Skip to main content

Full text of "BBC Monograph Number 53"

See other formats


BBC 



ENGINEERING DIVISION 



MONOGRAPH 



NUMBER 53: JULY 1964 



Aerial distribution systems for receiving 
stations in the l.f., m.f., and h.f. bands 

by 



J. B. IZATT, Ph.D., B.Sc.(Eng.) 

{Researeh Department, BBC Engineering Division) 



BRITISHBROADCA STING CORPORATION 

PRICE FIVE SHILLINGS 




BBC ENGINEERING MONOGRAPH 

No. 53 

AERIAL DISTRIBUTION SYSTEMS FOR RECEIVING 
STATIONS IN THE L.F., M.F., AND H.F. BANDS 

by 

J. B. Izatt, Ph.D., B.Sc.(Eng.) 

CResearch Department, BBC EngJneeriTig Division) 



JULY 1964 



BRITISH BROADCASTING CORPORATION 



FOREWORD 

THIS is one of a series of Engineering Monographs 
published by the British Broadcasting Corporation. 
About six are produced every year, each deahng 
with a technical subject within the field of television and 
sound broadcasting. Each Monograph describes work 
that has been done by the Engineering Division of the 
BBC and includes, where appropriate, a survey of eariier 
work on the same subject. From time to time the series 
may include selected reprints of articles by BBC authors 
that have appeared in technical journals. Papers dealing 
with general engineering developments in broadcasting 
may also be included occasionally. 

This series should be of interest and value to engineers 
engaged in the fields of broadcasting and of telecom- 
munications generally. 

Individual copies cost 5s. post free, while the annual 
subscription is £1 post free. Orders can be placed with 
newsagents and booksellers, or bbc publications, 35 

MARYLEBONE HIGH STREET, LONDON, W.]. 



CONTENTS 



Section Title 

PREVIOUS ISSUES IN THIS SERIES 

SUMMARY ..... 

1. INTRODUCTION .... 

2. DISTRIBUTION SYSTEMS 

3. DISTORTION IN VALVES AND TRANSISTORS 

3.1 Distortion in Valves 

3.1.1 General Comments 

3. 1.2 A Figure of Merit for Low Distortion in Valves 

3.1.3 Methods of Reducing the Intermodulation Products 

3.2 Distortion in Transistors . 

4. WIDE-BAND AMPLIFIERS 

4.1 The Two-valve Cascade Amplifier 

4.1.1 The Circuit 

4.1.2 Gain, Bandwidth, and Noise Factor 

4.1.3 Intermodulation Products 

4.1 .4 The Effects of Component Tolerances 

4.2 Push-pull Amplifier Arrangements . 

5. OCTAVE AMPLIFIERS. 

5.1 The Wide-band Amplifier with External Filters 

5.2 The 'Integrated' Octave Amplifier 



6. CONCLUSIONS 

7. ACKNOWLEDGMENTS 

8. REFERENCES 
APPENDIX I 
APPENDIX II 



Page 

4 

5 

5 

5 

6 
6 
6 

7 



10 
10 
10 
10 
10 
12 
12 

14 

14 
15 

17 

17 

17 

18 

19 



PREVIOUS ISSUES IN THIS SERIES 



No. Title 

1 . The Suppressed Frame System ofTelerecord'mg 

2. Absolute Measurements in Magnetic Recording 

3. The Visibility of Noise in Television 

4. The Design of a Ribbon Type Pressure-gradient Microphone for Broadcast Transmission 

5. Reproducing Equipment for Fine-groove Records 

6. A V.H.F.JU.H.F. Field-strength Recording Receiver using Post-detector Selectivity 

7. The Design of a High Quality Commentator's Microphone Insensitive to Ambient Noise 

8 . An Automatic Integrator for Determining the Mean Spherical Response of Loudspeakers and Microph 

