i.e. (2.66 - 1) or 1.66
Mc/s, to 3 Mc/s for the 405 -line system) and the same modulated
I' BANDPASS I'DELAY

1.66-3Mcis N + 1/4 CYCLES
-Q' AXIS
DETECTOR G'
GAIN 028
MODULATED
SUBCARRIER
INPUT
REFERENCE
Mc
0.' BANDPASS
212-3 +1.62
+Q' AXIS
DETECTOR 11' Y`
GA IN 0.96
0-66
0
+Q, REFERENCE
-1'55
-la' AXIS
-0-73 DETECTOR è.
GAIN 1.11
t
-Q REFERENCE
Fig. 9.30. Pre -detection chrominance signal matrix arrangement for

the 405 -line system
sub -carrier signal is also fed to a band-pass filter which accepts the
sidebands of the Q' signal (i.e. (2.66 - 0.34) or 2.32 Mc/s to 3 Mc/s
for the 405 -line system).
Now the output of the I' bandpass filter is passed through a delay
line whose electrical length is equal to an exact number N of sub-
carrier cycles plus a quarter cycle, the total delay being equal to that
required for the time coincidence of the I' and Q' signals. Note
that if this delay line output is fed to a Q' phase detector, the output
will be the I' modulation.
Now the red difference signal in terms of I' and Q' is given by
(R' - Y') = 0.96/' 0.63Q'
or
1.04(R' - Y') = I' + 0.66 Q'
and the blue difference signal is given by
(B' - Y') = - 1.11I' + 1.72Q'
or
- 0.9(B' - Y') = I' - 1.55Q'
Hence, by adding 0.66 of the output of the Q' band-pass filter to the
delayed output of the I' band-pass filter and detecting the result by a
Q' phase detector with a gain of 1 to 1.04 = 0.96, this detector
output will be equal to (R' - Y').
Similarly, by adding - 1.55 of the output of the Q' band-pass
filter to the delayed output of the I' band-pass filter, and detecting
the result by a - Q' phase detector with a gain of 1 to 0.9 = 1.11,
this detector output will be equal to (B' - Y').
The (G' - Y') signal may be obtained from the equation
(G' - Y') = - 0.281' - 0.64Q'
or
- 3.58(G' - Y') = I' ± 2.28Q'
Thus, the delayed output of the I' band-pass filter is added to 2.28
of the Q' band-pass filter output and the result is detected by a - Q'
phase detector having a gain of 1 to 3.58 = 0.28, to give (G' - Y').
In Fig. 9.30, the appropriate proportions of the Q' band-pass filter
output required to be added to the I' band-pass filter output are
shown, the additional quantity 0.73 being included to balance the
Q' channel load.
It will be appreciated that the I' channel delay line must be cut
sufficiently accurately to give the odd quarter cycle delay for the
quadrature phase shift. One way of doing this in practice is to use
two sections of line (both having the same characteristic impedance,
of course), one section having less delay per unit length so that it
may be used as a " fine control " of delay.
9.14. Summary
In this chapter we have seen that chrominance information may be
recovered from the modulated sub -carrier signal by synchronous
detection along the appropriate axes. This entails a " sampling "
of the sub -carrier signal by a reference sine wave locked in fre-
quency and phase to the transmitter's sub -carrier oscillator, the
relative phase of the reference determining the detection axis.
After synchronous detection, some form of matrix circuit is
required so that R', G' and B' signals can ultimately be fed to the
display tube. If full use is made of the transmitted information,
synchronous detection along the I' and Q' axes is required, and a
matrix circuit is necessary to convert the I' and Q' signals to the
difference signals (R' - Y'), (G' - Y') and (B' - Y'). The final
matrix operation of adding Y' to the difference signals is usually
carried out in the tube itself by feeding the difference signals to the
appropriate grids of the tube, and a - Y' signal to the cathodes.
An alternative decoding procedure is to detect synchronously
along two or three detection axes, the bandwidths of the detected
signals being made equal and rather larger than the Q' bandwidth.
This " wideband equiband " technique results in some crosstalk
between the higher frequency chrominance signals, but the benefit
of the increased definition compared with equiband Q' bandwidth
operation outweighs the undesired effects. Equiband working
allows a free choice of detection axes so that the difference signals
can be obtained directly from the synchronous detectors without a
matrix operation, or alternatively the axes may be chosen in such a
manner that a simple matrix circuit suffices to supply the required
difference signals, as in symmetrical and XZ detection.
Several synchronous detector and matrix circuits have been
described, and while these have been labelled as suitable for either
405 -line or 525 -line N.T.S.C. systems, the basic circuits are suitable
for any system provided the frequency limiting components of the
circuit are designed for the appropriate bandwidths.
CHAPTER 10
Colour Receiver
Reference Frequency Generators
10.1. Introduction
Synchronous detection of the modulated sub -carrier signal has
been described in Chapter 9, and there it was pointed out that, in
order to demodulate a quadrature modulated signal, a C.W. source
of sub -carrier frequency must be available which is locked in fre-
quency and phase to the unmodulated sub -carrier at the transmitter.
To enable colour receivers to generate this reference, a short sample
of unmodulated sub -carrier is transmitted during the line blanking
period, and this is usually referred to as the colour burst synchroniz-
ing signal. The device in the colour receiver which generates a
C.W. sub -carrier signal from the burst signal is called a reference
frequency generator.
The television engineer who is not yet familiar with reference fre-
quency generators is often understandably alarmed by the function
that they have to perform. Since vital hue information is carried
by the phase of the modulated sub -carrier signal, it obviously follows
that any spurious phase changes in the C.W. reference signal will
give rise to spurious hue changes in the reproduced picture. Further,
the burst signal has a very low duty ratio since only some 3.50 of
the transmitter's sub -carrier is transmitted only once during each
100µs of line scan (in the 405 -line system), and this would appear
to suggest that any spurious signals due to noise or interference
would have ample opportunity to produce hue errors in the picture.
These doubts about the practicability of generating a C.W.
sub -carrier source of sufficient stability from such a short synchroniz-
ing burst have been voiced by critics of the N.T.S.C. system, and
it is only fair to add that the authors themselves were originally
among these critics. However, the reader may rest assured that,
provided reasonable attention is paid to layout and design, the
generation of a C.W. reference from the burst signal is perfectly
practicable even under extremely adverse signal-to-noise ratio
259
conditions.In fact, satisfactory synchronization can be obtained
even when the signal-to-noise ratio is below usable level for the
monochrome signal.
The fundamental principle which permits the generation of a
sufficiently pure reference from the burst signal is integration.
The longer the integration time, the better will be the purity of the
reference output signal, and theoretically this process can be con-
tinued indefinitely. However, one practical limitation is the
stability of the sub -carrier source at the transmitter, since any
integrating device at the receiver must not integrate so effectively
that changes in the transmitter's sub -carrier frequency cannot be
accommodated.
This particular limitation is not usually of any consequence
because extreme frequency stability is normally achieved by the
transmitting authorities. A more realistic practical limitation is
economy in the colour receiver circuit.
For those readers who are not familiar with the mechanism of
integration processes, an analogy may be drawn by way of a simple
experiment in photography. Suppose that a synchronized sine
wave is displayed on an oscilloscope and suppose sufficient output
from a random noise generator is added so that the displayed
sine wave appears to be " lost " in the noise background. If a
long time exposure photograph, with a corresponding adjustment
of aperture, is now taken of the display, it will be found that the
sine wave is quite easily visible. The longer the exposure (i.e. the
longer the integration time) the more visible the sine wave will
become. In effect, the random noise integrates up to give a uniform
background, while the sine wave, being non-random, integrates
up to give its original waveform. Note that, if, during the exposure,
the frequency of the sine wave were varied, the result would be
spoilt. This is the equivalent of having a longer integration time
than is warranted by the stability of the transmitter's sub -carrier
frequency.
The integration performed by a reference generator may be
expressed in terms of a frequency response characteristic. Thus,
if a signal is integrated over a time Tm, then any noise fluctuations
present which have a frequency greater than 1 will not contribute
significantly to the output since their mean value will be near zero.
Hence, only fluctuations with frequencies less than will be signi-
T-m
ficant. Therefore integration over a time Tm is equivalent to passing
Effect of phase error of reference generator on reproduced hue;
(above) phase delay; (below) phase advance
The front view of a colour monitor with the cabinet removed. The
lower photograph shows the controls at the bottom of the set
0-- 0 ---o, 0-- fittD

REFERENCE FREQUENCY GENERATORS 261
the signal and noise through a filter having a pass bandwidth of
1
-T.Since the required output from a reference generator is of
sub -carrier frequency, the integration can be carried out by a filter
1
having a pass bandwidth of -T. with a centre frequency equal to the
sub -carrier frequency.
EQUIVALENT
NOISE
BANDWIDTH
Fig. 10.1. Definition of
equivalent noise bandwidth
of a filter
FREQUENCY
1
The bandwidth -T. is called the noise bandwidth, fry, and therefore
1
-=-f-N
T.
The equivalent noise bandwidth of a filter may be found by
plotting the square of the voltage output against frequency, and
then drawing a rectangular pass band characteristic having the
same peak value and enclosing the same area with the frequency
axis, as the physical response. The equivalent noise bandwidth
18
is then the width of the rectangular response, as shown in Fig. 10.1.
The square of the voltage output is used in order to obtain equiva-
lence on a power basis.
The reference frequency generator is basically an integrator.
It accepts a noisy sine wave signal and provides an output which is
that same sine wave but with much reduced noise content. It is
therefore equivalent to a narrow band-pass filter tuned to the sub-
carrier frequency.
10.2. Colour burst signal

The colour burst synchronizing signal for the 405 -line British
adapted N.T.S.C. system is shown in Fig. 10.2.
From the point of view of the reference frequency generator, the
significant factors are:-
The peak -to -peak amplitude of the burst is equal to that of the
line sync waveform.
The burst commences about 1.5µs after the back porch of the
line sync pulse, and lasts for not more than 3.75µs.
The burst is omitted during the 4 lines of field sync pulses.
The frequency of the burst is 2.6578125 Mc/s, ±8 c/s, with a
maximum rate of change of 0.1 c/s2.
PEAK
16 5 ±0.5its
WHITE
1.5 -10Ips
4-- 9±l/./s -0-1 1
0 U BLACK LEVEL
-4-- LIFT
0 ...------COLOUR BURST
9 ICis OF REFERENCE
FREQUENCY
SYNC. LEVEL
Fig. 10.2. Colour burst synchronizing signal

It is important to realize that the colour bursts are samples of a
continuous sine wave of sub -carrier frequency. That is, at the
transmitter a continuous sine wave of sub -carrier is taken and applied
to a gating circuit which selects the appropriate number of cycles
once every line, except during the field sync period. Note that
there is no phase specified for the start of the burst. This may vary
as the phase of the divided down line frequency varies relative to
the sub -carrier frequency. However, no matter where the time
position of the line sync pulses may be relative to the burst, and
no matter what the starting phase of the burst may be, it is always
possible to draw a continuous sine wave starting with any one burst,
which will pass through, and be exactly coincident with, all the other
bursts.
Since the sub -carrier frequency is locked to an odd multiple of
half the line frequency, the starting phase of the burst will change
by 180° every line. Thus, an oscilloscope locked to line frequency
will display the burst (or any other sub -carrier component) as two
sine waves 180° apart, i.e. interlaced. On the other hand, if the
oscilloscope is triggered once every other line, only a single sine wave
will be displayed.
It must be emphasized that the 180° phase change in the start of
the burst, and which occurs every line, is due to the fact that the
time interval between the gates contains an integral number plus a
cycle of sub -carrier, as a result of the frequency relationship between
the sub -carrier and line frequencies.
A Fourier analysis of the burst waveform is given in Appendix 1,
where it is shown that it is equivalent to a sine wave of sub -carrier
frequency together with sidebands separated from it by intervals
of the line frequency. The two nearest sidebands have practically
the same amplitude as the sub -carrier component.
10.3. Specification of reference generator performance

Since the phase of the modulated sub -carrier relative to the burst
is a function of the hue of the colour being transmitted, it is obvious
that any phase changes between the output of the reference genera-
tor and the burst will produce hue errors in the reproduced colour.
It therefore follows that the largest phase locking error between the
reference output and the burst should ideally be such that the
corresponding hue error is just not noticeable.
Neglecting the effects of noise for the moment, subjective tests
have shown that while some people are capable of observing hue
errors corresponding to only about I° of phase error, most people
will accept the range of hue errors corresponding to phase errors
of approximately ±5°. However, the authors are of the opinion
that reference generators should be designed to have a phase error
not greater than +IF (although to date the transmitter tolerance
may be as large as +100), to provide a reasonable safety factor.
This type of phase error is called the static phase error. It is the
maximum possible value of the phase error between the reference
output and the burst.
When noise is present with the burst signal, the reference output
phase will be modulated by it to give a dynamic phase error, as it
is called. It is usual to specify random noise, since this will inevit-
ably be present, and it is additionally the most difficult type to reject.
In a severe case, random noise in the burst signal applied to a
correctly designed reference generator will show as a low frequency
colour flutter on the reproduced picture. It is customary to specify
the noise performance of a reference generator in terms of the
r.m.s. value of the phase fluctuations obtained with a specified
signal to random noise ratio. Thus, a recommended noise perfor-
mance may be quoted as a 5° r.m.s. dynamic phase error for a
unity signal-to-noise ratio of peak to peak burst to r.m.s. noise.
This figure of 5° r.m.s. dynamic error for a unity signal-to-noise
ratio was originally suggested as a reasonable criterion, and exper-
ience has shown that it is, in fact, realistic. With such a noisy
signal, flutter effects due to phase errors in the reference output
are subjectively unimportant compared with the other effects
present, such as time base synchronization and brightness fluctua-
tions. For a given dynamic phase error and for a given signal-to-
noise ratio, it is possible to calculate the integration time T. and
therefore the equivalent noise bandwidth
1
fr'= 7;
which is required. For the British system, this works out to be
140 c/s. This is discussed in Appendix 2.
A further condition which should be satisfied by a reference
generator is that it should have a reasonable stabilization period,
so that when switching from one channel to another only a short
time is required for synchronization. A stabilization period of
1 second may be regarded as satisfactory, while one of about 3
seconds is tolerable and one of 10 seconds is poor.
A final desirable feature of a reference generator is that it should
provide a reasonably constant output amplitude for all levels of
signal input, and for all possible settings of the hue control. How-
ever, the constancy of output amplitude required will depend on
the type of synchronous detector used in the decoder.
10.4. Burst gating
The function of the reference generator is to provide a con-
tinuous sine wave output locked to the burst and to the burst only.
Since, in the chrominance signal, other sub -carrier components are
present which have various phase values depending on the hues of the
transmitted colours, it is necessary to separate the burst from these
components. The selection of the burst to the exclusion of all
other signals can be carried out by means of a suitable gating wave-
form which, in effect, switches the sub -carrier signal into the reference
generator only at those instants when the burst is present.
Thus, a typical composite video signal is shown in Fig. 10.3(a)
while the chrominance component is shown in Fig. 10.3(b). If a
gating waveform such as Fig. 10.3(d) can be generated, this can be
Ti T2
F
_IJ
ift
1
( a) COMPOSITE VIDEO
W -r
L _r
(b) CHROMINANCE COMPONENT
V
( C ) DIFFERENTIATED SYNC. SEPARATOR OUTPUT
( d ) BURST GATE
W Al
( t) SEPARATED BURST REPRESENTS
NINE CYCLES
Fig. 10.3. Burst gating waveforms

made to switch an amplifier fed with the waveform of Fig. 10.3(a)
or Fig. 10.3(b) to give the burst only output of Fig. 10.3(e).
In order to exclude unwanted sub -carrier components, the burst
gate should open not earlier than the start of the pre -sync porch
T, and should close not later than the end of the post -sync
porch T2. However, the longer the gate is open, the greater will
be the noise allowed through, and therefore the ideal gate should
open immediately before the burst and close immediately after it.
If the burst gate width exceeds the burst width by a factor of r to 1,
+250V
22k.n.
GATE OUTPUT
100y p -p
FROM SYNC
SEPARATOR VI
,,44rL\AAAAe_1100pF
10kil
10k£1
Fig. 10.4. Gate generator circuit
and the required noise bandwidth of the reference generator is fN,

then the generator itself must be designed to have a noise bandwidth
of -11±r Nevertheless, it is advisable to have a slight excess of gate
.67
width to allow for any relative time shift between the gate and the
burst. It is essential to ensure that the gate closes before the start
of the active line, otherwise disturbing hue errors will result which
will vary with the hue of colours in the extreme left of the picture.
A very satisfactory method for generating the burst gating wave-
form is to develop a pulse of the appropriate width by means of a
monostable multivibrator which is triggered by the trailing edge of
the line sync waveform. A convenient source for the latter is the
anode of the conventional sync separator, where negative going
mixed sync is available.
Differentiation of the sync separator output results in positive
going " pips " which are coincident in time with the trailing edges
of the line sync waveform, as shown in Fig. 10.3(c). Application
of these " pips " to the monostable multivibrator of Fig. 10.4
results in the output shown in Fig. 10.3(d) where the length of the
gating pulses can be adjusted by the time constant in the grid
circuit of V2.
The operation of the circuit of Fig. 10.4 is as follows:-
With no signal input, V2 is biased positively relative to V1 so that
V2 is conducting and holds V1 non -conducting by way of cathode
feedback. When a positive " pip " is applied to the grid of V1, V1
conducts and a negative pulse from its anode renders V2 non-
conducting. However, since the V1 anode is not D.C. coupled to
the V2 grid, the conduction period of V1 lasts only for a time deter-
mined by the time constant in the grid circuit of V2. When the V2
grid has reached a sufficiently positive value, V2 conducts again and
renders V1 non -conducting. The circuit then remains stable until
the next positive " pip " triggers V1.
The use of this type of gate generator ensures that the gate opens
before the start of the burst, in fact, about 1.5ps before. How-
ever, some delay can be incorporated by reducing the frequency
response to the grid of V1 by means of a series resistor, and this
usually leads to about his between the opening of the gate and the
start of the burst.
It is recommended that the time constant in the V2 grid be adjusted
to give a gate width of about 5µs so that the longest burst (10 cycles)
can be accommodated. For the most unfavourable condition of
an 8 cycle burst, this gives an excess burst gate to burst period ratio
of 1.66 to 1, so that the reference generator should be designed for
a noise bandwidth of 0.775fN to achieve an effective overall noise
bandwidth of fN.
It will be appreciated that the gate generator of Fig. 10.4 will
produce two gates per line during the four lines of field sync pulses,
where the burst signal is absent. This gives an excess burst gate
width of 210.5 to 202.5, or 1.04, which increases the noise band-
width by a factor of 1.02, which is not significant. However, if
the reference generator is at the extreme end of its pull -in range, the
lack of burst during the field sync period can lead to hue errors
at the top of the picture.
An alternative method for generating the burst gating waveform
is to derive it from the receiver's line time base. Thus, a suitable
winding on the line time base output transformer can be arranged
to supply a pulse of +100V or so during the flyback time. If such
a pulse is applied to an integrating circuit of about 3.31d) and 100pF,
the resulting pulse waveform can be used as a burst gate.
While derivation of the gate from the line time base is an econo-
mical technique, it suffers from certain disadvantages. For example,
the gate width is normally relatively long to allow for variations
in the synchronization time of the time base, and of course the
locking of the time base can affect the performance of the reference
generator.
The amplitude of the gating waveform required will depend on
the level of the chrominance signal existing where the gate is applied.
The amplitude must be large enough to exclude the largest chromin-
ance excursions (full saturated red and cyan) from the reference
generator. Thus, if gating is carried out in the final stage of the
burst amplifier, the burst input amplitude may be about 5V peak
so that the largest chrominance signal would be about 13V peak.
The gate amplitude should therefore exceed the cut-off bias by at
least 13V, leading to a value of about 20V peak. A minimum
gate amplitude of 30V or so usually allows for a satisfactory safety
margin.
The gate waveform should ideally have a flat top or " on "
period, and its leading and trailing edges should not have excessively
fast rise and decay times otherwise spurious rings will result which
may simulate the burst signal. Note that the gating waveform
applied to the gated valve will produce a gate waveform in the
anode circuit, and by differentiation " pips " corresponding to the
leading and trailing edges of the gate will appear with the burst
waveform.
The luminance signal is another source of spurious signals which
can simulate the burst. Thus, the leading and trailing edges of the
line sync waveform, and the leading edge of the line blanking when
there is a white on the extreme left of the picture, all consist of
step waveforms which occur close to the burst signal. Now, if a
receiver has a relatively narrow I.F. pass -band (with corresponding
compensation in the chrominance amplifier response) then the sync
and blanking edges can produce spurious rings having a frequency
close to that of the sub -carrier component of the burst. These
spurious rings are often called sync or video widgets. Note that
in a monochrome transmission in which the field frequency is mains
locked, a mains frequency change of only about 0.1 c/s can produce
a widget having a frequency exactly equal to the sub -carrier. (For
example, the 263rd harmonic of the line frequency for a field
frequency of 49.9 c/s in the 405 -line system is only about 260 c/s
from the sub -carrier frequency.)
Colour " unkilling " and spurious locking due to widgets can be
avoided by having a wide I.F. pass -band and an accurately positioned
and correct width burst gate. (See Appendix 1.)
10.5. Burst amplification

Most reference generators require a burst input amplitude of the
order of 50V peak, and this amount of burst is not normally avail-
able in the receiver's chrominance channel. The usual practice is
therefore to provide an amplifier which accepts the chrominance
signal from a suitable point, amplifies and gates it, and then feeds
its output to the reference generator.
The amount of gain required from a burst amplifier will naturally
depend on the burst take off point in the chrominance channel,
but it is unlikely to exceed 50dB, in which case the burst input
amplitude would be about 150 mV peak.
It should be noted that there are conflicting requirements as far
as the position of the burst take off point is concerned. On the
one hand, the take off point should be from a point as close as
possible to the synchronous detector input to minimize phase shift
variations between the reference signal and the chrominance sub-
carrier. On the other hand, the burst should be taken as remote as
possible from the synchronous detector to prevent the reference
signal (which is necessarily present, usually with large amplitude,
at the synchronous detector) being present with the burst, so
deteriorating the pull -in performance of the reference generator. In
the authors' experience this latter requirement is by far the more
significant, and it is good practice to take the burst from at least
one stage before the synchronous detectors.
In order to reduce the noise content of the burst signal which is
finally applied to the reference generator, the gating operation should
be carried out in the last stage of the burst amplifier. This eliminates
any noise which may otherwise be introduced by post gating stages
in the periods between bursts. Further, since the final burst
amplifier stage may often be a heavy current valve, H.T. current
drain is conserved since the valve is switched " on " for only about
5 % of the time.
It is important to realize that although the reference generator
itself is a very narrow pass -band filter, the burst amplifier should have
an appreciable bandwidth up to the gated stage. Thus, it is good
practice to design for a bandwidth equal to about 40 times the line
scan frequency (0.4 Mc/s in the British system) and centred about
the sub -carrier frequency. The reason for the relatively large
bandwidth is to ensure that the majority of the burst information
does actually reach the reference generator. Thus, the burst
waveform can be analysed into Fourier components, and the
power content of any number of harmonics in the vicinity of
the sub -carrier frequency can be calculated and compared with the
total power in the burst waveform. It will be found that the
power content of the sub -carrier component and the nearest 20
harmonics on each side of it is approximately 85 % of the total
GATED BURST
V2 OUTPUT
CHROMINANCE
INPUT
6.8pF
0.IAF
GATE INPUT -100V
Fig. 10.5. Gated burst amplifier circuit
power in the burst waveform. Hence, for a burst amplifier band-

width of 0.4 Mc/s (0.6 Mc/s in the American system) prior to the
application of the burst gate, the burst waveform is practically
undistorted so that most of the burst is included during the gating
interval. (Fourier analysis of the burst waveform is described in
Appendix 1.)
In the case of reference generators which are literally passive
narrowband filters, there is obviously no point in maintaining the
bandwidth of any post gated stage; but for A.P.C. loops, or any
generator employing a phase detector, the bandwidth should be
maintained even after gating to ensure that the maximum signal is
available for phase control.
A typical two stage burst amplifier is shown in Fig. 10.5. The
circuit is quite conventional, the anode loads being simple tuned
circuits tuned to the sub -carrier frequency, with suitable damping
to give the required bandwidth of 0.4 Mc/s. Note, however, that
because of the small duty ratio of the burst, the input impedance
of the phase detector which is fed from the burst amplifier output
is relatively low. The burst gate is applied to the lower end of a
6.8k0 resistor connected to the grid of V2, and some sub -carrier
decoupling (100pF) is included to prevent sub -carrier components
entering the gate generator.
If the gate is generated from the back edge of the line sync, as
described in Section 10.4, it is advisable to apply a negative bias
to the V2 grid, as shown. The object of this bias is to cut off V2
if mixed sync signals are absent, otherwise the amplifier would be
" open " all the time and may accept spurious signals or reference
output which would not have been significant during the short
gating periods.
Note that an LZ329 valve and an HN309 valve can be used for
both gate generation and burst amplification, to give a compact
unit.
10.6. Classification of reference generators

The preceding sections have described the performance required
of a reference generator, and also the principles of amplifying and
selecting the burst signal to the exclusion of other sub -carrier com-
ponents. The reference generator itself is therefore presented
with the burst only waveform of Fig. 10.3(e), from which it must
generate a continuous sine wave locked to the burst sine wave,
that is, it must " fill up the gaps " between bursts.
Broadly speaking, reference generators fall into two classes:-
(i) Passive integrators which select the sub -carrier frequency
component of the burst waveform by means of a narrow pass -band
filter. Note that it is not sufficient for the filter to merely exclude
the sidebands on either side of the sub -carrier frequency component.
Its pass -band must be narrow enough to give an equivalent noise
bandwidth of about 140 c/s (in the British system) so that dynamic
phase errors due to noise are subjectively acceptable.
(ii) Dynamic integrators which perform the filtering operation by
means of a low pass filter to which is fed a beatnote signal formed
by heterodyning an oscillator output with the burst signal. The
conventional dynamic integrator takes the form of an oscillator
whose output is frequency and phase controlled by a phase detector
which compares the oscillator and burst signals. This arrangement
is usually called an A.P.C. loop, that is, an automatic phase control
loop. Like the passive integrator, the dynamic integrator should
have a frequency characteristic equivalent to a noise bandwidth of
140 c/s centred about the sub -carrier frequency.
Normally, the reference generator performs other functions in
addition to the primary one of reference frequency generation.
BURST
AMPLIFIER AMPLIFIER LIMITER
SUBCARR I ER
FREQUENCY
CRYSTAL
oI
of I
REFERENCE
V2
OUTPUT
Fig. 10.6. Circuit for a passive integrator
These include colour killing and A.C.C. or automatic chrominance

control.
For reasons which will be discussed later, reference generators of
the A.P.C. type are almost always used in preference to passive
integrators. However, each type will be described in the following
sections.
10.7. Passive integrators
A simple passive integrator circuit is shown in Fig. 10.6.
The gated burst signal is fed to a bridge circuit which includes a
quartz crystal nominally tuned to the sub -carrier frequency. The
object of the balanced arrangement is to neutralize the effect of
the mounting capacitance which is unavoidably present across the
crystal, thereby preventing feed through of the burst sidebands.
The crystal rings at its resonant frequency in response to the sub-
carrier component of the burst waveform, and the resulting voltage
is amplified. Since the crystal voltage has an exponential decay
during the period between bursts, the amplified voltage is passed
to a limiter which provides a constant amplitude of sine wave
output.
The crystal Q which is needed in this circuit is obviously a func-
tion of the noise bandwidth required, and it can be shown that the
equivalent noise bandwidth of a simple tuned circuit is 2of its
3dB bandwidth. Hence, if the resonant frequency is fo c/s, the
3dB bandwidth is 1:j c/s, and the noise bandwidth is
7r fo
fN = -j: c/s
Hence
7r f
Q fN
For satisfactory noise performance, fN should be about 140 c/s
for the British system, so that the Q required is
2.66 x 106
Q-2 x
140
ct 30,000
if it is assumed that the crystal behaves as a simple tuned circuit.

This high value of Q accounts for the use of a crystal rather than
an LC circuit, and of course, crystals can be obtained which have the
required Q.
There are severe limitations to this type of integrator if it is
designed for the above noise bandwidth. The difficulty is that,
since the circuit Q is so high, a small change in either the crystal
tuning or the transmitted sub -carrier frequency can cause very
serious static phase errors. The static phase error has two com-
ponents. One is produced by the phase shift which is caused by
slightly detuning a high Q circuit and the other arises from the
slightly " out of step " effect of the crystal sine wave, which accumu-
lates during the line period before the next burst arrives. These
two sources of static phase error may be calculated as follows:
The equivalent circuit of the crystal itself is shown in Fig.
10.7(a), where L, C and r are the series components of the crystal,
and S is the mounting capacitance. Now a parallel resonance of
the crystal will occur when the applied frequency is slightly higher
than the series resonant frequency, so that the series L, C com-
ponents are then equivalent to an inductance / which resonates
with the capacitance S, as shown in Fig. 10.7(b).
The impedance of the circuit of Fig. 10.7(b) is given by
r jwl
Z
1 - co21S jwSr
coo
But if -277. is the parallel resonant frequency
1 Qr
IS=wo2, 1=
coo
1
cooQr
Substituting
r Qr
wo
Z= (02
1- +j
coo cooQ
and it is the phase of Z which is of interest, since variation of the
phase of the voltage across the circuit relative to the burst current
fed to it will determine the phase error. Hence, the phase 0, of Z,
c r
Fig. 10.7. Equivalent
circuit of a quartz
crystal
(a) (b)
which is the phase of the voltage across the circuit relative to the
total current through it, is the same as the phase of the vector
2
W1I W
[(' :f2) Q)+ ja70 _Q wo
to
1+J-Q(1- 0)022)
since Q > 1
Therefore, if f8 is the applied sub -carrier frequency, and f; is the
parallel resonant frequency, the above vector becomes
18 1l fo\ 2
1+ J-Qk
fo - fo2
Now suppose there is a tuning error Of so that fo = f8 + Af,
then
A A Of
fo ./. + Of
1
-
Af -
1 - A
-
1 +-
A
and
f8 2 24
7.' ' 1 -Ts
The above vector then becomes
Af - + ' 1 2 -AI Q
1-1-jQ(1- 78 27; 1
so that
Af
tan 0, = 2Q-A
or
01 = tan -1 2 Q-i.
Af
J8
This error 01 is the first of the phase shifts mentioned above, and
it may be caused either by a tuning error of the crystal or by a
change in the transmitter's sub -carrier frequency.
The " out of step " phase error may be calculated as follows:-
M-2
If the sub -carrier frequency is times the line frequency (M =
525 in the British system) and if there are 8 cycles of burst per line,
then between bursts there are (M-
2
8) cycles of sub -carrier. Each
cycle lasts for a time off seconds, so that the time between
(M - 16)
bursts is seconds.
2f8
Now if there is a crystal tuning error such that
A = A + Af
the number of " crystal " cycles in the time
(M2fs 16) seconds is (Al 16)(f8 + AD

s
However, the number of " genuine " sub -carrier cycles during this
(M 16)'
time is so that the total error at the start of a burst is
- 161 /A/ - 16) Af
(M 2f, )(fs + 2
- -(M
2f,
- 16) cycles
Since each cycle is a phase of 27r radians, this phase error is
Af
02 = 27rws(M - 16) radians
This expression may be simplified since M - Mf

16 M and fs =
where fr, is the line frequency.
Thus approximately
2irzfM
0. - _Am - 27r- radians
fL
The total phase error of the passive integrator is therefore
= 0, ± 0, = (tan-i 2Q3.18) 277.X
Since 2Q-Af will normally be small, this expression may be further

fs
simplified to
Af Af
2Q-- + 27r- rattans
.1,
or
0
37r
f fs
-) degrees
7r
Now if 0 is not to exceed 2.5°, then Of must not exceed about 2 c/s
for a Q of 30,000, i.e. a noise bandwidth of 140 c/s.
Since the transmitter frequency tolerance is +8 c/s, the required
static phase accuracy cannot be guaranteed for a noise bandwidth
of 140 c/s.
Of course, if the Q is reduced, a larger value of ifcan be tolerated.
Thus, if
Of= 20 c/s
The inside of a colour monitor showing the underside of the I.E. deck and
sync. separator. The two lower units are the tuner and reference oscillator
A typical complete 21 -inch shadow mask receiver of a type which
could go into production. This type of receiver will accept either
N.T.S.C. colour transmissions or monochrome signals
the value of Q for a static phase shift of 2.5° is about 2,100, which
gives a noise bandwidth of
7 2.66 x 106
-2 kc/s
2 2,100
This corresponds to a dynamic phase error of about 17° r.m.s.

for a peak -to -peak burst to r.m.s. noise ratio of unity, whereas the
desirable dynamic error is only 5° r.m.s.
The conflicting requirements for small dynamic and small static
phase errors cannot therefore be simultaneously satisfied. A com-
promise may be taken, of course, or alternatively an automatic
device may be included in the circuit to correct the static phase
error, as shown in Fig. 10.8.
In the circuit of Fig. 10.8, the integrator is designed to have
the required noise bandwidth regardless of static phase error.
The integrator output is then compared in phase with the burst
BURST CRYSTAL REACTANCE

INPUT FILTER VALVE
LIMITER
REFERENCE
OUTPUT
90°
PHASE
SHIFT
LOW
PHASE
PASS
DETECTOR
FILTER
Fig. 10.8. Passive integrator with static phase correction
signal in a phase detector whose output corrects the tuning of the

integrator by means of a reactance valve, thereby reducing the
static phase error. A phase shift of 90° must be included in one
of the sub -carrier feeds to the phase detector, since phase detectors
usually give a zero or equilibrium output when the two signals
being compared are 90° apart in phase. (Phase detectors and
reactance valves are discussed in Appendices 5 and 6). By including
sufficient gain in the feedback loop, the static phase error may be
19
reduced to the required figure, and provided the frequency pass
band of the loop is small compared with the noise bandwidth, the
noise performance of the integrator is not impaired.
It should be noted that in crystal filters or ringing circuits, the
output voltage decays exponentially between bursts by a factor
e-11/2/ where t is the time measured from the end of a burst.
Since -/ = the decay factor may be written as e-6)0020, where
wo = 271. x 2.66 x 108 radians/sec, and the maximum value of t is
approximately 100 x 10-6 sec. The decay factor is therefore e -2667o,
and this has a value of 0.9 for a Q factor of 8,350. Hence a 10
drop in amplitude occurs during the line period for a Q of 8,350.
For a Q of 30,000, the drop in amplitude during the line period is
only about 3 %. However, since some synchronous detectors are sen-
sitive to any amplitude change of the reference signal, a limiter after
the crystal is advisable to prevent colour changes across the picture.
Passive integrators have not proved popular in practice because
of the requirement for high Q and accurately cut crystals. To
BURST
-4-- OSCILLATOR -4- REACTANCE
VALVE
A.C.TRANSMISSION OF FILTER -m c
D.C. TRANSMISSION OF FILTER
x LOW-PASS
+X FILTER xR
RC = T
X WHEN X << I
Fig. 10.9. A.P.C. loop block diagram
overcome the static phase error, a phase loop is necessary (unless a

compromise is taken between static and dynamic phase errors)
and the circuit then becomes strictly comparable with an A.P.C.
loop from the point of view of the number of components.
10.8. Dynamic integrators or A.P.C. loops

A block diagram of a simple A.P.C. loop is shown in Fig. 10.9.
The output of a stable oscillator nominally tuned to the sub -carrier
frequency is fed to a phase detector where its phase is compared
with the burst waveform. The resulting beatnote output of the
phase detector is then passed through a low pass filter to a reactance
valve which controls the oscillator frequency.
The design of A.P.C. loops must cater for two conditions; the
loop must first pull -in to frequency and phase lock, and having
locked, the characteristics of the loop must be such that variations
of the incoming burst phase (due to noise or interference) do not
cause large variations of the oscillator output phase. Since the
formulae which describe the in -sync and pull -in performance of
A.P.C. loops require more advanced mathematics for their deriva-
tion, formulae will merely be quoted in the following sections.
However, a full account of A.P.C. loops is given in Appendices
3 and 4 and phase detectors and reactance valves are described in
detail in Appendices 5 and 6.
10.9. In -sync performance
Let us first consider the in -sync performance of the loop. Two
conditions must be satisfied. First, the static phase error must not
exceed a certain value, such as ±21°. Second, the noise perform-
ance must be such that the dynamic phase error does not exceed
about 5° r.m.s. for a peak -to -peak burst to r.m.s. noise ratio of
unity. This corresponds to a noise bandwidth of about 140 c/s
for the British system.
Before discussing the static and dynamic phase errors, it is
necessary to define the various parameters which determine the
loop performance. These are as follows:-
The sensitivity of the reactance valve, which is usually called
and is measured in c/s/V. Thus, one volt applied to the reactance
valve grid produces a frequency change of the oscillator of ac/s.
In reference generators the reactance valve is usually arranged to
" look like " a capacitor. That is, a positive D.C. voltage increase
on the grid reduces the frequency.
The sensitivity of the phase detector, which is usually called p.
and is measured in terms of volts/radian. The meaning of p, may
be illustrated by feeding a sine wave voltage of one frequency to
one side of the phase detector, and a sine wave voltage of a different
frequency to the other side. The phase detector output under these
conditions is a sine wave having a frequency equal to the frequency
difference between the signals being compared (i.e. the beatnote
frequency) and a peak amplitude of pc. volts. If the beatnote
frequency is co radians/sec, the phase detector output is tc sin cut,
and if this is plotted against cot this sine wave has a slope of
volts/radian at the origin. (The slope of a sine wave at the origin
is numerically equal to the peak value of the sine wave.)
The usual type of phase detector used in A.P.C. loops gives an
output which is proportional to the cosine of the phase difference
between the two signals being compared. The detector output is
therefore zero for a 90° phase difference, and it is usual to consider
phase errors in terms of the angle 0 by which the phase difference
differs from 90°. That is, if there is a phase error 0, the detector
output is µ cos(90 - 0), i.e. ti sin 0.
The final parameter to be considered is the transfer characteristic
of the loop filter. This is the ratio output voltage/input voltage of
the filter at any frequency (.0 radians/second, and may be called
F(w). For the filter shown in Fig. 10.9
1 + jwxT
F(w) -
1 + jw(1 x)T
where T =VC
x
This reduces to
(1 x) as woo, when it is called the A.C./D.C.
gain ratio, m.
10.10. Static phase error
If the oscillator has a tuning error of Of c/s, then a voltage -Af
must be applied to the reactance valve to correct it. This voltage
must come from the phase detector, that is
Of
II. sin A0 =-
where 6.0 is the static phase error.
Hence
sin AO = -Af =f
The quantity fc is the product pfl, and is usually called the D.C.
loop gain. It represents the maximum hold -in range of the loop.
The loop cannot generate the required correction voltage to tune
the oscillator if the oscillator error is greater than fc.
Since 6,0 is usually required to be not more than +2-k°, a reason-
able approximation for A0 is
Of
=C0c
= - radians =rfcIT
.1cr
- -
180 Afo 180 Acoo
where Au) = 27rAf and (D, = 2ITfc

10.11. Dynamic phase error
The dynamic phase error may be investigated by assuming a
noise -free input signal whose phase is changed sinusoidally at a
particular frequency and noting the amplitude of the corresponding
output phase change. By repeating this process at different fre-
quencies, a plot of output phase change divided by input phase
change may be made against frequency. This ratio is called the
phase transfer ratio, usually written Q(w), and it is analogous to
the voltage transfer ratio of a filter. In fact, it is shown in Appendix
r = Vukc
Voltage transfer characteristic Va coot! --r--o
L = 0 + x)Rfivc
V1
1/jwC xR 1
-1 to ,C jw(1 + x) RI co, ± 1 'PAC xR
xR
V2
I jaiXT 1 iil
1-(1 x)w2T1w, jco(xT±11coc)

Fig. 10.10. Equivalent circuit for the phase transfer ratio of the
A.P.C. loop shown in Fig. 10.9
3 that, for the loop filter of Fig. 10.9, the phase transfer ratio of the
loop is the same as the voltage transfer ratio of the circuit of Fig.
10.10. The noise bandwidth of the loop may be found by squaring
the amplitude of the Q(w) curve, and finding the equivalent rectan-
gular bandwidth which encloses the same area.
In Appendix 3 it is shown that
24
fN
iNN
wc[1+(l+ x)
(1
where fNN is the noise semi -bandwidth.

x2T(.0, 1
x MO
we (1 ± mxTeoc)
4 (1 -I- x7100
Note that the loop does not distinguish between noise fluctuations
which are above or below the sub -carrier frequency, so that the
area under the Q(w) squared curve leads to the noise semi -band-
width.
For the British system, fN = 140 c/s so that fNN = 70 c/s, for
5° r.m.s. dynamic error and unity peak -to -peak burst to r.m.s.
noise ratio.
It may be wondered why the resistor xR is included in the loop
filter. It is included to enable the static phase error and the noise
semi -bandwidth to be chosen independently.
Thus, if xR is zero (i.e. short circuit)
(oc
fNN = -4
But the static phase error is
AlpAu)
-
we
so that
1 Aw
fNN = 4 X
or
04
fNN&/ =
Thus, for a given tuning error Ow, the AO and fNN cannot be
independently chosen.
1.6
I IS K VALUE OF 0.25
- 1.2 2 IS K VALUE OF 10
3 IS K VALUE OF 00
08 f= PHASE MODULATION
FREQUENCY OF INPUT
0.6
6 0.8 1.0 1.2 14 1.6
fNN
Fig. 10.11. Phase transfer characteristics for

various K values
However, by including xR, we can first be chosen to give the

required figure for Ay6, and then two variables, that is, x and T,
are available for making fNN the required figure. Further, since
both x and T are available, one of these may be used to control
another characteristic of the loop. Notice that the equivalent
circuit of the loop, shown in Fig. 10.10, has L, C, R elements so
that ringing could occur under certain conditions. It is shown in
Appendix 3 that this ringing is a function of a damping factor K
defined by
+ xTcoc)2
K
4(1+ x)Twe
K> 1 corresponds to overdamping (i.e. no ringing), K < 1 corre-
sponds to underdamping, and K =1 corresponds to critical damping.
In fact, K controls the shape of the Q(w) curve, as shown in Fig.
10.11. To avoid ringing on impulse interference, K should be 1
or more.
Notice that if xR = 0
1
K = 4Tw,
and as shown earlier
to,
fNN =4-
Thus, for given tuning and static phase errors, fNN is fixed, and
changing the time constant T merely changes the value of K, i.e.
the shape of the Q(w) curve.
To summarize the in -sync performance, the formulae are
Static phase error
Of Ow
sin 64 =
Jc
=- we
we = 27r0
Dynamic phase error, i.e. noise bandwidth
710,
we [1 +1± 1 + x2Teoc
fN = 2fN N =
2(1 + xTwc) 2xT
since for the range of values which are of interest, x << 1 and
xTwc >> 1.
Damping factor K (preferably 1 or more)
(1 + x710,)2 x27-4,2
K= 4(1 + x)710, 4
It is important to notice that the in -sync performance is not

merely a function of the low pass filter constants x and T. The
in -sync performance is also a function of the loop gain wc.
10.12. Pull -in performance
The pull -in performance of the loop should be such that frequency
and phase lock occur reasonably quickly after the burst signal is
applied. If there is a small tuning error (strictly Of < mfc) the
loop will pull in to frequency lock immediately, and there will
be a small phase stabilization time which is usually only of the
order of hundredths of a second.
If there is a large tuning error
Af > fc V2m - m2
the loop will never pull in to frequency lock.
If there is an intermediate tuning error
mfc<Of<fcV2m-m2
the loop will pull in to frequency lock, though if Lf is close to
jec i/2m - m2 the time taken may be of the order of minutes.
When the loop has pulled in to frequency lock, rapid phase stabiliza-
tion occurs as for the case Af < mfc. (See Appendix 4.)
The mechanism of frequency lock for the above case may be
qualitatively explained as follows:-
Suppose for the moment that the reactance valve is disconnected.
The phase detector output will then be sinusoidal, with a frequency
equal to the tuning error. If the reactance valve is now connected,
it will try to increase the oscillator frequency during one half cycle
of beatnote, and reduce it during the other half cycle.
The beatnote waveform will therefore no longer be sinusoidal,
one half cycle corresponding to a low frequency beatnote and the
other corresponding to a high frequency beatnote. The beatnote
waveform will have a shape as shown in Fig. 10.12. This waveform
has unequal areas under the positive and negative half cycles, so
that it has a D.C. component. This is stored in the filter condenser
so that it reduces the mean tuning error. As time progresses, so
the D.C. component builds up as a voltage across the condenser,
and the mean tuning error is gradually reduced. Eventually the
tuning error becomes less than mfc, the loop then pulls in to fre-
quency lock immediately, and a phase transient follows.
It is interesting to note that the D.C. component will always act
in such a direction as to correct the frequency error. The " larger "
half cycle of beatnote necessarily corresponds to a shift of oscillator
frequency towards the burst frequency.
Another point to notice is that, quite apart from the loop filter,
the loop itself has a frequency characteristic. For example, if the
condenser C were short circuited, there would be no frequency
dependent components in the loop, yet the beatnote waveform of
Fig. 10.12 (which in this case would be a steady beatnote, so that the
loop would not pull in) would become more sinusoidal if the tuning
I.0
0.8
0-6
0.4
= 0
'10-2
-04
-06
- 04
- 1-0
10
08
06
a -A f -z o
04
= 0
-0.2-
iB 2T8 3TB 4TB 5 T' t
-0-4-
-06-
-0-8-
- 1-0-
1.0
0.8-
0.6 I
0 =A -=30
0-4- MiC
1)- 0.2
=
tn
0
TB 218 3 4 1 5 116 7 8
T-
97B
- 0-2-
- 0.4-
- 0.8
- 1.0 - D.C. COMPONENT = O -
Fig. 10.12. Detector beatnote output for constant loop gain (mfo = constant)
and various tuning errors. TB is the beatnote period. D.C. components are
shown dotted
error were increased. The effective loop gain in this simplified

loop is mf, c/s, which means that the loop is capable of shifting
1 radian of phase in a time sec. Now if the tuning error is large,
note1
27,mfc
the beatnote frequency is high, and the time of a half cycle of beat-
is small. If the time
Zninfc
sec is larger than the beatnote
period, the loop cannot change the phase much from one half cycle of
beatnote to the next so that the beatnote tends to become sinusoidal.
Referring again to the loop filter of Fig. 10.9, the maximum pull -
in range is
Afmax =fc A/ 2m - m2
This may be written as
AO)max WC V2x
for small values of m, since then m ^ x.

Now the noise semi -bandwidth has been derived earlier as
1 + x2Twe
fNN 4xT
or
4fNN 1 1 4fNN 1
coc
x x2T
The maximum pull -in range may therefore be expressed as

rfNN 1
Acomax wcV2x 1/2
1/x TV:Xi]
For a constant value of fNN, the pull -in range can therefore be
indefinitely increased by reducing x and increasing T appropriately.
This will increase the quantity
4fNN 1
TA/
.
and hence cue must be appropriately increased, since (0, of
Vx
the above expression. However, apart from economic considera-
tions, the maximum pull -in range cannot exceed -14 where ft, is the
line scan frequency. Remember that the sidebands adjacent to the
sub -carrier component of the burst are separated from it by fL, and
have a strictly comparable amplitude (Appendix 1). For tuning
errors greater than -fl, an ambiguous beatnote between the oscillator
and the nearest sideband can occur, and the oscillator may even
lock to the sideband. This would give an incorrect reference fre-
quency, of course, and this effect is called side -lock.
Quite apart from the above considerations, the pull in time would
be excessive. For example, it is shown in Appendix 4 that for an
optimum K value the pull in time is approximately given by
4 .2(Af ) 2
TF
fNN3
If Af = 5 kc/s, and fNN = 70 c/s, then TF, would be just over

5 min.
This example illustrates the limitations of the straightforward
A.P.C. loop shown in Fig. 10.9. Thus, for large tuning errors,
pull -in time will be excessive unless fNN is increased, that is, unless
the noise performance is deteriorated. For the required small value
of fNN, therefore, reasonable pull in times can be achieved only if
the tuning error Af is kept small. This immediately suggests the
use of a crystal controlled oscillator.
In fact, there are two solutions to the problem of good noise
performance consistent with short pull -in time. The use of a
crystal oscillator is one solution; the use of a two mode loop, or
quadricorrelator, is the other solution.
10.13. Crystal oscillator A.P.C. loops

A circuit diagram of a crystal controlled oscillator A.P.C. loop
is shown in Fig. 10.13.
The oscillator is a modified Pierce crystal oscillator with the
screen grid of the pentode acting as the " anode " of the oscillator
triode. The output is taken from the actual anode of the pentode,
and is therefore electron coupled to the oscillator, thereby providing
some isolation. The tuned circuit in the anode is tuned for maxi-
mum output.
The reactance valve anode is capacity coupled to the crystal,
and its tuned circuit is adjusted so that the beat between the refer-
ence signal and the sub -carrier is nominally zero when the reactance
valve grid is earthed. This condition is easily observed on a colour
picture, when all the colours will be seen to pass through the spec-
trum range at the beat frequency.
Some cathode D.C. feedback is included in the reactance valve
for stabilizing the valve characteristics, and heavy decoupling is
advisable to prevent hum modulation of the oscillator phase,
particularly for a series heater chain. It should be noted that for
crystal oscillators the sensitivity of the reactance valve is relatively
PHASE PHASE DETECTOR
DETECTOR REACTANCE CRYSTAL
DRIVER STAGE VALVE OSCILLATOR BUFFER
+250V
FROM ikn
BURST
iokn 2.2 RR 4/k11
AMPLIFIER WF 0.01/2F
-RI 0.01#F
130 pF ]
-200,uH
D77
pF :70/2M
m." g
100
:28mH
pF g
REFERENCE
OUTPUT
3.3 pr
4 50 pF
6
max.
ropF HUE
68kn CONTROL
470 RD. Z77
1.5 An
!pi
oIALF 68011 47 kn
100 k
470 pF 005/[F/ 2'66 Mc/s
IDO/IF
CRYSTAL
18kIL / 001µF
220n- 1 15011
--
0 01/2F .r
I TT
Fig. 10.13. Circuit of a 2.66Mcls crystal oscillator A.P.C. loop

low because a large capacity change is required to shift the crystal
frequency. The sensitivity is also a function of the crystal frequency
accuracy.
For example, if a large capacity is required to be added to achieve
the correct frequency, the reactance valve itself must be able to
provide a relatively large capacity change.
The oscillator drive to the phase detector is obtained from an
amplifying stage whose anode circuit includes one half of a centre
tapped auto -transformer. This transformer should have very tight
coupling between its two windings so that its push-pull output is
symmetrical about earth. A convenient arrangement is a Ferrox-
cube " E " and " I " core of B2 material, since the high permeability
ensures tight coupling and also reduces the self capacitance of the
winding.
A total of about 10 turns of wire (preferably bifilar wound) is
usually required to tune the transformer to 2.66Mc/s with an added
trimmer of 10pF.
It should be noted that an accurate balance is desirable in the
phase detector, otherwise a spurious D.C. unbalance voltage will
be fed to the reactance valve which, in turn, will detune the oscil-
lator.
Accuracy of balance is dependent on the symmetry of the two
transformer windings and the equality of the two leak resistors of
(in this case) 470k S2.
Balance accuracy may be checked by measuring the D.C. phase
detector output by means of a high impedance voltmeter connected
to the junction of the 681d2 resistor and 470pF capacitor. (Alter-
natively, shift of oscillator frequency may be measured). The
voltage should be measured as zero, firstly with the burst signal
absent and the oscillator signal present, and secondly with the
burst signal present and the oscillator signal absent. The first of
these measurements checks the equality of the resistors and trans-
former windings, and the second measurement checks that the
impedances of the two halves of the circuit have equal phase angles.
In this connection the trimmer should be adjusted for accurate
balance rather than maximum output, though normally these two
settings are close together. As a general rule, the unbalance
voltage should not exceed about 1 % of the smallest voltage applied
to the phase detector, and it is advisable to use matched pairs for
the leak resistors.
The hue control shown makes use of the phase shift between
current and voltage which changes as a tuned circuit is tuned
through resonance. Considering a simple circuit of L, C and R
all in parallel, a phase shift of 45° occurs when C is changed by
an amount
1
SC =
woR
from its resonant value, and the magnitude of the impedance is

then 3dB below its resonant value. Thus, if R = 31d1
SC = 19.7pF
for a 45° phase change at 2.7 Mc/s. If a 50pF trimmer is used, and
the circuit is tuned by adjustment of the inductance when the
trimmer is set half way, then a phase shift of more than +45°
can be made by adjustment of the trimmer. The use of a relatively
large fixed capacity in the tuned circuit enables lead capacity to be
accommodated. Screened lead is necessary and its length will be
dependent on the physical position of the hue control relative to
the reference oscillator.
The value of the load resistance R will depend on the type of
synchronous detectors used in the demodulators. 31d1 has been
chosen in this case as a fairly representative value. Coupling
between the hue control and the oscillator itself should be sufficiently
small to prevent change of oscillator tuning with hue control
setting. This is one reason for the buffer output stage, which also
isolates the load from the oscillator. However, with a carefully
laid out circuit the buffer stage can be omitted provided the output
is sufficiently large and of low enough impedance.
Note that the hue control may be included in the burst signal
side rather than the oscillator side.
The loop design is a function of the static and dynamic phase
errors which can be tolerated, the maximum tuning error to be
expected, and the pull -in time which can be allowed. Typical
values for a crystal controlled oscillator A.P.C. loop are as follows:-
Burst drive to phase detector 80V peak
Reference drive to phase detector 35V peak
p. = 35 volts/radian
13 = 150 c/s/V
fe = 5.25 kc/s/radian
Maximum oscillator drift +200 c/s
Static phase error +2.2°
Noise semi -bandwidth fNN = 100 c/s
Shunt time constant of filter xT = 18 x 10-3sec
A.C./D.C. gain ratio of filter m ^ x = 10 x 10-3
Pull -in time for maximum tuning error 0.25 sec
A very important practical consideration in A.P.C. circuits is
the amount of stray feedback from the oscillator side to the burst
side of the phase detector. If feedback is present, the loop will
try to lock to its own oscillator; that is, the phase detector will
tend to change the oscillator frequency so that a 90° phase difference
occurs across the phase detector. The feedback therefore shows as
PHASE PHASE
ADVANCE ADVANCE
VB
VR
VR 1\
/
/
I VB VR
(a) (b)
Fig. 10.14. Phase and burst signals for the circuit

shown in Fig. 10.13. YR is the oscillator drive to top diode. (a) shows unstable
and (b) stable positions
an unbalance which is a function of hue control setting, and it can

be checked by measuring the oscillator frequency as the hue control
is rotated through its entire range with no burst applied. In case
the hue control affects the frequency directly, this can be checked
by earthing .the grid of the reactance valve while the oscillator fre-
quency is measured for various hue control settings.
The feedback between oscillator and burst input can be reduced
by fitting a screen directly across the diodes. Note, however,
that this will only reduce internal feedback. External feedback
can occur if the burst take off point is contaminated with reference
signal.
It has already been stated that when an A.P.C. loop is locked,
a 90° phase difference exists between the signals applied to the
phase detector (apart from the small static phase error). Whether
this is a 90° lag or lead will depend on the circuit configuration,
but as an example, consider the circuit shown in Fig. 10.13. Sup-
pose the burst drive is represented by VB while the reference drive
to the top diode is VR. First, draw VB as a reference vector as
shown in Fig. 10.14(a). Next, assume that VR leads VB by 90°.
If this is the correct phase relationship, then if VR is displaced in
phase, the control action should restore it. Suppose VR is displaced
clockwise by an angle 0, then the control action should try to
advance the phase of VR to restore it to its initial position. How-
ever, because VR is now more nearly in phase with VR, the bottom
diode will provide the larger output voltage, and since the output
is taken from the cathode of the bottom diode the output voltage
will be positive. This will try to reduce the oscillator frequency,
that is, retard the phase of VR. The control action therefore acts
in the same direction as the initial displacement 0, and the stable
position of VR is consequently as shown in Fig. 10.14(b).
If the above procedure is applied to Fig. 10.14(b), it will be found
that the control action acts oppositely to the initial displacement.
This method of determining the sign of the 90° phase relationship
is useful for checking that the correct reference phase can be achieved
within the range of the hue control. It is also useful for determining
the sign of the D.C. output voltage in the case of two mode loops,
as discussed in the next section. Note that a 180° phase shift can
be obtained by connecting the anode of the driver stage to the other
end of the push-pull transformer.
10.14. Two mode loops or quadricorrelators
It has already been mentioned that simple A.P.C. loops have
conflicting design requirements in that good pull -in performance
cannot be achieved unless in -sync noise performance is sacrificed.
The crystal oscillator loop overcomes this problem by its inherent
stability which allows a fast pull -in for a relatively weak control.
However, if a two mode loop is used, one mode can be designed
to give good pull -in performance regardless of in-sync conditions,
while the other mode can be designed for good in-sync performance
regardless of pull -in performance.
In this case the pull -in performance may be designed to cater for
relatively large tuning errors, and a crystal oscillator is not required.
It is estimated that an oscillator stability of about ±2 kc/s is feasible
at 2.7 Mc/s for a carefully designed LC oscillator, so that a pull -in
range of +3 kc/s should allow a reasonable factor of safety. A
fr,
stability of ± -2-, where fi is the line scan frequency, is the maximum
allowable drift if sidelock is to be avoided.
The change over from one mode to the other can be carried out
by an automatic switch which recognizes the in -sync condition of
the loop, and acts accordingly. Since this " switch " usually operates
from a quadrature relation between two signals, the term " quadri-
correlator " is often used to describe two mode loops.
Fig. 10.15 shows a block diagram of a typical quadricorrelator.
It consists of a conventional A.P.C. loop which is designed for the
required in -sync noise performance regardless of pull -in con-
siderations, and a second or auxiliary phase detector to which
INPUT LOOP PHASE EACTANCE

OSCILLATOR
11 DETECTOR VALVE
LOOP FILTER
LOW-PASS
FILTER
HAUDI IARY
DETECTOR
90° PHASE
SHIFT
4
Fig. 10.15. Quadricorrelator block diagram
is fed the burst signal and a quadrature signal from the oscillator.
The object of the auxiliary detector is to produce a D.C. output
voltage only when the loop has synchronized, and a zero output
for all other conditions. After passing through a low pass filter,
the auxiliary detector output operates a " switch " which is closed
when the output is zero, i.e. when the loop is unlocked, and open
when there is a D.C. output from the auxiliary detector, i.e. when
the loop is locked. When the switch is closed, the transmission is
increased through the filter and the design is arranged for good
pull -in performance. As soon as the loop has locked, the switch
is opened and the loop filter is then suitable for good noise per-
formance.
It will be remembered that during pull -in, the beatnote output
of a loop phase detector is typically as shown in Fig. 10.12. The
beatnote output from the auxiliary detector is a quadrature version
of this, and while the conventional beatnote of Fig. 10.12 is pro-
portional to sin 0, the auxiliary detector output is proportional to
cos 0. This is readily deduced from sin 0, and is shown in Fig.
10.16. Notice that this waveform is symmetrical about the time
axis and therefore has a zero D.C. component. If the auxiliary
20
detector output is taken through a low pass filter to remove the
A.C. components, the output will be zero whenever a beatnote
exists, that is, when the loop is not locked.
When the loop has locked, however, 4. is zero (apart from the very
small static phase error) and the auxiliary detector output is then a
D.C. voltage equal to
ILA COS q ILA
where ILA is the sensitivity of the auxiliary detector. This D.C.

voltage can be positive or negative, as desired, the sign being
1.0
0.8
0.6
04
02
0
0 TB 3TB 2TB
2 2
-02-
-0 4- 1
a= -=I 05
rnrC
-06
- 1.0
Fig. 10.16. Quadrature detector beatnote
dependent on whether a 90° lag or lead is included. Normally a

negative D.C. output is required so that a valve can be biased off
to change the loop filter characteristic.
A typical quadricorrelator circuit, designed by the Hazeltine
Research Corporation, is shown in Fig. 10.17. This design is
OUTPUT I OUTPUT 2
a 3-I2pF a
+375V
.14ALF 14mHF, 100.
100pF ioopF
18kn 10 kn.
11
220pF 0.0022,0
6ALS
Vi(e)
1.5MI1
sivVvvv 2 70 pF
LOOP
DETECTOR +270V
22 kn.
271(11.
0.001/.4F 0.001,0 IOOpF

IMn 800 IMI1 V2 V3
8W LAR/2, 3pF
0.047,0 22M.F1
FRO M 560.0. 12AT7
BURST
AMPLIFIER "r2 2opF
0.001/cF ooo1,u5 Vs(a)
iMfl iMil 270k11
+12 AX
AUXILIARY 4ALF 47 kn
DETECTOR 1-
V5(b) 315Mcis
6AL5 oo005,12F R.F.C.
1.5AD. .-120
022,0
lek.11 1.2kfl
0.01,LtF 0.01/0
0-005/2F1 -T T
REACTANCE OSCILLATOR BUFFER
BY-PASS VALVE
Fig. 10.17. 3%58 McIs quadricorrelator circuit

based on the American system sub -carrier frequency of 3.58 Mc/s,
but of course it can be modified for 2.66 Mc/s operation by changing
the various tuned circuits appropriately.
The conventional A.P.C. loop consists of a detector V1, reactance
valve V2 and oscillator V3. A buffer stage V4 feeds a quadrature
network which supplies two quadrature phased reference outputs
for the receiver synchronous detectors.
The auxiliary detector V, is fed with the same burst signal as the
loop detector V1. but its reference feed is 90° out of phase with the
V1 reference. The auxiliary detector output is taken through a
simple low pass filter of 1001S2 and 0.22 ILF, so that any A.C. com-
ponents are heavily attenuated. As explained earlier, this auxiliary
detector output is zero when the loop is not locked, and (in this case)
a negative D.C. voltage when the loop is locked.
Now the conventional loop filter is designed so that the noise
bandwidth is satisfactorily small for in -sync performance, but a
cathode follower V6 is arranged so that it can increase the A.C.
transmission through this filter and thereby increase the noise
bandwidth for the unlocked condition.
The cathode follower grid resistor is returned to the auxiliary
detector output, so that in the unlocked condition of the loop,
the cathode follower operates and causes rapid pull -in. When
pull -in has occurred, the auxiliary detector output is a negative
D.C. voltage, and this cuts off the cathode follower so that the
original filter circuit is restored.
The essential additions to the circuit of a simple A.P.C. loop
therefore consist of an auxiliary phase detector and a cathode
follower by-pass. However, because the pull -in range is indepen-
dent of the in -sync design, very much greater oscillator drift is
permissible so that the oscillator need not be crystal controlled.
This type of quadricorrelator is often called the variable ratio
quadricorrelator because the A.C./D.C. gain ratio m (and therefore
mfg) is increased for pull -in. Instead, it is possible to increase the
phase detector sensitivity µ while keeping m constant, so that
= and therefore 'life, which controls pull -in, is increased.
This is the action of the variable gain quadricorrelator, and in this
case the auxiliary detector bias is used to reduce the gain of one of
the burst amplifier stages (and therefore the burst drive to the
phase detector) when the loop has locked.
It is perfectly practicable to employ the variable gain and variable
ratio techniques simultaneously, but in the authors' opinion the
loop gain fe should be high even for the in-sync condition to keep
the static phase error small.
Typical values for a variable ratio quadricorrelator circuit are as
follows:-
Burst drive to both detectors 80V peak
Reference drive to both detectors 35V peak
p. = 35 volts/radian
P = 2 kc/s/V
fe = 70 kc/s/radian
Maximum oscillator drift +3 kc/s
Static phase error +2.5°
In -sync noise semi -bandwidth fNN = 100 c/s
In -sync shunt time constant of filter 6 x 10-3 sec
In -sync A.C./D.C. gain ratio of filter 0.5 x 10-3
Out of sync A.C./D.C. gain ratio of filter 15 x 10-3
Pull -in time for maximum tuning error 1.5 sec
It should be noted that the increase in the value of the A.C./D.C.
gain ratio for the not -in -sync condition, produced by the cathode
follower by-pass, is equivalent to an increase in the value of the
shunt resistor of the filter, and not to a decrease in the value of the
series resistor. This is shown in Appendix 7.
10.15. Colour killing
If a black and white picture is displayed on a colour receiver
whose chrominance circuits are operative, spurious colour effects
will occur due to noise and luminance components which fall
within the chrominance pass band. If a burst is present, the effects
of the latter are subjectively reduced because of the frequency lock
between the receiver reference generator and the line scan, but in
the absence of a burst it is very desirable to switch off the chromin-
ance circuits. This is particularly true if A.C.C. (see Section 10.16)
is fitted, in which case the chrominance gain is a maximum when
there is no burst.
Since the presence of the burst is a reliable indication of a colour
transmission, it is reasonable to place the responsibility on the
reference generator for automatically switching the chrominance
circuits. Such a switching action is appropriately referred to as
colour killing.
In order to switch off the chrominance circuits, a negative bias is
generally convenient. Hence, the colour killer must provide a
negative D.C. voltage when the burst is absent and a zero voltage
when it is present. However, reference generators normally give the
converse of this; for example, the auxiliary detector output of a
quadricorrelator is zero for not -in -sync (i.e. no burst) and a negative
D.C. for in -sync. However, a negative voltage which appears
only when the reference generator has synchronized or when there
is a burst, can be utilised for operating a valve which will give the
required bias for killer switching.
A suitable circuit is shown in Fig. 10.18. A triode is used as a
grid controlled rectifier; that is, when its grid is earthed, the triode
anode and cathode operate as a rectifying diode, so that a negative
PULSE INPUT -10--1 KILLER OUTPUT
FROM LINE
TIMEBASE o D.C. BURST PRESENT
- D.C. ^ ABSENT
D.C. CONTROL
FROM REFERENCE
GENERATOR
- D.C. BURST PRESENT
o D.C. ABSENT
Fig. 10.18. Colour killer circuit
D.C. output is obtained if a waveform is applied to the anode.

This waveform may be a 50 c/s sine wave (e.g. the mains supply)
or preferably a pulse waveform from the receiver line time base.
The higher frequency of the line waveform allows smaller (and
therefore quicker action) time constants to be used in the smoothing
circuit. Positive going pulses are desirable since a negative D.C.
output is required.
Of course, when a negative bias is applied to the triode grid, the
rectification action is prevented and the killer output becomes
zero.
Sometimes the triode grid standing bias is made adjustable so
that the killer output becomes zero only when the D.C. control
from the reference generator has reached a sufficiently large negative
value; a spurious negative D.C. due to noise, for example, can
thereby be " backed off ". Such a bias control is usually called
the killer threshold adjust.
In the case of a simple A.P.C. loop, such as Fig. 10.13, the D.C.
control for the colour killer may be obtained from the anode of the
lower of the two phase detector diodes. The D.C. voltage at this
point is increased negatively when a burst is present. When there
is no burst the D.C. voltage here is equal to the peak value of the
reference drive. This method of deriving the killer control bias is
unreliable for poor signal-to-noise ratio signals, because it depends
on simple amplitude detection of the burst; any noise present can
also produce a D.C. voltage. Thus, a noisy monochrome signal
may be interpreted as a colour signal.
In the case of two mode loops, however, the D.C. output of the
auxiliary detector is a very reliable indication of the presence of a
burst. The fundamental reason for this is that the auxiliary detec-
tor, like the loop detector, operates as a synchronous, and not as an
amplitude, detector. Thus, when the loop is not synchronized,
noise present with the incoming burst will produce fluctuations
above and below zero, which, when integrated by the low pass
filter, will produce a zero D.C. output. When lock has occurred,
noise fluctuations will tend to reduce the D.C. output somewhat,
since fluctuations in the phase will always act in a positive direction;
the output is proportional to cos 0, where 0 is the phase error, and
as 0 varies about zero the value of cos 0 will vary from unity to
less than unity. However, the D.C. output can always be made
large enough to avoid ambiguity. Note that the use of the auxiliary
detector for colour killer bias means that the chrominance channel
is made operative only when the loop has locked. Spurious colour
is therefore avoided during pull -in.
It must be remembered that the killer bias must not be applied
to the chrominance channel prior to the burst take -off point, other-
wise the reference generator will never receive a burst signal. How-
ever, should this situation arise, it may be overcome by adding the
burst gate pulse to the killer bias so that, for monochrome trans-
missions, the killer action applies only to the active part of the line
scan. In general, this technique is not recommended since spurious
colour will be displayed on the left of the picture unless carefully
adjusted line blanking is employed.
10.16. Automatic chrominance control
It is possible to arrange an automatic gain control for the chromin-
ance channel of a receiver so that the chrominance output remains
practically constant as the chrominance input level changes. Such
an arrangement removes the probable need for a change in saturation
adjustment from one station to another, and it also tends to com-
pensate for any differential fading between the chrominance and
luminance signals. To distinguish this type of gain control from
the conventional A.G.C. which may be fitted to the receiver, it is
usually called A.C.C., or automatic chrominance control.
As in the colour killer, the burst is the only reliable measure of
the chrominance signal as it appears during colour transmissions
only and its amplitude is independent of picture content. In
fact, the negative D.C. voltage which is used to bias off the colour
killer valve whenever the burst is present may be used as an A.C.C.
voltage by feeding it back to control the gain of a chrominance
stage prior to the burst take -off point, provided that this negative
voltage increases with burst amplitude.
Thus, if the negative bias appearing at the anode of the lower
diode of the simple loop of Fig. 10.13 is fed back to the chrominance
stage which feeds the burst into the loop, then the burst input will
tend to remain constant and, of course, so will the chrominance
signals corresponding to a given picture. Manual saturation
control may then be applied after the burst take -off point.
The above method of A.C.C., like the killing action, will be
unreliable under poor signal-to-noise conditions because the D.C.
control is obtained from amplitude detection. This would mean
that when switching from a " clean " to a " noisy " transmission,
the saturation would decrease in the latter case, because the noise
would simulate a larger burst.
It follows that reliable A.C.C. is best accomplished by synchro-
nous detection, and the D.C. output of the auxiliary detector of a
two mode loop may be utilized for this purpose, as well as for colour
killer control, provided that the auxiliary detector output is arranged
to be substantially dependent on the burst amplitude rather than the
reference amplitude. This can be done by making the burst feed
to the auxiliary detector smaller than the reference feed.
10.17. Reference generator circuits

The circuits which have so far been given for passive and dynamic
integrators have deliberately not included colour killers or A.C.C.
for the sake of simplicity. However, more complete circuits will
now be given, and these will additionally illustrate various modifica-
tions which may be encountered in reference generator designs.
10.18. Passive/dynamic integrator

Fig. 10.19 shows a circuit designed by R.C.A. for their CTC-5N
series of receivers. It is a hybrid design in that the output of a
passive integrator is used to injection lock an oscillator whose
frequency is controlled by the crystal of the passive integrator.
The gated burst is fed to a transformer whose secondary is con-
nected to the oscillator grid via a quartz crystal. The mounting
capacitance of the crystal is neutralized by a 1.6pF capacitor
connected to the appropriate anti -phase end of the primary winding,
+300V
A.C.C. KILLER
THRESHOLD 2.5mn THRESHOLD
KILLER
2.5MO. 0.011,"
V3 11'400\ 114 PULSE

TIMEBASE
Ann
A.C.C. OUTPUT 470th
(TO FIRST STAGE
OF CHROMINANCE
,1 0 KILLER OUTPUT
AMPLIFIER) (TO SECOND STAGE
2.2 M.(1 o.oviF I1 OF CHROMINANCE
AMPLIFIER)
6AW8 13.1pf
+300V
I k.O.
0.0IpF INGO 1 X REFERENCE
OUTPUT
68k.O1
47pF
Z -REFERENCE
OUTPUT
pF 680pF
3300F
68k.ft
6008 H
CONTROL
500.O
BURST 3.58M cis OSCILLATOR

AMPLIFIER CRYSTAL
Fig. 10.19. Circuit of a .3.58McIs passive/dynamic integrator
and the circuit so far described is similar to a simple passive inte-

grator, the narrow bandwidth (quoted as 400 c/s) of the crystal
determining the noise performance. Apart from any noise com-
ponents within this pass -band, the signal fed to the oscillator grid
is substantially the subcarrier component of the burst.
The electron coupled oscillator is arranged so that the crystal
is included in its grid circuit, hence the oscillator frequency is
crystal controlled. Consequently, when a burst is present, the
sub -carrier component of it synchronizes the oscillator. The hue
control operates by changing the effective capacitative load on the
output tuned circuit.
When the burst is absent, the oscillator develops a constant
When a burst is present, however, the
negative bias at its grid.
sub -carrier component of it produces a larger bias voltage by
rectification at the oscillator grid. In addition, a crystal diode
voltage doubles this bias by conducting on the negative peaks of
the C.W. Hence, by backing off the standing oscillator bias by
means of threshold controls, the D.C. voltage output may be used
for colour killing and A.C.C.
Note that although the killer bias is developed by simple ampli-
tude detection, its noise performance is adequate because of the
narrow crystal pass -band.
It should be noted that the noise performance of this circuit is
entirely dependent on the crystal parameters. A very narrow noise
KILLER
+131W
13-4-- +75V PULSE FROM
0010 LINE TIMEBASE
6
150k11
KILLER
THRESHOLD
10 KILLER OUTPUT
loM11
REFERENCE
0'047/2F" ""0047/./F OUTPUT
6EA8
+220V
47 kfl 15kfl
+38W 2.2Mfl
1kfl
1,000pF 330pF
4701,11
4F
470kfl
VI
V3 1.5k11
33kf1
22011
68Mil
3.58Mcis 27k11
4 CRYSTAL
'132pF 47kfl
-HUE CONTROL 001,11F
68on 0611./F
Opf
BURST PHASE CRYSTAL REACTANCE
AMPLIFIER DETECTOR OSCILLATOR VALVE
Fig. 10.20. Circuit of a simple A.P.C. loop and killer

bandwidth can be achieved only by the use of high Q crystals which
would lead to large static phase errors.
10.19. Simple A.P.C. loop
A straightforward and effective crystal oscillator A.P.C. loop,
used by R.C.A. in their CTC9 series of receivers, is shown in Fig.
10.20.
The general principles of this type of circuit have already been
discussed, but note that here the burst signal is fed push-pull to the
detectors, and the detector output is taken from the junction of the
diode leak resistors. The hue control varies the phase of the burst
signal rather than the reference signal.
The bias for the killer valve is obtained from the negative voltage
appearing at the anode of the top diode, and since a negative
voltage due to the reference appears at this point when the burst
signal is absent, a positive backing off voltage is applied to the
killer grid from a killer threshold control. This control is also
used to back off any spurious D.C. voltage caused by noise which
may be present with the monochrome signal.
The positive bias action of the killer threshold control tends to
lift the anode of the top diode positive. This would normally
upset the D.C. balance of the phase detector, but a compensating
negative D.C. component is obtained by bleeding the cathode of
the lower diode to earth through 6-8M C2.
10.20. Quadricorrelator circuit
A quadricorrelator circuit used in the G.E.C. TTIV series of
receivers is shown in Fig. 10.21.
This circuit is designed for series run heater operation, and
certain precautions are advisable to prevent hum modulation of
the reference phase. Thus an earthed cathode Colpitts oscillator
is used, one end of the heater of the oscillator and reactance valve
(LZ329) combination being earthed. Next in the chain are the
filter by-pass and loop and auxiliary detector heaters. The cathode
of the filter by-pass is tapped down to reduce hum effects, since
any hum on this cathode will not be reduced when the valve is
cut off for in -sync operation.
The electron coupled oscillator has large swamping capacitors
in its tuned circuit to aid stability, and a negative temperature
coefficient capacitor provides compensation for warm-up drift.
The H.T. supply to the screen of the oscillator is obtained from a
potentiometer to reduce the dependence of oscillator frequency
on H.T. line variations. Coarse oscillator frequency adjustment
LOOP +200V
DETECTOR
MIL 1081.1 4.78.11.
3,000pF=
100,a14
68kni 47pF
2.2M17.
0.04pF.. 3LZ 329 V6 iLZ32

V4 33pF1 V5
,--HN 309- 470pF 10014
82011
6/./F
w - 251.".1
4.7pF 1pF
PAPER 3.3kfl
220811
3,000 Imr rT l,360pF
100k11
0.04/./F 4,700g
=pF 6F ,100kfl
CERAMICON
47pF 22k 3idi
4 7k1 2.2kn "kn
FROM To -04µF 10 .04/..a
BURST T T Tag
AMPLIFIER 1
REFERENCE FILTER REACTANCE OSCILLATOR
680pF DRIVER BYPASS VALVE
it
+2 OV
47p ISOk11
2.2811
680pF 230V
i A.C. INPUT
150k1.1
KILLER OUTPUT
OV IN SYNC
680k1.1 -200V OUT OF SYNC
0.5,a1 IOOpF
28mM
68OpF IWO 33k11
D77 H71309 REFERENCE
PH309 OUTPUT
VIO VII
47,r* 10pF."
0.01µF
IMII
80pF
_1/IF
680pF 50pF max.
0.04 HUE
.-ioopF 2201711 54 100 kn. AlF CONTROL
2 10.04µF 15011
A.C.C.
2015
TOME 004µF
me. AUXILIARY KILLER KILLER OSCILLATOR
DETECTOR AMPLIFIER BUFFER
Fig. 10.21. 2.66Mcls quadricorrelator circuit

is by way of a brass slug tuned coil, and a fine oscillator frequency
control can be made with a 25pF trimmer.
The loop detector is fed via an RC phase shifting circuit of 4.7pF
and 4.7kS2 so that the reference phase at the loop detector leads
the phase of the transformer push-pull output by about 70°. The
magnitude of the auxiliary detector output is therefore proportional
to cos 20° (or 0.94) instead of cos 0° (or 1) which would obtain for
a quadrature phase. The non-quadrature relation therefore only
slightly affects the magnitude of the auxiliary detector output.
The RC circuit of 4.7pF and 4.7kSI also reduces the amplitude
of the reference fed to the loop detector so that the detector output
is chiefly dependent on the reference signal amplitude. One of the
coupling capacitors is adjustable so that an accurate balance may be
obtained for the loop detector.
The reference feed to the auxiliary detector is large so that the
output is chiefly dependent on the amplitude of the burst signal.
This detector is arranged to give a detected burst pulse output
rather than a D.C. output so that the detected burst pulse may be
amplified by an " audio " amplifier. The anode waveform of this
amplifier therefore consists of positive going pulses whose amplitude
is proportional to the burst signal. These pulses are D.C. restored
one way by the diode D1 and the reverse way by the diode D2,
and the outputs of the two D.C. restorers are connected in series.
The result of this arrangement is that, when not in sync, the diode
circuits give equal but opposite D.C. outputs, that is, zero D.C.
output. When in sync, however, one diode circuit gives a larger
D.C. output than the other, and a net negative D.C. output is
obtained. Because of the appreciable gain of the " audio " ampli-
fier, a large D.C. output indicative of lock is readily obtained, and
reliable killer action results even for very small burst inputs; in
fact, provided the burst signal is large enough to achieve lock, the
killer becomes completely " unkilled ". This arrangement avoids
the occurrence of a " semi -killed " state in which, although the
loop is locked, the auxiliary detector output is not sufficient to
switch off the killer completely and so a reduction in saturation is
observed.
The amplified burst output mentioned above may also be used
for A.C.C. purposes, in which case very strong A.C.C. action is
obtained because of the high gain of the system.
10.21. Alternative possibilities for reference generation
Before leaving the subject of reference generators, brief mention
should be made of two techniques which may well prove useful.
The first of these is applicable to colour receivers which feed
difference signals to the display tube. Thus, since the burst phase
is -(B' - Y'), a correctly adjusted receiver should give a zero
output in its (R' - Y') output corresponding to the synchronously
detected burst. Therefore, if a gating circuit is arranged to " look
at " the detected (R' - Y') output during the burst period only,
and if this gated output is used to A.P.C. an oscillator from which
the reference signal is derived, then the oscillator reference will
automatically adjust its phase to give a zero output for the detected
burst in the (R' - Y') phase. This, of course, corresponds to the
correct hue setting, so that no manual hue control is required.
Provided the gain of the A.P.C. loop is adequate, the responsibility
for correct hue setting then rests entirely on the accuracy of the
transmitted burst phase.
Further, if a similar gating system is applied to the receiver's
detected (B' - Y') output, then a negative going detected burst
pulse should be obtained when the receiver's reference signal is
correctly synchronized. This signal may therefore be used (as
the equivalent of an auxiliary detector output) for quadricorrelator
switching of the A.P.C. loop filter, and also for colour killing and
A.C.C.
Preliminary tests carried out by the authors have indicated that
the above technique is feasible. It has the advantage of economy
in that the whole of the receiver's chrominance amplifier is used for
burst amplification, and of course no hue control should be required.
While on the subject of automatic control, it should be pointed
out that automatic setting of the saturation is practicable. Thus,
since the peak -to -peak burst amplitude is equal to the sync pulse
height when the chrominance gain bears the correct ratio to the
luminance gain, a circuit which compares the detected burst and
the sync pulse height may be arranged to automatically adjust the
saturation accordingly.
The second technique (which as far as the authors are aware
has not yet been investigated), takes advantage of the fact that in
the case of N.T.S.C. systems which use A.M. sound, the sound/vision
carrier beat is simply related to the sub -carrier frequency. Thus,
for the British adapted N.T.S.C. system, for example, the sub -carrier
frequency is i of the sound/vision carrier beat. Consequently, if a
strong carrier beat can be derived which is sufficiently free of sound
and vision modulation, the reference may be easily generated from
it by means of a simple regenerative divider.
The possible disadvantage of this technique is the' need to remove
the modulation, but a fundamental advantage (particularly for
positive modulation systems in which the vision carrier during the
active line period is never less than 30%) is the advantageous duty
ratio compared with that of the conventional burst. Note that this
technique is not applicable to F.M. sound N.T.S.C. systems.
10.22. Summary
In this chapter we have seen that the receiver's reference sine
wave which has to be locked in frequency and phase to the trans-
mitter's sub -carrier sine wave can be derived from the transmitted
burst signal. Certain specifications as to the accuracy of the static
and dynamic phase of the reference sub -carrier must be met to
reduce overall hue errors and hue fluctuations due to noise.
The transmitted burst signal must be gated out of the sub -carrier
waveform to prevent interference from the picture content, and
the resulting gated burst waveform is then suitable for reference
frequency generation.
The reference frequency generator itself is the equivalent of a
very narrow (of the order of 100 c/s) band-pass filter tuned to the
sub -carrier frequency, and it may be passive or dynamic in form.
Most generators used at present are dynamic integrators in the
shape of A.P.C. loops, and they are either crystal oscillator con-
trolled or two mode quadricorrelators. Either extreme oscillator
stability or two mode action is necessary so that fast pull -in or
stabilization time may be achieved simultaneously with good noise
performance.
Besides providing a reference signal, most reference generators
also supply a D.C. bias voltage which can be used for switching
off the receiver's chrominance channel during monochrome trans-
missions (colour killing) and for automatic gain control of the
chrominance channel (A.C.C.).
CHAPTER 11
Operation of the Shadow Mask Tube

11.1. Introduction
Most of the colour receiver circuits so far described can be used
with any type of display device, although it is usually more econo-
mical to operate single -gun tubes with their own special types of
decoding circuits. This chapter is concerned with those circuits
which are peculiar to the shadow mask tube. The basic principles
of construction and operation of this three-gun display tube have
been outlined in Section 3.8. Here we shall be concerned with the
details of the time base, convergence and bias circuits for a typical
shadow mask tube, or aperture mask tube as it is sometimes called.
11.2. Electrode voltages
Each of the three electron guns of the shadow mask tube is of
tetrode construction and electrostatically focused. Separate con-
nexions are brought out for the cathode, modulator grid and screen
electrode of each gun. The 6.3V, 0.6A heaters are connected in
parallel and the focus electrodes are all common to the one terminal.
The phosphor screen, mask and final anodes are operated at 20 to
25kV with a maximum total beam current of 800 to 1,500µA. The
focus voltage is typically 4,000V. Both focus and e.h.t. voltage can
be obtained from the line time base. Since purity and convergence
adjustments depend upon the e.h.t. voltage, whilst varying beam
current load can affect the line scan and boost voltage, it is necessary
to regulate the e.h.t. supply.
There is considerable choice of screen -cathode voltage, 400 to 600V
being usual, although higher screen voltages can give improved grey
scale tracking. The grid -cathode bias may be 50 to 150V, although
the actual D.C. voltage on the cathode with respect to earth will
depend upon the type of signal drive arrangement.
Typical circuits for providing bias controls are shown in Fig. 11.1.
11.3. Deflection yoke
The three guns of the shadow mask tube necessitate a larger
diameter neck than is usual with monochrome tubes. The R.C.A.
21CYP22 tube has a neck diameter of 2 in., which is to be compared
308
+385V
LUMINANCE SIGNALS
27 An
RED
SHADOW MASK
1200 GREEN
TUBE
-4- BLUE
910.n.
1.1mH
10
9n. 680pF
VERTICAL 56 kn. 0- 0 47/./ F 561,11
sot
BLANKING r IOkn
3.3 k 11 R Y
300,0 18 An 1-117.
390 lo L
-
V2
FOCUS E.H.T.
I MR
22 pF FROM LINE
12 BH 7A TIMEBASE
2M11
RED
12 BY 7A
8 2kR
IImH
-20V 680 kR
0.047/./F
CONTRAST 10 kn. 1001,11 100k.n. 100AR
MASTER
BRIGHTNESS G -Y
250k 11 F.v11-71.
500n. 390kfi.
V3
,;(zzoIl +385V
I Mn
180 All 12 BH 7A
330 pF"'" 2M11 B IG R
13.047A2F
220011' GREEN
I Mil M11 I M11
SCREEN
LUMINANCE CONTROLS
OUTPUT IImH I
0.047/1F
10k n. COLOUR DIFFERENCE SIGNALS
B -Y
V4
F.711:3
390kfi BOOST
VOLTAGE
12 BA TA I MR I0k11 +BOOV
Fig. 11.1. R.C.A. bias controls 2 Mll
for a shadow mask tube in receiver CHROM !NANCE BLUE GRID OR BACKGROUND
type CTC5T1 OUTPUT STAGES CONTROLS
with the 14 in. for a monochrome 70° tube. The deflection yoke
has, therefore, to fill a larger volume than normal with magnetic
flux. At the same time, if all three beams are to be treated alike,
field aberrations must be avoided over a larger volume than is
tolerated with black and white tubes, and graded windings are
essential. To further improve the uniformity of the field, the
deflection coil window area is usually made even larger than the
size fixed by the tube. Unwanted field effects due to the end turns
of the yoke must be minimized by bending them sharply away from
the tube neck. The angular separation of the three beams is only
two degrees and external magnetic and electrostatic fields must be
kept to a minimum To prevent any stray field from the deflection
coils reaching backwards into the convergence region, the deflection
+400V
BOOST
+700V
GATING
0 GANGED
PULSES TO O
RECEIVER O
LINE SHIFT
CONTROL
FROM LINE EFFICIENCY