9. The Application of Phase-coherent Detection and Correlation Methods to Room Acoustics 

10. An Automatic System for Synchronizing Sound on Quarter-inch Magnetic Tape with Action on 

35-mm Cinematograph Film 

11. Engineering Training in the BBC 

12. An Improved * Roving Eye' 

1 3. The BBC Riverside Television Studios: The Architectural Aspects 

14. The BBC Riverside Television Studios: Some Aspects of Technical Planning and Equipment 

15. New Equipment and Methods for the Evaluation of the Performance of Lenses of Television 

1 6. Analysis and Measurement of Programme Levels 

17. The Design of a Linear Phase-shift Low-pass Filter 

1 8. The BBC Colour Television Tests: An Appraisal of Results 

19. A U.H.F. Television Link for Outside Broadcasts 

20. The BBC's Mark II Mobile Studio and Control Room for the Sound Broadcasting Service 

21. Two New BBC Transparencies for Testing Television Camera Channels (Out of Print) 

22. The Engineering Facilities of the BBC Monitoring Service 

23. The Crystal Palace Band I Television Transmitting Aerial 

24. The Measurement of Random Noise in the presence of a Television Signal 

25. A Quality-checking Receiver for V.H.F. P.M. Sound Broadcasting 

26. Transistor Amplifiers for Sound Broadcasting 

27. The Equipment of the BBC Television Film Studios at Ealing 

28. Programme Switching, Control, and Monitoring in Sound Broadcasting 

29. A Summary of the Present Position of Stereophonic Broadcasting 

30. Film Processing and After-processing Treatment of\6-mm Films 

3 1 . The Power Gain of Multi-tiered V.H.F. Transmitting Aerials 

32. A New Survey of the BBC Experimental Colour Transmissions 

33. Sensitomeiric Control in Film Making 

34. A Mobile Laboratory for UHF and VHP Television Surveys 

35. Tables of Horizontal Radiation Patterns of Dipoles Mounted on Cylinders 

36. Some Aspects of Optical Lens Performance 

37. An Instrument for Measuring Television Signal-to-noise Ratio 

38. Operational Research on Microphone and Studio Techniques in Stereophony 

39. Twenty-five Years of BBC Television 

40. The Broadcasting of Music in Television 

41. The Design of a Group of Plug-in Television Studio Amplifiers 

42. Apparatus for Television and Sound Relay Stations 

43. Propagational Factors in Short-wave Broadcasting 

44. A Band V Signal-frequency and a Correlation Detector for a VHP/ UHF 

Field-strength Recording Receiver 

45. Vertical Resolution and Line Broadening 

46. The Application of Transistors to Sound Broadcasting 

47. Vertical Aperture Correction using Continuously Variable Ultrasonic Delay Lines 

48. The Development of BBC Internal Telecommunications 

49. Apparatus for Measurement of Non-linear Distortion as a Continuous Function of Frequency 

50. New Methods of Lens Testing and Measurement 

5 1 . Radiophonics in the BBC 

52. Stereophony: the effect of cross-talk between left and right channels 



Date 



JUNE 


1955 


SEPTEMBER 


1955 


OCTOBER 


1955 


DECENIBER 


1955 


FEBRUARY 


1956 


APRIL 


1956 


JUNE 


1956 


ones AUGUST 


1956 


NOVEMBER 


1956 


JANUARY 


1957 


MARCH 


1957 


APRIL 


1957 


JULY 


1957 


OCTOBER 


1957 


DECEMBER 


1957 


MARCH 


1958 


APRIL 


1958 


MAY 


1958 


JUNE 


1958 


AUGUST 


1958 


NOVEMBEJt 


1958 


JANUARY 


1959 


FEBRUARY 


1959 


MARCH 


1959 


JUNE 


1959 


AUGUST 


1959 


JANUARY 


1960 


FEBRUARY 


1960 


APRIL 


1960 


MAY 


1960 


JULY 


1960 


OCTOBER 


1960 


DECEMBER 


1960 


FEBRUARY 


1961 


FEBRUARY 


1961 


APRIL 


1961 


JUNE 


1961 


SEPTEMBER 


1961 


OCTOBER 


1961 


FEBRUARY 


1962 


APRIL 


1962 


JULY 


1962 


AUGUST 


1962 


OCIUBER 


1962 


DECEMBER 


1962 


FEBRUARY 


1963 


MAY 


1963 


MAY 


1963 


JULY 


1963 


SEPTEMBER 


1963 


NOVEMBER 


1963 


MARCH 


1964 



AERIAL DISTRIBUTION SYSTEMS FOR RECEIVING STATIONS IN 
THE L.F., M.F., AND H.F. BANDS 

SUMMARY 

The factors affecting the design of distribution systems for monitoring services or other receiving stations operating in the 
low-, medium-, and high-frequency bands (such as those for relaying transmissions) are considered and a number of ex- 
perimental amplifiers is described. The intermodulation products generated in the amplifiers are at a very low level, and 
the methods of achieving the linearity required are discussed. 