OSCILLATOR .000 DIODE
LINE SCAN
COILS
TO C.R.T.
ANODE
V34
V 55
6CB5
_ET E.H T
LINE OUTPUT E.H.T. RECTIFIERS REGULATOR
Fig. 11.2. Line time base of a Murphy colour receiver

OPERATION OF THE SHADOW MASK TUBE 311
yoke is screened from the convergence coils by means of a copper
shield and a ring of high permeability magnetic material.
The equilaterally spaced guns are not symmetrical with respect
to the scanning fields and the defocusing effects which occur with
uniform deflecting fields produce convergence errors which are
different for the three guns. Since the blue image has the least
luminance and produces the least distressing effects when it is in
error, the guns are arranged to ease the convergence of the red
and green spots by making the red and green convergence errors
similar. Capacitive unbalance effects may be different for the three
guns and ringing on the line time base may be more marked thari on
monochrome.
Circuit characteristics of a typical R.C.A. deflection yoke are:
Horizontal deflection coils 11.8mH, 70 and a deflection angle of 70°
for a peak -to -peak current of 1.7A.
Vertical deflection coils 115mH, 5511 and a deflection angle of 55° for
a peak -to -peak current of 0.5A.
11.4. Line time base
The basic design of the line time base can be conventional.
Either directly synchronized or flywheel oscillators can be used and
the output stage can utilize an energy recovery diode circuit. Sample
designs of line time bases are given in Figs. 11.2 and 11.3
The e.h.t. supply of the Murphy receiver is obtained from the
voltage doubler circuit of V37 and V39, the third rectifier V38 main-
tains the average positive potential on the anode of V at a value
slightly below that of V39. This arrangement is more efficient than
using a high resistance in place of V38. The e.h.t. voltage is stabilized
by the shunt regulating action of V36, whose grid voltage varies as
the beam current of the cathode-ray tube alters the potential on the
cathode of V39. The focus voltage is also obtained from the cathode
of the first rectifier V39. The heaters of two of the rectifiers are
fed from a separate transformer as there is not room on the core of
the output transformer for more than one winding. In this parti-
cular receiver the line time base also provides pulses for gating out
the colour burst and for the colour killer circuits.
In the G.E.C. receiver the e.h.t. is produced by a single rectifier
and the focus voltage by another rectifier. The shunt regulator
triode is controlled by a fraction of the boost voltage since this
varies with the beam current loading on the output stage.
It is not practical to obtain picture shift by the simple means
often adopted in monochrome receivers, such as a magnetic shuffle
plate. The shift field must be uniform over all three beams and
LINE BLANKING PULSE 800V BOOSTED H.T.
TO LUMINANCE TO VERTICAL TIMEBASE 4-H.T. 400V
OUTPUT STAGE COLOUR TUBE SCREENS 10011
SHIFTLINE 1 FROM VERTICAL
TIMEBASE
47k11 50pFT
.
LINE
PI LINEARITY--
100011 151(1).
47011 380mH
220k11 0.25g gaV2
I 6AU4
FOCUS Iii
LINE VI INN
li'l
SYNC n ISOLATING
INPUT0 1/..tF
III- -INDUCTANCES .,
22mH
, i
1
.' II
. t
a II 1 VLF
I. 250pF- 250pF LINE
,-.115//F E.H.T.
.i- 1,DEFLECTION
,
23kV
I'g 310.0.
.T. 470 1.1 . COILS
250pF
C I/IF
I
0.01pF ME lc/ 0.002,0
4711 *it --,,/ (CAPACITY OF
o
-II-, V WIDTH°gm
SWITCH WIDTH 2M.O. POLYTHENE
TZ2pF CHOKE "BOOT")
100k11
481
.----I 10kn
10011 6470kfl 4700.(1
1330pF TO DISPLAY TUBE
b b FOCUS
1.5mn0.005/2F
ff
11
vs LINE
1 "In
TO
DRIVE FUSE
CONVERG.
tokri OUTPUT 6.8mn \
015/IF-. 900pF TO ISOL TED
=0001,0 E.H.T. MAINS WI NDING
LINE "".1A .,..7714. ADJUSTER
6 .8mn 4701,11 (500 V)
HOLD 500pF .--,o °A° 21
k
470011- 3.3011 oo/
loon: ,.._.... 6.8mn
_
- - E.H.T.
- - --
BLOCKING FOCUS
OSCILLATOR OUTPUT RECTIFIER DIODE STABILIZER
Fig. 11.3. Line time base of a G.E.C. colour receiver

must not interact with the convergence fields. Vertical and hori-
zontal shifts of about + 5 % of the picture size are normally obtained
by passing D.C. through the field and line deflection coils. In
Fig. 11.2 two ganged potentiometers are cross -connected to allow
a picture shift either side of the central position. The D.C. used
for picture shift is kept out of the line transformer by isolating con-
densers, and the shift circuits are prevented from short circuiting the
deflector coils by isolating inductances.
The design of the output transformer follows the general principles
developed for good quality monochrome receivers, but it is pre-
ferable to keep the flux density as low as is practicable. The anode
circuit to deflector coil turns ratio may be 1.6 to 1 and the secondary
inductance may be of the order of 75mH. The e.h.t. overwinding
needs careful design and insulation, of course, and may be encased
in a silicone rubber moulding.
All the e.h.t. points and connectors are housed in small anti -
corona shields and pressings so that they will not corona under
humid and dusty conditions.
Any high voltage stream of electrons will generate X-rays when
the electrons collide with metallic electrodes and similar structures.
The use of X-ray absorbing lead glass for valve and cathode-ray
tube envelopes minimizes the emission of X-rays outside the receiver
and all commercial designs are quite safe in this respect, giving con-
siderably less than the maximum permissible dosage even in the
immediate vicinity of the cabinet.
However, the receivers should not be operated for any length of
time without the protective glass over the tube face and service
engineers should not spend long periods close to unshielded line
time base rectifiers and shunt stabilizers when the chassis is out of
the cabinet.
The final anode coating on the inside walls of the R.C.A. 21CYP22
tube is split into two sections which are connected together through a
resistance of 501a2. This is to safeguard the cathodes, should a
momentary internal arc occur. The e.h.t. supply must always be
connected to the outer of the two coatings.
The e.h.t. smoothing is obtained by the capacity between the
final anode coating on the tube bulb and the graphite coating on the
outside. A typical e.h.t. regulation is 5 %. An interesting method,
due to R.C.A., of stabilizing the e.h.t., is to inject positive pulses
into the grid of the line output valve so that it conducts during the
flyback periods and absorbs some of the e.h.t. power. The regulat-
ing pulse amplitude is controlled by the boost line and effectively
maintains the e.h.t. voltage constant. However, this arrangement
BOOSTED H.T. SUPPLY FIELD BLANKING PULSE TO
FROM LINE TIMEBASE LUMINANCE OUTPUT STAGE
FIELD
SYNC 6 f1 NJ. SUPPLY TO
INPUT
0 LINE TIMEBASE
15 kn 50aF, 5oV
HEIGHT 0
+H.T.
6w
420V
IM11
RI
470 kJ.). t;
!i; VERTICAL
;;IDEFLECT ION
100kn 100k.n. ,' COILS
CO OI/IF 47n ,,,
it
0.02/IF
11 500,aF N
VERTICAL
SHIFT
VI V2
D77 0.1,0
13N olii ,
16.6 \II 11.1
i
1
o
1
7 ')F.12 OUTPUT SOCKET

0.08/IF i
I
VERTICAL 12 " 10 !TO CIRCUITSCONVERGENCE
470k11. 33k11 -LINEARITY B 0 ti
I I,
IM11 100kn
VERTICAL
500/LFT LINEARITY A
VERTICAL
HO D
33kn. 47011 100k.n.

6W 10kn.
FLYBACK
SEPARATOR OSCILLATOR CONTROL OUTPUT
Fig. 11.4. Field time base of a G.E.C. colour receiver

calls for increased dissipation in the output valve and the total time
base current varies with the picture content.
Line time base interference, both radiated and mains borne, needs
special preventive measures. The mains leads should be filtered
and both the time base and the deflector coils should be as completely
screened as is practicable.
11.5. Field time base
The vertical deflection circuits are conventional except that
increased output is called for, compared with 70° monochrome
circuits. A representative circuit is shown in Fig. 11.4. The output
transformer turns ratio is typically 7 to 1, with an output valve
anode current of about 40mA.
11.6. Purity
The alignment of the three electron beams in a shadow mask
tube is readily carried out by adjusting the purity magnets and
deflection yoke so that the electrons from any particular gun hit
only one colour of phosphor dots (see Section 3.8). The geometry
of the holes in the mask and the phosphor screen dots is such that
the electrons must appear to come from one of three points, called
the colour centres, if each beam is to produce a uniform picture in
one of the primary colours. The process of deflection is such that
any deflected beam appears to have passed through a point about
half way along its undeflected path in the deflection yoke and such a
point is called the deflection centre for that beam (see Section 11.7).
The purity magnets produce a transverse field of up to 10 gauss in
the tube neck and are used to move the three beams bodily about in
the neck. The deflection yoke must also be moved axially along the
neck until the plane of the deflection centres coincides with the plane
of the colour centres, when the purity magnets can be used to make
the deflection centres coincide precisely with the colour centres.
This coincidence can be judged by the colour purity of the screen
with two of the guns blacked out. It is usual to leave the red field
on, since this has the highest beam current, until the purity is as
good as possible, finally checking for a compromise position on all
three fields in turn.
The purity magnets correct both for tolerances in manufacture
which may cause the initial paths of the electron beams to be in
error so that they do not pass through the colour centres in the
deflection plane, and for positional errors caused by stray magnetic
fields, including the earth's field. Since the purity magnets act on
all three beams they can only correct for uniform errors. The
effect of the earth's magnetic field depends upon both the strength
of the field and the orientation of the tube in that field and purity
may need to be corrected each time the receiver is moved to a new
position. However, these effects are normally small and the beams
are masked by the aperture mask to a size smaller than the phosphor
dot separation so that there is a reasonable tolerance on purity.
Non -uniform effects may be produced after the beams have been
deflected since they change direction relative to the earth's magnetic
ABOUT 1,000 AMPERE TURNS
[77 A.C.
Fig. 11.5. Degaussing coil
field as they are scanned across effects may also

be caused by large metallic masses near the tube and by inaccuracies
in deflection yoke manufacture. They can be largely cancelled out
by an empirical adjustment of the field equalizing magnets, which
need only produce movements of the beam of ± 5 thousandths of
an inch at the edges of the picture.
If these adjustments do not result in satisfactory purity the tube
and its mounting should be demagnetized. This is readily carried
out with a degaussing coil and should always be done when installing
a new tube. A suitable degaussing coil (of 18 in. diameter), Fig.
11.5, consists of 28 turns of copper wire (14 S.W.G.) carrying an
alternating current of 30A, which may conveniently be derived from
a mains transformer. Alternatively, about 500 to 3,000 ampere -
turns can be produced by using higher impedance coils operating
from higher alternating voltage sources.
The degaussing coil, with the alternating current flowing through
it, should be slowly moved around the front and sides of the tube
and then gradually withdrawn several feet away from the receiver
before switching off. In this way the alternating magnetic field is
gently reduced to zero, which gives the most thorough demagnetizing
effect. The purity adjustments should be checked again after the
static convergence has been carried out.
GETTER
ASSEMBLY
BEAM -CONVERGING
POLEPIECES
GRID BR 4
GRID FIR 3
(FOCUS)
BLUE -POSITIONING
POLEPIECES
GRID NO2
GRID No I
(a)
BLUE
GUN
(b)
GREEN RED
GUN GUN
Fig. 11.6.(a) electron gun and (b) base

connections of an R.C.A. shadow mask
tube. Pin connections are:
1 Heater 7 Green screen G2
2 Red modulator G1 9 Focus grid G3
3 Red screen G2 11 Blue screen G2
4 Red cathode 12 Blue modulator GI
5 Green cathode 13 Blue cathode
6 Green modulator G1 14 Heater
11.7. Convergence
The convergence yoke must be so placed on the tube neck that the
poles sit immediately over the internal pole pieces of the electron
guns and the blue lateral shift magnet must sit over its corresponding
pole pieces (see Fig. 11.6), which may be clearly seen through the
glass neck.
The static convergence (see Section 3.8) is adjusted until the three
beams precisely overlap at the centre of the picture, using if necessary
the blue lateral shift magnet. The magnetic field strengths neces-
sary to achieve static convergence may be up to 7 gauss and can be
produced either by permanent magnets or by passing D.C. through
the vertical convergence coils.
Converging the three electron beams to produce coincident spots
at the screen centre in the undeflected position is not, in itself,
sufficient to ensure that the three spots remain coincident as the
beams are scanned over the raster. The electron guns are separated
by about 1 in. and are tilted a little over 1° towards the tube axis
so that the beams tend naturally to converge roughly at the screen
centre. The three electron beams therefore have slightly different
paths through the magnetic deflecting fields and this results in the
three spots separating out as they scan across the picture area. The
three beams may be thought of as parts of a very thick electron
beam which suffers deflection defocusing as it is swept across the
raster. This lack of dynamic convergence has to be corrected by
applying A.C. magnetic fields to the convergence pole pieces so that
the angle between the beams is reduced as the deflection increases.
To understand how the scan distortions vary between the three
electron beams and why the convergence yoke currents have their
particular waveshape, briefly consider the mechanism of magnetic
deflection. Non -uniformity in the deflecting fields and their
interaction with stray magnetic fields may produce scan distortions
as in monochrome television but only the case of three electron beams
traversing a uniform deflecting field will be considered.
Any one electron beam experiences a force proportional to the
current in the deflecting coils, and this force acts in a plane perpendi-
cular to the direction of the magnetic field and at right angles to the
direction of motion of the electrons. The electrons therefore move
along a circular path whilst they are in the deflecting field, as in
Fig. 11.7(a).
When they leave the deflecting field they fly off in a straight line
again, tangentially to their circular path. The electrons then appear
to have come from a point C inside the deflecting field half way
along their deflected path, and for smallish angles of deflection this
O\
RADIUS OA
PERPENDICULAR
TO INCIDENT BEAM \ CIRCULAR PATH
ELECTRON BEAM UNDEFLECTED RAY
DEFLECTION CENTRE
LIMITS OF
(a) MAGNETIC FIELD
WHICH IS PERPENDICULAR TO PAPER
LINE SCAN
DEFLECTING FIELD
(b) FLAT SCREEN
Fig. 11.7. Magnetic deflection

point is a fixed position whatever the angle of deflection. This
fixed point is called the deflection centre.
Consider a beam CD which is given a fixed vertical deflection, as
in Fig. 11.7(b), and is then scanned horizontally. The beam traces
out part of the surface of a cone whose apex lies at the deflection
centre and whose axis is parallel to the magnetic field producing the
horizontal deflection. If the beam strikes a flat screen, the line
traced out on the screen is only straight for the special instance of a
ray such as CE which is perpendicular to the magnetic deflecting
field. In all the other cases the flat screen intersects the suface of the
cone in a curve. This conic section is a hyperbola and the amount
of curvature increases as the vertical deflection increases. Similar
considerations apply to all four sides of the raster and give rise to
the well-known pincushion distortion. If the screen is a sphere
with its centre at the deflection centre then the pincushion distortion
disappears and a truly rectangular raster is obtained. Practical
tubes have spherical screens whose centre of curvature lies some
way behind the deflection centre and the resulting raster produced
by a uniform deflecting field shows some pincushion distortion, but
less than with a flat screen.
For the case of the three -gun colour tube, each red, green and blue
raster shows such pincushion distortion. The blue horizontal scan
only traces a straight line for the special case when the scan passes
through the horizontal projection on the screen face of the blue
deflection centre, i.e. when the electron beam is perpendicular to the
vertical magnetic field which is producing the horizontal scan.
Similarly, vertical lines on the blue raster are only straight in the one
case for which a vertical line passes through this same projection of
the blue deflection centre.
Since the three guns have different deflection centres, the green
and red guns are tracing out curved lines when the blue gun is
tracing out a straight line, and so on, as in Fig. 11.8. Thus, although
the three beams may produce one spot in the undeflected position,
the paths traced out by each gun show a curvature which depends on
the position of the spot with respect to the projection on the screen
of the appropriate deflection centre.
For example, when the blue beam has such a fixed vertical deflec-
tion that it strikes the point X, Fig. 11.8, the horizontal scan then
makes the blue spot trace out the solid curve shown. However,
when the green gun is firing at the point X, or X', for which it needs
a rather greater vertical deflection than the blue gun did, the hori-
zontal scan traces out a green line which has rather greater curvature
than the blue, and whose lowest point is offset from X to X' and is
R,,
BLUE GREEN BLUE

AND RED APEX AND GREEN
X X'
RED AND GREEN

BLUE SCANNING
SLIGHT CURVED
LINE STRA4GHTN.,
0
_ _ _ Ro oG - - - - - - -
PROJECTION 01,t THE SCREE/
OF DEFLECTION CENTRES
BLUE
RED APEX
BLUE
,,
AND GREEN AND RED
BLUEAPEX
APEX FOR
R AND G
R G
Fig. 11.8. Scan shape variations due to offset of

deflection centres. The actual displacements are of
the order of several millimetres
vertically above the green deflection centre. Similarly, the red trace
through the point X has the same curvature as the green trace but is
displaced to the left of the blue trace. For the case of a 70° shadow
mask tube with a flat screen and the deflection centres indicated, the
sideways shift of the red and green traces with respect to the less
curved blue trace results in the red trace separating out on one side
and the green trace on the other. For rather greater separation of
EXTENT OF CIRCULAR PATHS
DEFLECTING HAVE SAME
FIELD CURVATURE
GREENS I RED
BLUE
(a) ELECTRON BEAMS
EXTENT OF
DEFLECTING FIELD
FIELD SCAN
BLUE
ELECTRON
BEAMS
GREEN
AND
RED
VIEW. FROM THE

SIDE, OF VERTICAL
SECTION
8
(b)
Fig. 11.9. Beam displacements due to tilt of guns (a) during horizontal scanning;
(b) during vertical scanning
the deflection centres the green trace on the right hand side may fall
below the blue line and the red trace on the left hand side may also
fall below the blue trace.
For vertical lines traced out with constant horizontal deflection,
the blue lines cross over the other lines. The red and green traces
touch at their common apex, but on the right hand side of the screen
the red trace has the greatest curvature whilst on the left hand side
the green trace is more curved than either the blue or red.
A further effect arises due to the tilt of the electron beams. As
the beams scan horizontally, they do not stay vertically above each
other as they trace out their curved horizontal paths, the red scan
leading the blue at any instant while the green spot lags behind. This
effect is illustrated in Fig. 11.9 (a). The blue beam is symmetrical
viewed from above and is in line with the tube axis. The tilt on the
red and green guns causes the red and green beams to converge on the
blue beam before the screen is reached and then to diverge again so
that the spots separate out on the screen face. Notice that the blue
spot is not midway between the red and green spots. Any two of
the beams intersect in a point which traces out a paraboloid inside
the tube bulb as the raster is scanned.
Similarly in the vertical direction, illustrated in Fig. 11.9(b), a
timing error also occurs. The red and green beams are equally
offset and tilted and suffer the same vertical displacement. The
blue gun is tilted by a larger amount and the blue spot always appears
below the red and green spots, the displacement being greater at the
bottom of the scan than at the top.
The three beams trace out their respective rasters as in Fig. 11.8,
but the time at which they pass any particular point depends upon
the alignment of the electron guns.
Fig. 11.10 illustrates the combined displacement of the beams on
a raster, when the red, green and blue spots converge at the centre
of the screen. The four triangles of dots give the relative positions
of the three beams at instants when the scan is at the centres of the
raster edges. By applying appropriate currents to the convergence
yoke coils the spots can be moved radially so that they are more
nearly coincident. The required current waveforms have approxi-
mately the shape of flat parabolas and are also sketched in Fig. 11.10.
For vertical deflection the red and green vertical convergence
waveforms are the same, but the blue convergence current wave-
form is of opposite sign. Both are asymmetrical with respect to the
scan centre, or tilted. In the horizontal case the red and green
convergence currents are different, but both are still opposite in sign
to the blue waveform. Theoretically, perfect correction is not
VERTICAL DYNAMIC
BEAM DISPLACEMENTS OVER RASTER CONVERGENCE CURRENT
0 +
2 I
6
DEFLECTION BLUE
CENTRE APEX
POSITIONS B-
10
R AND fl
APEX
12
14
16
ONLY RADIAL CORRECTIONS ARE 18
POSSIBLE WITH CONVERGENCE YOKE R AND G B
20
HORIZONTAL DYNAMIC CONVERGENCE CURRENT
BLUE MS TIME
APEX
B S
0
R---------- ---- --
RED GREEN
G R
APEX APEX
litS TIME
0 10 20 30 40 50 60 70 80 90
Fig. 11.10. Total convergence errors and required corrections
possible in the horizontal direction because the colour dot triangles

near the ends of the line scan are not isosceles. The dynamic
convergence coils and their associated pole pieces can only move the
dots along radial lines and these lines do not meet at one point.
This is analogous to the problem of obtaining static convergence
with the blue lateral shift magnet. To overcome this defect, some
receivers use a dynamic blue lateral shift coil wound around the blue
lateral shift magnet, as in Fig. 11.11.
In practice the time base sweeps are not precisely linear with time
and the convergence waveforms which are derived from them must
correct for the actual displacements of the spots. Similarly, con-
vergence adjustments should only be made after the time bases have
been set up. Normally parabolic waveforms are derived from the
time bases and a tilt or sawtooth waveform added to produce the
required asymmetry (see Fig. 11.12). A few receivers have used
sine wave convergence waveforms for the horizontal correction but
this is only just tolerable. The circuits used for convergence vary
considerably, from those using several valves to those using none,
and depend upon the type of convergence yoke used. The basic
principles of the wave shaping circuits used are familiar enough in
general electronics and are summarized in Fig. 11.13. Integration
of a sawtooth waveform produces a parabolic waveform and
repeated parabolas may be conveniently derived from various points
in the time bases. Early yokes used the same pair of bifilar windings
for both vertical and horizontal convergence, and isolating inductors
BLUE HORIZONTAL SHIFT MAGNET
DYNAMIC SHIFT
COIL
BLUE GUN POLE PIECE

NECK OF TUBE
BLUE ELECTRON GUN
DIRECTION OF FLUX
IN ABOVE ARRANGEMENT
Fig. 11.11. Blue dynamic lateral shift
were then needed to keep the line waveforms out of the frame cir-
cuits (Fig. 11.14). Wave shaping circuits in the anode circuit of
the field output valve derived both parabolic and sawtooth wave-
forms and the latter could be reversed in sign. Although only one
set of coils and controls is shown in Fig. 11.14 there were, of course,
three sets. For the horizontal convergence, voltage pulses from a
22
PARABOLA i=kt2
SINE WAVE APPROXIMATION
AMPLITUDE 0.8 OF PARABOLA
0
-40 -30 -20 -10 0 10 20 30 40
TIME t IN 11 SECONDS
FROM CENTRE OF LINE SCAN
SAWTOOTH t= Fit
o-
-
2
-40 -30 -20 -10 0 10 20 30 40
7-
6- COMPOSITE WAVEFORM OF
Al
PARABOLA PLUS SAWTOOTH
5-
4 -I
3-
ANY NEGATIVE VALUE IS CORRECTED

WITH STATIC CONVERGENCE
-40 -30 -20 0 10 20 30 40
Fig. 11.12. Convergence waveform tilt by adding sawtooth to parabola

iA CIRCUIT ANALOGY
INTEGRATION OF
CONSTANT CURRENT
ii= ki
GIVES A SAWTOOTH t
WAVEFORM
!id/ = kit
INTEGRATION OF
SAWTOOTH
12411
GIVES A PARABOLIC
WAVEFORM
12dt=2kIt2
SAWTOOTH VOLTAGE PARABOLIC CURRENT

L-d-.V
i
dt v
Ei =j; dt
0
SAWTOOTH CURRENT PARABOLIC VOLTAGE
cq yl
=Cpt
q is CHARGE ON C
Fig. 11.13. Derivations of parabolic waveforms

Ikn TO VERTICAL
+400V OUTPUT
TRANSFORMER
100/1F
VERTICAL
PARABOLA
AMPLITUDE
DYNAMIC
CONVERGENCE COILS
(ONE SET ONLY SHOWN)
---ISOLATING INDUCTORS
IOkfl HORIZONTAL
PARABOLA
AMPLITUDE
V2
56011
HORIZONTAL
CONVERGENCE
AMPLIFIER
PULSES FROM
LINE OUTPUT HORIZONTAL
50Bn PARABOLA
TRANSFORMER
PHASE
Fig. 11.14. Early R.C.A. convergence circuits

special winding on the line time base output transformer were
converted to sawtooth waveforms, amplified, and then further
amplified and integrated to produce parabolas.
Later convergence yokes have separate windings for the horizontal
and vertical convergence, as shown in Fig. 11.15, and this obviates
the need for isolating inductors. The - ratio of these coils is high
so that they provide a degree of current integration directly. A
simplified version of the convergence circuits used in one receiver
with this type of convergence yoke is shown in Fig. 11.16. Positive
going flyback voltage pulses from the line time base transformer
(see Fig. 11.3) are integrated at the point A by L1, C1 and R1. The
resulting sawtooth voltage is applied to the convergence coil LL
where it produces a parabolic current waveform. At the line
scanning frequency the impedance of L2 is of the same order as that
of R1, whilst at harmonics of the line frequency the impedance of
L2 becomes relatively larger. The result is to shift the phase of the
fundamental of the parabola relative to its harmonics and produce
a tilt on the waveform.
A sawtooth voltage derived from the cathode circuit of the frame
output valve (Fig. 11.4) produces in the convergence coils LF a
parabolic current waveform.
Small pulses of either positive or negative polarity from the tilt
windings on the frame output transformer (Fig. 11.4) are also
applied to the convergence coil LF to produce an adjustable sawtooth
tilt to the vertical parabola.
D.C., conveniently obtained from the line output valve, is passed
through the centre tap of the tilt winding to the vertical convergence
coil to provide static convergence. The static convergence may be
adjusted either side of zero by R2.
The complete convergence circuit is shown in Fig. 11.17 and should
be read in conjunction with Figs. 11.3 and 11.4. The blue lateral
shift sawtooth convergence current waveform is provided by pulse
voltages from the line time base which produce a sawtooth magnetic
field acting vertically on the blue lateral pole piece. The permanent
magnet in the blue lateral shift assembly is used to bias the dynamic
field about the required mean position.
To ease the setting up of the convergence circuits, it is possible to
D.C. re-insert on the apices of the parabolic waveforms so that the
static convergence does not need constant readjustment as the
dynamic fields are being adjusted. Fig. 11.18 shows an R.C.A.
circuit, in simplified form, which does this. The essentially similar
red and green horizontal convergence adjustments are common.
FERRITE CORE
LAYER WOUND VERTICAL

CONVERGENCE COIL
WAVE WOUND HORIZONTAL

CONVERGENCE COIL
GLASS WALL OF TUBE NECK
INTERNAL POLE
PIECES
Fig. 11.15. Dual winding convergence magnet
LINE TIMEBASE VERTICAL TIMEBASE

OUTPUT VALVE OUTPUT VALVE
VI V2
5/iF VERTICAL loon 0'5,0

STATIC
SHIFT AMPLITUDE
VOLTAGE
WAVEFORM
loon
411)
VERTICAL OUTPUT
TRANSFORMER
loon
CONVERGENCE
COIL ASSEMBLY
(ONE SET OF
COILS ONLY SHOWN)
HORIZONTAL HORIZONTAL
PARABOLA PARABOLA
AMPLITUDE TILT
POSITIVE PULSES FROM LI CI

LINE TIMEBASE
OUTPUT TRANSFORMER 05µF
RI Ion
Fig. 11.16. Simplified passive convergence circuit

PLUG TO VERTICAL
I -K) 111 .2 ! TIMEBASE
D.C.
TIT' CONVERGENCE COILS
BLUE GREEN RED
RED
10011
PLUG TO LINE . ./
TIMEBASE I
GREEN BLUE 1
7-1 11 LATERAL
'
I I AMPLITUDE
1 loon
23
r_.__. BLU
HORIZONTAL(
AMPLITUDE., LATERAL'_COI
BLUE
BLUE ',RED loGREEN 1r8LUE
loon IO5MF
BLUE
125/Lr 1-4
10I1 ILO LATERAL,
SHAPE
GREEN
-4"
fRE GREEN BLUE 100.11.-
loon
HORIZONTAL
TILT
RED
100.11 i -1-2914Q;
PII ,,8 5
I A10 f 7 4 1 11
VERTICAL TILT
BLUE GREEN RED SOCKET
VERTICAL
AMPLITUDE
100n I00n loon
Fig. 11.17. Complete passive convergence circuit

11.8. Grey scale tracking
A typical curve of modulator grid -cathode voltage against beam
current or screen brightness for a shadow mask tube gun is given in
Fig. 11.19. The red, green and blue screen grid potentials are used
to adjust the gun gm characteristics so that they have the same
shape. The modulator grid brightness controls or sit or back-
ground controls as they are variously termed, are then adjusted so
that the three characteristics overlap, so to speak, and all three guns
black out together.
The drive waveforms to each grid -cathode space are proportioned
to the ratios required by the differences in phosphor efficiencies.
Typical ratios are R to G to B = 1 to 0.8 to 0.6. The modulator
grid input capacitance is 7pF for each gun and the cathode input
BLUE HORIZONTAL CONVERGENCE

AMPLITUDE
0Iig
PULSES FROM HORIZONTAL
LINE OUTPUT CONVERGENCE
TRANSFORMER
loon YOKE
RED 8 GREEN RED

HORIZONTAL CONVERGENCE
AMPLITUDE m
HORIZONTAL
CONVERGENCE
0.1/2E COILS
AMPLITUDE
PULSES FROM BALANCE
LINE OUTPUT
TRANSFORMER
GREEN
PHASE BALANCING
VERTICAL CONVERGENCE
RED, GREEN 8 BLUE
TILT AMPLITUDE
VLRT IC AL
OUTPUT
VALVE
Fig. 11.18. Simplified R.C.A. convergence circuits using D.C. restoration

= 600V
/ 02=6000
1-
CATHODE MODULATION
GRID MODULATION
20 30 40 50 60 70 80 90 100 110
DRIVE VOLTS ABOVE BLACK -OUT
Fig. 11.19. Typical input-output curve for shadow mask

tube. Final anode volts are 25 kV and the tube white is
cold blue white. V2 = screen volts above cathode
for zero drive
capacitance is 16pF for the three guns combined. When the tube
itself is used to matrix the luminance and colour difference signals
to the tristimulus values, it is normal to apply the luminance signal
to all three cathodes and the colour difference voltages to the grids.
In this way the relatively large average luminance signal does not
produce varying currents in the three colour difference circuits,
which may have different impedances. Also, the colour difference
circuits are narrow band and would provide frequency selective
feedback if used in the cathode circuit.
Adjustment of the grey scale tracking is the most critical of the
setting up procedures with the shadow mask tube. Particular care
must be taken that the dark greys are neutral in colour, with the
chromaticity of Illuminant C, since errors here produce a more
subjectively objectionable effect than colour casts in the white high-
lights. Although setting up can be carried out on a picture, a
gradation step wedge such as the centre blocks of Test Card C is
preferable. The eye is very sensitive to changes in white balance
over the brightness scale and it is difficult to measure the changes in
gun gm which produce noticeable visual effects. A detailed account
of grey scale adjustment is given in Chapter 13.
11.9. Summary
The practical circuit details inherent in shadow mask tube opera-
tion have been discussed. The absorption of electrons by the mask
necessitates the use of a high e.h.t. voltage to obtain adequate
brightness. This, together with the larger tube neck, calls for
increased scanning power. The time bases must be reasonably linear
if good convergence is to be obtained and slight errors in convergence
make a noticeable difference to picture definition before the colour
fringing effects become objectionable as such. It must be remem-
bered that magnetic fields affect the colour purity and although a
permanent magnet waved near the tube face produces the most
interesting colour patterns, some of these patterns will probably
stay there when the magnet is removed.
Grey scale tracking is very important and the grey colour chosen
must be close to the reference colour, Illuminant C, if the colour
reproduction is to be correct. In particular, green casts have
objectionable effects on the complexion of the actors in a scene and
such incorrect flesh tones must not be corrected by means of the hue
control.
The reader is reminded to be careful about the danger of X-rays.
CHAPTER 12
Colour Receiver Test Equipment

and Performance Measurements
12.1. Introduction
Many of the instruments required for measuring the performance
of colour receivers will already be possessed by organizations which
are equipped for adequate performance testing of monochrome
receivers. The additional apparatus required for colour work is
chiefly concerned with accurate determination of frequencies and
phases in the region of the sub -carrier (e.g. 2.7 Mc/s in the 405 -line
system).
In this chapter a list is given of the apparatus required for colour
work, from the point of view of the repair shop and the performance
testing laboratory. The use of the equipment is then discussed.
It is anticipated that all the apparatus mentioned in what follows
either is or will be commercially available, but in case this is not so,
two essential items, namely, a crystal controlled sub -carrier frequency
oscillator and a calibrated phase shifter, will be described in detail
so that the engineer may construct them for himself.
It is, of course, assumed that a radiated Test Card C or D and a
colour signal will be available.
12.2. Servicing equipment required

From the point of view of the service engineer, the following
apparatus will be necessary. The list is not necessarily arranged in
order of importance.
1. General purpose high impedance (20,000 ohms/V) meter for
voltage, current, and resistance measurement.
2. Valve voltmeter.
3. " Battery box ", or variable source of D.C. voltage up to about
10V.
4. Video oscillator covering the frequency range from about 50
c/s to 5 Mc/s, output at least 1V.
335
5. Signal generator covering the R.F. and I.F. bands, output about
100mV, and preferably capable of 100 % modulation by an
audio frequency square wave.
6. A swept oscillator or " wobbulator " for the R.F. and I.F.
bands, with a frequency sweep adequately covering the video
band.
7. An oscilloscope with a Y amplifier response at least up to the
highest video frequency, and with an X amplifier input terminal.
Ideally, the X amplifier should have as good a frequency response
as the Y amplifier, though an X amplifier bandwidth up to about
0.5 MO may suffice. Time calibration " pips " of 1µs are
very useful for checking burst gate widths.
8. An accurate wavemeter (of the BC221 type, for example) for
measuring sub -carrier frequencies.
9. A (monochrome) television waveform generator which can
provide
(a) About 5 different luminance levels or " greys " for checking
colour balance tracking.
(b) A thin white grid on a black background, consisting of
about 16 equally spaced vertical lines and 12 equally spaced
horizontal lines. This type of signal is essential for con-
vergence adjustment.
10. A crystal controlled sub -carrier frequency oscillator having an
output of at least 1V.
11. A calibrated phase shifter giving 360° maximum phase shift
at sub -carrier frequency.
12. A degaussing coil for demagnetizing shadow mask tubes. This
is easily constructed, as described in Section 11.6. Construc-
tional details for items 10, and 11, which are easily made, are
given below in Sections 12.4 and 12.5.
12.3. Performance measurement equipment required

For thorough investigation of colour receiver performance, the
following items should be added to the above list.
1. A group delay equipment, or some means of measuring phase
characteristics.
2. A travelling microscope for time base linearity measurements.
3. A colorimeter, such as the Harrison colorimeter, for purity and
colour balance measurements.
TEST EQUIPMENT AND PERFORMANCE MEASUREMENTS 337
4. An Illuminant C reference for setting the correct white chroma-
ticity.
5. A local source of colour signal, such as a colour bar signal,
available at video and R.F. and with the correct -frequency
lock between the synchronizing and sub -carrier signals. C.W.
sub -carrier should also be available, and a sound carrier with
the correctly locked frequency spacing from the sub -carrier.
(This is discussed in Chapter 6.)
6. For design work, a colour picture source such as an R, G and
B flying spot scanner is highly desirable.
7. A vectorscope for measuring the amplitude and phase of the
modulation carried by the quadrature modulated sub -carrier.
This instrument is chiefly of use for checking the accuracy of
transmitter encoders.
A block diagram of one form of vectorscope is shown in Fig. 12.1.
The composite video waveform to be checked is passed through a
band-pass filter which accepts the modulated sub -carrier, and this is
then synchronously detected in two detectors operating along the
R' Y'
DETECTOR
OSCILLOSCOPE
R -è R FERENCE
A PHASE
SUBCARRIER
COMPOSITE
CHROMINANCE REFERENCE
VIDEO -0--
INPUT
BAND PASS
FILTER
SOURCE
Y
W B' -Y' REFERENCE

PHASE
B=Y'
DETECTOR
Fig. 12.1. Vectorscope block diagram
(R' - Y') and (B' - Y') axes. The reference source for the
detectors may be obtained either from the transmitter's C.W. sub -
carrier oscillator, or from a reference generator locked to the
incoming burst.
If the two detectors have equal gains, then the oscilloscope display
will show the appropriate amplitude and phase of the original
chrominance signal as plotted on the (R' - Y') and (B' - Y')/
1.78 diagram.
Normally the vectorscope is used for displaying the vectors
corresponding to the colour bar waveform, and this gives a quick
and accurate check of the encoder if a scale calibrated with the
SUBCARRIER OUTPUT Fig. 12.2. Sub -carrier

FREQUENCY
CRYSTAL frequency crystal oscillator
correct amplitudes and phases of the three primary and their com-
plementary colours relative to the burst is placed over the display.
12.4. Crystal controlled sub -carrier oscillator
A suitable circuit for a crystal controlled oscillator is shown in
Fig. 12.2. The oscillator has an electron coupled output from the
anode of the pentode, the oscillation occurring between the control
and screen grids. The circuit shown gives about 5mA peak of
anode current for a 2.65781 Mc/s crystal, suitable for the 405 -line
N.T.S.C. system. Normally, the oscillator is required to feed a
load of about 70 ohms, so with a ten to one step down transformer
the anode load is about 7k11, which results in an output of about
3.5V peak, or approximately 2.5V r.m.s. This oscillator is of most
use if the crystal can be made to oscillate within a few cycles of the
actual sub -carrier frequency, since it can then be used for testing
reference generators as well as for phase measurements.
12.5. Calibrated phase shifter
A calibrated phase shifter can easily be made from suitable lengths
of co -axial cable of the type often used for television aerial down-
leads. The principle of such a phase shifter relies on the time
delay which occurs when a signal is passed down a characteristically
terminated cable. If a sine wave voltage is applied to a cable of
such a length that a delay equal to the period of one cycle of the sine
wave occurs, then the cable output sine wave will lag the input by
one cycle, or 360°. Clearly, fractions of 360° may be obtained by
using shorter lengths of cable, and a convenient arrangement con-
sists of various cable lengths which can be switched in or out between
input and output.
A diagram showing one possible switching method is shown in
Fig. 12.3(a), in which the total length of cable used is equal to one
cycle of delay, or 360°. By appropriate choice of switches, any
phase delay from 0° to 360°, in 1° steps, may be obtained. However,
a rather more convenient arrangement which saves some mental
arithmetic is shown in Fig. 12.3(b), in which the total length of
cable required is equivalent to 365°. The extra 5° of cable is certainly
well worth while.
The total physical length of cable required for a 360° phase shift
will depend on the effective transmission velocity in the cable used,
and the actual frequency at which the phase shift is required. It is
important to remember that the cable phase shifter will read phase
correctly for only one value of frequency, though a correction factor
can be employed for other frequencies. For example, for a fre-
quency equal to twice that for which the device is designed, the
phase readings must be multiplied by two. Also, the phase reading
will be correct only if the cable is characteristically terminated. It is
shown in Appendix 8 that a 1 % mismatch produces a maximum
phase error of about -1°.
The transmission velocity of most types of co -axial cable is about
0.6 x 3 x 108 m/s, that is, 180 mhus. For a sub -carrier frequency
1
of 2.66 Mc/s, the delay required for one cycle is so that the
2--66/is,
80
length of cable required for a 360° shift is m, or about 75 yd.
266
However, for any given cable it is advisable either to consult the
manufacturer concerning the transmission velocity, or alternatively
a rough check on the delay per yard may be made by measuring the
width of the rectangular waveform (on a time calibrated oscillo-
scope) produced when a square wave input is applied to a known long
length of short circuited cable. This width will equal twice the
delay of the line, as shown in Fig. 12.4.
Calibration of the phase shifter can be carried out by first accurately
determining the physical length of line required to give a 360° phase
INPUT OUTPUT
(&)
INPUT OUTPUT
(b)
Fig. 12.3. Delay cable phase shifter (a) shows a 360° total delay and (b) a 365° total delay. Both show a 236° delay
shift, and then cutting the line in proportionate measured physical
lengths; for example, for 1°, a fraction (3:0) of the 360° length is
required. This makes the reasonable assumption that the cable has
uniform characteristics.
The physical length of cable required for a 360° phase shift is best
determined by measuring, at the appropriate frequency, the reactive
component of the input impedance of a short circuited length of
cable which is known to be near to (obviously preferably in excess
INCIDENT SQUARE -WAVE
REFLECTED SQUARE -WAVE SHORT-CIRCUITED LINE

AT INPUT
SQUARE -WAVE
T GENERATOR
SUM OF INCIDENT
& REFLECTED WAVES
Fig. 12.4. Approximate measurement of the delay of a coaxial cable
of) the length which corresponds to one wavelength at the required

frequency. The correct length is obtained when the input impedance
is purely resistive.
(The input impedance of a short circuited line having a character-
istic impedance Zo is
Zo(tanh a 1 j tan T27r 1/ I
1 j tanh a / tan T /
where a is the attenuation constant, A is the wavelength of the signal
in the line, and 1 is the length of line. If / = A, the reactive terms
become zero.)
However, if an impedance bridge is not available, an alternative
method of measurement consists of first arranging for a straight line
Lissajous figure to be displayed on an oscilloscope using a voltage
source of crystal controlled sub -carrier frequency. Some phase
shift may have to be introduced (for example, by way of a suitable
R.C. circuit) to allow for phase shift in the oscilloscope circuits.
Having arranged for a straight line Lissajous figure to be displayed
from a source of sub -carrier frequency, the estimated 360° length of
23
correctly terminated cable is introduced into the X or Y path to the
oscilloscope, and the length is adjusted until a straight line Lissajous
is again obtained.
12.6. Luminance channel performance
Satisfactory operation of the luminance channel of a colour
receiver requires adequate amplitude versus frequency response and
a sufficient output voltage to drive the display tube. In addition,
the luminance delay must be such that luminance transients are
centred as accurately as possible within the corresponding chromi-
nance transients.
The receiver frequency response up to the conventional diode
detector may be measured in the same way as for conventional
monochrome receivers, such as by feeding an R.F. signal into the
aerial socket and measuring the diode current for various fre-
quencies covering the R.F. band. Alternatively, of course, a
wobbulator may be used.
Again, measurement of the luminance video stages may be carried
out by using the techniques of monochrome video stage measure-
ment but it is recommended that a wobbulator method be employed
so that any reflections caused by the luminance delay line may be
easily observed, together with the luminance notch introduced by the
sub -carrier frequency rejector which is required to remove large area
sub -carrier dot structure.
Direct measurement of the luminance video channel is difficult
in practice because it requires a video signal to be introduced in the
diode detector circuit, and the appropriate generator source impe-
dance corresponding to the detector impedance must be included.
This impedance is not easily determined and it is therefore better to
derive the characteristic of the luminance video stages by taking an
overall characteristic from aerial input to luminance output, and
then to subtract from this the response of the R.F. and I.F. stages
up to the detector.
The overall response may be conveniently measured by first
investigating the D.C. and low frequency response, and then the
high video frequency response. In these measurements, the A.G.C.
action must be rendered inoperative by connecting a suitable
negative bias to the A.G.C. line.
To determine the D.C. and low frequency response, a vision carrier
frequency signal is fed into the aerial socket, and the luminance
output is displayed on a D.C. coupled oscilloscope. The D.C.
change produced by switching the signal on and off is then noted,
and compared with the peak -to -peak value of the displayed output
when the same R.F. signal is modulated 100 % by an audio frequency
square wave, for example. The ratio of these two measurements
gives the D.C. to A.C. gain ratio of the receiver. Note that if a D.C.
coupled oscilloscope is not available, the D.C. change may be
measured by means of a high impedance voltmeter, and of course in
this case the oscilloscope trace of the modulation waveform must be
calibrated. It is advisable to arrange for the D.C. output change to
be about half that produced by a typical picture to ensure that
overload is avoided.
The high video frequency response may be compared with that at
low frequencies by feeding two signals into the aerial socket and
measuring the luminance output. Thus, a vision carrier frequency
is added in a resistive pad to another carrier whose frequency can
be varied over the R.F. pass -band, and the corresponding beatnote
luminance output may be measured by means of a valve voltmeter
or oscilloscope, either of which must be capable of registering the
highest video frequency. An alternative and more satisfactory
arrangement is to replace the variable frequency generator by a
wobbulator, when an oscilloscope display of the luminance output
will have an envelope corresponding to the overall response, again
provided that the oscilloscope frequency response is adequate.
The maximum voltage output of the luminance amplifier may be
determined by feeding a video signal into the detector circuit and
measuring the corresponding output on an oscilloscope or valve
voltmeter. While this method will give no reliable indication of
luminance video frequency response (because the impedance of the
feed will almost certainly not be correct), the maximum output
performance will not be affected. The measurement should be carried
out for a D.C. video input and also for an input of the highest
video frequency, for it is possible for a video amplifier to have good
high frequency response for low amplitude signals but not for large
amplitude signals. This effect can occur if a cathode compensated
video valve does not have a large enough current swing available
to charge the capacitance by-pass.
The delay introduced by the luminance delay line can be measured
by feeding a square wave or sharp pulse into the detector circuit,
and displaying simultaneously on an oscilloscope the original signal
and the luminance output. A single -beam oscilloscope may be
used if the two signals to be compared are added together in a
suitable high impedance capacitance pad.
However, the actual value of the time delay required will depend
on the bandwidth of the chrominance channel, and the best way to
check the timing is to use a colour signal from a picture source which
consists of a vertical strip, the top third of which is red, the middle
third green and the bottom third blue, all on a black background.
The width of the strip should be such that the chrominance signals
produced correspond to the highest chrominance frequency which
the receiver is capable of resolving. By looking at the picture dis-
play of such a signal, the relative positions of the luminance and
chrominance signals can be easily checked to see if the delay is too
long or too short, and whether it is the same for all three colours.
12.7. Chrominance channel performance
As in the luminance channel, the amplitude versus frequency
characteristic and the maximum available output voltage are of
interest in the chrominance channel of the receiver, that is, the
section of the receiver which accepts the modulated sub -carrier signal
from the composite video and ultimately provides difference signals
either to a matrix circuit or to the display tube.
Additionally, the effective gain and detection axes of the
synchronous detectors and the A.C.C. performance, need to be
investigated.
It should be noted that the measurements required of a chromi-
nance channel will depend on the type of circuits employed, and the
bandwidths of the various amplifiers and synchronous detectors will
be different for I' Q' and equiband working. However, it is very
probable that the majority of receivers will use the display tube as
the adding matrix for the colour difference and luminance signals,
and in the measurement techniques to be described it will be assumed
that the difference signal outputs are available at the display tube.
In addition, it will be assumed that equiband working is employed,
though if this is not the case the various modifications to the measure-
ments should be fairly obvious: for example, in I' Q' working the
bandwidths of the I' and Q' detector outputs must be individually
checked, as must the following I' bandwidth channel and Q' band-
width channel. Again, I' channel delay must be such that I' and
Q' transients occur symmetrically.
The amplitude versus frequency characteristic may be con-
veniently determined by a similar method to that used for the
luminance channel. A vision carrier signal and a wobbulator output
are added in a resistive pad and the result is fed to the aerial input
socket. An oscilloscope display of each of the colour difference
outputs in turn will then give the individual overall chrominance
responses.
Note that a zero beat will be observed between the sweeping video
frequency and the reference generator frequency. This frequency
mixing occurs at the synchronous detectors, and the zero beat
corresponds to the D.C. difference signal position on the trace.
The bandwidth is easily determined by feeding a video oscillator
output into the oscilloscope as well as a difference signal output,
when a second zero beat will be observed corresponding to the video
oscillator frequency.
The maximum output of the chrominance channel is best deter-
mined by feeding a colour signal of colour bars into the aerial socket,
and observing each colour difference output on an oscilloscope. By
increasing the saturation control, the point at which the waveform
COLOUR RECEIVER
SUBCARRIER CHROMINANCE DIFFERENCE

SYNCHRONOUS
OSCILLATOR AMPLIFIER DETECTORS
-t- SIGNAL
OUTPUT
PHASE
SHIFTER
REFERENCE
GENERATOR
Fig. 12.5. Test set up for the determination of synchronous detector gains and
detection axes
limits indicates maximum drive without overload. Note that the

colour difference signals have positive and negative excursions
depending on the particular colour, and the overload may occur in
either direction. For example, for full pure red, the relative
difference signal amplitudes are,
(R' - Y') = 0.7, (G' - Y') = -0.3
and
(B' - Y') = -0.3
while for full pure cyan,
(R' - Y') = -0.7, (G' - Y') = 0.3 and (B' - Y') = 0.3
A suitable arrangement for measuring synchronous detector
performance is shown in Fig. 12.5.
A crystal controlled sub -carrier oscillator is connected to a suitable
point in the chrominance channel, and also to the receiver's reference
generator burst input via a calibrated phase shifter. It may be
necessary to feed a monochrome signal into the receiver to generate
a burst gate, otherwise the reference generator may not lock to the
sub -carrier oscillator input.
By connecting a valve voltmeter to each difference signal output
in turn and by adjusting the phase shifter, the phase reading corre-
sponding to the maximum and minimum output from a difference
signal channel may be determined, and hence the corresponding
detection angles may be derived. For example, if maximum
(B' - Y') output occurs at a phase shifter reading of 100° and a
maximum (R' - Y') output occurs at a reading of 10°, then the red
difference axis leads the blue difference axis by 90°. Measurement
of minimum as well as maximum values checks the symmetry of
detection for negative and positive difference signal components.
The relative amplitudes of the maximum (for example) values of
the difference signal outputs will give the relative gains of the
detection axes. In this connection, it must be remembered that
certain factors are introduced by the transmitter, and also by the
receiver to allow for the unequal phosphor efficiencies.
For example, the transmitter factors are -1 -1 and -1

1.14' 2.03 0.7
for the
red, blue and green difference signals respectively, so that a receiver
which uses equal phosphor efficiency drives will give maximum
difference signal outputs (using the above method of measurement)
in the ratio of 1.14, 2.03 and 0.7, or 1 to 1.78 to 0.615 for red to
blue to green respectively. If the receiver has a phosphor efficiency
drive ratio of 1 to 0.6 to 0.8 for red to blue to green, respectively,
then the above method of measurement will give maximum dif-
ference signal values in the ratio 1 to 1.07 to 0.49.
A quick check on the synchronous detector performance may be
made by feeding a colour bar signal into the receiver and observing
the display in each of the primary colours in turn. If the colour
bar signal has the usual form of vertical colour stripes of white,
yellow, cyan, green, magenta, red and blue, then the red content
should have the display shown in Fig. 12.6(a), the blue as in Fig.
12.6(b) and the green as in Fig. 12.6(c). These displays may be
observed either by switching off two guns at a time (e.g. to observe
the red display, switch off blue and green, etc.), or by observing the
complete display through suitable red, blue and green filters. The
" on " periods of any one waveform should be equal, and all the
" off " periods should be equally " black ". This test is quite
severe and the correct displays will be obtained only if the saturation
ON
OFF RED
(a)
ON
OFF BLUE
(b)
ON
OFF GREEN
(C) WHITE YELLOW CYAN GREEN MAGENTA RED BLUE
Fig. 12.6. Individual gun displays for the correct reception of

colour bar signal
and hue control settings, and the relative detection axes and gains,
are all correct.
The automatic chrominance control performance is best deter-
mined by the use of a local source of colour signal, such as a colour
bar signal. If the receiver is adjusted to receive such a signal
normally, and one of the difference signal outputs is displayed on
an oscilloscope, the amount of chrominance plus burst attenuation
required at the encoder to reduce the difference signal output by 6dB
may be determined. This attenuation figure is then the A.C.C.
figure of merit.
The amount of chrominance gain in hand may be determined by
the above method by turning the saturation control to maximum
and measuring the amount of attenuation which can be introduced
at the encoder before the difference signal output drops below its
normal level.
12.8. Reference generator performance
Measurements which are indicative of the performance of colour
receiver reference generators include frequency drift and pull -in
performance, maximum static phase error and range of hue control,
dynamic phase error due to noise and phase shifts due to inaccuracy
of burst gating. A measurement which gives an indication of the
amount of stray feedback from reference generator output to burst
input is also advisable.
Apart from its primary function of reference frequency generation
for the synchronous detectors, the reference generator also provides
auxiliary services such as colour killing and A.C.C. The measure-
ment of A.C.C. performance has been described in the previous
section, but killer performance will be discussed below.
Frequency measurements on reference generators are best carried
out with a precision wavemeter, and this must be as loosely coupled
as possible to the reference output to prevent " pulling " of the
wavemeter, and to ensure that no signal from the wavemeter inter-
feres with the generator.
Frequency drift versus time after switch on and frequency shift
due to mains voltage changes are easily measured. Pull -in per-
formance is best measured by noting the generator frequency when
no signal is applied to the receiver, and measuring the time for pull -in
immediately after applying a colour signal. The time to pull -in
from various detunings can be determined by deliberately altering
the tuning of the generator, and before each measurement some
time min) should be allowed to permit the generator to become
stable at the frequency to which it is tuned, since long time constant
circuits may be present in the generator circuitry.
It should be noted that some idea of pull -in performance can be
obtained by feeding a stable but variable frequency into the burst
input, and a monochrome signal into the receiver aerial socket to
operate the burst gate, if necessary. However, this type of measure-
ment is of doubtful accuracy for two reasons. Firstly, if the line
scan is not correctly locked to the burst, spurious frequencies closer
than a spacing equal to half the line scan frequency may be produced
next to the sub -carrier component, and this will impair pull -in per-
formance (see Appendix 1). Secondly, since some receivers may
have an excess of burst gate width, the effective burst duty ratio is
greater than it should be and this will give an optimistic pull -in
performance. It is therefore advisable to measure pull -in per-
formance by using a correct burst waveform which is locked to the
line scan in accordance with the N.T.S.C. specification.
Maximum static phase error may be determined by means of the
set up shown in Fig. 12.7. A stable but variable frequency generator
is set to the sub -carrier frequency, and its output is fed into the
reference generator at the burst input. It may be necessary to addi-
tionally feed a monochrome signal to the receiver in order to operate
the burst gate. The reference output is fed to the Y -plates of an
VIDEO REFERENCE
OSCILLATOR GENERATOR
OSCILLOSCOPE
PHASE
SHIFTER
Fig. 12.7. Reference generator phase measurement set up
oscilloscope while a second output from the signal generator is fed to

the X -plates via a calibrated phase shifter. Adjustment of the latter
should enable a straight line Lissajous figure to be obtained when
the signal generator is set to the sub -carrier frequency. Next, change
the signal generator frequency until it differs from the sub -carrier
frequency by an amount equal to the limit of the reference generator
pull -in range (which has already been determined) and measure the
phase change required to restore the Lissajous figure to a straight
line. The measurement should be carried out for both positive and
negative tuning errors, and the signal generator frequency is best
measured by a wavemeter since small frequency changes are involved.
The above measurement will tend to give an optimistic result if
there is an excess of burst gate width. An accurate measurement
can be made if a correct burst waveform and a continuous sub -
carrier sine wave are both available from the transmitter, in which
case the burst waveform can be used for synchronizing the reference
generator, while the continuous sine wave may be used for the
oscilloscope X -plates.Detuning of the reference generator may be
carried out by adjustment of its frequency trimmer. It is possible
to use the burst waveform for the X -plate deflection instead of a
continuous sine wave, but because of the low duty ratio of the burst
a rather " faint " Lissajous results.
The set up shown in Fig. 12.7 may be used for measuring the
phase shift range of the hue control by noting the phase reading
required to obtain a straight line Lissajous for one extreme end
setting of the hue control, and subtracting the phase reading required
to restore the straight line Lissajous at the other extreme end of the
hue control.
An estimate of the dynamic phase error may be made by operating
the receiver from a colour bar signal, and feeding the composite
video to the Y -plates of an oscilloscope and the reference output to
the X -plates via a calibrated phase shifter. Each colour bar will be
displayed as an ellipse, and as the signal to noise ratio is reduced (by
attenuating the signal input to the receiver, for example) so each
ellipse will show a rotational ". blur " due to noise. Measurement of
the phase shift required to produce comparable peak to peak phase
changes for a " clean " signal can be made by adjustment of the
phase shifter. Better accuracy can be obtained by this method if a
" clean " composite video signal is always available for the oscillo-
scope display, but of course this is possible only if a colour bar
generator is available.
Phase shifts due to burst gating inaccuracies are normally only
significant in receivers which employ line time base gating. Using
the above arrangement of a colour bar Lissajous display, the phase
RED
GREEN CYAN MAGENTA Fig. 12.8. Burst gate accuracy

test picture
BLUE
shifter can be adjusted until a chosen ellipse becomes a straight line.

The phase shift caused by variation of the line time base hold control
can then be measured. The most critical condition for burst gating
accuracy obtains when the colour bar on the extreme left of the
picture is blue, since this has a phase almost opposite to that of the
burst.
A very sensitive subjective test of burst gating accuracy can be
made if a signal corresponding to the colour picture shown in Fig.
12.8 is available. If the burst gate is too long, for example, the
cyan and magenta bars will show hue changes which correspond to
the red, green and blue sections. Note, however, that similar hue
changes can occur due to insufficient D.C. gain in the chrominance
channel, but this signal provides a severe overall test.
A final measurement of reference generator performance which is
recommended concerns the effect of the hue control setting on the
free running generator frequency. With the receiver operating on a
monochrome signal, and with the killer disabled, the peak -to -peak
change of generator frequency should be measured as the hue control
is slowly rotated through its range. There are two possible causes
for this frequency change.
Firstly, direct coupling between the hue control circuit and the
oscillator tank circuit, and secondly, stray feedback from the gener-
ator output back to the burst input. Disabling the killer may tend
to emphasize the latter effect, if stray coupling is by way of the
chrominance channel, for example.
12.9. Colour killer performance
Two features of colour killer action are of importance: one is the
reliability of the interpretation of a noisy monochrome signal as a
monochrome signal and the other is sufficient sensitivity to recog-
nize a burst for the smallest colour signal for which the chrominance
channel will operate satisfactorily.
While operating the receiver from a monochrome signal, this may
be attenuated and the receiver gain correspondingly increased until
the killer " unldlls ". Knowing the input signal level and the
receiver noise factor, the signal to noise ratio at which the killer
fails may be estimated.
It should be noted that there are two general methods for deriving
a killer bias. One involves amplitude detection of the burst, in
which case noise can be confused with burst signal for relatively high
signal-to-noise ratios.
The other involves synchronous detection of the burst (this type
is normally found only in two mode A.P.C. loops), and is reliable
under extremely adverse signal-to-noise conditions. In fact, it is
not likely that the killer can be made to fail for the latter case unless
a considerable noise voltage is added to the receiver input at
maximum receiver gain.
The sensitivity of the killer circuit may be checked by operating
the receiver from a colour signal and reducing the signal level until
the receiver has no gain in hand. Having adjusted the saturation
control for a satisfactory colour picture, the killer bias should be
shorted out. If the saturation then increases, the killer detector is
not producing sufficient output to remove the killer bias.
12.10. Performance of time bases
Measurement of colour receiver time base linearity does not
involve techniques unfamiliar in monochrome receivers. A white
grid on a black background signal is convenient for linearity measure-
ment by some form of travelling microscope, and in order to simplify
the procedure it is convenient to have only one gun (e.g. the green) in
operation. It should be noted that time base linearity is particularly
important in a colour receiver since the convergence circuits will not
operate correctly unless a certain standard of linearity is achieved.
12.11. E.h.t. supply performance
Adequate e.h.t. voltage, current, and regulation are essential for
good picture quality, and it is recommended that a regulation curve
of e.h.t. voltage versus current be plotted. An electrostatic meter may
be used for voltage measurement, while a lmA full scale meter will
suffice for current measurement of the cathodes of the tube.
12.12. Convergence accuracy
For checking the convergence of the three images, a convenient
signal is a white grid on a black background. The maximum
misalignment of any two colours along a given grid line, both
horizontal and vertical, should be measured. The measurements
should be carried out for two areas of the picture, the first is an
inner area enclosed by a circle having a radius equal to half the radius
of the tube face, and the second is an outer area extending from the
inner zone to within about an inch of the edge of the tube. The
extreme edges of the picture may normally be ignored.
The measurements above should be carried out about two minutes
after switching on, and repeated after about one hour to check the
convergence stability.
12.13. Purity and colour balance performance
Lack of purity on the tube face is easily noticeable if a plain raster
is displayed. Areas of impurity can be measured by a colorimeter
after the central area of the tube has been set to the colour of Illumi-
nant C, and a plot of the spurious colours can be made on a C.I.E.
diagram. Drift of overall colour balance, due to temperature
effects or mains voltage variations, may be determined by measuring
the change of chromaticity of the central area of the tube face.
12.14. Summary
This chapter lists the apparatus considered necessary for servicing
and for measuring the performance of colour receivers. As far as
servicing is concerned, very little additional apparatus is required
to be added to existing monochrome receiver test gear, and details
are given of a crystal oscillator and calibrated phase shifter.
Various techniques are described for the measurement of phase
shifts at sub -carrier frequency, and for checking demodulator and
reference generator performance. In addition, measurements which
should be carried out on the remaining colour receiver circuits are
discussed.
CHAPTER 13
Receiver Installation
13.1. Introduction
The installation of a colour receiver is sometimes regarded with
some trepidation, even by the most competent monochrome tele-
vision engineer. But provided that he understands the basic
principles involved, an engineer who is competent enough to install
a monochrome receiver satisfactorily is perfectly capable of instal-
ling a colour receiver. The basic principles are outlined below.
The major portion of the apparatus in a colour receiver is con-
cerned with the provision of a black and white picture. In fact,
most of the circuits used in this part of the receiver are identical
with monochrome practice, the essential difference being the display
tube itself. The shadow mask tube may be regarded as the equiva-
lent of three tubes, one red, one green and one blue, " wrapped
up " as one tube. In order to achieve a satisfactory black and white
picture, it should be realized that two conditions must be satisfied.
Firstly, each gun must excite only one colour. The red gun
must provide a pure red raster over the whole scanned area, and
similarly the green and blue guns must provide pure green and pure
blue rasters. This requirement is referred to as purity.
Secondly, the relative brightness of the three rasters must be
such that a white raster is subjectively observed. Furthermore
this white must not change colour from the lowlights to the high-
lights. This adjustment is called grey scale tracking.
With similar video signals applied to all three guns, a black and
white picture will result if the above conditions are satisfied, but
the presence of a picture entails a further requirement, namely that
the three pictures be superimposed one on the other, or registered.
This registration of the images is called convergence, and it has two
general adjustments. Static convergence which may be looked
upon as picture shift of each complete raster relative to the other two
and dynamic convergence which varies across each raster and in-
creases with distance from the centre of the tube.
The remainder of the adjustments for a black and white picture
are exactly the same as for a monochrome receiver, i.e. adjustment
354
RECEIVER INSTALLATION 355
of time bases, brightness and contrast, spot limiting, focus, and
oscillator tuning.
It cannot be emphasized too strongly that one of the most im-
portant features of colour receiver installation is the attainment of a
good black and white picture. A good colour picture cannot be
obtained from a receiver which exhibits a poor monochrome picture.
In case the reader is alarmed at the requirements for purity,
grey scale tracking, and convergence, it should be noted that it is
only the purity which will usually need attention, since the grey
scale and convergence adjustments will have been set by the manu-
facturer. The purity can be affected by stray magnetic fields,
however, so that the optimum adjustment at the final location of
the receiver may be different from that at the factory, where a field
free room may be used. However, provided the receiver has not
been subjected to severe mechanical shock, the purity adjustment
will consist merely of degaussing and edge purity correction, as
described below. In view of the fact that the purity can be in-
fluenced by stray magnetic fields, it is recommended that the receiver
should be installed as remotely as possible from large masses of
iron or steel, such as radiators or girders.
The remainder of the circuitry in a colour receiver is the chromin-
ance circuit which decodes the colouring information contained
in the transmitted signal, and applies the appropriate signals to the
three guns so that the otherwise black and white picture is displayed
in colour. There are essentially three components in a complete
chrominance circuit.
The first is a chrominance amplifier which accepts the sub -carrier
signal (2.7 Mc/s in the 405 -line system) and its sidebands. This
amplifier usually consists of one or two stages with simple tuned
circuit loads.
The second is a synchronous detector which demodulates the
amplified sub -carrier signal, and ultimately provides three difference
signal outputs which are fed to the appropriate grids of the tube,
while the ordinary monochrome signal is fed to the cathodes of
the tube.
The third is a reference frequency generator which is synchronized
by the transmitted colour burst signal, and whose output is used
by the synchronous detector as a phase reference.
There are usually two controls associated with the chrominance
channel. The saturation control which alters the output of the
chrominance amplifier and therefore the " strength " of the colour
(with this control turned right down a monochrome or nearly
monochrome picture will result, while a " too coloured " picture
will be obtained if this control is turned right up), and the hue
control which determines the hue of the displayed colours. Turned
one way, flesh tones will be reproduced too green and turned the
other way, they will be too magenta.
Before considering installation procedure in detail, it must be
remembered that different manufacturers will have different designs
and therefore different installation instructions. Naturally, the
manufacturer's instructions must be strictly observed, but the
following are likely to be typical. Furthermore, it is assumed that
no faults are present as fault conditions are considered in Chapter 14.
13.2. Aerial requirements
A special aerial is not required for a colour receiver. An aerial
installation which provides a good quality signal as displayed on a
monochrome receiver should be perfectly satisfactory for operating
a colour receiver.
There are, however, one or two points to bear in mind. Remem-
ber that while a video frequency in the neighbourhood of 2.5 Mc/s
in the 405 -line system corresponds to fine detail in a monochrome
picture, and may not be regarded by some as being essential for
an acceptable monochrome picture, the sub -carrier in a colour
signal is about 2.7 Mc/s and this carries low frequency colour
information which is obviously essential to a colour picture. To
take an extreme example, if an aerial installation were to provide
a flat video frequency response up to 2.5 Mc/s with a sudden cut-
off above this, then 2.5 Mc/s bars of Test Card C would be visible
but the signal would be quite unsuitable for colour reception.
Again, if a close echo or " ghost " is present, this would have a
relatively small effect on a monochrome receiver but it could have
a disastrous effect on a colour receiver if the phase of the spurious
signal were such as to reduce the sub -carrier signal appreciably.
However, such a situation is rather unlikely, and in the authors'
experience, if a ghost is present which can be tolerated on a mono-
chrome receiver, then for 90% of cases it can also be tolerated on a
colour receiver. While on the subject of ghosts, it may be advisable
to point out that although the sub -carrier phase can be altered by a
ghost signal, the hue control can always be adjusted so that the large
area hues in a colour picture are correct. Hue errors are therefore
usually confined to vertical edges of objects, the width of the in-
correct hue being equal to the echo displacement.
As a general rule, therefore, an aerial installation which can
provide up to 3 Mc/s definition on a monochrome receiver should
be perfectly satisfactory for colour reception. Normally, this
Colour beats on a picture produced by (above) colour bar signal
with reference generator out of lock; (below) colour bar signal
with reference generator in a side -lock
Typical displays from the Marconi Vectorscope are for checking the performance
of the encoding equipment. The top photograph shows a colour bar display
showing vectors representing the six primary and complementary colours lying
within the appropriate boxes. The photograph on the left shows the same
display with the Q component removed. The photograph on the right shows
the same display with the I component removed. (Courtesy of Marconi Ltd.)
entails the correct choice of array for any particular location,
careful positioning and orientation of the array, and preservation
of the impedance matching between aerial and receiver. However,
the conscientious engineer will recognize the foregoing as routine
procedure for a monochrome receiver aerial installation.
Any ghost images which may be present, if acceptable on a mono-
chrome receiver, should also be acceptable on a colour receiver.
An interesting echo effect is sometimes observed in the case of the
405 -line system when a monochrome to colour (or vice versa)
change is made at the transmitter. Since the vision carrier has a
slightly different frequency in the two cases, it is possible for a posi-
tive echo in one case to change to a negative echo in the other for a
fixed aerial position. This effect can be observed on a monochrome
receiver, of course.
133. Monochrome adjustments
The first task in colour receiver installation is to obtain a satis-
factory black and white picture.
A large number of the receiver circuits required for monochrome
operation are identical in form with those of standard monochrome
receivers, the essential difference being the display tube and its
associated apparatus.
Apart from the usual time base, contrast, tuning, focus and
brightness adjustments, the additional adjustments which may be
required (but in any case must be checked) are concerned with raster
purity, convergence and grey scale tracking.
13.4. Purity
With no signal applied to the receiver, adjust the brightness
control so that the raster is visible and use the focus control to
obtain clearly defined scanning lines.
The raster should be approximately white in colour, but the
important point to notice is whether the colour is uniform all over
the raster. If it is not, the tube must be demagnetized by means of
a degaussing coil (see Section 11.6). Move the degaussing coil
slowly around the sides and front of the receiver, and while main-
taining a circular movement of the coil, gradually move back from
the tube face to a distance of five or six feet before switching off.
The purity should now be satisfactory, and a sensitive check can
be made by switching off the appropriate guns to obtain an all -
red raster. There may be some residual impurity near the edge of
the raster, in which case the nearest edge purity magnet should be
adjusted (see Figs. 13.1 and 13.2).
24
156,000 OHMS
TUBE SHIELD :7NOTE:-
/' ULTOR LEAD MUST
BE CONNECTED AS
SHOWN
PULL-UP RING ASSEMBLY s'
:\ /
ULTOR ANODE
LEAD
EDGE PURITY (COLOUR EQUALIZER)

MAGNETS -IN NEUTRAL POSITION
DEFLECTION YOKE
YOKE MOUNTING CLAMP CONVERGING COIL AND POLE PIECE

ASSEMBLY
STATIC CONVERGENCE
MAGNET ADJUSTMENT LATERAL MAGNET ADJUSTMENT
PURITY MAGNET
ADJUSTMENT TABS ana.t. p LATERAL MAGNET ASSEMBLY
TUBE SOCKET AND

I-4 BOARD ASSEMBLY
Fig. 13.1. Tube adjustments and components
If the degaussing process and the edge purity magnet adjustments

do not result in good purity, the purity magnets on the neck of the
tube and the deflection yoke assembly position must be investigated.
Set the purity magnets on the tube neck to neutral by moving the
tabs together, and also set the edge purity magnets to neutral so
that the two magnets in each position are in line one above the
other. Move the deflection yoke as far back as possible and
switch off the green and blue guns. If these guns cannot be extin-
guished by turning down the screen and background controls,
connect 1001dI resistors between the grids of the green and blue
guns and ground.
Now rotate the purity magnet around the tube neck and at the
same time adjust the tabs relative to one another until there is a
uniform red area at the centre of the tube. Move the deflection
yoke forward until a position is found which gives the best overall
red raster consistent with absence of neck shadow. Check the
purity of the green and blue guns independently; always adjust so
that purity errors show only at the edges of the screen. Correct
any edge impurity by means of the edge purity magnets.
13.5. Time bases
The time base adjustments are similar to the conventional mono-
chrome procedure, i.e., height, width and linearity controls should
be adjusted as required, while observing a grid or Test Card circle.
However, it should be noted that time base adjustment can affect
convergence and hence the time bases should be set up before con-
vergence is attempted. Time base adjustment can be conveniently
carried out by observing one display only, e.g., the green.
13.6. Convergence
Before attempting to set up the convergence, ensure that both
time bases are correctly adjusted. A generator which provides a
Ni4TOP OF CABINET
\\ STATIC CONVERGENCE
MAGNET ADJUSTMENT /
/ 4//
k /
I I1111°'
MO la BLUE
\REEN RED
\\NN444.-.""----- --/ \ \
1-
\ EDGE PURITY MAGNETS
JIN NEUTRAL POSITION)
/ ri------_,_ ____---- \\
/ /
/ / \\ \
41
Fig. 13.2. Edge purity and centre convergence magnets
cross -hatch of fairly fine white lines is preferable for dynamic con-
vergence adjustment.
A preliminary static convergence adjustment should first be
made by applying the cross -hatch signal to the display and adjusting
the bias on the guns so that all three colours are operative. After
adjusting the focus, and with all the dynamic convergence controls
set to mid -range, adjust the red, green and blue centre convergence
magnets (Fig. 13.2) and the lateral magnet (Fig. 13.3) so that the
three images are registered at the centre of the screen. The direction
of movement of the images is shown in Fig. 13.4(a) for the magnets,
CONVERGENCE MAGNET
POLE PIECES CENTRED
OVER TUBE POLE PIECES
FRONT EDGE OF HOLDER

TO BE IN LINE WITH FRONT
EDGE OF BLUE POLE PIECE
LEFT EDGE OF HOLDER

TO BE IN LINE WITH LEFT
EDGE OF BLUE POLE PIECE
LATERAL MAGNET
ADJUSTMENT SLEEVE
Fig. 13.3. Location of convergence and lateral beam magnets
(a) (b)
Fig. 13.4. Image movement pattern

while the lateral magnet produces a movement (Fig. 13.4(b) ) in

which the red and green movement is opposite to the blue.
Next, check the purity, and re -adjust if necessary. For vertical
convergence adjustment, apply the cross -hatch pattern, and observe
the vertical centre bar of the pattern. Adjust the red and green
tilt controls to produce equal displacement of the red and green
lines at the top and bottom of the line, as shown in Fig. 13.5(a).
Next, adjust the red and green amplitudes to produce straight,
parallel, red and green vertical lines. Now gradually reduce the
red and green amplitudes to converge the red, green and blue
lines adjusting the red and green tilt controls as necessary, to
keep the lines parallel. It should be possible either to make the
three images overlap, or show a slight displacement of red at one
side and green at the other with all lines parallel from top to bottom.
In the latter case, re -adjust the static convergence magnets so that
the three images are superimposed to give a vertical white line,
Fig. 13.5(b).
For setting up the blue vertical convergence, examine the centres
of the extreme top and bottom horizontal bars of the cross -hatch.
Adjust the blue vertical tilt and amplitude controls to produce an
equal downward displacement of the blue line from the top and
bottom bars, as shown in Fig. 13.5(c). Reduce the blue vertical
amplitude to converge all horizontal bars at the centre, from top to
bottom, adjusting the blue vertical tilt as necessary, to give the
result shown in Fig. 13.5(d).
The horizontal convergence adjustment will depend on the
type of circuits employed, but as an illustration a typical procedure
for a recent design will be given. In this, six controls are provided,
three right hand, Bl, RG1 and RG2 and three left hand, B2, RG3
and RG4.
With the cross -hatch pattern applied, adjust Bl to make the blue
line at the right hand centre a straight horizontal line, as shown in
Fig. 13.6(a). Now adjust B2 for a straight horizontal blue line
at the left hand centre, as shown in Fig. 13.6(b).
Adjust RG1 so that the vertical lines at the right hand side con-
verge, Fig. 13.6(c).
Adjust RG2 to make the horizontal red and green lines converge
at the right hand side as indicated in Fig. 13.6(a). Re -adjust B1
to make the blue line at the right centre fall on the converged
red and green lines, and re -adjust RG1 for convergence of the
vertical lines on the right hand side.
Adjust RG3 to make the vertical lines on the left hand side con-
verge, as in Fig. 13.6(d).
BLUE
WHITE
MEM 3
MI=
(a) (b) (C.)
B
(d)
Fig. 13.5. Vertical convergence patterns

Adjust RG4 to make the horizontal red and green lines converge
on the left hand side, Fig. 13.6(b). Repeat adjustment of RG3
to compensate for any interaction.
Re -adjust B2 to make the blue line on the left hand centre fall
on the converged red and green lines.
The cross -hatch pattern should now be converged on all parts
of the screen, as in Fig. 13.6(b).
13.7. Grey scale adjustment
The purpose of the grey scale adjustment is to provide a constant
white chromaticity at all drive levels, and a convenient signal is
one containing about five steps from black to peak white. Six
controls have to be adjusted, three of which are control grid or
4111111111111111
4111111111111111 111111111111111111
11111111111ME B110111111111,
1111111111=1/ 111111111111111F
(C)GBR GBA
(a)
411111111\
111111.11.11
MEM.
G
(b)
111111111/
(d)
Fig. 13.6. Horizontal convergence patterns
background controls used for setting the highlight colour and three
screen controls for setting the lowlight colour. In order to achieve
as bright a picture as possible, at least one of the screen controls
should be at a maximum after the setting up is completed.
With the three screen controls at maximum and the background
controls at minimum, set the contrast and brightness controls to
give a satisfactorily contrasted picture. The first step is to decide
which of the three screen controls should remain at maximum, and
this can be done by adjusting the three background controls to give
white in the highlights, and noting the lowlight colour.
If the lowlight colour is cyan, the red screen should remain at
maximum. Similarly, for a magenta lowlight, the green screen
should remain at maximum, and for a yellow lowlight the blue
screen should remain at maximum. Thus, the complementary
lowlight colour is the screen which should remain at maximum.
If the lowlight colour is red, green or blue, then gradually turn
down the screen of the predominant colour. If the lowlights
become grey, leave the remaining two screens at maximum and
adjust the background of the reduced colour to give white high-
lights. Check the lowlight colour again, if necessary adjusting the
screen which was turned down. In this case the tracking is com-
pleted.
If the lowlights become cyan, magenta or yellow rather than grey,
then this indicates that the red, green, or blue screens should remain
at maximum.
Having determined which screen should remain at maximum,
do not adjust either the screen or background controls for this
colour. Adjust the remaining two screen controls for grey low -
lights, and the remaining two background controls for white high-
lights, re -adjusting the two screens for lowlights if necessary.
13.8. Colour adjustments
The chrominance channel adjustments are chiefly concerned with
the saturation and hue controls, but first it is advisable to check the
operation of the reference frequency generator and its associated
circuits.
13.9. Reference frequency generator
The reference frequency generator has at least two and sometimes
three functions in a colour receiver. It provides a constant fre-
quency reference which is locked to the colour burst. It recognizes
the presence of the colour burst and switches the output chromin-
ance stages on or off depending on whether the burst is present or
absent, respectively (colour killing, as it is called) and in some cases
it provides automatic chrominance control, or A.C.C.
The free running frequency of the reference generator can be
checked by observing a colour picture, removing the burst feed to
the generator (by grounding it, for example) and also the killer bias
which ordinarily disables the chrominance output. The colour
content of the picture will then appear as horizontal colour beats
if the saturation control is turned up sufficiently. As the fre-
quency control of the generator is adjusted, so the horizontal beats
will reduce in number until finally, when the correct setting has
been reached, the colour beat becomes almost zero and drifts
through zero beat. Application of the colour burst should now
bring the generator into lock.
In some receivers, the burst gate which selects the colour burst
from the chrominance signal may be operated from the line time
base. In this case, adjustment of the line hold control will affect
the operation of the reference generator, and the manufacturer's
instructions should be closely followed on this point.
The colour killer may be affected by noise entering the receiver
with the signal, so that a noisy monochrome signal may switch
on the chrominance output stages. In order to prevent this, a
killer threshold adjustment is sometimes provided, and this is
usually set so that with a normal level of monochrome signal
applied, the cross -colour just disappears.
13.10. Saturation control
The saturation control is usually a customer operated control,
since the " colouring " of the picture is to some extent a matter of
taste. Also, newcomers to colour television tend to require rather
more than the correct saturation but after some hours of viewing a
lower and more nearly correct saturation is preferred by most people.
However, while the acceptable range of the saturation control
is quite large, there is strictly only one position for which the colour
receiver will reproduce the transmitted colours most accurately
and it is really the duty of the installation engineer to set the control
at this position.
Before attempting to adjust the saturation, first turn the control
right down and, if necessary, short out the burst input to the refer-
ence generator so that a monochrome picture is displayed. Check
that the purity, grey scale, contrast and brightness are acceptable.
Now slowly turn up the saturation while observing flesh tone and,
if possible, any obviously saturated colours. Eventually a point
will be reached where the saturation is too high. Now reduce the
saturation until the coloured content seems reasonable. For the
present, ignore any incorrect hues in the picture, for example too
green flesh tones.
13.11. Hue control
Having set the saturation control as above, adjust the hue control
so that flesh tones are neither too green nor too magenta. It
is recommended that only flesh tones should be observed while
adjusting the hue control, since this is a colour known by everyone,
and occurs most frequently. Further, it seems to be the most
sensitive colour in a subjective sense.
In view of the results of subjective tests carried out by the authors
on the relationship between hue control setting and viewer prefer-
ence, it would seem that while correct flesh tone lies between slightly
too green and slightly too magenta, it is preferable to err on the
side of magenta rather than green. After setting the hue control,
re -adjust the saturation control, if necessary.
13.12. Effect of other adjustments on saturation and hue
It should be pointed out that the degree of saturation of a colour
television picture can be appreciably altered by adjustment of the
master brightness control. Turning the brightness down will tend
to increase saturation; turning brightness up will decrease satura-
tion. Note that this effect is not caused by an interaction of
circuits. It is simply that a particular brightness level requires a
particular chrominance level.
Again, if the white balance is set too green, for example, and the
hue control is set to give acceptable flesh tone, then what should be
a pure red will be reproduced with a magenta cast.
The above points help to illustrate the importance of achieving
correct black and white reproduction before adjustment of the
chrominance channel is attempted.
13.13. Summary
In this chapter we have seen that the first task in receiver installa-
tion is to obtain a satisfactory black and white picture. This
necessitates a good aerial, a pure raster (i.e. a uniform colour over
the whole raster), a satisfactory grey scale and accurate registration
or convergence so that the three pictures are superimposed. Apart
from the adjustments one would make to a monochrome receiver,
it is likely that the only additional requirement will be degaussing
the tube, since the grey scale and convergence adjustments will
probably not need attention, though they should be checked.
Having achieved a satisfactory black and white picture, it is
advisable next to check the free running frequency of the reference
generator. Finally, the saturation control should be set to give a
picture which is judged to be sufficiently but not excessively coloured,
while the hue control should be set to give correct flesh tone. The
engineer may rest assured that the adjustment of the saturation
and hue controls is very simple, and calls for little practice.
CHAPTER 14
Colour Receiver Fault Finding

14.1. Introduction
Colour television receivers contain more components and cir-
cuits than monochrome receivers but are no more difficult to service.
In general the colour display tube provides more information on the
location of any fault condition and this extra information offsets
the increase in the number of stages. Experience, a systematic
approach and adequate test instruments are all valuable aids to the
For some years to come
colour service engineer, as in all servicing.
there will be considerable design differences between the various
makes of receiver and the service man must have a clear picture
in his mind of the block schematic diagram of the particular set
he is repairing. Whenever possible he should work with the
manufacturer's service handbook beside him. There are basic
differences between receivers using RGB amplifiers and those using
luminance and colour difference channels and these differences
affect the appearance of symptoms. The picture on the cathode-
ray tube will quickly localize the seat of the trouble however, and
the usual monochrome techniques of measuring voltages, currents
and waveforms will soon identify the specific fault if the general
principles of operation of colour reception are borne in mind.
The previous chapter gives an outline of the basic arrangement
of colour receivers and describes various adjustments which can be
made to the specifically colour parts of the circuits. The reader is
recommended to study Chapter 13 and to familiarize himself with
the colour test techniques given in Chapter 12.
Throughout this chapter the bandwidths, etc., quoted refer to the
405 -line system.
14.2. Safety precautions