1. Introduction 

It has been found convenient at monitoring stations to 
supply many receivers from one aerial rather than to pro- 
vide one aerial for each receiver, because aerials for the 
range 100 kc/s to 30 Mc/s are very large physically. 
Methods of distributing the signals from the aerials to the 
receivers are considered in this report and, in particular, 
possible types of distribution amplifier are examined in 
some detail. Except for certain arrangements employing 
octave amplifiers, the only amplifiers considered are those 
which provide the full frequency range at each outlet. 

If the number of receivers is not too great, passive distri- 
bution systems have many advantages. Systems of this 
type are considered in Section 2. On the other hand, as the 
number of receivers is increased, amplifiers become neces- 
sary in order to overcome the distribution loss. The ampli- 
fiers may be either wide-band or restricted -band depend- 
ing on the type of aerial,* and both types are considered. 
Several factors govern the choice of active element em- 
ployed for amplification, and consideration of the merits 
of valves and transistors is given in Section 3. 

When amplifiers are used at monitoring stations where 
very weak signals mnst be received, or at relay stations 
where good reception is required in the presence of strong 
signals from nearby transmitters, the intermodulation 
products must be at a very low level. The discussion of 
amplifier performance will therefore be mainly concerned 
with this aspect. Experimental work on wide-band and 
octave amplifiers is described in Sections 4 and 5. 

2. Distribution Systems 

Because the available signal power from the aerial must 
be shared among a number of receivers, the use of passive 
distribution systems must result in some insertion loss 
between the aerial and any particular receiver. Further 
loss may also arise in the resistors which are included for 
matching purposes and in transformers. The overall effec- 
tive noise factor when a receiver is connected to an aerial 
depends not only on the insertion loss but also on the 

* In the past, the aerials have generally been either of restricted 
bandwidth (e.g. one octave) and omnidirectional or wide-band and 
directional. 



aerial-noise factor^ F* and the receiver-noise factor F,. 
The case of a single receiver connected directly to the 
aerial with zero insertion loss is taken as a reference. Fig. 1 
then shows the loss Z.; that may be permitted between aerial 
and receiver if the degradation is not to exceed 1 -5 dB and 
Fig. 2 indicates the permitted loss for a degradation of 3 
dB. 

Median values of f^have been tabulated,^ and the design 
of a system can make allowance for the variation in F^ 
found inpractice. For example, if i^„ is less than ISdBfora 
negligible fraction ofthetime,and the receiver-noise factor 
is 5 dB, it can be seen from Fig. 1 that the insertion loss 
may be as much as 7 dB without causing a degradation 
exceeding 1 • 5 dB. For an 'ideal' distribution system with 
no inherent losses, the use of five receivers fed from one 
aerial would be permissible (provided that the receivers 
have the necessary reserve of gain). 

Although a number of forms of distribution system is 
possible, only two are in widespread use. Fig. 3 (a) shows 
one of these constructed from hybrid transformers. If the 
transformers are lossless, the insertion loss of each hybrid 
is 3 dB. In practice, the insertion loss per hybrid is about 
3-5 dB and a minimum isolation of 20 dB can be main- 
tained over the whole band. The hybrid system is most 
useful where the mmiber of receivers is small (say two or 
four) but becomes complicated and expensive for a larger 
number of receivers, for which the transformer system 
shown in Fig. 3 (b) is more suitable. In the latter system, 
with m outlets, the insertion loss between the aerial and 
any one receiver is 

101ogio(2/n-l)dB (I) 

and the isolation between receivers is 

20 logio(2m - \) dB (2) 

Expressions (1) and (2) have both been derived assuming 
matched conditions throughout but, although this is not 
realized with practical receivers, the effects of receiver 
mismatch are not great. If, for example, m \ outlets are 
short-circuited, the drop in signal level at the remaining 
outlet cannot exceed 3 • 5 dB. From equation (1) it may be 

*Fa is the noise power available from the aerial relative to the tlier- 
nial noise power that would be available from the aerial if its tem- 
perature and the radiation temperature of the surroundings were 
both 288°K. 