Most of these will be second nature to the service man from his
experience with monochrome receivers. The fumes from faulty
selenium rectifiers should not be inhaled and the rectifiers should
not be brought into contact with the bare skin. The power supplies
will usually have a lower impedance than is normal with mono-
chrome receivers and shocks can be rather more serious. This is
367
particularly true of the e.h.t. supply which will be over 20kV with a
stabilized output of a milliamp or so. Colour tubes are heavier
and more awkward than black and white ones and may have more
energy stored in their screen capacity, so the e.h.t. terminals should
be discharged before handling. The risk of implosion is the same
as with black and white tubes but the replacement cost is higher.
Out of its cabinet, the chassis may generate some X-rays when
it is operating. The most likely sources are the front face of the
tube when the safety glass has been removed, the e.h.t. rectifier
and the e.h.t. stabilizer valve. Portable meters which indicate the
strength of such fields are available on the market, or small photo-
graphic film strips in suitable light -opaque wrapping may be left
near the suspected source or worn by the engineer. Subsequent
development of the film will indicate, from a knowledge of the
exposure time, the seriousness or otherwise of the amount of
radiation. In general, no hazard to health is likely to arise but the
dangers should be appreciated and a balanced view adopted.
14.3. Initial examination with no signal input

Before making any detailed examination of the receiver a cursory
and rapid inspection of the general receiver functions should be
made, as in normal service procedure. Indeed, so much of the
servicing of colour television receivers is similar to monochrome
work that most of the procedure will be given in brief note form,
suitable for quick reference at the work bench. The work bench
itself should be sited well clear of anything likely to disturb the
magnetic field, such as large iron or steel objects, transformers, etc.
(a) Inspect interior for signs of fire or damage.
(b) Check mains taps.
(c) Switch on and scrutinize for signs of smoke, voltage break-
down or corona.
(d) Raster should be uniformly white.
(i) If no raster or insufficient raster-treat as monochrome
and check e.h.t. and other tube electrode voltages and
time bases.
(ii) If raster uniformly coloured, check that all three guns are
operating and adjust screen and grid voltages for white
screen. Check gun cathode potentials.
Yellow raster indicates insufficient blue electron beam.
Magenta raster indicates insufficient green electron beam.
Cyan raster indicates insufficient red electron beam.
COLOUR RECEIVER FAULT FINDING 369
(iii) If raster has coloured patches, follow installation pro-
cedure for purity adjustments, field edge equalizer magnet
adjustments and if necessary degauss.
(e) All three beams should go through optimum focus at same
setting of focus control. If one beam is significantly out of
focus at optimum setting, only cure is to replace tube.
(f) Turn up contrast and check that luminance and sound
channels have sufficient gain to produce noise-treat as
monochrome.
14.4. Examination with monochrome signal input
Monochrome test patterns are an essential part of colour tele-
Linearity and convergence patterns, resolution
vision testing.
wedges, Test Card C or D, etc., are all useful (see Chapter 12),
O 0
/".
(a) (b)
Fig. 14.1. Incorrect static convergence. The limit spot positions

are shown with the convergence field (a) too large, (b) too small
as are normal monochrome picture transmissions, but check that

aerial has adequate response over 21 Mc/s to 3 M/cs region.
(a) Check luminance gain-as for monochrome.
(b) Re -check purity and focus.
(c) Check raster size and linearity, as monochrome. If conver-
gence is incorrect it may be convenient to adjust linearity
with only one gun on (green is the brightest). Unless linearity
is good it may not be possible to converge rasters accurately.
(d) Check static convergence (see receiver installation procedure).
If faulty, check magnets, including blue lateral shift magnet
and/or D.C. current through convergence yoke windings.
If blue spot cannot be moved below green and red spots,
convergence field or current is too high. If blue spot cannot
be moved above other two spots, convergence field or current
is too low (Fig. 14.1). Try reversing magnets in their holders.
As last resort, change tube.
(e) Check dynamic convergence (see receiver installation).
(i) If one beam cannot be controlled, check convergence
yoke winding.
(ii) If centre but not the sides can be converged, check
horizontal convergence amplitude control and current
through it, with oscilloscope. Check input waveform
from line time base.
(iii) If raster comes into convergence on one side of screen
before the other, check horizontal convergence tilt or
phasing control.
(iv) If raster will converge at sides but not at top and bottom,
check vertical convergence controls, waveforms, and
yoke windings. Check input from field time base.
(v) Look out for misleading effects. Poor frequency
response in one colour channel may cause fringe of colour
on one or both sides of convergence spot. Similarly,
poor focus of one beam or low emission in one gun, will
produce colour halo around convergence spot. Inspect
primary pictures one at a time. Sudden jumps in con-
vergence positions may be due to faulty e.h.t. stabiliza-
tion.
(f) Check grey scale tracking (see installation procedure, Section
13.7).
(i) Adjust tube grid (background) controls so that highlights
are neutral and a reasonable match to Illuminant C. If
dark greys are then coloured (say red), reduce screen
volts of appropriate gun (i.e. red) or increase screen volts
of other two guns if possible, and re -adjust grid volts.
(ii) If RGB amplifiers used, check relative RGB gains and
maximum outputs.
(iii) If neutral grey scale not possible, check range of grid and
screen volts.
(iv) Check tube cathode volts and check video output stages,
including colour difference outputs if used. Remember
stages may be D.C. coupled. Check gains of RGB
amplifiers.
(v) Check matrix resistance values, unless these only control
colour difference signals.
(vi) Check spot limiters.
(g) Check frequency response and R.F.-I.F. alignment as
monochrome.
(i) If one block of resolution bars, say 2 Mc/s, is missing
from Test Card C, sub -carrier rejector is off tune.
(ii) If marked echoes at intervals of about 1µs-check
termination of luminance delay cable.
(iii) Check that the luminance response is fairly flat to past
the sub -carrier frequency when the sub -carrier rejector
circuit is disconnected (or shorted as appropriate).
(iv) Incorrect convergence will markedly affect picture
resolution.
(h) Check colour killer operation.
(i) Adjust saturation to maximum. If coloured patches
appear on picture where there is fine monochrome
detail, then colour killer is not operating.
(ii) Check killer threshold adjustment, if any.
(iii) Check bias on chrominance stages.
(iv) Check D.C. output from auxiliary phase detector in
reference generator or its equivalent.
(j) Check chrominance gain. This is more conveniently done
on colour transmission. However, if killer voltage is shorted
or removed, and saturation turned to maximum, cross -
colour effects should show on monochrome transmission if
chrominance channel is amplifying.
14.5. Examination with colour signal input
Do not attempt to check the colour performance until the mono-
chrome performance is correct. If only a colour signal is available,
operate the colour killer or turn down the saturation or bias -off
the chrominance channel, so that the receiver reproduces a mono-
chrome picture. Good colour reproduction is not possible if the
monochrome grey scale tracking is faulty, or if the picture white is
not Illuminant C.
Be sure that the aerial and any distribution system in use has
adequate response, without peaks, in the vicinity of 2.7 Mc/s.
The colour bar waveform is a convenient signal for fault tracing
as the service engineer will soon recognize the video and chromin-
ance waveforms to be expected on this signal (Fig. 5.2).
(a) If monochrome picture but no colour.
(i) Check fine tuning.
(ii) Short out or disable colour killer. If coloured beats,
reference drive exists but oscillator is not locked. If
not even coloured beats, there is no reference drive.
(iii) Check saturation control voltage and bias on chrom-
inance stages.
(iv) Check gain of chrominance amplifier.
(v) Check that chrominance is being fed to reference
generator.
(vi) Check burst gate is properly timed and burst is reaching
phase detector.
(vii) Check D.C. bias on reference oscillator as test that
oscillator is oscillating.
(viii) Check oscillator is approximately on tune.
(ix) Check sufficient amplitude of reference is reaching syn-
chronous detectors.
(x) Check waveforms at inputs and outputs of synchronous
detectors.
(xi) Check waveforms through matrix circuits.
(b) No colour lock. With killer disabled, horizontal bands of
colour moving vertically across the picture show that the
reference oscillator is slightly off tune. The nearer it is on
tune, the broader and fewer these bands become.
(i) Check burst input to loop detector.
(ii) Check burst gating.
(iii) Check bias at grid of reactance valve. If bias is large,
earth grid of reactance valve. If colour bars then roll
slowly near in -tune position, loop discriminator is
faulty. If colour remains badly out of sync, then
oscillator or reactance tube is faulty.
(c) Long pull -in time. With killer disabled, if horizontal bands
of colour do pull into sync but take a minute or so to do this,
check these three points.
(i) Tuning of oscillator.
(ii) If quadricorrelator circuit, check auxiliary discriminator
output.
(iii) Operation of variable time constant in feed to reactance
valve.
(d) Incorrect colour. The colour bar signal is a great aid in
diagnosing faulty colour response.
Ii pical frequency characteristics of a colour
receiver measured at the luminance detector
with the sweep input applied to different points.
In the top photograph the signal is applied to the
third I.F. stage grid and shows the sub -carrier
marker on the left. In the centre and lower
photographs the input is applied to the mixer
grid and aerial input respectively. (Frequency
sweep is from right to left.)
Receiver image misregistration; (above) effect caused by excessive delay in
receiver luminance channel: (below) effect on black -and -white picturelrepro-
duced on a shadow mask tube with static convergence grossly out of adjustment
(i) Note absence of any colours. This will usually localize
a fault in a particular synchronous detector or its subse-
quent amplifiers and matrix, if any.
Orange and cyan colours only-insufficient Q' signal
(insufficient (B' - Y') is similar).
Green and magenta colours only-insufficient I' signal
(or (R' - Y').
When the I' signal fails there is usually very little colour
left in a normal outdoor scene.
If demodulator or colour output valve dies, the D.C.
coupling to the display may give picture a colour cast
similar to the colour axis of the decoder, e.g. failure of
(R' - Y') demodulator can give overall surplus of red
in picture.
If reference oscillator pulls into side lock the wrong
colours will be obtained and these will give the appearance
of lack of drive.
(ii) Check setting of hue control. If all colours are present
but occupy wrong positions, check operation of hue
control, reference feed, phase shift circuits, and matrix
(see Chapter 12 test methods).
Critical test is to view colour bar output from one gun
only (black out other guns or view through sharp cut
optical colour filter).
Adjustment of hue, contrast and saturation controls
should make all visible bars the same brightness. If this
is not possible, check demodulation angles of syn-
chronous detectors, and matrix. Check all three guns.
(iii) If only pale and desaturated colours are present-
check fine tuner, chrominance gain, A.C.C. voltage from
reference generator, R.F.-I.F. alignment, side lock of
reference oscillator.
(iv) Coloured hum bars. Check heater -cathode leakage in
appropriate colour channel, reactance valve and oscil-
lator; check also for hum on grid of reactance valve.
(e) Faulty colour transitions.
(i) Colours smear. Check fine tuner and chrominance
bandwidth. Check sufficient reference drive to synchro-
nous detectors.
(ii) Colour displaced from luminance outline of object.
Check delay circuits. Colour too far to viewer's right
is insufficient luminance (and I') delay.
25
( i i i) Incorrect colour edges. Insufficient luminance high
frequencies. Too wide Q' channel response.
Too wide chrominance bandwidths in equiband receivers.
Incorrect chrominance channel alignment.
Feedback from luminance sub -carrier rejector circuit to
chrominance channel.
Note that the colour bar generator is not, in general, a
realistic test of colour transitions-especially for the
green -magenta transition where all of the three basic
rectangular waveforms are changing, the green signal is
being switched off and the red and blue signals are being
switched on. Note also, from Chapter 16, that the
N.T.S.C. system has certain deficiencies in its transition
performance, although these are relatively minor on the
average programme scene.
(f) Coloured beats.
(i) Check reference oscillator is in -sync.
(ii) Check 3.5 Mc/s rejector in chrominance channel is on
tune.
(iii) Check sound I.F. rejection.
(iv) Check input level not high enough to cause cross -
modulation.
(v) Carrier interference at R.F. or I.F. near to sub -carrier
frequency.
Note that such interference may come from harmonics
of reference drive or chrominance.
For example :
The 13th harmonic of the sub -carrier is 34.5515625 Mc/s.
The vision carrier I.F. is normally 34.65 Mc/s.
The 14th harmonic of the sub -carrier is 37.209375 Mc/s.
The standard sub -carrier I.F. is 37.31 Mc/s.
(g) Monochrome beats on colour transmission.
(i) Check sound rejector circuits in luminance channel.
(ii) Consider possibility of c.w. interference.
(h) Picture on sound on colour transmission only.
Insufficient rejection in sound channel at sub -carrier frequency.
(j) Colour appears and disappears.
(i) Check killer threshold setting.
(ii) Check A.C.C. circuit.
(iii) Check level of burst signal to reference generator.
(iv) Check timing of burst gating.
(v) Check intermittent oscillation of reference oscillator,
or frequency jumping.
(k) Colour varies with picture content.
(i) Check grey scale tracking.
(ii) Check burst gate timing.
(iii) Check for differential phase effects in common lumin-
ance-chrominance stages.
(iv) Check A.C.C. is operating and that chrominance signal
is not limiting before saturation control.
(v) Check D.C. level maintenance.
(1) Colour accuracy varies over picture area.
(i) Check purity.
(ii) Hum in phase discriminator, reactance valve or reference
oscillator.
(iii) Hum in chrominance channel.
(iv) Check constancy of reference output from reference
generator.
14.6. Miscellaneous
14.6.1. NOISE
In fringe area reception, random noise may appear in both the
luminance and colour channels. It will generally be more notice-
able in the white parts of the picture where it can produce coloured
flecks. In areas of saturated primary colour, such as green, it will
change the colour from one green to another green and the only
obvious fluctuation will be in the luminance noise (see Section 4.14).
However, aside from this effect, faulty reference oscillator sync
can give rise to hue fluctuations, particularly in the subtractive
primary colours yellow, cyan and magenta, due to noise. This
may be due to insufficient integration in the reference generator, or
the quadricorrelator may not have changed over to the in -lock
position.
14.6.2. APPEARANCE OF BURST
A vertical bar of green on the left hand side of the picture will
probably be the colour sync burst appearing on the horizontal
flyback due to insufficient line blanking.
14.6.3. LINE TIME BASE RADIATION
The increased energy in the line time base makes it even more im-
portant to keep the line time base shielding and mains filtering intact.
14.6.4. REFERENCE OSCILLATOR RADIATION
This is a new source of television receiver interference with other
services. Screening and decoupling of reference generator and
demodulators must be above reproach.
It is also important to prevent the reference signal from spraying
back into the burst signal path and eventually reaching the input
to the reference generator; this can cause the reference oscillator to
go well off frequency (see Section 10.13). When replacing com-
ponents, wiring or screening, these points must be borne in mind.
14.7. Summary
In servicing the colour television picture much information can
be obtained from the picture itself. Always work for a good mono-
chrome picture before trying to analyse an obscure colour fault.
Become familiar with the picture appearance when either one of
the primary colours or one of the chrominance signals is missing.
Once the approximate location of the fault has been established,
the normal monochrome techniques of voltage, current and wave-
form measurements will pin -point the faulty component.
CHAPTER 15
Monochrome Reception
of N.T.S.C. Signals
15.1. Introduction
The N.T.S.C. system is designed in such a way that the colour
signal on an unmodified monochrome receiver will be displayed as
a black and white picture of the transmitted colour scene. This
compatibility of the N.T.S.C. signal is an extremely important feature
because it allows colour to be added to an existing monochrome
system, without modification of existing channel allocations, and
also the need to duplicate the transmission for colour and mono-
chrome receivers is removed.
In a perfectly compatible system, the operation of a monochrome
receiver by a oolour signal should be indistinguishable from its
operation by a monochrome signal, but while the N.T.S.C. system
does approach this ideal, the monochrome display of an N.T.S.C.
signal i$ not identical with that of a conventional monochrome signal.
The N.T.S.C. signal has been described in Chapter 4, and it will
be remembered that it consists of a monochrome signal to which a
sub -carrier signal carrying the chrominance information has been
added. Now consider the effects of an N.T.S.C. signal as displayed
by a monochrome receiver.
15.2` Asynchronous working
Apart from the sub -carrier, the N.T.S.C. monochrome signal
differs from its conventional counterpart in that the scanning
frequencies are derived by division from the sub -carrier frequency,
and consequently the line and field scanning frequencies are crystal
controlled and therefore independent of the mains supply frequency.
Thus, the N.T.S.C. signal entails asynchronous working of all
receivers, so that sufficient smoothing must be included to prevent
undulations on the picture and brightness hum bar modulation.
The reader may wonder why it is necessary to crystal control the
sub -carrier frequency, and hence the scanning frequencies, at the
transmitter. Why not have a nominal sub -carrier frequency of
377
2.66 Mc/s, divide down to the scanning frequencies, and compare the
field frequency with the mains supply frequency and thereby control
the sub -carrier frequency to give mains synchronous working?
If this scheme were adopted, colour receivers would not necessarily
be able to generate a sub -carrier reference frequency which is essen-
tial in order to demodulate the sub -carrier signal. For example, if
the supply frequency changed by 2 % (from 50 c/s to 49 c/s), the sub -
carrier frequency would also change by 2 % and 2 % of 2.66 Mc/s
is about 53 kc/s. Now it is shown in Section 10.12 that the gated
nature of the colour burst synchronizing signal puts a limit, for the
405 -line system, of + 5 kc/s on the frequency difference between the
transmitted sub -carrier frequency and the free running frequency of
the receiver's reference frequency generator. Hence, conventional
synchronous working would be possible only if the mains supply
frequency were held to within about 0.2 %, and for economical
reasons in the colour receiver, preferably much less than this. A
change of only 0.02 % in sub -carrier frequency (about 530 c/s) would
be difficult to accommodate in the colour receiver, and would severely
limit the choice of reference generator.
However, one method of synchronous working has been suggested
which could be adopted in colour transmissions. The sub -carrier
frequency is crystal controlled in the usual way so that colour re-
ceiver reference frequency generation is unaffected, but a mains syn-
chronous, or very nearly mains synchronous, field frequency is derived
by changing one of the dividing factors in the system. Thus, instead
of varying the master oscillator frequency in sympathy with mains
frequency variations, a division factor is changed appropriately. In
this case, the line frequency could be crystal controlled by dividing
down from the sub -carrier frequency, and the field factor 405 could
be changed appropriately, but always to an odd number in order
to preserve interlace. Alternatively, the sub -carrier frequency to
525
twice line frequency factor (-4 for the 405 -line system) could be
changed (again to an odd number of quarters to preserve the dot
interlace), and in this case the number of lines would not alter, since
the line and field frequencies would change together.
Although maintaining mains synchronous operation may involve
complicated circuit arrangements, since the transmitter and not the
receiver is involved, the complication is probably justified.
15.3. Carrier frequencies
In the 525- and 625 -line N.T.S.C. systems the vision and sound
carriers are the same as in their corresponding conventional
MONOCHROME RECEPTION OF N.T.S.C. SIGNALS 379
monochrome systems. This is necessary in order to maintain the
same frequency difference between the vision and sound carriers so
that intercarrier sound operation is not affected on existing mono-
chrome receivers.
In the British 405 -line N.T.S.C. system, however, the difference
between the vision and sound carriers is changed from the conven-
tional 3.5 Mc/s to approximately 3.54 Mc/s (strictly, 3.54375 Mc/s).
This change is a result of locking the frequency difference between
the sub -carrier and the sound carrier to an odd multiple of half the
line frequency so that any beat between the sound carrier and the
sub -carrier has minimum subjective visibility.
The usual practice is to keep the same sound carrier frequency as
in the conventional monochrome case, and to increase the vision
carrier frequency by about 40 kc/s. By this means the sound carrier
traps in existing receivers are still correctly tuned, while the small
increase in vision carrier frequency causes a very slight drop in carrier
response at the nominal 6dB down point.
Thus, the 405 -line N.T.S.C. vision carrier is very slightly higher
than its conventional value, and this change has a negligible effect
on monochrome receivers.
15.4. Sub -carrier dots
The most significant difference between a conventional mono-
chrome signal and an N.T.S.C. type of signal is the presence of the
sub -carrier dots in those areas of the picture where the colour is not
the Illuminant C reference white. These dots produce two direct
effects on a monochrome receiver: firstly, they appear as an inter-
ference effect, and secondly, they produce an increase in luminance
due to rectification by the non-linear light output versus voltage
input characteristic of the cathode-ray tube. Both these effects
increase with the purity times luminance product of the transmitted
colour, since it is this product which determines the amplitude of the
sub -carrier signal.
As far as the interference effect of the dot structure is concerned,
a BBC report on compatibility tests on the 405 -line system states
that 2.7 % of a total of 624 engineer observers found the dot pattern
" somewhat objectionable ", while 1 % found it definitely
objectionable ", on a motion picture test. Furthermore, the picture
size influences the results to the extent that more than one grade of
subjective difference exists between pictures under 9 in. high com-
pared with pictures over 12 in. high. Thus, if an observer were to
grade the dot structure as " somewhat objectionable " on a small
picture, he would grade it as rather more than " definitely
objectionable " on a large picture. The subjective grading used in
these tests was as follows:
Criterion Grade
Imperceptible I
Just perceptible 2
Definitely perceptible but not disturbing 3
Somewhat objectionable 4
Definitely objectionable 5
Unusable 6
It is interesting to note that the results obtained for static slides

showed that 2.2 % of observers found the dot pattern " somewhat
objectionable " while 0.6 % found it " definitely objectionable ".
Since the colorimetric purity of scenes in the motion picture was not
higher than that of the slides, it is suggested that the more unfavour-
able results for the moving pictures may have been caused by
strobing effects which upset the dot integrating effect, and also there
is the possibility that observers may have based their comments for
the moving pictures on those scenes in which the dots were most
conspicuous.
The dot structure on a monochrome display can be reduced either
by rolling off the video response before the sub -carrier frequency, or
by including a notch filter tuned to the sub -carrier frequency. The
former method naturally severely restricts the picture definition, and
is therefore not recommended, but it has been found that a notch
filter with a rejection of about 12dB at the sub -carrier frequency and
with a 3dB bandwidth of about 400 kc/s considerably reduces the
large area dot structure in the case of the 405 -line system, and pro-
duces only a slight degradation of definition. Thus, the 21- Mc/s bars
of Test Card C have reduced contrast, but the 3 Mc/s bars and edge
definition are scarcely affected.
It should be noted that only the large area dot structure can be re-
duced since the dot structure along edges corresponds to sidebands
of the sub -carrier frequency which fall outside the greatest rejection
range of the notch filter. Complete rejection of all dot structure
can be achieved only by limiting the video response to about 1.7
Mc/s in the case of the 405 -line system, with unreasonable conse-
quences to definition.
In view of the results obtained from compatibility tests, the BBC
recommend that a notch filter is essential in the case of the larger
displays, if these are not to become an obstacle to the introduction
of a 405 -line N.T.S.C. system.
15.5. Sub -carrier dot rectification
Besides appearing as an interfering pattern in coloured areas, the
subcarrier also produces a contribution to the luminance in such
areas owing to the non-linear law of the monochrome display tube.
Thus, consider a coloured area having a monochrome signal
value Y' = 0.30RY 059GY + 0.1 1 BY and a sub -carrier signal
LUMINANCE
OUTPUT
MEAN LEVEL
DRIVE VOLTAGE SPURIOUS "D.C:' LUMINANCE
CORRECT LUMINANCE LEVEL
GRID -CATHODE
VOLTAGE OF
C.R.T.
Fig. 15.1. Luminance increase by

sub -carrier rectification due to C. R.T.
SUBCARRIER SINE WAVE characteristic
SIGNAL
.4F --LUMINANCE SIGNAL
S sin wt, displayed on a monochrome tube whose luminance output

L is related to the signal drive V according to the equation
L = KVY
where K is a constant.
For a square law display, the luminance output for the colour
signal will then be given by
L = K (Y ' S sin (.002
S' S'
= K[ (Y 'y + 2 Y'S sin wt - cos 2 cot]
The mean value of the displayed luminance is then
L K[(111)2 + 2-
21
Hence, rectification of the sub -carrier signal produces a con-

tribution to the displayed luminance which is proportional to the
square of the sub -carrier amplitude in the case of a square law dis-
play. Fig. 15.1 shows diagrammatically how the sub -carrier sine
wave voltage produces a distorted luminance output waveform
which has a spurious " D.C." component.
In practice, most monochrome tubes have a power law of about
2.7, so that a closer approximation to the display luminance would
be
L KV3 = K ( I" S sin cot)3
Using the identities
2 sins B = 1 - cos 20
and
2 sin3 0 = sin 0 - sin 0 cos 20
sin 0 -4 [ sin (0 - 20) + sin (0 + 20)]
it follows that the mean luminance is
3 '
(Y)3 + 2 S2
It is interesting to evaluate the display luminance for the full pri-
maries and their complementaries for the case of square and cube
law displays. Table 15.1 illustrates the results.
It can be seen that the luminance contribution of the sub -carrier
makes the displayed luminance approach the correct Y value for
Table 15.1
RELATIVE LUMINANCE VALUES OF SATURATED COLOURS
Colour 012 (n2 + S; Y (r)3+3r (Y')3
Red 0.63 0.090 0.288 0.30 0.205 0.027

Green 0.59 0.348 0.522 0.59 0.513 0.205
Blue 0.44 0.012 0109 011 0.033 0.001
Cyan 0.63 0.490 0.688 0.70 0.760 0.343
Magenta 0.59 0.168 0.342 0.41 0283 0.069
Yellow 0.44 0.792 0.889 0.89 0.965 0.705
both the square law and the cube law displays, and in fact, in the
square law case, the displayed luminance is nearly correct for all six
colours.
Notice that if the sub -carrier luminance contribution were absent,
the displaced luminance would be proportional to the figures in the
( Y')2 or ( Y')3 columns, depending on the display law. These show
considerable variations, and the effect on the luminance display is
to produce a non -panchromatic characteristic. Thus, a red of
luminance 0.3 would be displayed as a luminance of 0.09 or 0.027
(depending on the display law), while a green of luminance 0.3
would be displayed as a luminance of 0.188 or 0.082 for square
and cube law displays, respectively, and for y = 2.2.
However, the BBC compatibility tests indicated that viewers
show no marked preference for or against the presence of the sub -
carrier signal, as far as its effect on tonal gradation is concerned.
15.6. Transient effects
The above discussion on the luminance contribution of the sub -
carrier applies to large picture areas into which the relative frequency
characteristics of the luminance and sub -carrier signals do not enter.
But in the case of a sudden change from one colour to another, or
even from one luminance level to another without a chromaticity
change, it is obvious that the transient response will be affected by
the restricted bandwidth of the sub -carrier signal. Since, for all
colours other than Illuminant C, the sub -carrier carries some of the
luminance information, then clearly this luminance component is
restricted in bandwidth to the same extent as the chrominance
signals, and hence the sub -carrier luminance contribution must have
a relatively poor rise time. All luminance transients (except grey
ones) are therefore built up of a fast Y' transient and a slow sub -
carrier luminance transient.
In addition, it is possible for a monochrome receiver to display
a luminance transient when the original transition is between two
different colours of equal luminance. This effect can occur when
the chromaticity passes near the white point during the transition,
for then the sub -carrier luminance contribution is necessarily smaller
than it was at the start of the transient, or will be at the end of the
transient. Thus, the sub -carrier luminance contribution has a
minimum value or a " notch " when the transition passes nearest
to the white point, and because of the bandwidth limitation of the
sub -carrier this notch is broadened out.
The initial and final values of a luminance transient, as displayed
on a monochrome receiver, will depend on the actual colours between
which the transition occurs, and during the transient the luminance
will also be dependent on the chromaticity path taken by the
originating transition. As an illustration of the processes involved,
consider a transition between a full pure red and a pure green of
0.51
( b)
Bi
(c) 0
Y L
03
(d)
1.0
Fig. 15.2. Instantaneous tristimulus values and luminance
of transition from full red to pure green of equal luminance
before encoding
equal luminance. For simplicity, assume a gamma of 2 and a square

law display so that the displayed luminance is proportional to
( n2+ _S 2
2
as derived in Section 15.5.
Let us further assume that the colour transition being televised
leads to instantaneous tristimulus values which vary linearly with
time, so that if the time t is measured from the start of the transition
which is complete after a time T
_t 0.30 t
Gi = and Bi = 0
T 0.59 T
are the instantaneous numerical tristimulus values.
These values are plotted in Fig. 15.2 and notice that the true
luminance Yi remains constant throughout the transition.
Now the displayed luminance will have two components, one due
S2
to ( Y')2 and one due to -2
'
The luminance produced by the Y' signal will be
( Y')2 = [0.30 Rii + 0.59 Gii + 0.11 Bel2
= [0.30 (1 -x)i + 0.59 (00:3509x)12
= 0.09 + 0.087x + 0.6 'A/0.177x (1 - x)

where x = -T.
This luminance contribution is plotted in Fig. 15.3 (a) for values

of x between 0 and 1, i.e. during the transient.
Since the sub -carrier amplitude S is given by
(R' - Y')2 (B' - Y')2
S2 - 1.142 2.032
and since
(R' - Y') = 0.7 (1 - x) - (0.177x)i
and
(B' - Y') -0.3 (1 -- x)i - (0.177x)i
it follows that
S2 049 - 0.313x - 1.4 A/0.177x (1 - x)
2 1.142.2
0.09 + 0.087x + 0.6 A/0.177x (1 - x)

2.032.2
which gives the instantaneous luminance contribution of the sub-
carrier. This is plotted in Fig. 15.3(b) assuming that the transient
0.3
0.z
LUMINANCE
DISPLAY OF
(YI)2
SIGNAL
0.1
(a)
02-
2
01 .- LUMINANCE
DISPLAY OF
SUBCARRIER
FULL BANDWIDTH
0
(b)
0.2
2
z0.1 LUMINANCE
DISPLAY OF
SUBCARRIER
RESTRICTED
BANDWIDTH
(c)
0.3
TOTAL
DISPLAYED
LUMINANCE
0.2
(y I)24.s2
2
0.1-
(d)
T '1
3T
Fig. 15.3. Monochrome display of the transition from full red to pure green of
equal luminance for a square law display. y = 2
is complete after a time T. However, since the sub -carrier signal has
a restricted bandwidth, the curve has been re -plotted on the assump-
tion that the transient takes three times as long, Fig. 15.3(c).
The total displayed luminance, which is the sum of the ( Y')2 and
2 values, is shown in Fig. 15.3(d).
There are several points of interest in Fig. 15.3. Notice that,
although the correct luminance is constant, not only does ( Y')2
start and finish at different values but it also exhibits an overshoot.
Hence, even if the sub -carrier were not transmitted, an entirely
spurious luminance transient would be displayed.
S2
Again, the graph of ( Y')2 -y starts and finishes at different
values, but these are closer to the correct luminance level of 0.3 than
However, a slow and entirely
are the initial and final values of ( Y')2.
spurious luminance transient is displayed.
Note that, in Fig. 15.3, it has been assumed that Y' is delayed at
the transmitter so that the centre of the Y' transient coincides in time
S2
with the centre of the slower -2 transient. This makes the transient
more symmetrical and therefore subjectively more acceptable.
Unlike the large area sub -carrier luminance contribution, the
transient sub -carrier luminance effects cannot be removed by a
sub -carrier rejector filter. This is because the sidebands of the
sub -carrier are significant during fast transients, and they would fall
outside the rejection range of a notch filter. Sub -carrier luminance
transient effects can be removed only by limiting the video response
of the monochrome receiver to exclude all possible sidebands of the
sub -carrier, and this would lead to a video response of only about
1.7 MO instead of the usual 3 Mc/s for the 405 -line system.
15.7. Burst visibility

Another effect associated with an N.T.S.C. type signal arises from
the presence of the colour burst synchronizing signal, which is
transmitted so that colour receivers may generate a continuous
sine wave locked to the transmitter sub -carrier frequency for syn-
chronously detecting the quadrature modulated sub -carrier. This
burst occurs during the post line synchronizing period, which usually
forms part of the line flyback period of a monochrome receiver,
and consequently the sub -carrier dots due to the burst are " stretched
out " by the relatively fast flyback scan. The effect of this is to
produce vertical lines down the left hand side of the picture as shown
in Fig. 15.4. In general, these burst lines are visible only on those
pictures which contain a black area on the left-hand side, and their
visibility is a function of the D.C. gain of the receiver, the line
flyback time, and the setting of the black level control. Complete
invisibility of the burst lines can be achieved by applying a line
blanking pulse to the C.R.T. or video stage, in a similar manner
Fig. 15.4. Display of colour burst on a monochrome receiver. The effect has
been emphasized for clarity
to the field blanking which is often employed to remove the line

scan during the field flyback.
Burst visibility on monochrome receivers does not appear to be
a significant effect, for the BBC compatibility tests showed that
only 1.2% of the general public, and 1.8 % of engineers in the tests,
found burst visibility " very annoying ".
15.8. Synchronizing effects
The sub -carrier signal occupies an appreciable part of the sync
pulse range of the video waveform, particularly for full saturated
red and blue, and for the colour burst. However, the colour burst
occurs some time after line synchronization has taken place, and its
position does not change with picture content. One would have
expected to find greatest sensitivity to line synchronizing trouble for
full saturated red or blue on the extreme right of the picture, where
the sub -carrier signal would be only about 1ps from the leading edge
of the line synchronizing pulse, and, of course, for greatest sensitivity
to field synchronizing trouble, saturated red or blue at the bottom of
the picture would appear to be the severest test.
The BBC compatibility tests indicate that synchronizing difficulties
were mentioned by about 5 % of observers, but the severity of the
effects is difficult to assess. Previous tests have shown that colour
transmission causes little trouble with receiver synchronizing circuits.
In the authors' experience, synchronizing trouble has been found
only when the transmitted signal is not correctly balanced. Thus,
if sub -carrier is present on the synchronizing waveform (which, of
course, is not in accordance with N.T.S.C. standards), then the
receiver line time base synchronization is " pulled " so that sub -
carrier dots on the picture are arranged in vertical lines, that is,
the dot interlace is destroyed.
15.9. Sound buzz
It will be appreciated that the sub -carrier which is present in a
colour signal is necessarily blanked at the line and field frequencies,
in the same way that a conventional monochrome signal is blanked.
Therefore, the sub -carrier and its sidebands are modulated with line
and field frequency components. The field frequency component
can, under certain circumstances, cause a buzz in the sound channel
of a receiver. There are two mechanisms by which this buzz may
enter the audio channel; either by cross -modulation in the R.F.-
I.F. channel, or the sub -carrier sidebands in the upper video region
may be accepted directly by the sound channel.
Adequate sound channel selectivity and freedom from cross -
modulation should ensure satisfactory sound reception. The BBC
compatibility tests showed that 4.4 % of the general public and 1.8 %
of engineer observers regarded sound buzz as being either moderately
or very annoying. The difference between the general public and
engineer observer results may be due to misalignment of the general
public receivers; for example, an incorrectly tuned sound I.F. could
produce sound buzz.
15.10. Summary
An N.T.S.C. colour signal consists of a monochrome signal to
which a sub -carrier signal, carrying the colour information, has been
added. When this signal is received on a monochrome set, the black
26
- 4 r

and white picture produced differs in certain respects from a picture
produced by a monochrome transmission.
The monochrome component of the colour signal differs from a
black and white signal in two ways. Firstly, the line and field
frequencies are constant and so the field frequency of the colour
signal is not mains synchronous. Secondly, except for white and
shades of grey, it is not a true measure of luminance. This difference
occurs because the luminance signal of a colour transmission is
formed by adding together the appropriate proportions of R, G and
B signals, which have been individually gamma corrected, whereas
the luminance signal of a black and white transmission is gamma
corrected as a whole. The effect of this difference on reception is
to cause spurious luminance transients.
The sub -carrier appears as an interfering signal which causes a
dot pattern on those areas of the scene which are coloured when
televised, but since the sub -carrier frequency is locked to the scanning
frequencies, this pattern has minimum visibility.
The action of the sub -carrier on the non-linear light output versus
voltage drive characteristic of the C.R.T. produces an increase of
luminance. This increase tends to correct the luminance distortion
which is caused by the monochrome component signal but this
correction is only effective over large picture areas, because the sub -
carrier has a limited bandwidth.
All colour transitions, except those between shades of grey, will
be displayed as a fast transient superimposed on a slower transient
which is approximately complementary in shape.
Minor effects caused by the monochrome reception of a colour
signal can include visibility of the burst, scan synchronization effects
and sound channel buzz.
CHAPTER 16
Shortcomings of N.T.S.C.
Systems
16.1. Introduction
An N.T.S.C. colour picture suffers from certain defects of a second
order, and it is the purpose of this chapter to point out how they
arise. It must be emphasized that these defects are not serious
enough to qualify as objections to the N.T.S.C. system as such, and
in fact many of them would occur in other types of system to a
more marked degree. Most of the defects in the N.T.S.C. system
can be removed by allowing additional complication at the receiver,
though modification of the transmitter only can be effective in some
cases. It should be noted that the term " defect " used in this
section refers to a departure, however slight, from a theoretically
perfect picture. The reader can be assured that, from the authors'
personal viewing of N.T.S.C. pictures, the colour picture which
suffers from all the defects mentioned below would nevertheless
be described as " very good ", and would have a colour quality
rather better than that obtained with good colour photographs.
The N.T.S.C. system discussed below is that which satisfies
the specification given in Chapter 4. That is, three primary signals,
R, G and B are each gamma corrected individually, then encoded
in terms of full bandwidth Y' and differentially bandwidth limited
I' and Q' signals having bandwidths of approximately one third
and one ninth, respectively, of the Y' signal.
Criticism of such a system can cover many features. For example,
only three primaries are used, with the result that some colours,
particularly in the blue-green region, cannot be reproduced with
full saturation. Again, the " detail colour blindness " quality of
human vision is relied upon to produce the subjective effect of full
definition in each primary colour. These defects are of a very small
order, of course, and are small compared with the fact that the
picture is necessarily quantized in a line structure.
Defects of a more noticeable order are concerned with constant
luminance failure and distortion of transitions between colours,
391
coupled with non -uniform noise performance. All these are a
result of the type of gamma correction used which, while producing
these defects, is nevertheless economical from the point of view of
the receiver design.
Again, bandsharing of the video spectrum by the luminance and
chrominance information, and single sideband transmission of the
upper part of the I' spectrum, both contribute to spurious effects
which are further complicated by the presence of noise.
Further defects arise from the use of asymmetric sideband trans-
mission of the main vision carrier, or single sideband distortion,
as it is often called. This particular effect is also a function of the
sign of modulation used, while the use of F.M. or A.M. of the sound
carrier can also produce different subjectively noticeable beat effects,
particularly with regard to the relatively low frequency beat between
the sound carrier and the R.F. chrominance sub -carrier. The
relative performances of a negative modulation, F.M. sound N.T.S.C.
signal and a positive modulation, A.M. sound N.T.S.C. signal are
briefly discussed below, but in the authors' opinion the latter is
unquestionably the better system as far as picture quality is con-
cerned.
Let us consider some of these defects in more detail.
16.2. Constant luminance failure
One consequence of individual gamma correction of the R, G
and B signals before encoding is that the luminance signal Y'
then does not carry all the luminance information. The balance
is therefore carried by the chrominance parameters I' and Q'
(or (R' - Y') and (B' -Y ') ) so that any noise which may be
present in the chrominance channel can affect the displayed
luminance. Hence a degradation of subjective noise performance
is produced.
Let us consider the case of a large area colour, so that the effects
of differential bandwidth limiting do not enter. Then the receiver
display will have the signals Y ' , (R' - Y'), (G' - Y'), and
(B' - Y') applied to it, and if the display has a y law then the
displayed colour will have the following tristimulus values:-
Red
( Y' R' - Y')7 = (R')'Y = R
Green
(Y' G' -Y ')v = (G')Y = G
Blue
(Y' B' -Y = (B')Y = B
SHORTCOMINGS OF N.T.S.C. SYSTEMS 393
Hence, the displayed colour will have the correct chromaticity
and luminance. However, let us investigate how much of this
luminance is carried by Y', and how much is carried by the
difference signals. That is, let us find the displayed luminance in
terms of the Y' and colour difference signals.
The displayed luminance is
0.30R + 0.59G + 0-11B = Y
the correct value. Hence
Y= 0.30[Y' + (R' -Y r 0.59[Y' + (G' -Y
--I- 0.11[Y' + (B' - Y')]Y
For simplicity, assume that y = 2 (strictly y = 2.2), so that
Y = 0.30[(r)2 + (R' - Y')2 + 2 Y '(R' - Y')]
+ 0.59[012 + (G' - Y')2 + 2 Y'(G' - Y')]
+0.11[(Y')2+ (B' - Y')2 + 2 r(B' - Y')]
Collecting terms
Y.= (Y')2 + 0.30(R' - Y')2 0.59(G' - Y')2 + 0.11(B' - Y')2
+ 2 Y10.30(R1 - Y') + 0.59(G' - Y') + 0.11(B' - Y')]
But
Y' = 0.30R' + 0.59G' + 0.11B' = (0.30 + 0.59 + 0.11) Y'
.'. 0.30(R' - Y') + 0.59(G' - Y') + 0.11(B' - Y') = 0
and hence
Y = (Y')2 + 0.30(R' - Y')2 + 0.59(G' - Y')2 + 0.11(B' - Y')2
y,)2 + y82
It can therefore be seen that, although the displayed luminance
has the correct value of Y, some of this luminance that is, Y82, is
carried by the difference signals. The fundamental reason for this
is that
1 1 1
Y' = 0.30RY + 0.59G7 + 0.11 BY

instead of
Y' = (0.30R + 0.59G + 0.11B)Y = YY
This point has been further discussed in Section 5.3. Notice
that there is one exception to this failure of constant luminance,
i.e. if the transmitted colour is Illuminant C, then the difference
signals are zero and Y = ( Y')Y. This arises because, if R = G = B
(which is so for Illuminant C), then
1 1 1 1 1
Y' = 0.3ORY + 0.59 GY + 0'11 = (0.30 + 0.59 + 0.11) RV = RY

and also
1 1
= (0.30R + 0.59G + 0.11B)Y RY
The fraction of the true luminance carried by the Y' signal is

(r)2
given by for a y of 2. This fraction is sometimes called the
constant luminance index, and its ideal value is unity.
Therefore
( Y')2 (0.30R1 + 0.59G1 + 0.11B1)2
Y 0.30R + 0.59G + 0.11B
and if numerator and denominator are divided by R G B,
then
(r)2 (0.30r1 + 0.59g1 + 0.1161)2
Y 0.30r + 0.59g + 0.11b
This equation shows how constant luminance failure varies with
(r)2
chromaticity; thus, for Illuminant C, r = g = b and - 1,
but for pure blue
( Y')2 (0.1161)2
= 0.11
Y 0.11b
so that for pure blue the Y' signal carries only 11 % of the true
luminance. On the other hand, for pure green the Y' signal
carries 59 % of the luminance.
It follows that the greater the purity of a colour, the greater is
the failure of constant luminance, and the nearer the dominant
wavelength is to the lowest luminance primary, that is blue, the
greater is the failure of constant luminance. Thus, constant
luminance failure is zero for Illuminant C, and a maximum for
pure blue. It would therefore appear that the amount of luminance
carried by the chrominance parameters is a function of both the
purity and the dominant wavelength of the colour.
16.3. The elliptical sub -carrier
The present N.T.S.C. specification leads to an elliptical relation
between sub -carrier amplitude per unit Y' and sub -carrier phase,
for a given constant luminance index and a gamma of two. This
may be shown from the equation
Y 0.30(R' - Y')2 + 0.59(G' - Y')2 + 0.11(B' - Y')2

-1+
(Y')2 (Y')2
which was derived in the previous section, by substituting
0.59(G' - Y') = - 0.30(R' - Y') - 0.11(B' - Y')
(R' - Y') = 1.14S sin 0
and
(B' - Y') = 2.03S cos 9
where S and 9 are the sub -carrier amplitude and phase.
Hence
S2 0-30
( -1 --1- (y,2(0 30 x 1.142(1 + 0-.59) sin2 0
)2
0.11
+ 011 x 2'032( + 059)C°S2 0
+2 x 0.3 x 0.11 x 1.14 x 2.03
sin .9 cos 0)
0.59
S2
=-- 1 ± ( y,)2 (0.59 sin2 9 ± 0.54 cos2 9 + 2 x 0.13 sin 0 cos 0)
1
.*. 0.59 sin2 0 + 0.54 cos2 0 + 0.13 x 2 sin 0 cos 0 -
( y) 2
where K - is the constant luminance index.