£6 

ffl 

■D 
^-24 

S 
1 22 

E 

tn 
11)20 



?18 



1.16 
p 



'M 



12 



o 
c 10- 



1. _ 



6 

4) 

S 4 

ui 

E 





























































/ 




























A 


A 


























/ 


/ 


Y 
























/ 


/ 


/ 


J 




















c 


/ 


/ 


/ 


J 


/ 




















// 


/ 


/ 


v 


















/ 


A 




'- 


/ 


A 


















y 


/" 


^ 




A 


y 


















y 


/ 


y 




y 


c 
















^ 


A 


Y 


y 


*^ 


y 
















^ 


d^ 
























•— 1 



























10 12 



16 IB 20 22 24 26 28 30 



aerial noise factor Fa dS 
Fig. I — Permissible insertion loss £, between aerial and receiver for a degradation of\-5dB 

O Receiver-noise factor F„ 2 dB 
X Receiver-noise factor F„ S dB 
• Receiver-noise factor F„ 10 dB 



deduced, for example, that three receivers could be used 
for an insertion loss of 7 dB. As already indicated, the 
degradation in signal-to-noise ratio would not then ex- 
ceed 1 ■ 5 dB if the receiver-noise factor is 5 dB and F^ is 
15 dB. In noisy situations the minimum value of F^ may 
well be higher, say 25 dB, There is then a moderate in- 
crease in the permissible loss and this allows a large in- 
crease in the number of receivers. Tlius, for a minimum 
value of F« of 25 dB, an insertion loss of 16 dB is per- 
missible and twenty receivers may be used. 

Where the tota 1 in sertion loss including the tran smi ss ion- 
line losses exceeds the allowable value, an amplifier must 
be placed before the distribution network. Assuming that 
the power gain A of the distribution amplifier is equal to 
the succeeding loss. Fig. 1 or 2 may be used (as before) to 
find the permissible transmisslon-iine loss between aerial 
and distribution amplifier if a correction term k is sub- 
tracted from the loss in decibels as indicated by the graph, 
where 



/:-101og.(l+f^-l.)dB 



(3) 



and Fi is the noise factor of the distribution amplifier. 

Although distribution amplifiers can improve the over- 
all noise factor and sensitivity of a receiving system, they 
do introduce intermodulation products of the various 
input signals. This fact is the dominating problem in the 
design of such amplifiers. When a large number of re- 



ceivers has to be supplied it may be difficult to design a 
single distribution amplifier with sufiiciently low inter- 
modulation products. Improved performance may then 
be obtained by using a small number of similar amplifiers 
supplied from the aerial, with each amplifier feeding a 
small number of receivers. 



3. Distortion in Valves and Transistors 

3. 1 Distortion in Valves 
3.1.1 General Commen ts 

The transfer function of a valve is inherently non-linear, 
and this effect was found to be the principal source* of 
distortion in the amplifier examined. One method of 
assessing non-linearity is to measure the harmonics 
generated in the valve when the input is a pure sine-wave 
but there is some difficulty in generating a sufficiently pure 
wave at frequencies above 100 kc/s. A more convenient 
method is to apply at the input terminals two e.ra.f.s of 
different frequencies /j and /a and to measure the inter- 
modulation products (i.p.s). Neither generator need have 
a very pure waveform for measurements of the second- 
order i.p.s, viz,/i dzf2, but for measurements of the third- 
order i.p.s, viz. 2/i i/a, 2/2 zt/i, either one generator 
must have little or no second harmonic or there must be 

* Hysteresis loss in the output transformer Juigtit be expected to 
cause distortion but with modern ferrite cores of high resistivity this 
was not found to be so. 




10 12 14 16 18 £0 
aerial noFse factor Fg^ dB 



22 24 26 26 30 



Fig- 2 — Permissible insertion loss £, between aerial and receiver for a degradation of7> dB 

O Receiver-noise factor F„ 2dB 
X Receiver-noise factor F„ 5 dB 
• Receiver-noise factor F„ 10 dB 



little or no second-order non-linearity in the equipment 
under test. Witti these mild restrictions, measurements of 
the second- and third-order i.p.s are straightforward: 
higher order i.p.s are generally negligible by comparison 
and have not been considered. 

Let /q be the mean anode current in the valve, /j be the 
peak current swing due to either of two equal inputs at 
frequencies /i or /a, /jj, be the (peak) amplitude of either 
second-order i.p., and let 7^, be the (peak) amplitude of 
any of the third-order i.p.s. Then to a first approximation : 



^B2 
1 1 



a' la 



and 






(4) 



(5) 



where a and ^ are constants. If the valve were an ideal 
space-charge-limited device obeying the three-halves 
power law,* a would be 6 and /9 would be 72. Fig. 4 shows 
some measured second-order i.p.s for three valves and it 
can be seen that a = 6 is an excellent approximation. Re- 
sults for the third-order i.p.s showed much greater varia- 



tion and iS was usually less than 12. It was rarely less than 
20, however, and the latter value may be used to estimate 
the order of magnitude of the third-order i.p.s. 