Y
Now the polar equation to an ellipse of semi -axes a and b, the
former of which is rotated through an angle II. relative to 0 = 0, is
given by
1 12 1
cos2(0 - µ) + - si n2(0 - µ)) = -i
r
a b
or else it may be given by
(.1 sin2 -1/1-2 cos2 ,u) sin2 0 + cost -1g2. sin2 µ)cos2 0
a2
+
1
a
- -) sin IL cos µ x 2 sin 0 cos
1
b2
-1
r2
S
Y'
Hence, the equation between and 0 is an ellipse whose
/- I 1
constants tk, a and b, can be determined from the equations

1 1
sin2 172- cos2 µ = 0.59
1 1
cos' µ + sin2 µ = 0.54
and
- µ, cos p. = 0.13
Hence
= 50.50, a = 1.2 and b = 1-5
The equation between 0 and (y,) may therefore be written as
1 1 1
c°82(6 + 1-52 sin2(64 /4) [ 1 S 2
Y'
NK-1
or
1
i S \2 (1 -K K) {1.1
2222
cos2(0 - 1.52 sin2(0 -
kyl
(2,"-1
sin2(0-µ)
151K)
1K) 2cos.(0_0+ (.4/ 2
For each value of the constant luminance index therefore, the
sub -carrier amplitude per unit Y' describes an ellipse as 0 varies,
and the smaller the value of K, the larger are the semi -axes of the
ellipse, as Fig. 16.1 shows. This illustrates that constant luminance
failure becomes more severe as the chromaticity approaches the
perimeter of the primary triangle. Note that the largest ellipse
(R' -Y')
1.14Y'
2.03 Y'
Fig. 16.1. Constant luminance

index loci in the chrominance
plane. y = 2
which can be drawn to include a physical chromaticity passes through

the blue primary, showing that constant luminance failure is most
severe for the pure blue chromaticity.
16.4. The circular sub -carrier
If the sub -carrier amplitude per unit Y' were circularly related
to the sub -carrier phase for a given value of the constant luminance
index, then the latter would be independent of the sub -carrier phase.
That is, all the luminance carried by the sub -carrier would be a
function only of the sub -carrier amplitude per unit Y'. In this
case, it would be possible for the transmitter to derive a correction
signal from the sub -carrier amplitude, and add this to the Y' signal.
The modified Y' signal would then carry all the luminance, so
constant luminance operation would be achieved, and in addition,
monochrome receivers would operate panchromatically, provided
they were fitted with a sub -carrier rejector. Colour receivers would
require some additional circuitry for deriving the conventional
Y' signal from the corrected signal, of course, otherwise they would
display too much luminance. Remember that, as far as a colour
receiver is concerned, correct large area luminance is achieved
when the conventional Y' signal is transmitted.
A circular sub -carrier can be achieved by a modification of the
ting(R
sub -carrier make-up at the transmitter. Suppose instead of modula-
with
one sub -carrier with
(B2
Y'
and the quadrature sub -carrier
1.14)
we modulate one sub -carrier with
. 03 '
[(R' - Y') A(B' -Y ')]C
and the quadrature sub -carrier with (B' - Y')D, where A, C, and
D are suitable constants.
Then, if S and B are the sub -carrier amplitude and phase
[(R' - Y') A(B' -Y ')]C = S sin
and
(B' - ")D = S cos 0
Now if y = 2, the luminance contribution of the sub -carrier is
Y.2 = 0.30(R' - Y')2 0.59(G' - Y')2 0.11(Le - Y')2
0.30 0'11
= (0.30 + 0-.592)(R' - Y')2 + (0.11 + 0-.592)(B' - Y')2
0.066
-CTT-(IV Y')(B'
But
(B' - Y ') -S cos

D
and
(R' - Y') - S sin 0 - CA(B'

C
- I")
CAS
S sin 0 - T)-- cos 0
C
)
=--
sin
- -D cos
A
0
Substituting in the equation for Y82, it follows that
5,2in
si 0
Y82 = 0.30(1 m 0.59) k C - ----
D
cos 0\ 2
j
0.11)S2
+0.11 (1 + 0.59D2 COS2 0
0.066 S2 sin 9 A
+ 059
x
D C
-D- cos ()cos 0
0.59 Y82 1
- D2(0 267A2 + 0.077 - 0.066A) cost 0
52
0.267 1
+ C2
sine B + -DC(0.066 - 0.534A) sin 0 cos 0
If this is to be independent of 0, then

(0.267A2 + 0.077 - 0.0661) 0.267
D2 C2
and
0.066 - 0.534A = 0
in which case
0.59 Y82 0.267
S2 C2
which is independent of 0.
Solving the two equations, A = 0.124 and C = 1.91D
The sub -carrier make-up for circularity is therefore
1.91 DKR' - Y') 0.124(B' - Y')]
modulated on to one sub -carrier, instead ofthe conventional

(R'- Y'),
1-14
and D(B' - Y') modulated on to the quadrature sub -carrier, in-
)
stead of the conventional (B2.03
The circular sub -carrier amplitude is given by
DA/1.912[(R' - Y') 0.124(B' - Y')]2 (B' - Y')2

where the constant D may be chosen to limit the sub -carrier excur-
sions in the video range. The greatest excursion must occur for one
of the full purity complementaries, and evaluating for these three
colours, we have the results shown in Table 16.1.
Table 16.1
CHOICE OF CIRCULAR SUB -CARRIER WEIGHTING FACTOR
Colour Y' DA/1 -91'[(R'- Y')+ -124(B'- Y')12±(B'- Sum

Cyan 0.70 1.3D 0.70+1-3D
Magenta 0.41 1 -4D 0-41+1-4D
Yellow 0.89 0.89D 0.89+0-89D
If the greatest sub -carrier excursions are limited to 1 times the

0 to 1 video range, as in the conventional case, the possible values
of D are given by the equations
0.70 + 1.3D = 1.33, or D = 0.485
0.41 + 1.4D = 1.33, or D = 0.655
0.89 0.89D = 1.33, or D = 0.495
The smallest value of D must be chosen, i.e. 0.485, so that the
maximum excursion of 1.33 is not exceeded for any colour. (For
example, if D = 0.655, then 0.70 + 1.3D = 1.55, which exceeds
Table 16.2
MAXIMUM SIGNAL EXCURSIONS FOR SATURATED COLOURS USING CIRCULAR
SUB -CARRIER SIGNAL
Colour Y' Y' S Y' -S S

Red 0.30 0.93 (0.93) -0.33 -0.33) 0.63 (0.63)
Green 0.59 1-27 (1.18) -0.09 0.00) 0.68 (0.59)
Blue 0.11 0.54 (0.54) -0.32 -0.33) 0.43 (0.44)
Cyan 0.70 1.33 (1.33) 0.07 (0.07) 0.63 (0.63)
Magenta 0.41 1.09 (1.00) -0-27 (-0.18) 0.68 (0.59)
Yellow 0.89 1.32 (1.33) 0.46 (0.45) 0.43 (0.44)
1-33.) The excursions for all six full purity colours can now be
calculated, assuming D = 0.485 (see Table 16.2).
The conventional N.T.S.C. figures are given in brackets, and it
can be seen that apart from blue and yellow, the sub -carrier ampli-
tude is greater than or equal to the conventional values. Hence,
slightly improved signal-to-noise ratios should be obtained except
for blue and yellow, but it can be seen that the differences are quite
small.
For decoding the circular sub -carrier at the receiver, the syn-
chronous detection axes must be altered slightly compared with
those required for the conventional elliptical sub -carrier.
The transmission axes are shown in Fig. 16.2. The output of a
synchronous detector operating along the 0 axis shown will give
an output
0.485[1.91(R' - Y') 0.124 x 1.91(B' - Y')] cos 0
- 0.485(B' - Y') sin 0
= 0.485[1-91 cos 96(R' - Y') (0.124 x 1.91 cos 0
- sin 0)(B' - Y')]
If this is to be a pure (R' - Y') signal, then
sin 0 = 0- 124 x 1.91 cos 0
whence
= 13.3°
The output will then be
0.485 x 1.91 cos our - Y') = 0.9(R' - Y')
Hence, if two synchronous detectors are used to obtain the red and
blue difference signals, they must operate with a phase difference of
0485 1.91 (R -Y) +0424 x1.91 (E1'-`1)]
0.9(RLY')
Fig. 16.2. Transmission axes

for a circular sub -carrier
03=13.3.
0 485 (8' 10
103.3° instead of the 90° separation required for a conventional

elliptical sub -carrier.
16.5. Effects of non-linear encoding of chrominance parameters on
displayed chromaticity
The conventional gamma correction leads to constant luminance
failure, and hence a subjective degradation of performance results
when noise is present. This effect is brought about by the luminance
carrying property of the chrominance signals, and it would seem
that these signals tend to change luminance rather than chrominance
whenever severe constant luminance failure is occurring. This
immediately suggests that the extent of chrominance control by the
chrominance parameters will vary so that for chromaticities where
constant luminance failure is severe, the control is weak, and the
chrominance signals will not have much effect on the chrominance
of the display. Since there is no large area colour distortion, the
control of the chrominance signals over the displayed chrominance
must be correspondingly greater for chromaticities where there is
only mild constant luminance failure. Thus for low purity colours,
a given change in a chrominance signal will produce a relatively
large change in displayed chrominance but only a small change in
displayed luminance.
For high purity colours a given change in a chrominance signal
will produce a relatively small change in displayed chrominance
but a large change in displayed luminance.
The effects of the luminance controlled by the chrominance
signals have been considered, so now it remains to examine the dis-
played chrominance controlled by these signals; and more par-
ticularly, the relationship between the sub -carrier amplitude and the
displayed purity, and between the sub -carrier phase and the dis-
played dominant wavelength will be investigated, so that the effects
of noise fluctuations on the received chromaticity can be estimated.
16.6. Sub -carrier amplitude control of displayed purity

To examine the control of sub -carrier amplitude over the purity
of the displayed colour, consider a colour having a total tristimulus
value D and a chromaticity
r =a +p, g =a, b =a
Since r + g + b = 1, it follows that
3a + p = 1
so that
a = (1 3-p)
The chromaticity is therefore
(1 + 2p) 1-( P)
r=a+p- , b -a - 3
Referring to Fig. 16.3, this chromaticity has a dominant wave-
length equal to that of the red primary, and asp is varied the purity
will vary but the dominant wavelength will remain constant. If
p = 0, the purity is zero and the chromaticity is Illuminant C.
If p = 1 the purity is 100 % and the chromaticity is pure red. Now
let us find the relationship between the sub -carrier amplitude S and
the quantity p, and determine how variation of one will affect the
Fig. 16.3. Variation of p changes purity,

while dominant A remains constant at red
0 10-0
3
0
other. Notice that the dominant wavelength will remain constant

at red so that the sub -carrier phase does not enter as a variable.
(This is explained in the next section.)
It has been shown earlier that the sub -carrier amplitude is given by
S = Va2(R ' - Y')2 + /32(B' - Y')2

where
1 1
=-.
1.14' ' 2.03
Since r = D-R' or R = rD, where D = R + G + B (see Chapter 2) it
follows by substitution that

S2 1 1 i i2
= oc2kry - 0.30r7 - 0.59gY - 0.11bY)
DY
1 1 2
132(bY - 0.30rY - 0.59gY - 0.11bY)
Substituting the values of r, g and b in terms of p we have
2
375'2 r
a21.0.70(1 2p)Y - 0.59(1 - - 0.11(1 - p);} 2
DY
1 1 112
+ /32[0'89(1 - p)Y - 0.30(1 2p)Y - 0.59(1 - p)YJ
1 1 1
0.702a2[(1 2p)Y - (1 - p)Y12 + 0.30132[(1 + 2p)7
i.e.
1 2
31S r, 1,2
= Lt1 21117 - (1 - P)Y1 (0.702m2 ± 0.302/32)
DY
or
D1 p)Y]
31
where
K = V0.702a2 0.302/32 = 0.63
Now find how S changes when a change is made in p, assuming
that D remains constant all the time.
Differentiating
1
dS KID '-1
dp y 3)71-
1.2(1 2p)7 1 (1 p);
If p =0
dS
dp P =0
- DY3 7
Y
If p -> 1
dS
-> 00
dp 1
(since
(1 - p); 1 - 1
- oc. if y > 1 and p 1)
(1-P)
Hence, for small values of p, S increases slowly as p increases.
For large values of p, S increases rapidly as p increases.
A plot of S against p is shown in Fig. 16.4, and it can be seen
that equal increments in p, produced by changing the purity of the
transmitted colour, produce small changes in S for low values of p,
and large changes in S for high values of p. At the receiver, a
change in the received value of S has a large effect on p for low
802
01 0-2 03 0-4 05 0-6 0.7 08 09 10
Fig. 16.4. Relation between sub -carrier amplitude and p for cases where
D = 1, y = 2; and D = 1, y = 1
values of S, and a small effect on p for high values of S. Conse-

quently, noise will change p more markedly at low values of p,
and less markedly at high values of p.
dS
In the linear case, y =1 and S = KDp and -dp = K D.
Hence, noise fluctuations will produce equal effects on p, indepen-
dent of p. That is, noise performance will be uniform for y = 1.
16.7. Sub -carrier phase control of displayed dominant wavelength

Let us now examine the relationship between the phase of the
sub -carrier and the dominant wavelength of a colour, and hence
determine the relative control of one over the other.
27
If 0 is the phase of the sub -carrier relative to the blue difference
axis, then
tan 0 _
a(R' - Y')
P(B' -Y
where «= and
1.14 2.03
Substituting 0.30R' 0.59G' -F 0.11B' for Y', it follows that

1 1
a (0.7ORY - 0.59GY - 0.11BY
tan 0 - 1 1 1
0.89B7 - 0.59GY - 0.30RY
Dividing numerator and denominator by (R + G

1
- tan 0 - 0.70r' - 0.59g1 - 0.11b7

1 1 1
0.89bY - 0.59g7 - 0.30rY

1 1 1
0.70(rY - gY) - 0.11(bY - gY)
1 1 1 1
0.89(bY - gY) - - gY)

Hence, for a given value of 0, r can be plotted against b, remember-
ing that r + g + b = 1. By repeating these plots for various
values of 0, a family of curves is obtained, as shown in Fig. 16.5.
Notice that these curves converge towards the primaries and diverge
away from the complementaries, so that a change in the dominant
wavelength of a chromaticity produces a relatively large change
in the sub -carrier phase angle in the region of a high purity primary
dominant wavelength and a relatively small change in phase angle
in the region of a high purity complementary. In terms of receiver
noise performance, therefore, the displayed dominant wavelength
of high purity complementaries will be more affected by phase
fluctuations than will the dominant wavelength of high purity
primaries.
For low purity chromaticities, the relationship between dominant
wavelength and sub -carrier phase angle is similar to the linear
or y = 1 case.
For consider a chromaticity r = + 81, g = -} + 82, b = 83,
where the 8 quantities are small compared with 1.
1 1 1 i 1
38
Then rY = a + 81)v = (})Y(1 + 380Y = (})7(1 + -;), by the Bino-
mial expansion.
38
Similarly, 8-7 = (k)7(yy 1 _)
382
Y
and by = (-1)Y(1
Substituting in the expansion for tan 0 given earlier

«
0.70 ( 3-b -1-- 3`) -0.11

Y
(1+ 3-18 -1- "2)
/
-ce tan 0 - YI+
0.89 (1± 383 -1- 382) -0.30 (1+ 381 -1- 382)
Y
0.70(81 - 82) - 0.11(8, - 82)

0.89(83 - 82) - 0.30(81 - 82)
0.70(r - g) -0.11(b - g)
0.89(6 - g) - 0.30(r - g)
which is the same as the linear case.
Notice that in the linear case, a linear relationship exists between
tan 0 and the chromaticity co-ordinates, so that a particular value
of 0 will completely specify the dominant wavelength. In the case
where y r 1, however, this is true only for the three particular
cases r = g, r = b, and g = b. These three equations give straight
line loci on the chromaticity diagram which pass through the white
point and the appropriate primary and complementary points.
The dominant wavelengths of all chromaticities lying on these loci
are therefore completely specified by the sub -carrier phase angle,
but for all other chromaticities the phase angle is not a sufficient
specification of the dominant wavelength, and another parameter,
such as the sub -carrier amplitude for a given luminance, is required
as well. This leads to the conclusion that, unless the displayed
dominant wavelength is that of a primary or complementary, the
displayed dominant wavelength will be a function of both the
saturation and hue control settings of the receiver.
It should be pointed out that Fig. 16.5 shows the connection
between chromaticity and sub -carrier phase angle, and if the sub-
jective effects of phase variations are required, the non-linear
property of the conventional chromaticity diagram in terms of
equally perceptible chromaticity differences must be taken into
account, as described in Section 2.23.
16.8. Effects of gamma correction on transient response
The preceding sections have been concerned with the effects of
gamma correction on the display of large area colours for which
the limited bandwidth of the chrominance signals is not involved.
However, for relatively small area colours, or for transitions
between different colours, it is to be expected that certain distur-
bances will occur as a result of constant luminance failure.
Basically, distortion of transients arises because the luminance
is carried partly by the high definition Y' signal and partly by the
relatively low definition chrominance signals. Hence, if there is
any sudden change from one colour to another which also involves
a change in the luminance contribution of the chrominance signals,
the resulting displayed colour change will have a fast luminance
transient controlled by the Y' signal superimposed upon a relatively
slow luminance transient controlled by the bandwidth limited
chrominance signals. In addition, the instantaneous chromaticity
103-5° RED
I.0
PHASES MEASURED RELATIVE TO (-6Y1 AXIS
120° (RLY9
I. 4
08
ao°
140°
0.6- 203
160°
1675° 60.7° MAGENTA

YELLOW
180°
O4-
200°
02-
220° 20°
o°
0
347 5° BLUE
02 0.4 08 I.0
240.7° 2600 2835° no° 320° 340°
GREEN CYAN
Fig. 16.5. Chromaticity loci for constant sub -carrier phases, y = 2

of the originating transient will not be faithfully reproduced, but
since the human eye is more sensitive to luminance changes these
will provide the greater subjective effects.
The exact form of transient distortion will depend on the actual
colours before and after the transient, and also during the transient.
In Section 16.2, it was shown that, for the case y = 2, the displayed
luminance is given by
y = )12 + y82
Hence
dY = 2 Y'dY' 2Y8 dY,
so that for a luminance change d Y in the original picture, this will
be correctly displayed if Y' and Y8 are present at the display tube.
But if d Y is a fast transient, only Y' will actually reach the receiver
as a full bandwidth signal, and the ratio
luminance displayed d( Y')2 Y'd Y'
luminance in original picture - d Y Y'd Y' Y8d Y8
gives a measure of the distortion. The idai value of this ratio is

unity, in which case there would be no distortion, but this can
occur only for the cases IT, = 0 or d Y8 = 0. Thus, there is no
transient distortion when the transient is between shades of grey,
i.e., when Y8 = 0, or when it is between colours which always lie
on a contour of constant Y8 throughout the transition, i.e., when
d Y8 = 0.
In all other cases some distortion will occur, and this will be a
function of the initial and final colours and also the intermediate
colours, of the originating transient.
In order to illustrate the mechanism of luminance transient
distortion, consider three separate examples. In each case, assume
that the colour transient being televised is accomplished linearly
with time, and that the Y' bandwidth is sufficient to accommodate
the Y transient.
Let us assume that the restricted bandwidth of the sub -carrier
signals increases the duration of the transient threefold, and in
addition let us assume that there is no differential bandwidth
restriction of the chrominance signals.
EXAMPLE 1
Saturated full blue to a grey of Y = 0.04. The relevant wave-
forms are shown in Fig. 16.6. Since the original transient is taken
0I
R
0.11
0
1.0
H COLOUR
TRANSITION
BEFORE
CAMERA
0,
of
)2
YI
DISPLAYED
LUMINANCE
0
OF Y SIGNAL
201
Ys
FULL
BANDWIDTH
0
0.1
s2 DISPLAYED
LUMINANCE
RESTRICTED
OF CHROMINANCE
01 SIGNALS
CY 1)2+ yd TOTAL
DISPLAYED
0 LUMINANCE
3T 111
Fig. 16.6. Example 1. Blue to grey transition

as a linear function of time, the tristimulus values at any instant are
given by
= 0-04-T ' Gi = 0.04-

and
= 0.96 (1 --T 0.04
where t is measured from the start of the transient, which is com-

pleted after a time T. Hence the instantaneous luminance is
0.30Ri 0.11Bi
and similarly
Y't = 0.59Gil
Then
y812 = (y/02
Notice that the luminance carried by Y' increases during the

transient, while Y decreases. Also the initial and final thirds of
the total displayed luminance period are entirely dependent on the
Y82 response is concerned.
EXAMPLE 2
Saturated full blue to a grey of Y = 011. In this case, illus-
trated in Fig. 16.7, the instantaneous tristimulus values are
Ri =0.11T, =0.11- and Bi = 0.89(1 --T 0.11
Although there is no change in Y throughout the transient,

there is a considerable disturbance in ( Y')2 and Y82, and since
the latter is bandwidth limited, these two disturbances are not
complementary in shape so that an entirely spurious luminance
transient is displayed.
EXAMPLE 3
Pure red of R = 0.403 to a greenish blue R = 0, G = 0.018
and B = 1. The instantaneous values are
Ri = 0.403 (1 - yt , G =10 0 1 8T and = -
and the waveforms are shown in Fig. 16.8.

01
0 COLOUR
TRANSITION
BEFORE
CAMERA
DISPLAYED LUMINANCE
OF 1' SIGNAL
(Y92
2 0.1
VS
FULL
BANDWIDTH
01
2 0.1 -
Ys
RESTRICTED
DISPLAYED
LUMINANCE
BANDWIDTH
OF CHROMINANCE
0 SIGNALS
TOTAL
DISPLAYED
LUMINANCE
-a1
3T
Fig. 16.7. Example 2. Blue to grey transition at constant luminance

1,0
0.1
0 COLOUR
TRANSITION
BEFORE
10 CAMERA
0I -
0.1
DISPLAYED
(vi)22 LUMINANCE
OF YISIGNAL
0
0I
rs2
BANDWIDTHlL
0
DISPLAYED
01 LUMINANCE
OF CHROMINANCE
Y52
SIGNAL
RESTRICTED
BANDWIDTH
0
TOTAL
01 DISPLAYLD
LUMINANCE
(Y1)2* y52
3T
Fig. 16.8. Example 3. Transition between colours

having equal Y, (Y')2 and Yee values
In this case, in spite of the fact that Y, ( Y')2 and Y82 each have
the same initial and final values, and although Y is constant through-
out the transient, nevertheless an entirely spurious luminance
transient is displayed. Notice that a drop in luminance occurs
on each side of the centre of the transient owing to the falling
luminance contribution of the slow Y8 signal. Notice also that
the displayed luminance is correct at the centre of the transient
because it is here that the values of the restricted bandwidth and
full bandwidth Y82 signals are equal.
It should be pointed out that in all the above examples the sharp
transitions of the displayed luminance will be rounded off as a
(r? -0
1 14
Fig. 16.9. Contours of con-

stant Ys' in chrominance
plane. Equal time intervals
are indicated on the I' and
Q' axes 2.03
consequence of the finite luminance channel bandwidth, but this

practical point has been omitted from the figures for the sake of
simplicity.
It is interesting to examine the above type of transient in more
general terms. In Section 16.2 it was shown that
Y82 = 0.30(R' - Y')2 0.59(G' - Y')2 + 0.11(B' - Y')2

')-
For constant values of Y82 may be plotted against
(R'1.14
(B
2.
03Y) in the chrominance plane. The contours of constant
1782 are ellipses, as shown in Fig. 16.9. (Note that these ellipses
differ from the loci of Fig. 16.1 in that the latter are plotted for
(Y')2
constant values of and since ( Y')2 = Y - Y82, it follows
,
Y
that the ellipses of Fig. 16.1 correspond to constant values of the
ratio Y82 to Y.)
The larger ellipses of Fig. 16.9 correspond to the larger values of
Y82, and it will be seen that a transient between two colours such
as A and E, each of which has the same value of Y82, will have a
straight line path AE if the red and blue difference axes have equal
bandwidths. For a wideband I', narrow band Q' receiver, the
transient will follow the path ABCDE. In either case, the sub -
carrier luminance contribution at the centre C of the transient is
smaller than at A or E, so that a sub -carrier luminance transient
occurs although the transient is between colours of equal sub -
carrier luminance. This luminance transient is broadened by the
bandwidth limitation of the chrominance signals so that a darken-
ing occurs at the transient which is not cancelled by the corre-
sponding opposite polarity luminance contribution of the higher
definition Y' signal. The dip in luminance contribution of the
sub -carrier during the transient is often called a luminance notch.
Notice that the transient passes near the white point so that
desaturation occurs at the centre of the transient. If the transient
could be made to follow the elliptical path between A and E, the
luminance notch would be removed completely and the saturation
would be higher. This corresponds to the case d Y8 = 0.
16.9. The I' Q' fallacy
Another consequence of constant luminance failure which is of
interest is concerned with the underlying principles of wideband
I', narrow band Q' operation. In a conventional I'Q' system, cross-
talk between I' and Q' or vice versa does not occur over the double
sideband frequency spectrum common to I' and Q'. Within the
single sideband region of the high frequency I' components, Q'
is absent because it has been bandwidth limited at the transmitter,
and therefore Q' cannot crosstalk to I'. However I' can crosstalk
to Q' over this region, but such crosstalk can be eliminated by
including a Q' pass filter at the receiver.
Now the need for a filter to remove I' crosstalk from the Q'
channel may be questioned; since high frequency components of Q'
produce colours on the Q' axis, these colours should not be visible
owing to the effective tritanopia of the central fovea of the human
eye. Put another way, since the high frequency Q' components
should not be visible, neither should the crosstalk of I' to Q' be
visible. The fact that a Q' filter is required in practice suggests
that the spurious luminance components of I' to Q' crosstalk are
subjectively more noticeable than the spurious colour components
of crosstalk, in which case it follows that Q' bandwidth limitation at
the transmitter removes luminance information which would be
visible, although the corresponding chrominance components
would not be visible.
It will be shown that the rejection of high frequency Q' com-
ponents by the transmitter produces a " blind " luminance axis so
that for colours lying on this axis the single sideband or high fre-
quency components of I' do not affect luminance, and since the
high frequency Q' components are absent, the axis is luminance
sensitive only to the low frequency Q' components.
Referring to the equation for the displayed luminance in terms
of Y' and the chrominance signals
Y = ( Y')2 0.30(R' - Y')2 0.59(G' - Y')2 011(B' - Y')2
if y 2
Substituting
(R' - Y') = 0.961' 0.63Q'
(G' - Y') = - 0.281' - 0.64Q'
(B' - Y')= -1.111' ± 1.72Q'
it follows that
Y = (Y')2 + 0.46(11)2 -I- 0.69(Q)2 0.15/'Q'
If Y' and Q' are held constant, the control of Y by I' is given by
dY
0.92/' 0.15Q'
d/'
Similarly, if Y' and I' are held constant, the control of Y by Q'
is given by
dY
dQ'
1.38Q' 0.15/'
dY dY
The axesd/'-= 0 and d-Q'
= 0 are plotted on the chrominance
plane diagram in Fig. 16.10. The equations of these axes are
0.15, 1.38
0.92'' and = - 0.15Q' respectively.
Fig. 16.10. Luminance control by I' and Q'
dY
For colours lying on the axis - 0, the I' signal has no control
'
over luminance at any frequency. Such control would normally
come from the Q' signal and since Q' is bandwidth limited, there
is no luminance control by any high frequency chrominance
component along this axis. Thus, luminance transient response
dY
is limited for colours on the -= 0 axis, and for colours near this
d/'
axis the single sideband components of I' will have little effect on
luminance.
16.10. Effects due to multiplexing luminance and chrominance signals
An essential feature of the N.T.S.C. system is the sharing of part
of the video band by the luminance and by the sub -carrier and its
sidebands. While a colour television system using out of band
chrominance could be envisaged, this would necessitate an appreci-
ably wider bandwidth than that required for the corresponding
monochrome system and one of the great advantages of N.T.S.C.
operation, i.e., the ability to add colour to an existing monochrome
system, would be lost. Additionally, the provision of increased
R.F. and I.F. bandwidth in colour receivers, and the possibility
of frequency dependent propagation effects, may present economic
difficulties. However, multiplexing the luminance and chromin-
ance signals produces certain spurious effects, as follows :-
16.10.1. CROSS -COLOUR

The luminance channel accepts the sub -carrier and its sidebands
as a luminance signal, giving a dot structure interference which itself
produces undesirable luminance effects. Similarly, the chrominance
channel accepts those components of luminance which lie in the
region of the sub -carrier, and interprets them as chrominance infor-
mation. This effect is usually called cross -colour.
The N.T.S.C. system is designed so that the above spurious
effects have reduced visibility in a subjective sense. Thus, the
sub -carrier signal is zero for white, and is small for the statistically
frequent pastel colours. Also, locking the sub -carrier frequency to
an odd multiple of half the line scan frequency produces an
integrating effect which reduces dot visibility.
Both these properties
of N.T.S.C. systems reduce the spurious effects of chrominance to
luminance crosstalk. Again, spurious colour effects produced by
luminance to chrominance crosstalk tend to integrate out because
of the frequency relationship between the sub -carrier and scanning
frequencies, so that a spurious colour on one picture is added to its
complementary on the succeeding picture.
It is interesting to note that, for a monochrome transmission
(i.e., one in which the colour burst is absent), the receiver reference
frequency generator is necessarily unlocked so that the optimum
conditions for reducing spurious colour effects do not apply. In
other words, cross -colour effects are more serious when the burst
is absent, and these spurii are emphasized subjectively because the
picture is nominally monochrome. Furthermore, if the receiver is
fitted with A.C.C., the chrominance gain will be set to maximum.
The need for some colour killing mechanism which switches off the
receiver chrominance channel during " burstless " reception is
therefore obvious.
16.10.2 PARC NOISE
In the same way that luminance components in the frequency

region of the sub -carrier are interpreted as chrominance information
by the receiver, giving rise to cross -colour, so any noise components
in the appropriate frequency band of the luminance signal are
similarly displayed as " coloured noise " or parc noise as it is some-
times called. The noise in the luminance signal which produces
parc noise can originate either in the picture source at the trans-
mitter or in the transmission path to the receiver, or both. Notice
that cross -colour and parc noise, while originating in the higher
frequency video band (somewhere near 2.7 Mc/s in the British
N.T.S.C. system) are nevertheless displayed as low frequency
chrominance information because of the heterodyning action of the
reference frequency generator. This tends to increase their
visibility.
16.11. Transmitter notch filter

Cross -colour and parc noise can be reduced by including a
rejection filter in the transmitter luminance channel which attenuates
luminance components in the region of the sub -carrier frequency.
The exact characteristics of such a filter are best determined by
experiment, as too much attenuation will spoil luminance defini-
tion while too little attenuation will be ineffective. An attenua-
tion of 6dB at sub -carrier frequency, with a 3dB bandwidth of
±200 kc/s, is likely to be near optimum. This type of luminance
notch filter must not be confused with the luminance notch filter
in a colour receiver. The latter, it will be remembered, is included
to remove the large area dot structure of chrominance to luminance
crosstalk.
16.12. Effects of gamma characteristic on cross -colour
Although cross -colour visibility is reduced by locking the sub -
carrier frequency to an odd multiple of half the line scan frequency,
in point of fact the display tube has an approximately square law
relation between light output and voltage input, and this produces
relatively high visibility terms which would be absent for a linear
display. The generation of these higher visibility components is
probably best illustrated by an example.
Suppose a " monochrome " sine wave of amplitude M, and
oscillating about a mean luminance level M, is transmitted. Then
the luminance signal is
Y' = M(1 + sin wit)
where -2.n. is the frequency. Suppose that a burst is transmitted,

so that the receiver chrominance channel is operative, and the
receiver reference frequency generator is correctly locked. Let us
further suppose that no chrominance information is transmitted.
If -w2 is the sub -carrier frequency and if falls in the acceptance
27r 27r
-
band of the chrominance channel, the colour receiver will derive
the chrominance difference signals as follows:
(R' - Y') = 1.14M sin(w, - w2)t
(B' - Y') = 2.03M cos(w, - (02)t
(G' - Y') = -M[0.58 sin(w, - (.02)t --I- 0.38 cos(wi - W2)1]
The (R' - Y') output is obtained from the (R' - Y') synchronous
detector by heterodyning the incoming Msin wit signal with a reference
signal A sin I wet -2), while the (B' - Y') output results from hetero-
dyning M sin wit with a reference A sin wet. The higher frequency
terms are eliminated by the receiver chrominance filters, and the
gains 1.14 and 2.03 are applied by the receiver to compensate for
the reduced chrominance signals in a colour transmission.
The luminance signal is
Y' = M(1 + sin wit)
but if a luminance notch filter is included in the receiver (as it must
wi
be), then if this has an attenuation of -m to 1 at the frequency -
27r'
the luminance signal actually reaching the display will be M(1 +
m sin wit)
For a square law display, the displayed colour will be
R = [M(1 + in sin wit) + 1.14M sin(wi - w2)/2

G = [M(1 m sin wit) - M(0.58 sin(wi - w2)t
+ 0.38 cos(coi - (02)0l2
B = [M(1 + in sin wit) + 2.03M cos(wi - ai2)tP
In these expressions, in addition to the terms M2(1 + in sin (011)2

which would be obtained if the chrominance channel were inopera-
tive, terms having pulsatances of (w, - w2), 2(w, - (2w, - w2)
and w2 are obtained.
If we assume that wi is a multiple N of the line scan frequency
L, so that the signal M(1 m sin wit) appears as vertical stripes,
and if we take w2 = (2P ± 1)L/2, where P is an integer, in accord-
ance with the N.T.S.C. specification, then
w, - w2 =- NL - (2P + 1)f, = (2N - 2P - = Odd x

2(w, - (02) --- (2N - 2P - 1)L = Odd X L
2wi - w2 = 2NL - (2P + 1)f = (4N - 2P - 1) = Odd x
Hence terms involving (w1 - w2), (2w1 - w2) and w2 have low
visibility, while those involving 2(w - w2) have relatively high
visibility.
(Note that if the display were linear, the chrominance terms would
all be functions of (w, - WO, which has low visibility.)
If we assume that the low visibility terms are, in fact, not visible,
then the displayed output will be
R = M2[(1 m sin (0,02 + 0.65(1 - cos 2(w - w2)t)]
G = M2[(1 m sin wit)2 + 0.24 - 0.10 cos 2(w, - w2)t
± 0.22 sin 2(aq - w2)t]
B = M2[(1 + m sin wit)2 + 2.06(1 + cos 2(w, - w2)t)]
Notice that the factor m reduces some of the low visibility terms,
but does not affect the high visibility 2(w, - w2) terms.
Since
m2
(1 M sin wit)2 = 1 + 2m sin wit + -y (1 - cos twit)
m2
and since w1 is well above I cut off, we may write 1 + for
(1 -I- m sin w11)2 if we are considering chromaticity rather than

luminance.
The chromaticity co-ordinates of the displayed colour are there-
fore
r = 0[1.65 + - 0.65 cos 2(w, - (00d

m2
g =x[1.24 + -2- - 0.10 cos 2(w, - w2)t + 0.22 sin 2(w, - (00d
28
-111.
b = -1 [3.06
'
-
n2 2
2.06 cos Awl - w2)t]2
where
3
D= 5.95 2m2 + 0.22 sin 2(wi - w2)t + 1.31 cos 2(wi - w2)t
It is interesting to evaluate the chromaticity co-ordinates over a

complete cycle of 2(wi w2)t for the extreme case of m = 0.
The result is shown in Fig. 16.11.
Since m = 0, the transmitted A.C. component of luminance,
m sin wit, is not reproduced, and apart from the D.C. component
Fig. 16.11. Locus of chromaticities of

\ high visibility 2(wi - w2)t cross -
0.6 \ colour beat
ILLUMINANT C S.
0 0I 02 03 0-4 0.5 06 0-7 0.8 0.9 I0
M2, the display is entirely composed of spurious terms. Thus,

vertical coloured bars having a frequency of (w - (00 are obtained,
IT
and the chromaticities vary between an orange of r = 0.81 and
g = 0.19 with a purity of 34 %, and a near blue of b = 0.98 and
g = 0.02 with a purity of 57 %, as shown in Fig. 16.11.
It will be appreciated that if a genuine chrominance signal is
present in addition to luminance components of the type M sin wit,
then further spurious terms will be generated. However, the sub-
jective effect of such spurii is likely to be reduced by the presence
of the genuine colour information in the picture.
16.13. Spurious colour in white areas
If a white area is present in an otherwise coloured picture, any
noise accepted by the chrominance channel will produce coloured
flecks in the white area. For reasonable signal conditions, e.g.
greater than 100/LV in Band I, such " coloured " noise is sub-
jectively acceptable, but for fringe area reception it may reach
Fig. 16.12. Sideband and carrier vectors for

double and single sideband
7-- my
2
my \
wt wt
el 92
8=w
(a) (b)
annoyance level. Hence, a subjective improvement might be

obtained in fringe areas if a device is incorporated in the chromin-
ance channel which rejects all chrominance information which is
less than a certain pre-set level. Then white, and all pastel or low
luminance colours, would be reproduced as a white chromaticity
without " coloured " noise.
16.14. Single sideband distortion
If an amplitude modulated R.F. signal is applied to a conven-
tional amplitude sensitive detector, the detected output waveform
differs from the original modulating signal if one sideband is
omitted. Since all television receivers now operate single sideband
over a substantial part of the higher frequency video spectrum, the
video waveform appearing at the conventional second detector of a
receiver is distorted. This distortion increases with modulation
depth, and is often called single sideband (or sometimes quadrature)
distortion. It is of some importance in N.T.S.C. colour trans-
missions, since the sub -carrier and its sidebands fall entirely within
the single sideband region.
In negative modulation systems the effective modulation depth is
greater for the black to white range than it is for positive modulation
systems and single sideband distortion is therefore more serious in
the former case.
The mechanism of the distortion may be illustrated by con-
sidering the simple case of a sine wave modulated carrier which is
applied to a conventional detector after the removal of one side-
band. Fig. 16.12(a) shows the double sideband case with two
equal vectors rotating in opposite directions at the modulating
frequency, the carrier amplitude being V and the modulation depth
index m, i.e. 100m % modulation. The detector measures the
magnitude of the vector sum of the three vectors, which can be
seen to lie always in phase with the carrier.
In the single sideband case of Fig. 16.12(b), if the sideband is
boosted 6dB relative to the carrier, the vector sum is equal to the
double sideband case only at the two instants when the sideband
is in phase or in anti -phase with the carrier. At all other instants,
the detector output, which is proportional to the magnitude of the
vector VT, will not be the same as the double sideband case, and
distortion will result.
Notice that in the single sideband case VT has a component in
quadrature with the carrier (hence the term quadrature distortion)
which is absent in the double sideband case because here the quadra-
ture components of the two sidebands are equal and opposite, and
therefore cancel.
The magnitude of VT is given by
v7,2 = V2 + m2V2 + 2mV2 cos 6,
by the trigonometric cosine rule.
Hence
VT = V(1 + m2)i (1 + ±2m cos BY
If m < 1, then by the Binomial expansion

M
VT = V(1± M2)1(1 + cos 0) L't V(1 + M cos 0)
which is the same as the double sideband case in which
m 711V
VT = V + -VCOSU
2
+ -2 cos 0
and is undistorted.
But suppose the modulation is 100 %, i.e. m = 1, then
VT = V2i(1 + -22 cos 0)1 = V2i cost 211

since
1 -I- cos 0 = 2 cos22
Hence
0
VT = 2V cos 2
(in which it must be remembered that VT is a magnitude so that the

0
magnitude of 2V cos must be taken for any value of 0) while the
2-
undistorted value is VT = V + V cos 0. These two cases are

shown in Fig. 16.13(a) and Fig. 16.13(b), respectively.
The original sine wave modulating signal is distorted to the same
shape as that of a full wave rectified sine wave. It has a D.C.
4V
component of - instead of the correct value V and a fundamental
IT
8 V-
component amplitude of instead of the correct value V.
37r
If one considers the case of a video waveform in which a sub -
carrier sine wave is oscillating about a luminance level, the sine
wave mentioned above may be taken as an example of the
sub -carrier sine wave. Hence, for a 100 % modulation case, the sub -
carrier amplitude is reduced since only the fundamental of the
sub -carrier can be accommodated in the video frequency range and
in addition a spurious D.C. component appears which affects the
luminance level. In positive modulation systems the D.C. com-
ponent increases the luminance level, while in negative modulation
systems the luminance level is decreased by the spurious D.C.
component, since in this case an increase in D.C. level at the
detector causes a decrease in displayed luminance.
It has been shown that the distortion is small for m < 1 and a
maximum for in 1. For intermediate values of m the distortion
itself will have an intermediate value which can be calculated in
terms of elliptic integrals, and these are convenient as they are
extensively tabulated.
The distortion of a sine wave modulating signal (such as the
sub -carrier signal) can be investigated by making a Fourier analysis
of the waveform. The waveform has a shape which is somewhere
between Fig. 16.13(a) and Fig. 16.13(b), depending on the value
0 27T 377
(a)
(b)
Fig. 16.13. Detector output for 100% modulation for (a) single
sideband and (b) double sideband
of m. It is symmetrical about 0 = ir and can therefore be expressed

as a cosine series, and it is fortunate that this symmetry applies for
all values of m, otherwise a phase shift (and therefore a hue shift)
would be produced which would vary with modulation depth.
The detector output VT is given by
VT= + M2)1(1 + 1 2m
+ m2COS
9)1
which may be expressed as a Fourier series

VT = ao + al cos 0 + a2 cos 20 . . . .
If we are considering the case of distortion of the sub -carrier, then
0 = 2pfst
where f, is the sub -carrier frequency, and we are concerned only with
the D.C. term ao and the fundamental amplitude al, since the higher
harmonics fall outside the video band and will be rejected by the
video output stage.
Then
2 ir
no =2a0
- VT dO and a1= 2- VT cos dO
7r 0
Substituting for VT in terms of 0, it can be shown that

n/2
ao =2V
-77 (1 + f
0
[1
(12V+771m)2 sin2o]
'P
2V
7T k
and
a1
327,1:1
+ mifff /21 2 Vrn1 2 sin201 +
Jo L ±
02 2,67-1
1 + m2 J0 + m)2
=3 V (1 ± 0(1 miE (1 m)2
K]
1 + m,
where
K = f'2[1. (i2±Anm)2 sin24
and
E = Er/ 2 [1 (12+71m)2 sin2d dy5
are complete elliptic integrals of the first and second kind

respectively, and may be looked up in tables for given values
of
12.6"\m/2
For any given colour, the value of m can be calculated by taking
the ratio of the sub -carrier amplitude to the mean carrier level
--
A 1001
CARRIER
B
I I
A
8
/
MODULATION DEPTH INDEX= M =
(a) (b)
Fig. 16.14. Modulation depths for negative and positive modulation systems
about which the sub -carrier is oscillating. Fig. 16.14(a) shows the
negative modulation case, and Fig. 16.14(b) the positive modulation
case. Notice that the greatest modulation depth is always smaller
in the positive modulation case because the synchronizing pulses
lift the level of carrier about which the sub -carrier signal oscillates.
By using the formulae quoted above, it is possible to calculate
the relative signal voltages applied to the display tube when a
colour bar signal is transmitted. For the case of the transmission
of the three full primaries and their full complementaries, Fig.
16.15 shows the signals at the red, green and blue displays for the
positive and negative modulation systems. The undistorted signals
are either 0 or 1 for each colour. In the negative modulation case
it was assumed that peak white was at 12.5% carrier, and that cyan
and yellow were clipped at 0 % carrier. Thus for cyan and yellow
the distortion introduced by clipping has been included in addition
to the single sideband distortion.
It can be seen that, in the negative modulation case, single side-
band distortion tends to reduce luminance and therefore increase
saturation. However a further effect is produced by the clipping
of the sub -carrier for yellow and cyan which tends to reduce satura-
tion, and in the case of yellow this reduction in saturation pre-
dominates over the increase in saturation caused by single sideband
distortion. For cyan, the single sideband distortion effect is the
greater so that the saturation is preserved in this case. If fully
saturated colours are transmitted, as in Fig. 16.15, the saturation
is unaffected by single sideband distortion since the saturation
cannot be further increased, i.e., the negative going signals applied
to the display are no more effective than zero signals. For non -
fully saturated colours, however, the negative going signals will
tend to increase the saturation of the displayed colours.
In the positive modulation case, no clipping occurs provided
the transmitter can accept satisfactorily the additional drive require-
1.2-
RED OUTPUT GREEN OUTPUT BLUE OUTPUT
I.0-
0.8-
0.6-
0.4
02-
411...
- -+ r
L 1 -- - -
Y C GMR B WY C GMRB W Y C GM R B
Fig. 16.15. Voltage output single sideband distortion for colour bars. Con.
tinuous lines show positive modulation and dotted lines negative modulation.
Undistorted levels are either 0 or 1
ment of the sub -carrier peaks. Single sideband distortion in-

creases the luminance and reduces the saturation, and is greatest
for red since this gives the largest m value. Notice that the greatest
single sideband distortion effect in the negative modulation case
occurs for cyan, the complementary of red.
Some idea of the subjective effects of single sideband distortion
can be obtained from Fig. 16.16, which shows the relative light
outputs in the case of negative and positive modulation systems,
assuming a square law relation between light output and voltage
input for the display device. That is, Fig. 16.16 is the " square "
of Fig. 16.15. The squaring process further reduces the luminance
in the negative modulation case, and increases the saturation in the
positive modulation case, and it would appear that the subjective
effects of single sideband distortion are reduced when positive
modulation is used. However, it must be pointed out that a full
luminance colour bar signal is the most severe test of single side -
band distortion effects, although bright saturated yellows are quite
likely to occur in practice.
16.15. Effects of method of modulation of vision carrier and sound carrier

The techniques employed in the N.T.S.C. colour system, while
originally designed for the American 525 -line standard which uses
negative modulation and F.M. sound, have been readily adapted
to positive modulation, A.M. sound systems, such as the present
British 405 -line system. It is instructive to consider some of the
more important differences between these two methods of vision
and sound modulation. It will be appreciated that differences
in line number or picture frequency are relatively minor from a
theoretical viewpoint, though there may be practical economic
considerations in the provision of wide video bandwidths in receivers.
Apart from this detail, the 525 -line and 625 -line systems may be
regarded as substantially similar, but this similarity does not include
the 405 -line British system because of its different vision and sound
modulation processes.
Before comparing the colour performance of negative and
positive modulation systems, it may be advisable to point out the
most important of the very significant differences between the
corresponding monochrome systems, since these are often not
generally appreciated. These differences may be listed as follows,
in which negative modulation, F.M. sound systems and positive
modulation, A.M. sound systems are designated N and P,
respectively:-
(a) Line and field synchronization are more susceptible to
noise or interference in N systems. Thus, flywheel line time
bases are required in 'N system receivers for B and I signal
RED OUTPUT GREEN OUTPUT

1 2 BLUE OUTPUT
1.0
0.6-
0.4-
02-
-0.2 WY CGMR B WY CGMRB WY COMRB

Fig. 16.16. Light output single sideband distortion for colour bars. Con-
tinuous lines show positive modulation and dotted lines negative modulation.
Undistorted levels are either 0 or 1
levels below about lm V, whereas for P system receivers the
corresponding signal level is about 200µV.
Again, field synchronization in N systems can be upset by
interference so that a field " bounce " or even collapse of the
field scan is produced. This effect is almost unknown in P
systems.
(b) N systems give black display of ignition interference, P
systems give white. Note that grey is probably the optimum.
(c) A.G.C. is apparently easier with N systems, since each line
sync pulse is a sample of 100 % carrier. However, gating is
normally essential to prevent A.G.C. action by interference,
and A.G.C. in N systems is then easier only on the grounds
of greater inherent gain.
(d) N systems tend to flatten picture whites as modulation drives
the carrier towards zero. P systems can have a large peak
white excursion for short periods while the black level remains
constant, but on N systems this can be done only by changing
the black level correspondingly.
(e) Single sideband distortion affects mainly picture content in
N systems, and mainly synchronizing waveform in P systems,
but the synchronizing properties are not impaired.
(f) D.C. maintenance is more difficult in N systems since absence
of the carrier drives the display above peak white if simple
D.C. coupling is used.
(g) Inter -carrier sound can be used in N systems.
Before discussing this point it may be as well to describe the
inter -carrier sound technique. In inter -carrier sound operation,
receiver sound rejection is normally not greater than about 20dB
so that an appreciable amplitude of the beat between sound and
vision carriers appears at the vision detector. Since the sound is
frequency modulated, the audio information does not produce
L.F. interference on the picture, and the sound carrier -vision
carrier beat which does appear is a high video frequency outside
the normal video band, and therefore not normally objectionable.
In any case, this beat can be removed from the video by means of a
suitable tuned rejector, if desired. The beat between sound and
vision carriers can be passed through a limiter and discriminator
to yield the audio information.
The advantages normally claimed for inter -carrier sound are
economic ones, in the sense that vision and sound I.F.'s are amplified
together, and that performance is relatively independent of local
oscillator tuning. However, it must be remembered that
economically the cost of a limiter and discriminator in N systems
must be weighed against the cost of the I.F. sound amplifier and
an A.M. detector in P systems.
As far as oscillator drift is concerned, the inter -carrier technique
does have some advantage, but this is not as great as one might
suppose. The reader may believe that since the sound -vision
carrier beat is not subjectively worrying on the picture, and since
the beat amplitude will increase rather than decrease for reasonable
detunings, then local oscillator stability is unimportant. But
while it is true that the audio output will be relatively independent
of tuning, it must be remembered that picture quality will not be,
for L.F. or H.F. video boost will result depending on the direction
of the tuning error. Again, if the detuning is sufficient to bring
the sound carrier to the edge of the sound rejection characteristic,
demodulation will result and produce audio breakthrough on the
picture. Strictly speaking, therefore, inter -carrier operation allows
a larger oscillator drift to the extent that the bandwidth over which a
20dB rejection can be maintained is greater than that over which a
40dB rejection can be maintained, 40dB being the approximate
figure for P systems. Also, a stable discriminator is easier to design
for frequencies in the region of the sound -vision carrier beat (about
5 Mc/s) rather than for sound I.F. frequencies (about 38 Mc/s).
Turning now to the colour performance of N and P systems,
it seems that the former suffers from these disadvantages :
(a) Full saturated yellow and cyan cannot be transmitted without
clipping the sub -carrier, unless the sub -carrier amplitude is
reduced relative to the vision carrier.
(b) Single sideband distortion is more significant (see Section
16.14).
(c) D.C. maintenance is essential in colour receivers to prevent
colour distortion, and this is more difficult in N system
receivers (see Section 16.15(f) (Monochrome effects) ).
(d) Accurate burst gating is more difficult in N systems because
of synchronizing difficulties.
(e) Unlike the sound carrier -vision carrier beat, the beat between
sound carrier and sub -carrier is a relatively low frequency
and therefore has high visibility. Considerable sound
rejection (about 60dB) is required in N system receivers to
remove this. In P system receivers, the monochrome figure
of 40dB rejection is adequate, since the sound carrier -sub -
carrier beat can be locked to an odd multiple of half the line
scan frequency, and this gives the subjective effect of approxi-
mately 20dB rejection compared with the unlocked case.
Although it is possible in principle to lock the unmodulated
sound -vision carrier beat in an F.M. sound system, in practice
this is not very successful.
Note that the greater sound rejection required in N systems
reduces the permissible oscillator drift compared with P systems,
even if inter -carrier sound is used (see Section 16.15(g) (Mono-
chrome effects) ).
It would appear that while some schools of thought may regard
the difference between N and P monochrome systems as being
marginal, or even in favour of N systems, there is overwhelming
evidence in favour of positive modulation, A.M. sound systems for
N.T.S.C. colour use.
16.16. Summary
In this chapter we have seen that individual gamma correction
of the red, green and blue video signals before encoding at the
transmitter, leads to constant luminance failure and hence a
degradation of noise performance. Gamma correction also entails
a non-linear relation between the transmission parameters and the
chromaticity of the colour being televised, so that the effects of noise
on the reproduced chromaticity are greater for pastel colours than
for saturated colours, and greater for complementary than for
primary hues. While there is no large area colour distortion, con-
stant luminance failure coupled with the bandwidth limitation of
the chrominance signals (and hence of the luminance carried by the
latter) causes distortion of transitions between colours and in some
cases produces luminance transients which are entirely spurious.
Bandsharing or multiplexing of the luminance and sub -carrier
signals leads to crosstalk between them which, when applied to a
non-linear display, produces spurious effects.
Single sideband distortion at the vision detector reduces the
sub -carrier amplitude and generates a spurious D.C. component
which tends to increase saturation in negative modulation systems
and decrease saturation in positive modulation systems. Because
the effective modulation depth is greater in the former case, negative
modulation systems suffer the greater distortion.
The sign of vision modulation and the type of sound modulation
used in an N.T.S.C. system have an appreciable effect on the per-
formance of colour receivers, and there are clear indications that
positive modulation and A.M. sound are to be preferred.
APPENDIX 1
Analysis of burst waveform

A.1.1. Calculation of Fourier coefficients
The burst waveform for the shortest burst of 8 cycles is shown in
Fig. A.1. It has been drawn over a period of 2 lines, during which
the sub -carrier completes 525 cycles in the British system. The
fundamental frequency is half the line frequency since the waveform
is repetitive after two line periods.
The waveform has been drawn symmetrically about 7r so that only
cosine terms will be present in the Fourier series.
The equation of the burst waveform during the burst is S cos Nx,
coLt
where N = 525, and x =
where WL = 27r times the line frequency.
If an is the amplitude of the nth harmonic, then
ir+ce
gran = S cos Nx cos nx dx S cos Nx cos nx dx
2ff
f 2 IT --Ot
S cos Nx cos nx dx
Since
cos Nx cos nx = 2 [cos(N n)x cos(N - n)x]

then
2iran [sin(N + n)x sin(N - n)xr
N+nS N-n Jo
[sin(N n)x sin(N - n)xlff
N+n N -n j 77 -et
[sin(N n)x sin(N - n)x 2*
N n -n J2,-ce
434
APPENDIX 1 435
2 LINE PERIODS
(525 CYCLES OF SUBCARRIER)
8 CYCLE BURST
0 iT
t
=
2
Fig. A.I. 8 cycle burst waveform over two line periods
Hence
27ra.
=0 for n even
and
27ran 4 sin(N n)a 4 sin(N - n)a
for n odd
S N n N -n
The harmonics of interest are those close to the sub -carrier
frequency, i.e. the values of n close to N.
Hence, approximately
-Iran 2 sin(N - n)a
S N -n
Putting m = (N - n), (m is always even)
Iran sin ma
2S
where m = 0 gives the sub -carrier frequency component, m = 2
gives the nearest harmonic lower than the sub -carrier frequency, and
m = -2 gives the nearest harmonic higher than the sub -carrier
frequency.
Note that the sub -carrier component has an amplitude aN which
is given by
TraN
2S
Limit sin ma -
436 APPENDIX 1
where « is the fraction of the twice line cycle occupied by half the
burst, i.e.
a- 4
525
X 27r
87r
=525
Hence
2Sa 16S
aN = 525
= 0.0305S
The burst waveform may therefore be represented by the series
0.03055 cos cost
+ I2cos(ws - 204,)t +
2S[sin 2«
cos (cos WL)t +sin44«
2S [sin 2« sin 4«
+ 2
cos(ws + wL)t + cos(w8 + 20.0t +
7r 4
82co8
where « = = 27r x sub -carrier frequency
and wL = 27r x line frequency.
Note that the amplitude of the nearest harmonic to the sub -
carrier component, and on each side of it, s of the amplitude
2a
of the sub -carrier component. This ratio is very nearly unity, since
sin 2a 2
2a
1 --3 «2= 1 - 0.00153
Care must be exercised in reference generator designs to avoid

locking to these adjacent sidebands. This is called side -lock.
The above series clearly illustrates the frequency interleaving
principle of the N.T.S.C. system. Thus, the fundamental com-
ponent of the burst waveform has a frequency of -tvL (the 262nd har-
47
monic counting downwards from the sub -carrier component) which
is half the line frequency, and all the harmonics are spaced from this
at intervals of the line frequency. These harmonics therefore fall
half way between the luminance harmonics, whose fundamental
frequency is equal to the line frequency (neglecting field frequency
components) and whose harmonic spacing is again equal to the line
frequency.
APPENDIX 437
A.1.2. Burst amplifier bandwidth
The mean square value of the waveform is given by the series
1 1 4S2rsin2 2a sin2 4« sin2
-(0.0305S)2
2 2 7T 2 22 42 62
1 4S2 sin2 2« sin2 4« sin2 6a

2 7r2 22 42 62
1 4S2 [sin2 2a sin' 4«+ sin2 6a

= 2(0.03055) + 7,2 22 42 62
for any number of harmonics.

The total mean square value can be calculated as follows :
The area under the square of the burst waveform for values of x
from x = 0 to x = a is given by
S2 S2 sin 2Nx-1°'
J S2 cos' Nx dx = -2 (1 + cos 2Nx) dx = -2 x ,
o 0
S2
=
The total area under the waveform over a two line period is there-
fore 2S2«, so that the mean square value is
2S2« 2S2 X 87r 8S2
0.0152S2
27r 27r x 525 525
This is the mean square value for an infinite number of harmo-

nics. If the sub -carrier component and 20 harmonics on each side
are taken, their mean square value is 85 % of this total. The
efficiency is shown in Fig. A.2.
The above waveform analysis has assumed that the burst is present
once every line, but strictly speaking there is also a field fre-
quency component because the burst is omitted during the field sync
pulses. However, since only 4 lines are affected in 2021, the field
frequency harmonics are very small.
It should be pointed out that, as far as a colour receiver is con-
cerned, the above analysis of the burst waveform assumes that the
gating circuit which separates the burst does not itself introduce a
spurious signal having a line frequency repetition rate. If the gating
circuit does produce a ring, or allows through a sync widget, then
such a spurious signal will have a fundamental frequency equal to
the line frequency and it will therefore have harmonics which are
29
438 APPENDIX 1
integers of the line frequency. These harmonics will be spaced by
f1/2 from the burst waveform sidebands and consequently, if large
enough, may limit the pull -in range of a reference generator to
+fL/4 instead of +fL/2 which applies in the case of a pure burst
waveform. Such spurious signals can be reduced by ensuring that
100
90
80
.I 70
0
0
11 60
50
S 40
u+
CI
CI
< 30
20
10
0 10 20 30 40 50 60
TOTAL NUMBER OF SIDEBANDS ACCEPTED
Fig. A.2. Graph of the efficiency of recovery of burst

energy against the bandwidth of the burst channel
the gating waveform does not have very fast rise and fall times,
and by positioning the gating waveform accurately to exclude
widgets.
If a colour signal is used in which the line frequency is not correctly
locked to the sub -carrier frequency (or alternatively is not held to
within about 0.01% of its nominal value) spurious line harmonics
may very appreciably impair the pull -in performance of a reference
generator and may cause side -lock to a frequency very close to the
APPENDIX 1 439
correct sub -carrier component. This is in no way a function of the
burst waveform itself which may be regarded as amplitude modula-
tion of the sub -carrier by a rectangular waveform of line frequency
repetition rate which has an " on " period equal to the burst dura-
tion. This modulating waveform has harmonics which are multiples
of the line frequency, so that after modulation the frequencies
produced will be equal to the sub -carrier frequency plus and minus
multiples of the line frequency. Thus, regardless of the numerical
relationship between the sub -carrier and line frequencies, the
nearest sidebands to the sub -carrier component will be spaced
from it by the line frequency. On the other hand, since a slight
change in the fundamental frequency of a spurious line frequency
component can produce an appreciable change in the absolute
frequency of a high order harmonic, then if the gating circuit in-
troduces spurious line frequency components into the burst wave-
form, pull -in performance can be reduced. For example, if the line
frequency were 20 c/s high at 10,145 c/s (about 0.2 % high) then the
262nd harmonic would be 2,657,990 c/s, which is only 177.5 c/s
higher than the sub -carrier component of 2,657,812.5 c/s. Such a
close spurious signal could seriously impair pull -in performance, and
it is therefore necessary to use a correctly locked line and sub -
carrier colour signal for a fair test of reference generator pull -in
performance. This would not apply to reference generators which
do not introduce spurious line frequency signals into the gated
burst signal.
29*
APPENDIX 2
Relation between
dynamic phase error, S/N ratio and
noise bandwidth
From the point of view of reference generator design, it is useful
to express the r.m.s. phase error for a given signal-to-noise ratio
in terms of equivalent noise bandwidth.
Thus, if the burst amplitude is S, the signal during the burst is
S cos wo, where co, is 27T times the sub -carrier frequency. The
presence of random noise will cause fluctuations in this signal, and
the phase errors which result are of particular interest as far as the
reference generator is concerned. If the instantaneous noise
which is present with the signal after integration by the reference
generator is N(t), this may be considered as the sum of two noise
components in quadrature such that
N(t) = a(t) cos cost b(t) sin cost
If a and b are typical instantaneous amplitudes of the quadrature

components, the total noise power will be given by
N2 = (a cos wit b sin (0802
which represents the mean square value of the cosine and sine
components, the amplitudes of which have a random variation with
time.
The mean square value of cosine or sine waveforms is half the
square of the amplitude, so that
2
N = -2- +
71 2 7; 2
where 1/2 is the mean of the squares of all the a components and b2
is the mean of the squares of all the b components.
440
APPENDIX 2 441
But by symmetry,
a2 =
so that
N2 = J2v
That is, the r.m.s. value of each of the quadrature components

is equal to the r.m.s. noise N, or
aRMS = bRMS = NRMS
Now the instantaneous amplitudes of signal and noise may be
represented by the vector diagram of Fig. A.3 in which the instant-
aneous phase error 4, is given by
tan -S b
a
which reduces to qS = - if the noise is not excessive. Hence, approxi-

S
mately
bRMS N
ORms -
The r.m.s. noise N above is that which is present with the signal
after integration by the reference generator. That is, before inte-
gration, the noise has a value N. in the video band fw (3 Mc/s in
Fig. A.3. Derivation of phase error

due to noise components
S+ a
a=INSTANTANEOUS NOISE COMPONENT IN PHASE WITH SIGNAL S

b=INSTANTANEOUS NOISE COMPONENT IN QUADRATURE WITH SIGNAL S
the British system), so that the noise power is N.2 for a bandwidth
N.2
fw, or -ff. per unit bandwidth. Consequently, if the reference
generator has an integration time T. corresponding to a noise
442 APPENDIX 2
Nw2
bandwidth IN, the noise power after integration will be - fN,
and this is equal to N2.
Hence
N Nw irfri
ORMS = 757; =
However, since the signal is present for only a fraction d of the time,
where d is the duty ratio of the burst, the effective integration time
is dry., i.e. the effective noise bandwidth is f -c-7. Allowing for the
burst duty ratio, the r.m.s. phase error becomes
Nw fN
RMS= 247 radians
- 180 Nw
7T
fN ,
-S- uegrees
fw
16
In the American N.T.S.C. system, fw = 4.3 Mc/s and d =
455
so that
Nw
ORMS = 0.148 A/TN
S
16
In the British N.T.S.C. system, fw = 3 Mc/s and d so that
525
Nw
S6RMS = 0.189 V.ITy
S
Note that the adverse tolerance of an 8 cycle burst is chosen. The
duty ratio of the burst is derived by noting that during 2 consecutive
line periods there are 455 or 525 cycles of sub -carrier (depending on
the system) and 16 cycles of burst.
For a dynamic phase error of 5° r.m.s. and a signal-to-noise ratio
S 1 .
of
. = (i.e. peak -to -peak burst equal to r.m.s. noise), the
2-
required noise bandwidth is
IN = 285 c/s for the American system
APPENDIX 2 443
I0
450 C/S
ISO C/S
50 C/S
0.1 0.2 03 04 05 06 0.7 08 09 10

S PEAK BURST
Nw R.M.S. NOISE IN VIDEO BAND
Fig. A.4. R.m.s. phase error as a function of signal-to-noise ratio for noise
bandwidths of 50 c/s, 150 c/s and 450 c/s (British system)
and
fN = 175 c/s for the British system
If excess burst gating width is to be allowed, these figures must be
reduced. (See Section 10.4.) Thus, for an excess burst gate of 1.5
to 1, the noise band -widths become
fN = 232 c/s for the American system
and
fN = 143 c/s for the British system
In Fig. A.4, the r.m.s. phase error (in degrees) has been plotted
against signal-to-noise ratio for three different noise bandwidths.
For any given noise bandwidth, the r.m.s. phase error is inversely
proportional to the signal-to-noise ratio, so that a design for a
444 APPENDIX 2
particular S to N. ratio automatically ensures less dynamic error
for higher S to N. ratios.
For a given signal-to-noise ratio, the dynamic phase error is pro-
portional to the square root of the noise bandwidth. Thus, reducing
the noise bandwidth 9 times (from 450 c/s to 50 c/s, for example)
reduces the dynamic phase error by a factor of 3.
APPENDIX 3
In -sync performance of an
A.P.C. loop
Let us assume that the loop is synchronized in frequency and
phase, and it is required to determine the static and dynamic phase
errors in terms of the loop parameters.
The loop parameters include the sensitivity of the reactance
valve, that is, the frequency shift of the oscillator in cycles per
second produced by one volt applied to the reactance valve, usually
called )9 c/s/V; the sensitivity of the phase detector, that is, the
voltage output of the phase detector for a phase difference of one
radian between the two signals being compared, usually called
µ volts/radian; and finally, the output voltage versus input voltage
characteristic of the loop low pass filter, which we shall call F(.0),
to indicate that this voltage ratio is a function of the frequency -fw
of the voltages involved.

As far as the in -sync performance is concerned, small values
of static or dynamic phase errors are of interest in which case the
phase detector output is virtually linear with phase angle. Strictly,
the usual type of phase detector gives an output which is proportional
to the sine of the angle by which the phase difference between the
signals being compared differs from 90°.
Thus, if one signal is sin(w8t 0) and the other is
sin(w8t + c + 90°)
the phase detector output is µ sin(0 - 0). However, if (0 - 0)
is small, then
sin(0 - 0) -
A.3.1. Static phase error
If the oscillator has an initial tuning error of Af c/s, then the
Af
reactance valve requires an input voltage of - to correct this error.
Since this is a static error, the correction voltage is D.C. and the low
445
446 APPENDIX 3
pass filter transmission is therefore 100 %, so that the phase detector
output voltage must be -Af Hence, if the phase difference between
.
output and input signals is ,60/4, which is the static phase error, then
µ sin A4,0 = -Af nt: /A AO,
that is
Af Af
"° - TLW - 7;
where fc = me is usually called the D.C. loop gain. (Note the
paradox that although fc is the D.C. loop gain, it is measured
in cycles per second per radian since IL = volts/radian and
I) = cycles/second/volt.)
Thus, the static phase error is directly proportional to the initial
detuning error, and inversely proportional to the loop gain. For a
given detuning error, therefore, the static phase error may be
reduced as desired by making µ13 large enough.
As far as dynamic phase error is concerned, the variations of
output phase due to noise fluctuations of the input phase are of
interest.
A simple approach to this problem is to assume a noise free input
phase which is given a sinusoidal variation with time, and to calcu-
late the magnitude of the corresponding output phase at various
frequencies. This is analogous to finding the voltage transfer
characteristic of a filter, except that here phase angle changes with
frequency are of interest instead of voltage amplitude changes.
Having calculated the phase transfer or phase following character-
istic of the loop, the equivalent noise bandwidth and hence the
dynamic phase error may be determined.
A.3.2. Phase transfer ratio

Assume that the loop is frequency and phase locked and that the
input phase is changed by an amount AO at a time t. If the corres-
ponding output phase change is AO, the phase detector output change
will be µ(A0 - 4). The voltage output from the loop filter will
therefore be µ(All - As9F(w), where F(w) is the voltage transfer
ratio of the filter. This voltage will produce a frequency shift of
µ(6,0 - Ach)F(a))13 c/s by way of the reactance valve, i.e. a shift of
fc(A0 - 601)F(w) c/s or toc(A0 - 64)F(.0) radians/second, where
APPENDIX 3 447
we = 2nfc. This frequency shift, however, must be equal to the
rate of change of output phase, since the loop is frequency locked.
Hence
dA0
coc(6,0 4l9F(w)
= (V-
PAO
where p represents the operator 4

dr
Therefore
PAO
AO = AO +
wcF(w)
or
AO F(w) p/wc
If the input phase variation is sinusoidal, then p = jw and
F(co) amplitude of output phase change
AO F(w) joi/wc - amplitude of input phase change
This ratio is often called Q(w), to indicate that it is, like F(w), a
function of the frequency 2 at which the phase is varied.
By using the above relation, it is possible to calculate the phase
following ratio Q(w) for any given A.P.C. loop. For example, if
the low pass filter of the loop is as shown in Fig. 10.9, then
Output voltage of filter 1 + jwxT
Input voltage of filter - w) - 1 + jw(1 x)T
where
T = RC
Then
F(w) 1 + jwxT
Q(`-") - F(w) jw/w, 1 - w2 (1 x)T/we jw(xT 1/we)
It must be remembered that Q(w) is a ratio of output to input

phases and not voltages. However, since voltage transfer character-
istics are usually more familiar, the expression for Q(w) above is the
448 APPENDIX 3
same as the voltage transfer characteristic of the circuit of Fig. 10.10,
1,,
which has a resistance of in series with an inductance of
wet.
(1 + x)1 in the top arm.
we
It is very important to notice that the expression for Q(w) is not
only a function of the filter characteristic F(w). The loop constant
we = 2irf, is also involved. While the low pass filter will modify
the output voltage of the phase detector, the effect on the output
phase will clearly depend on the magnitude of this voltage (i.e.
the phase detector sensitivity) and its frequency shifting capabilities
(i.e. the reactance valve sensitivity).
A.3.3. Dynamic phase error

A plot of I Q(w) I against frequency is shown in Fig. 10.11 for the
filter configuration of Fig. 10.9.
As far as phase variations are concerned, the A.P.C. loop behaves
like a narrow band filter tuned to the sub -carrier frequency, and it
is possible to find the equivalent noise bandwidth by squaring the
1Q(w)1 curve, and finding the equivalent rectangular bandwidth of
the same height and which encloses the same area, as described in
Section 10.1.
Notice, however, that the filter shape of Q(w) responds equally
well to frequencies above and below the sub -carrier frequency, so
that the 1 Q(co) 1 2 curve will correspond to the noise semi -bandwidth,
usually called fNN, such that
= 2-
INNfN
Hence
CO
IN = 2INN = 2 fo I Q(.)12 df
where the integral is the area under the squared Q(w) curve.
/NN = co. ,Q(cLi),2 dG7r1
00
= 2a o
1Q(0))12cla,
Substituting for I Q(w)12 according to

APPENDIX 3 449
1 + w2x2T2
IQ(w)21 -
[I - a2( I ± X)T1 2 + (02 (XT
(.0 CO C
it will be found that

2 no c
-1- (x1 + x)j
5- = /NN - 4(1 ± xTo.) c)
Notice that, for a given value of we, two variables are available
for satisfying this relation. Hence, apart from the noise band-
width, some control may be exercised over the shape of the Q(w)
curve to prevent resonant ringing due to impulse interference. To
investigate this, consider the operational form of Q(w) which is
1 + pxT
Q(P) - 2
1 +pwe-(1 + x)T p(xT ± w1/
This is the general expression for Q(w), in which the quantities are
not necessarily sinusoidal. The previous expression in which
p = jw, applies only to sinusoidal variations. Hence, if AO and AO
are the output and input phases, respectively, then
Act) 1
= Q(p) or
AO Q(P)4= A°
The complementary function of this differential equation is given by
Q(p)
=0
that is
2
1+ we
x)T p(xT + 1)
we
=0
Or
(1 + x)Tp2 4- (1 +xTwc)p+we =0
Solving this equation for p, the quantity under the square root sign
is
(1 + xTco ,)2 - 4(1 + x)T(.0,
(1 + x Tcoc)2
If a damping factor K is defined by K - then
4(1 + x)Tco,,
450 APPENDIX 3
if K> 1, the quantity under the square root sign is positive
and over damping occurs ;
if K <1, the quantity under the square root sign is negative
so that oscillatory ringing occurs ;
if K = 1, the damping is critical and near optimum conditions
are obtained in practice.
Hence, having chosen we so that the static phase error is within
the required limit for a given detuning error, x and T may be chosen
so that the required noise bandwidth is obtained. Further, since
x and T may be chosen independently, some control is available over
the shape of the Q(w) curve.
For the range of values which are of interest, x < 1 and
x71.0c > 1 so that
wc(1 x2Twc)
INN
4xTcoc
1 + x2Twc
4xT
1 xwc
= 4xT -F 4
and
(x11002 x2Twc
K
4Tcoe - 4
so that
1 + 4K
INN 4xT
or
xT (IC ±*)
INN
and
4K
x2T
we
Hence
x2T 4K .INN
X
xT (K -F
APPENDIX 3 451
and
T=X
1
XT
toc(K 4-
4KfNN, (K
fNN
1)
co,(K 1)2
4KfNN2
Hence, x and T may be calculated for a given noise bandwidth

and for given values of we (which determines the static phase shift
for a given detuning) and K (which determines the shape of the
response curve).
Note that since
1 xwe
fvN 4xT + 4
a value of x can be found which makes fNN a minimum. This
value xrain is given by
1
xmin -
1loc
in which case the damping factor for least noise bandwidth is
Kmin - x2min Mk
4
-4 1
and the minimum value of fNN is
fNNmin - 4xminT
2
-2
1 jc W
-T
However, to avoid ringing on impulse type interference, it is

advisable to use a K value nearer to 1.
APPENDIX 4
Pull -in performance of an

A.P.C. loop
In this section, formulae relating to the pull -in performance of the
loop shown in Fig. 10.9 will be derived.
The problem is to determine the time taken for the loop oscillator
to lock in frequency and phase to the burst signal, when the loop
oscillator has an initial tuning error of Ato rads/sec.
A.4.1. The loop parameters
The phase detector sensitivity, 12. volts/radian.
The reactance valve sensitivity, /3 c/s/V.
The D.C. loop gain, G = pg.
The operational form of the voltage transfer ratio of the loop
filter, F(p), which is equal to the ratio of the output voltage to the
input voltage of the filter.
A.4.2. The loop differential equation
The loop performance will be determined by a differential equation
which may be derived as follows:
Suppose the instantaneous voltage output of the loop filter is v
volts, then this will produce a frequency shift of fiv c/s in the oscil-
lator frequency. If the initial oscillator error was Af c/s, it is now
reduced to (of - /3v) c/s or Ow - 243v) rads/sec by the action of
the reactance valve.
If the instantaneous phase error between the input and output
signals is 91, rads, the instantaneous frequency error is d-dt rads/sec.
Therefore
dye,
- 27713v
dt =OW
v
But the phase detector output is ju sin 0, and - F(p)
14 sin 0
452
APPENDIX 4 453
Eliminating v from these two equations
d0
= - 277/3v = Ow- 277743F(p) sin 0 = Ow - co,F(p) sin
dt
Or
+ cocF(p) sin 0 =
This is the required equation which is the relation between 0 and t.

However, for the loop filter of Fig. 10.9 the operational form of
d .
F(w) in which p = Jo, is
1 ± pxT
F(P)l +P(1 ± 47.
and if this is substituted in the differential equation, the resulting
equation cannot be solved except by numerical methods.
A.4.3. The simplified loop
An approximate solution can be found by making use of the
results for a simple loop filter. Thus, if the loop filter is merely a
resistive network as shown in Fig. A.5, the voltage transfer ratio is
Fig. A.S. Simple resistive loop

filter
xR
VOLTAGE TRANSFER RATIO = m
simply given by F(p) = m. The equation for this simplified loop is

therefore
dcb
Imo c sin 0 = AUJ
dt
or
dt - ,60.0 - mac sin 0
Hence
d
mwct - + constant
Duo
Mcoc
-sink
454 APPENDIX 4
This may be evaluated, and there are two solutions depending on
whether -Ow is less or greater than one. If -Au) < 1, and if the
mwc mwc
phase at t = 0 is then the solution is given by
tan 2- - tan -2 tan - tan

- 9N
2 2
e -M0.) cos
cbT
c.
tan - cot tan -2 - cot -2 4,00
where Co is the value of 96 after an infinite time, and is given by

Aw
sin q. = -.mw,
That is, sb. is the ultimate static phase shift.
The above expression is rather similar to that for the voltage decay
curve of a discharging condenser
v = Ve -t/CR
and the " time constant " is
mwc cos Co
mcoc,f1 -( mw,,
)2
VOnwer (Aw)2
However, the value of ¢ at any time t will be a function of the initial

phase 00, since the " V" is a function of (/),. As an example of phase
stabilization time, suppose that 00, = 2° and 4,c, = 90°.
Then, if t3 is the time taken for the phase error to reach 3° (i.e.
within 1° of the ultimate phase), it will be found that
mw,t3 = 4.75
A typical practical value for mwc would be about 200 radians/
second/radian so that
4.75
13 - sec
or about 24 msec.
The important point to notice about the above solution is that ck

approaches Co without changing by more than 27r. In other words
APPENDIX 4 455
there is no frequency slip, only a phase transient. The design of
A.P.C. loops is usually more concerned with the case where fre-
quency slip does occur, since the time for frequency pull -in is very
much greater than that required for the, phase to approach reason-
ably closely to Co.
The " frequency slip " case occurs when -Au) > 1, for which the
nuo,
solution is
2
(a tan - 1) (a ta n 41° - 1)1
177wet tan -1 tan --1
a2 -1 Vat -I 1/a2 -1
where
Aw AT
a = mu, = info
The meaning of this equation can be clarified by writing
a tan 2- 1 a tan -2 -
J'= v a2 - 1 Y-
" V a2 - 1
and
27r
incuc 1/a2 =
TB
Then, the equation becomes

7T
-TBI = tan -'y - tan -1 Y
tan
7rt y -Y
TB- I yY
or
Y + tan irt/TE
I - Y tan 7rtITB
that is
y = tan (1rtITB (1))
where tan 'b = Y

456 APPENDIX 4
Now since
a tan -2 - 1
Y
v a2
then
tan
0
-
1 yA/a2 -I
2 a
But
2 tan 0/2
sin W 1 + tan2q/2
2a[1 y A/a2 - 1]
a2 + (1 V A/a2 - 1)2
2a [1 + - 1 tan (krt +OA

sin 0 - 2
a2 [1 + A/a2 - 1 tan (7t
IB
+0)
0 is proportional to the phase detector output,
which isµ sin 0. Hence, the above expression gives the waveform
of the detector output, and notice that while sin has the same
value for 0 as for (0 + 27r), the right hand side of the equation has
the same value for t as for (t TB). That is, sin 0 is cyclic with a
period
27T
TB -
my, \/a2 - 1
Af
Since a -mfc
TB - Af 2 VA/2 (mfc)2
nifC4AnTic) 1
The waveform of sin 0 is shown in Fig. 10.12 for three different

values of Notice that this beatnote waveform is not sinusoidal,
but tends to become sinusoidal as -LS,f increases. Notice also that the
mfc
peak positive and negative excursions are equal, and the waveform
APPENDIX 4 457
becomes inverted if the sign of " a " is negative, i.e. if the oscillator
error Af is negative.
To summarize the performance of the simplified loop, if -Of < 1
mfc
the loop pulls in without slipping a cycle, and the phase error ulti-
mately reaches a value equal to the static phase error. The phase
detector generates a control voltage proportional to i.e. propor-
tional to the tuning error, so that no frequency slip occurs.
Af
If -> 1, this loop does not pull in, and a steady beatnote
mfc
output results from the phase detector. This waveform has a D.C.
component, however, which reduces the mean tuning error, but for
large tuning errors the waveform tends to become sinusoidal so that
the D.C. component is then small. For various values of Of
(greater than mfc) it would be possible to measure the D.C. com-
ponent in terms of its frequency shifting capabilities, and obtain a
curve of D.C. control against Of. This D.C. control may be calcu-
lated by finding the D.C. component of the beatnote waveform.
Thus, the differential equation of the simple loop is
d0
+ mwc sin 0 = Ow
dt
and if this is integrated over a cycle of beatnote
t+ TB dO TB
dt dt + mwc sin 0 dt = AWdt
i.e.
t+ TB
27 ± ft mwc sin 0 dt = Aw TB
The integral in this equation is the area under a cycle of the phase
detector filter output, and the D.C. component of the output is
therefore this integral divided by TB. Calling this D.C. component
mwc sin 0
mwc sin 0 =0w -

-TB
27r
= L\w - 27 V6,f 2 - onfo2

Or
mfc sin 0 = of - -VAp - (mfg)2
30
458 APPENDIX 4
Hence, for a positive tuning error (oscillator frequency too high) a
positive D.C. component is generated which tends to reduce the
oscillator frequency. If the tuning error is negative, the opposite
sign of the square root applies and
mfc sin = - [Af - 412 - Onfe)9

that is, the D.C. component is negative. In this case the beatnote
waveform of Fig. 10.12 is inverted.
A plot of mfc sin 0, which is the D.C. component of the waveform
applied to the reactance valve (measured in terms of frequency) is
shown in Fig. A.6. Notice that the control is strong up to
Af = mfc, but for values of Of > mfg the control becomes pro-
gressively weaker as the beatnote waveform becomes more like a
D. C.
COMPONENT
-mfc
mfc AL, f
Fig. A.6. D.C. component of beatnote waveform as a

function of tuning error
sine wave. Notice also that although the transfer characteristic of

the simplified loop filter is not frequency selective, the loop behaves
as though there is a falling frequency response in that the control
voltage reduces as the beatnote frequency increases. The reason
for this behaviour is the loop time constant -1
mfc .
Thus, if mfc is very
1
large, - is very small and can change the phase appreciably during
mfc
the short time of one cycle of beatnote. On the other hand, if
mfc is very small very little phase change can occur during a beat -
note cycle.
The results obtained above for the simplified loop can be applied
to the loop which uses the filter shown in Fig. 10.9, in which a
condenser is included in series with the shunt resistor. As in the
APPENDIX 4 459
simplified loop, if the tuning error is small so that -Af < 1,

mfc
the loop does not slip a cycle and there is merely a phase transient.
If > 1, however, a beatnote output from the phase detector
m-fc
results, and the D.C. component of it is stored in the condenser.
The resulting mean beatnote frequency is thereby reduced, and a
larger D.C. component is therefore obtained, which again reduces
the mean beatnote frequency still further. This process continues
(provided that the tuning error, though greater than mfc, is not
excessive. The limits for the tuning error will be derived later)
until the tuning error has been reduced to mfc. The loop has then
frequency locked, and a phase transient occurs as in the simplified
loop when Af < mfc.
A.4.4. Quantitative analysis of the loop
The differential equation of the loop, quoted earlier, is
cocF(p) sin 4. = Aw
For the loop filter of interest (Fig. 10.9),

1 pxT
F(P) 1 + p(1 x)T
Putting x= -m m- 1
x
x
F(p) -
1 -m pmT
which may be written as

1 - 2m + m2
F(P) m ± 1 -m pT
Substituting this in the equation above, it follows that
d¢)
mc
co sin
± [1 - 2m + p WC Sin 4, = Aw
dt -m 1
or
d[1 - 2m ± 2i
mw c sin A,,,
m +mwc sin 9i
460 APPENDIX 4
This equation is the same as that for the simplified loop except that
here the right hand side of the equation is as quoted above, while for
the simplified loop it is merely iw.
The quantity
Fl _ 2m +
ACO COC si n0
L m + pT
therefore represents an instantaneous tuning error which reduces as
pull -in progresses.
Writing
11 - 2m ±
=A we sin 0
CO GO
L -m + pT
the differential equation becomes
d96
+ ?mac sin 56
dt
Using the results obtained for the simplified loop, the mean value of
4. over a cycle of beatnote
mcoe sin = w - A/c -T.)2 - (mcoc)2
where 7.0 is the mean tuning error. In the simplified loop this
quantity was, of course, constant. Here, however, it increases as
Z.0 reduces.
From the equation

_ 2m + m2]
w= Ac c sin 0
L -m + pTFl
the mean value of 0.) is given by
F1_ 2m 4-
Aco m2] coc sin 0
T./.)
I_ 1 -m pT
The quantity on the extreme right hand side is the mean value of the
frequency shift produced by the condenser voltage, and since the
time constant T of the loop filter is normally very much greater
than the beatnote period, it would seem to be legitimate to assume
APPENDIX 4 461
that this voltage is a result of applying the D.C. component of the
phase detector output w, sin 0 to the loop filter.
That is
_ 2m +
co , sin 0
P- 2m+ M21 Olc sin 9S
[1 -m + pT -m + pT
This may be shown as follows:
The operational impedance of the loop filter is
Zp = (1 x)R
For an applied voltage co, sin 0, the instantaneous current is
co, sin 0
Zp
and the voltage across the condenser is
co, sin 0 1 co c sin

Zp pC 1 4- p(1 + x)T
This voltage is applied to the reactance valve through a potentio-
meter of xR and R, so that a voltage
R to c sin
(1 x)R 1 + p(1 x)T
is applied to the reactance valve from the condenser.
1
Putting 1 + x -
-
from the condenser is
m,
the voltage applied to the reactance valve
(1 - m) co, sin 0 (1 - 2m + m2) co, sin 0

pT 1 -m + pT
1+
(1 - m)
From the two equations
co, sin 0
1
171 [co
_
- VT -02 - (mco,)2]
462 APPENDIX 4
11 -m pT
sin = [1 - 2m +
we m21 (Aw (7))
that is,
Aco w T dw
we sin C6 (1 - m) (1 - m) (1 - m)2 dt
co Aco
After eliminating we sinqS and putting p = and po =
mw c
it follows that
(1 - m) dp
mT
dt -
(1 - m) P2 - 1
m Po P
If TF is the time required to pull in from w = A w tow = mw,

(that is, TF is the frequency pull -in time) the corresponding values of
p are p = Po to p = 1, and therefore
(1 - m)TF 11 dp
mT j Po m Po + (1 - m) P2 - 1
A.4.5. Pull -in range
Before attempting to carry out this integration, it is possible to
calculate the maximum pull -in range. Thus, the above integral is
1
the area enclosed between the curve of [
mPo-P + (1 -nl)VP2-11
plotted against p, and the p axis, between the limits p = Po and p I.
Now if a real value of p can occur between Po and 1 which makes
[mpo -p (1 - m) V p2 - 11 zero, the reciprocal will be infinite
and so will the area under the curve. This will correspond to an
infinite value of TF.
Now if
mPo P (1 - M) VP2 - 1 = 0
then
m(2 - m) p2 - 2m PoP (1 -m)2 + m2 Poe = 0
and therefore
2mpo 4m2p02 - 4m(2 - m) - m)2 m2P021

P- 2m(2 - m)
APPENDIX 4 463
This cannot have a real value if
4m2p02 < 4m(2 - m) [(1 no2 m2p02]
that is
mPo2 < 2 m
or
Po < - - 1
The pull -in time TF will therefore be finite provided that
Af max< 4/ 2
mfc
or
Af max <fc 1/2m - m2
where Af max is the maximum tuning error.
A.4.6. Pull -in time
The above integral can be evaluated, but the solution is cumber-
some. However, we are normally interested in loops in which
m < 1, and making this approximation, we have
(1 -
TF
mT =
P.
dp
-1-
f1 p2
dp
(Vp2 1)2 p2
that is
TF r P0
(I - mT = (V P2 - 1 ± P) dP
-
[PO Poe - + Pe - 1 - cosh 1pd
Now provided pc is not near 1, the quantity on the right hand side
is approximately equal to p02.
For example, if po = 5 then
TF
(1 - m)-T = 23.1 ^ 52)
Therefore
f. 2
(1 TF
mT "2 mfc,
464 APPENDIX 4
This expression does not include the restriction that TF becomes
2 m,
infinite when po2 but by writing
(1 - m)TF Po2
mT
1
(2 m) P°2
this restriction is included. Since

1 - M 1
= , it follows that
x
xT po2
TF =
m 2
(2 - m) Po
For the range of values which are of interest
m x <1
so that
TF xT p 02 = xT (12
xfc
T 4\2
xkfc
Using the expressions
x ând
i)
T4KfNN
coc(K
coc(K +1)2
4KfNN2
which were derived at the end of Appendix 3 the pull -in time may
be expressed as
wc2(1( + An2
TF
421(2/NN3 kfs
/7.\ 2 (K (0n2
2) K2 fNN3
For given values of Of and fNN, K may be chosen to make TF as

(K 1)3
small as possible. The minimum value of occurs when
K2
K = and then
APPENDIX 4 465
1 2 27 (AD2
TF -671 4
fNN3
2(AD2
fNN3
If K = 1
TF 48
(of)2
fNN3
and if K =
(AD2
TF 4.93
fNN3
It is therefore preferable for K to be greater than -} rather than

less than I, and since the shape of the Q(w) response is optimum
near K = 1, it would appear that a K of 1 gives (very nearly) best
in -sync and best pull -in performance.
APPENDIX 5
Phase detector
A typical phase detector circuit which is often used in reference
oscillators is shown in Fig. A.7(a).
One signal to be phase compared is fed push-pull to two diodes,
while the second signal is fed to the anode and cathode junction of
the diodes. The output is taken either from point X (through a
-.--Vi sin (wt + CO
--*-- -V1 sin (wt + 0)
(a)
(b) (c)
Fig. A.7. Basic phase detector circuit
suitable resistor to prevent capacitance loading of the preceding

circuit) while Y is earthed, or from point Y while Xis earthed through
a suitable resistor.
A.5.1. Sensitivity of the phase detector
If V1 is the peak value of half the total push-pull drive, and V2 is
the peak value of the second signal, then the peak value of the
voltage across diode D1 is the vector difference between V2 and V1.
If this difference is V3, as shown in Fig. A.7(b), then
466
APPENDIX 5 467
V32 = V12 + V22 - 2V1V2 cos 9 --- V12 [1 11;21)2]
2 V2
x cosVl 0
1 + (-
V2)2
V1
V2 I
( - is small corn -
Now it is possible to make V1 > V2, so that I
pared with 1. Under these conditions, V3 is approximately given by
V2
= V1 [1 + Fil
(02
(44.1()2 ]+ 1 COS 0
V3
1+
V2 cos 0
= VI [1 + (i117)11-
[1
(V)2]
Since V3 is the peak value of the vo tage across diode DI the point
P1 will have a positive D.C. voltage relative to X of V3, where V3 is
given by the above expression.
Similarly, point P2 will have a negative D.C. voltage relative to X
which will be equal to the vector sum of the peak values of V2 and
V1, as shown in Fig. A.7(c). Calling this sum V4,
V42 = V12 + V22 - 2V1V2 cos (1800.- 0) = V12 + V22 + 2V1V2 cos 0
and therefore
V2 cos 0
V4 = VI [1 + (/iV +r 1 +
where V4 is a negative voltage relative to X.
Now the total D.C. voltage across the resistors R is V3 + V4 and
V4.
the D.C. current through them is V3 2R The voltage drop across
2R
the top resistor is therefore
V3 + V4 V3 ± V4
2R 2
468 APPENDIX 5
The point Y is therefore positive relative to X by a voltage of
(1/3 + V4\ V3 -V4
V3
2 / 2
V2 COS 0
+ (TT
i.e. the point Y is negative to X by approximately V2 cos 0. Hence,
if Xis earthed through a resistor, the output at Y will be - V2 cos 0.
On the other hand, if Y is earthed and the output is taken from X
through a resistor, the output will be + V2 cos 0.
Note that the output is proportional to the cosine of the phase
angle between the signals being compared, and to the amplitude of
the smaller signal, which is numerically equal to the sensitivity of the
phase detector. In the above example, therefore, if V2 had been
made greater than V1, so that -v. was less than 1, the output would
2
have been V, cos 0. In reference generators it is consequently
advisable to make the burst signal applied to the detector larger
than the oscillator signal. The phase detector output sensitivity is
then proportional to the oscillator drive, which is generally " cleaner "
than the incoming burst signal.
A.5.2. Effect of gated nature of burst
There is another reason for making the burst signal larger than the
oscillator reference signal. Thus, it has so far been assumed that the
reference and burst signals are both continuous sine waves. How-
ever, the burst signal is actually a short sample of sine wave and
this fact changes the characteristics of the phase detector somewhat.
Thus, referring to Fig. A.8(a), this shows a large push-pull signal
and a small " single ended " signal. The waveform vp, is the
instantaneous voltage at the point P1 of Fig. A.7(a), vp, is the instan-
taneous voltage at P2 and vx is the instantaneous voltage at X,
assuming Y to be earthed, and 0 = 180°.
If the larger signal is the burst, then after the burst has occurred
the bias on the diodes will be at + V, and -V, volts, D.C. These
bias voltages are sufficiently large to prevent diode conduction
during the period between bursts, that is, the phase detector
operates as a peak detector. On the other hand, if the small
" single ended " signal is the burst signal, then any D.C. voltage
developed during the burst will be discharged by the conduction of
+V,
-V2
-v,
(a)
Fig. A.8. Phase detector drive waveforms for

(a) large push-pull signal, (b) small push-pull
signal
470 APPENDIX 5
the appropriate diode on the peaks of the reference signal, after the
burst has occurred.
A similar situation obtains if the " single ended " signal is larger
than the push-pull signal, as shown in Fig. A.8(b). A little thought
will show that provided the peak burst signal is greater than twice
the peak reference signal, the detector will tend to operate as a peak
detector in either case. Thus, a greater efficiency results by making
the peak burst signal exceed twice the reference signal, whichever
signal is made the push-pull drive.
A.5.3. Justification for approximations
The reader may query the approximations made under the
assumption that V, > V2 (or vice versa). In practice, " very
much greater than " may, in this instance, be taken to mean " two
V2
to one " or more. In fact, if x is put for -, in the example of
Section A.5.1 the approximate detector output is
-x V, cos 0
while the exact output is given by
Vs -V4 Vi V1 ± X2 - 2x cos 0 -Vj V1 + x2 + 2x cos 0

2 2
The ratio
Approximate output
Exact output
is a maximum when 0 = 90° or 270°, and it then has a value of
+ x2.
If Vl = 2 V x = 0.5 and the above ratio is A71 -.7B-, or 1-12. Thus,
for a two to one amplitude ratio, the approximate output is only
12 % high even under the worst conditions.
It is interesting to note that while 0 = 90° or 270° gives the
maximum value of the above ratio, the approximate and exact
outputs are in fact both equal (i.e. zero) when 0 is exactly 90° or
270°. However, 0 = 90° or 270° gives the limiting value of the
maximum ratio.
APPENDIX 6
Reactance valve
A typical reactance valve circuit is shown in Fig. A.9, in which
only those components relating to the A.C. voltages and currents
are shown.
The purpose of the reactance valve is to vary the tuning of the
tank circuit by means of a D.C. voltage applied to the grid of the
triode. If C and R are both small, their series impedance is sub-
stantially capacitative and the voltage across R therefore leads the
Fig. A.9. Reactance valve circuit
applied voltage by approximately 90°, that is vg leads va by 90°.

The anode current ia, which is gmvg, therefore leads va by 90°.
vais
The impedance therefore capacitative, the value of which is a
Za
function of the gm of the triode. The gm is itself a function of the

D.C. bias on the triode, and hence varying the D.C. bias will alter
the tuning of the tank circuit.
A.6.1. Value of the impedance 7
The current through the C, R circuit is
va jcoCva
1 1 jcuCR
R+ .
joiC
Hence
jwCvaR
vg - 1 ± jcoCR
471
472 APPENDIX 6
and since is = gmvg (provided that the triode Ra is large) it follows
that
jwCRgmva
la - 1 ± jcoCR
Hence
jcoCva
i..=-- la ±
1 + jcoCR
- 1 jcoCva
+ jwCR
[gmR + 1]
and therefore
va 1 ± jcoCR
i - (gmR + 1)jaiC
- (gmR1+ :--
1) [R + jw1 C]
The triode and its CR circuit therefore behaves like a resistance

R
gmR + 1 in series with a capacitance of C(gmR + 1), that is a
resistance
R
(gmR + 1) (1 + 00122)
in parallel with a capacitance
(gmR + 1)C
1 + w2C2R2
The resistance damps the tank circuit without affecting the tuning,
gmRC
while the capacitance has a component 1 ± w2c2R2 which is pro -
portional to gm.
For values of interest in reference generators wCR is small com-
pared with 1 so that the effective shunt capacitance due to gm may
be taken as gmRC.
A.6.2. Sensitivity of reactance valve

The effect of the reactance valve on the tuning of the tank circuit
may be calculated:
APPENDIX 6 473
If the tank circuit is tuned to a frequency CO radians/second, with
an inductance LT and a capacitance CT, then
0)2 - 1
LTCT
We wish to know how w changes if a small change is made in CT.
Hence differentiating with respect to CT
2w
dw
dCT
- LTCT2
1 w2
CT
Therefore
df
dCT = - 2CTdco
-w - 2CT7
where f =
(The negative sign indicates the direction of tuning, i.e. increasing

dCT produces a negative effect on frequency, thereby reducing it.)
Now suppose for example, the tuning needs to be altered by
2 kc/s at 2.7 Mc/s, the tank circuit capacitance being 500pF.
Then
2x 103
dCT =- 2 x 500 x 10-12 x 2-7 x 106 0 74pF
That is, we require a tank circuit capacitance change of 0.74pF.

If the reactance valve changes its gm by 0.1mA/V for a bias
change of 1 V, then numerically
0.1 x 10-3RC = 0.74 x 10-12
i.e. RC -= 7.4 x 10-9
If R is made 2.21d2, then C = 3.35pF. These values will there-
fore give a reactance valve sensitivity of 2 kc/s/V.
It is usual to include a decoupled cathode feedback resistor in the
reactance valve to stabilize the valve characteristics against ageing.
The effective gm of the valve with a feedback resistor RF is
gm
1 + gmRF
and it is this value which must be used in calculating the sensitivity.
31
474 APPENDIX 6
For example, a typical valve characteristic would give a
gm = 2.7mA/V for a bias of - 1 V, and a gm = 1.6mA/V for a
bias of - 2 V. If a decoupled 1142 cathode feedback resistor
is used, the effective gm is
2.7
1 + 2.7
-0 73mA/V
at - 1 V, and
1+16 - 0.61mA/V
1.6
at -2 V. The change in gm for a 1 V bias change is then

0.73 - 0.61 = 0.12mA/V
It is important to remember that the above type of reactance valve
reduces the tank circuit frequency as the bias voltage is made more
positive. If C and R were interchanged (a suitable blocking con-
denser being included in series with R) the reactance valve would
behave as a variable inductance, and making the bias voltage more
positive would reduce the inductance and therefore increase the
frequency. This inductance type of reactance valve is not normally
used in reference generators, since the oscillator has a high C to L
ratio for stability reasons, and is therefore very sensitive to changes
in inductance.
APPENDIX 7
Quadricorrelator equivalent filter

The standard A.P.C. loop filter is shown in Fig. A.10(a). This
has a voltage transfer characteristic
V2 1 + jcuCxR
V,- 1 + jc0C(R xR)
When used in a variable ratio quadricorrelator circuit, the trans-
mission through this filter is increased by applying the voltage at
nrC7
P R P R
F\A'AAA17
Vi
Q V2
xR 1j AxR
V2' = AV2
(a) (b) (c)
Fig. A.10. Equivalent circuit of a quadricorrelator filter
P to Q via a cathode follower. Hence, no phase shift is introduced

and the equivalent filter circuit becomes that shown in Fig. A.10(b),
where S is a resistance whose value is such that
V2' = A V2
i.e., the filter transmission is increased by a factor A.

From Fig. A.10(b)
jwCxR(S R)
1+
V21 S xR
VI joJC[RS xR(S
1+
S xR
2' 2
Comparing the two ratiosV,.- andVI-, note that
475
476 APPENDIX 7
+ 7sR)xR
xR is now
xR
1+
and
R xR is now
RS + xR(S R)
S xR
i.e.
1
R is now
1+ xR
Normally, x is less than 0.01 and the increase in transmission
through the filter is at least 10 to 1. That is, A = 10 (at least), so
that - ft 10 also, i.e. -^ A
Thus
(1 A)xR
xR is now AxR
1 xA
and
R is now
1+
Hence, for those values of x and A which are of interest, the equiva-
lent circuit of the increased transmission filter is as shown in Fig.
A.10(c).
APPENDIX 8
Effect of resistive mismatch on the

value of the phase shift along a
characteristically terminated cable
In the co -axial cable phase shifter described in Section 12.5,
the need to accurately terminate the cable to prevent errors in
phase reading is mentioned. The error produced by a given mis-
match may be calculated:
For a line of length 1, propagation constant P(= a + ji9), charac-
teristic impedance Zo, termination load ZL, the ratio of input to
output voltage for a sine wave signal is given by
Zosinh
= cosh PI + PI
Vo
If the line is assumed to be lossless, a = 0 and therefore P = jfl.

The characteristic impedance Zo becomes purely resistive, i.e.
Zo = Ro.
Then
.Ro
Vo = cos /3/ jz, sin /3/
If 0 is the phase by which Vo lags 11/, then

Ro
tan0=-tan
ZL
Now suppose the load impedance ZL is purely resistive, and
equal to RL, and suppose that
Ro
RL= 1 + e
that is, E is the fractional mismatch. Then if the change in phase
477
478 APPENDIX 8
due to the mismatch is S such that the actual phase shift 0 is given
by
= 00 + 8
where 00 = flu is the phase shift for correct termination, then
tan(90 ± 8) = (1 + c) tan 00
Expanding the left hand side,
tan 00 + tan S
1 - tan 00 tan 8 - (1 ± E) tan 00
If S is small, then tan S = S so that
tan 00 + 8
(1 + c) tan 00
1 - 8 tan 00
whence
e tan 00
8r_t.-
sec20o
= -2 sin 200
=2 sin 2131
The error angle S is therefore a function of the mismatch and the
length of line. It is a maximum when sin 2f31 = ± 1, i.e. when
371.
2f31 =7-; or
The maximum value of the error S is therefore given by

90E°-
S = radians or
2- 7r
ir 3 7r
and it occurs for phase shifts of 45° and 135°, i.e. when flu = - or 4-
4
Hence, for a 10 % mismatch, for example, e = 0.1 and S 3°.
It is therefore recommended that the mismatch be limited to not
more than 1 %, in which case the maximum error is only about one
third of a degree.
Bibliography
For a comprehensive bibliography on colour television up to August 1956
see:-
" A Bibliography of Colour Television " (with two appendices), The
Television Society, London.
1. Proc. I.R.E., 39, No. 10 (Oct. 1951), 1st Color Television Issue.
2. Proc. I.R.E., 42, No. 1 (Jan. 1954), 2nd Color Television Issue.
3. " Specification of Colour Television Standards for Experimental Trans-
missions from Alexandra Palace ", B.B.C. Engineering Division (June
1955).
4. Atkins, I. R., Stanley, A. R. and Watson, S. N., " A New Survey of the
B.B.C. Experimental Colour Transmissions ", B.B.C. Engineering Divi-
sion Monograph No. 32 (1960).
5. " Colour Television Standards. N.T.S.C. ", Ed. by Fink, D. G., McGraw-
Hill Book Co. Inc. (1955).
6. " Television Engineering Handbook ", Ed. by Fink, D. G., McGraw-Hill
Book Co. Inc. (1957).
7. Davidse, J., " Experiments on adapting the N.T.S.C. Colour Television
System to the European 625 line Standard ", N.T.Z. 11, 461 (1958).
8. Carnt, P. S. and Townsend, G. B., " Subjective Effects of A.N.T.S.C.
Colour Television Receiver Tolerances ", Acta Electronica 2, 77 (1957).
9. Maurice, R. D. A., " A Review of Colour Television in the U.K. ",
Electronic Eng., 32, 68 (1960).
10. Davidse, J., " N.T.S.C. Colour Television Signals ", Electronic Radio
Eng., 36, No. 10, 370 (Oct. 1959).
11. Sproson, W. N., " Compatibility of the N.T.S.C. Colour Television
System adapted to the 405 -line Standard: Field Trials ", B.B.C. Engineering
Division Report No. T-062 (1956).
12. Radio Corporation of America Service Data for CTC-5, CTC-7 and
CTC-9 series of Colour Receivers. Prepared by Commercial Service,
R.C.A. Service Company.
13. Lectures on the Design of Colour Television Receivers. Hazeltine
Corporation Report No. 7,149 (Oct. 29, 1953).
14. A Colour Synchronizing Circuit Design. Hazeltine Corporation Report
No. 7,155 (June 9, 1954).
15. A Colour Decoder for Television Receivers: Part 1 -Theory and Measure-
ments. Hazeltine Corporation Report No. 7,192-1 (June 24, 1958).
16. Lind, A. H., " Operational facilities in the R.C.A. Colour Television
Tape Recorder ", Brit. Inst. Radio Engrs. Convention, July 1959.
17. Freeman, K. G., " A Gating Circuit for Single -gun Colour Television
Tubes ", Brit. Inst. Radio Engrs. Convention, July 1959.
18. Barnett, G. F., Bingley, F. J., Parsons, S. L., Pratt, G. W. and Sadowsky,
M., " A Beam -Indexing Color Picture Tube -The Apple Tube ", Proc.
I.R.E., 44, 1,115 (1956).
19. Clapp, R. G., Creamer, E. M., Moulton, S. W., Partin, M. E. and Bryan,
J. S., " A New Beam -Indexing Color Television Display System ", Proc.
I.R.E., 44, 1,108 (1956).
479
20. Maurice, R. D. A., " Colour Television Standards ", Wireless World, 66,
No. 11 (1960).
21. Carnt, P. S. and Townsend, G. B., " Random & Impulse Noise in
Colour Television ", Acta Electronica, 2, 230 (1957).
22. Lance, T. M. C., " A Medium Screen Colour Projector ", Brit. Inst.
Radio. Engrs. Convention, July 1959.
23. Fairhurst, H. A., " Development of Experimental 21" Colour Receiver ",
J. Television Soc., 8, No. 2, 71 (1956), & 8, No. 3, 95 (1956).
24. James, I. J. P., " A Vidicon Camera for Industrial Colour Television ",
J. Brit. Inst. Radio Engrs., 19, 165 (1959).
25. Chaste, R., Cassagne, P., " Henri de France Colour Television System ",
Ed. by SECAM system, Inst. Elec. Engrs. Paper No. 3,251 (April 1960).
26. The British Contribution to Television: Inst. Elec. Engrs. Convention.
April -May 1952.
27. Hunt, R. W. G., " The Reproduction of Colour ", Fountain Press, London
(1957).
28. Wright, W. D., " Researches on Normal and Defective Colour Vision ".
Henry Kimpton, London (1946).
29. " The Science of Colour ", Committee on Colorimetry, Optical Society of
America, New York. Thomas Y. Crowell & Co. (1953).
30. MacAdam, D. L., " Projective Transformations of I.C.I. Colour Specifica-
tions ", J. Opt. Soc. Am., 27, 244 (1957).
31. Wentworth, J. W., " Colour Television Engineering ", McGraw-Hill
Book Co. Inc. (1955).
32. Judd, D. B., ' Colour in Business, Science and Industry ". Chapman &
Hall (1952).
33. Land, E. H., " Colour Vision and the Natural Image ", Proc. Nat. Acad.
Sci., 45, 115 & 636 (1959).
(See also Judd, D. B., " Appraisal of Land's work on Two -Primary
Colour Projections ", J. Opt. Soc. Am., 50, No. 3, 254 (1960).
34. Gunzler, H., " Metal Interference Filters ", Zena Review. 4 year, 188,
(1959).
INDEX
A.C.C. Automatic phase control 271
See Automatic chrominance con- Automatic saturation control 306
trol
Achromatic point 47
Additive process 27 Background control
Adjustments See Brightness control
interaction 366 Baird 59, 80
receiver. See Receiver installa- Balanced modulator
tion I' 160
Aerial requirements 356-57 Q' 163
Afterglow 55 Beat frequency. See Difference
A.G.C. See Automatic gain control frequency
Alychne 45, 136 Beatnote 284, 456
Amplifiers ambiguous 286
burst. See Burst Benham 12
chrominance. See Chrominance Bifilar-T trap 200
amplifiers Black 25
colour receiver 196-212 Blind luminance axis 416
luminance. See Luminance Blue difference matrix 235
Amplitude response Blue lateral shift 66, 324, 360
I.F. stages 197, 199 Blue/grey transition 410, 412
R.F. stages 196 Brightness 22
A.M. sound 430 control 332
A.P.C. See Automatic phase con- measurement 27
trol photometric 29
A.P.C. loops 271, 278-307 Brown 26
analysis 459 Burst, colour 6, 107, 186, 262
crystal oscillator. See Crystal amplification, 269
oscillator A.P.C. loops amplifier bandwidth 437
damping factor 282, 449 analysis 434-39
D.C. gain 280, 452 appearance 375
differential equation 452 circuits 162, 266, 270
equivalent circuit 281 duty ratio 442
feedback 291 energy recovery efficiency 438
in -sync 279, 445 Fourier series 436
limitations 287 gating 265, 270, 432
parameters 452 insertion 164
phase errors 280-83 phase 121-22
pull -in 283-87, 452-65 sidebands 263
quadricorrelator. See Quadricor- specification 262
relator take off point 269
simplified loop 453 tests 350
transfer characteristic 280 visibility 387-88
Aperture mask tube. See Shadow waveform, 109, 435
mask tube
Apple tube 63
Asynchronous operation 111, 186, Calibrated phase shifter 338-42
377 resistive mismatch 477-78
Automatic chrominance control 211, Camera
272, 299-300 circuits 72
performance 347 colorimetric analysis 74
Automatic gain control 431 colour 3, 69-77
Automatic hue control 306 R.C.A. 70
481
482 INDEX
Camera -Continued Colour -Continued
signals 105 flicker 60
spectral response 75 fringing 60
vidicon 72 mixtures 36, 40-41, 47-50
Carrier non -spectrum 24
frequencies 172-73, 378 of objects 26
oscillator 170-72 perception 10-11
spacing 111, 379 receiver. See Receiver
suppressed 98 reference. See Reference
Chromacoder 73 space 38-40
Chromaticity 33 spurious effects 297, 423
changes 51 subjective 11-13
co-ordinates 33-35 test stripes 114
diagrams 34, 45-46, 51, 139-54 transitions 197, 373, 410-15
Maxwell diagram 39 transmission. See Transmission
of colour mixtures 36 triangle 35, 36, 134-38
of phosphors 56 Colour bar
of spectrum colours 40 circuit 180
transitions 151 generator 178-81
Chromaton 61 signal 122-25
Chromatron 61 waveform 124-25, 169, 178, 371
Chrominance 50, 78 Colour blindness 10
bandwidth 100-02 Colour difference
buzz 193 amplitude reduction 93
demodulators 182 axes 93, 95, 150, 228
Chrominance amplifier 208-12 signals 5, 87-91, 106, 223
circuits 209, 210 weighting 94, 125
function 355 Colour killer. See Killer
Chrominance channel
bandwidth 189-91 Colour killing. See Killing
performance 344-48 Commission Internationale de
response 208 l'Eclairage 43-46
Chrominance signal colours 46
axes 150, 152 reference stimuli 43
decoding. See Receiver standard illuminants 45, 84
detection 185 Compatibility 2, 78
effect of gamma correction 133-54 reverse 2
matrixing 255-57 Complementary colours 25, 47
on C.I.E. chromaticity diagram 139 Cones 9
transmitted 103, 106 Constant luminance
C.I.E. See Commission Inter- failure 392-94
nationale de l'Eclairage index 394
Clamp source 163 operation 91-93
Clamping principle 86-87, 121
in XZ detection 245 Convergence 66, 318-32, 354
of I and Q signals 163 adjustments 359-63
Coding 3 circuits 325, 328, 330-32
circuits 156-81 dynamic. See Dynamic con-
transmitter 78-116 vergence
Colorimeter 30 errors 311
Colorimetry 22, 30 horizontal 361, 363
Colour magnets 66, 360
addition 47-50 measurement 352
adjustments 364 static. See Static convergence
balance 352 vertical 361, 362
break up 60 waveforms 325-27
centres 315 Cross colour 192, 210, 418
controls 194 effect of gamma 419-23
faults 372-75 Cross modulation 200
INDEX 483
Cross talk chrominance 187 Divider chain, frequency 174-77
in XZ detection 244 Dominant wavelength 23, 47
I'Q' 197, 199 dependence on hue and saturation
with luminance 227, 418 407
Crystal effect of sub -carrier phase 405
equivalent circuit 274 Dot interlacing 110
filter 272 Dot sequential system 81
oscillator 172 Dots, sub -carrier 379-81
sub -carrier oscillator 162, 338 interference 379
Crystal oscillator A.P.C. loops rectification 381
287-92 structure 380
balance 289 Dynamic convergence 66, 318, 354
circuits 288, 302 adjustments 361
design 290 servicing 370
Dynamic integrators. See A.P.C.
loops
D.C. components Dynamic phase error 264, 279, 281-83
effect of modulation system 431, 432 derivation 446, 448
in A.P.C. loops 457 in terms of noise bandwidth 440-44
in decoding circuits 224, 247 measurement 350
in transmission circuits 163
Decay factor 278 Eccles Jordan divider 176
Decoding 5, 213-58, 223 Echoes 356, 371
Deflection, for shadow mask tube E.H.T. 313
308-15 measurement 352
centres 66, 315, 320 Electronic masking 76
horizontal. See Line time base Encoder 158-70
magnetic 318 circuits 161-62
vertical. See Field time base equiband 181
yoke 308. See also Time base Encoding, non-linear 401
Degaussing 68 Envelope delay. See Delay
Degaussing coil 316, 357 Equal energy white. See White
Delay Equiband
characteristics 113 operation 210, 226-28
encoder 170 receivers 191
envelope 17, 114 Excitation purity 47
of coaxial cable 341 Eye
Delay line, performance 343 acuity 7
Demodulator 182 adaptation 8
axes 189, 191 limitations 6
Detection
axes 216, 228. See also Decoding
Detector output 426 Fault finding
Dichroic mirrors 58 See Receiver fault finding
Dichromatism 10 Fechner 8, 13
Difference, colour. See Colour Field 13
difference equalizing 316
Difference frequency removal gate 165
sound/sub-carrier 158, 193, 200, 211 sequential systems 80
vision/sound 111, 379 Field time base 315
Differential gain distortion 205 circuit 314
Differential phase distortion 205 Film scanners 54-77, 74
Diode detection 216 F.M. sound 429
Display, colour 5, 54-69 Focus mask tube 61
primaries 55-56 Focus voltage 311
rotating disc 59-60 Foot lambert 29, 47
sequential 54 Fourier
simultaneous 54 coefficients 434
three -tube 57 components 18
484 INDEX
Fovea 9 Integration time 260
Frame. See Field Intercarrier sound 431
Frame time base. See Field time base Interference
Frequency dot. See Dots
difference. See Difference fre- ignition 431
quency pattern 6
interleaving 113, 436 I'Q' fallacy 415
lock 284
relationships 108, 158 Judd 52
slip 455
spectra 19-21, 113 Jumbo 57
Killer, colour 297

Gain control, receiver 200 adjustments 365
Gamma, unity 133-45 performance 351
Gamma correction 84-86, 156 reliability 299
effect on transient response 408 servicing 371
Gamut, of chromaticity 55 threshold 298
Gate generator, burst 266-69 Killing 297-99
Gated amplifier 270 purpose 418
Ghosts. See Echoes
Goethe 8
Grassman 31 Land 8
Green difference axis 228 Lateral shift. See Blue lateral shift
Green -magenta axis 96 Lawrence tubes 60-62
Grey scale adjustments 363-64 Light
Grey scale tracking 332, 354 characteristics 46
servicing 370 monochromatic 24
Group delay 17 nature 23-24
standard source 29
Line sequential systems 81
Herringbone pattern, coloured 236 Line time base 311
High level detection 231-38 circuits 310, 312
circuits 234, 237 flywheel 430
Horizontal deflection. See Line time radiation 375
base Loops, A.P.C. See A.P.C. loops
Hue 23 Lumen 29
effect of noise 406 Luminance 23, 29, 47
errors 263, 266, 356 amplifiers 196, 205-08
Hue control 194, 289 channel performance 342-44
adjustments 365 constant. See Constant luminance
automatic 306 control by I' and Q' 417
effect on generator frequency 351 delay 183, 206
function 356 effect of dot rectification 381-83
measurement 350 enhancement 200, 203
line 45
notch 415
I' axis 95 of burst signal 122
I signals, for unity gamma 143 signal 78, 86-87, 118-21
I.F. stages, receiver 197-205 transitions 154, 383-87
circuits 201, 202 Luminosity
Illuminant coefficients 33, 44, 86
normalizing 84 function 28
standard 45 relative 28
Image 11 Luminous efficiency of phosphors 55
Index strip 64 Luminous flux 29
Indexing frequency 64
Installation. See Receiver installa- Magenta 26
tion Magnetic deflection. See Deflection
INDEX 485
Masking, electronic 76 N.T.S.C. See National Television
Matrix System Committee
circuit 185, 219, 222
design 218-23
equations 166 Orange -cyan axis 96
network 160 Oscillator
three -tube. See XZ detection carrier 170-72
Maxwell colour triangle 35 crystal. See Crystal
Measurement drift 432, 433
colour 22-53 reference. See Reference generator
equipment 336 sub -carrier 162, 338
receiver performance 335-53 vision carrier 172
Minimum phase 18
Mismatch, resistive, in terminated
cable 477-78 Parc 192, 418
Misregistration 318. See also Con- Passive/dynamic integrator 300-03
vergence circuit 301
Mixer detection 229 Passive integrators 271, 272-78
Modulation circuit 272
depth 428 limitations 273
effects of method 429 with phase correction 277
waveforms 79 Perceptibility of chromaticity
Moire 69 changes 51-52
Monochromatic light 24 Perception. See Colour
Monochrome Persistence
adjustments 354, 357 phosphors 55
compatible signal 120 vision 7
reception of N.T.S.C. signals Phase 15-18
377-90 automatic control. See Automatic
servicing 369 phase control
systems 430 burst. See Burst
Multiplexing, effects 417 distortion 205
encoder measurements 168
errors 263, 279-83
Narrowband chrominance receivers following ratio 447
191 minimum 18
National Television System Com- response of I.F. amplifiers 197-99
mittee 2, 82 response of R.F. tuners 196
American system 442 transfer ratio 281, 446
bandwidths 100 zero reference 98
British system 442 Phase detector 271, 466-70
colour specification 117-55 approximations 470
outline of system 3 effects of gated burst 468
primaries 55, 117 sensitivity 279, 452, 466
reference stimuli 83 Phase shifter. See Calibrated phase
shortcomings of system 391-433 shifter
video signal 104, 156 Phosphors 55
Negative modulation 429 luminous efficiencies 55, 332
Noise, chrominance channel Photometer 28
bandwidth 261, 264, 273, 281 Photopic vision 9, 28
effect on hue 406 Picture period 13
effect on purity 405 shift 311
formulae 450 Picture tubes 54-77
power 441 drive ratio 226
random 260, 375 Positive modulation 430
relation to dynamic phase error Post deflection focus tubes 61
440-44 Power supplies 194
Noise factor of tuner 196 Pre -detection matrix 256
Notch filter 380, 419 Prevost, Benedict 11
486 INDEX
Primaries Reference stimuli 30
display 55 C.I.E. 43-46
non-physical 43 N.T.S.C. 83
Primary colours 30, 117 Reflections in delay cable 205
Purity 23, 47, 315 Retina 9
adjustments 354-55, 357-59 R.M.S. dynamic phase error 264
effect of noise 405 Rods 9
effect of subcarrier amplitude 402 Rooter 85
magnets 67, 315 Rotating disc. See Display
measurements 352
Purple 26
Saturation 23
Saturation control 211, 306
Q' axis 95 adjustment 365
Q signals, for unity gamma 143 function 355
Quadrature Scotopic vision 9, 28
crosstalk 101 Sensing tubes 62
distortion 424 Servicing equipment 355. See also
modulation 215 Receiver fault finding
signal 188 Shadow mask tube 54, 65
Quadricorrelator 287, 292-97 adjustments 358
beatnote 294 electrode capacitance 332
block diagram 293 input/output characteristics 333
circuits 295, 303-05 operation 69, 308-34, 358
equivalent filter 475-76 Side lock 436
variable gain 296 Signal/noise ratio 440-44
variable ratio 296 Simulquential pick up 73
Sine wave convergence 325
Single sideband distortion 423-29
Raster, faults 368-69 Sit control. See Brightness
R.C.A. shadow mask tube 54 Sound 111
Reactance valve 279, 471-74 buzz 389
circuit 471 carrier mixer 172
inductance type 474 modulator 173
sensitivity 279, 452, 472 rejection 197, 200-03, 432
Receiver R.F. carrier 158
amplifiers 196-212 Spectral
block diagram 183 energy 25, 30
decoding circuits/213z58 response of camera 75
design 182-95 Spectrophotometer 41
installation 354-66 Spectrum 22
performance. See Measurement colours 24
Receiver fault finding 367-76 Spot limiter 208
colour input 371 Stabilization period 264
initial examination 368 Static convergence 66, 354
monochrome input 369. See also adjustments 360
Servicing servicing 369
Reference colour 31 Static phase error 264, 279, 280, 445
Reference generators 182, 259-307 measurement 349
adjustments 364 Sub -carrier, colour 5, 78, 98
alternative 305 amplitude control of purity 402
classification 271 beat 158, 193
function 186, 355, 364 buzz 200
performance 348-51 circular 397
radiation 376 dots. See Dots
specification 263. See also A.P.C. elliptical 394
loops; Quadricorrelator frequency 109
Reference signal 215 harmonic radiation 236
effect of absence 223 maximum excursions 129, 132
INDEX 487
Sub -carrier, colour-Continued Trinoscope 57
modulation 96 Tristimulus values 31
phase control of dominant wave- N.T.S.C. 107
length 405 Tuners, R.F. 196
rectification 207 Two mode loop. See Quadricorre-
rejection 207 lator
resultant 97
with unity gamma 144
Subtractive process 27 Vectorscope 337
Symmetrical detection 239-41 Vertical deflection. See Field time
circuit 240 base
Sync widget. See Widget Video
Synchronizing signal, colour. See amplifiers 203
Burst circuits 204-07
Synchronous detection 213-18, composite waveforms 165, 169
229-55 Vision
Synchronous detector colour 9
circuits 217, 222 detectors 203-05
function 355 modulator 170-72
measurement 345-46
Synchronous working 378
Wave shaping circuits 325
T -unit 32 Wavelength, dominant 23, 46
Test equipment 335-53 Weber-Fechner 8
Three tube display 57 White
Time base balance stability 224, 238, 247
adjustments 359 equal energy 25, 46
field. See Field time base light 25
line. See Line time base N.T.S.C. system 117
measurement 352 raster 354
synchronizing effects, 388, 430 receiver 84
Time delay 16 reference 33, 47
matching of signals 102 spectral energy 25
Transformation 41-43 studip 84
Transient Wideband chrominance receivers 191
distortion 409 Wideband equiband operation 191,
effects 383 210, 227
Transient response, effect of gamma Widget'268, 437
correction 408
Transmission
early methods 79 X-rays 313, 368
signals 106 XZ detection 241-48
Transmitter axes 242, 252
block diagram 156 calculation 248-55
coding 78-116 circuits 242, 246
response specifications 101 matrix 242, 246, 247
tolerances 114
Trichromatic units 32
Trichromats 10 Y' luminance signal 118-21
r

Colour Television NTSC Carnt Wireless World 1961

Uploaded by

Copyright:

Available Formats

Colour Television NTSC Carnt Wireless World 1961

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Colour Television NTSC Carnt Wireless World 1961

Uploaded by

Copyright:

Available Formats

Colour Television

N.T.S.C. SYSTEM, PRINCIPLES AND PRACTICE

P. S. CARNT, B.Sc.(Eng.), A.C.G.I., A.M.LE.E. and

LONDON ILIFFE BOOKS LTD

© P. S. Carnt and G. B. Townsend, 1961

About this book - Introduction - Why N.T.S.C.? - Outline of

3 COLOUR PICTURE TUBES, CAMERAS AND FILM

APPENDIX 7 QUADRICORRELATOR EQUIVALENT

The Hazeltine Research Corporation

The Institute of Radio Engineers

Marconi W.T. Ltd.

Murphy Radio Ltd.

John Wiley and Sons Inc.

black and white brightness picture. The resulting effect is similar

Fig. 1.2. Sub -carrier in vision band

HALF SILVERED MIRROR

HALF SILVERED MIRROR

Fig. 1.3. Three -tube colour display

a special colour synchronizing signal during the back porch of the

1.10. The acuity of the human eye

to a plain white background will then produce an apparent image

BRIGHT A SHEEN RED

NEUTRAL FILTERS TRANSPARENT FILTER

Fig. 1.4. Sharp brightness change superimposed on an

WHITE LIGHT VIEWED

lines. In American nomenclature such a half picture of 2024 lines

Fig. 1.6. Phase for a sine wave oscillation

sin 0 = -12-C cos 0 = 9--D tan 0 = 13C

Fig. 1.7. Phase response curves

Fig. 1.8. Fourier sine wave components of a repetitive waveform

in the same way as the amplitude response of the channel. Some-

The group delay at an angular frequency w is the quantity which

1.19. Fourier components

1.20. Distortionless transmission

adjacent to the ends of the bands, with a corresponding distortion

Fig. 1.9. Frequency characteristics for a distortionless transmission

particular scene being transmitted. Usually there is a marked

sufficient correlation between line signals for those 25 c/s harmonics

kAAA 4A AAA Adkv

r,f, (n+ I) ft.

Fig. 1.11. Frequency spectrum for a picture which is changing (Reproduced

This is the most noticeable quality of colour. The light from a

The third quality of colour, saturation, is the variation in depth

700M/.1 600mµ 500mµ 400 My

FREQUENCY. MC/5. I 102 104 106 108 106) le 1014 1016

WAVELENGTH, 300m 3m 3CM 0.3 mm 3.I0 8M 3.108M 3.16110 3.ici'in 3.16'm

Fig. 2.1. The spectrum of electromagnetic radiation

The colours mentioned in Fig. 2.1 are called spectrum colours

BLUE GREEN YELLOW RED

Fig. 2.2. Spectral energy distribution of lights

also be produced by mixing amounts of the red, green and blue

Fig. 2.3. The sensation of brown. (a) has a black surround

Fig. 2.4. Photometer. (a) shows two types of visual field

Since light at a wavelength of 555 MIL appears brighter per watt

BLUE GREEN YELLOW RED

Fig. 2.5. The curve of relative luminous efficiency (VA) for

curve varies from one human to another, so an average curve called

2.6.2. THE LUMEN

Luminous flux is the name given to the total quantity of light

R-FG-PB C_-= R+G±BRH-R-FG-I-BG+R-FG-FBB

represent the green reference stimulus G. A point such as C

Similarly, TS meets RG at P, through which all the lines of

1(Z) = k7R k8G* k9B*

Fig. 2.13 shows a set of 3 such primaries, X, Y and Z*, on a

R, G and B* are measured in T -units, so that for a given reference