3.1.2 A Figure of Merit for La w D ist onion in Valves 

The second-order i.p.s are usually the more important, 
and a figure of merit may be constructed which will de- 
scribe the suitability of a valve for the present application. 
Assuming that the output stage is transformer-coupled to 
theload, it can be seen from equation (4) that by increasing 
the step-down ratio of the transformer so that the peak 
output-current swing for a given load power is reduced, 
the second-order i.p.s are reduced. The transformer ratio 
is limited by the shunting effect of stray capacitance at 
high frequencies, however, and thus the second-order 
i.p.s are proportional to the square root of the valve 
output capacitance, Co- Also from equation (4) the level 
of second-order i.p.s is inversely proportional to /^ and 
hence a simple figure of merit would be 



VCo 



If, however, negative feedback is applied in the manner 
described in Sections 4 and 5, the feedback is roughly 



proportional to the mutual conductance, g„, and a better 
expression for the figure of merit is 



VC, 



(6) 



Applying expression (6) to some particular valves yields 
42 for type EF80, 124 for type E180F, 865 for type E8I0F, 
and 882 for type E55L. It can be seen that the E810F and 
E55L are much better than the others and, in fact, no other 
valves were found to be comparable. 

3.1,3 Methods of Reducing the Intermodulalion 
Products 
Once the best valve has been found, the effects of valve 
non-linearity may be reduced further by suitable circuit 
design. The three techniques of most interest are (1) the 
octave principle, (2) cancellation, and (3) negative feed- 
back. 



In the octave principle, the bandwidth of the amplifier is 
deliberately restricted to one octave both at the input and 
at the output. Thus all harmonics and both second-order 
i.p.s fall outside the band and the only i.p.s of any in- 
terest are the third and higher orders. As these are generally 
at a much lower level, the apparent linearity of the ampli- 
fier is greatly improved. The technique is most useful with 
restricted-bandwidth aerials but could, in principle, be 
used to provide a wide- band amplifier by paralleling suit- 
able combinations of the different octave amplifiers. 

For a simple two- valve wide-band amplifier, the seco ad- 
order i.p.s may be cancelled either by using a push-pull 
stage or by using two stages in cascade in which the con- 
tribution from the first stage equals the contribution from 
the second stage (when allowance has been made for the 
difference in signal level between the two stages). The main 
disadvantage of both systems is the difficulty of main- 
taining adequate cancellation throughout the life of the 



hybrid networks 





Fig. 3 — Two possible forms of distribution network 
8 



-0 






































§-ic 












































































^-20 




























































^^ 


^ 


' 










o 

■o 

^ -30 

L 

LI 






















x*^,^ 
^0.^-'^ 


































cvi^^^ 




























o 


V 




^ 


> 



























*• .^ 


^ 


.^ 
































u. 


^ 
































^ 50 






































> 












































































-7d-^ 







































0-01 



0-1 

f1 



10 



f ii'g^. 4 — ■ Measurements of second-order intermodulation products in valves 

O E810F No. 1 X E810F No. 2 
• EF80 7i/6/o 



amplifier. The push-pull amplifier is generally easier to 
balance and also lias the very minor advantage that all 
other even-order i.p.s are cancelled. The cascade circuit 
suffers from the additional disadvantage that cancellation 
can never be as good at the high frequencies owing to the 
phase shift in the interstage coupling, although this is not 
very important in practice because the largest interfering 
signals occur at frequencies below 2 Mc/s. The principal 
advantage of the cascade circuit for a two-valve amplifier 
is the much greater gain-bandwidth product which means 
that greater use may be made of negative feedback. It will 
Ik found that a push-pull circuit is best for low gains (10 
to 15 dB), but that a cascade circuit is best for high gains 
(20 to 30 dB). 

The overall bandwidth of conventional feedback ampli- 
fiers is always greater than the loop bandwidth and there- 
fore the feedback provides no protection against distortion 
at the highest frequencies. This difficulty can be overcome 
in the type of circuit shown in Fig. 5 where the feedback 
makes the output current proportional to input voltage;^ 
by using a sufficiently large anode load resistor, subject to 
the time constant giving sufficient bandwidth, the overall 
bandwidth can be made less than the loop bandwidth. The 
distortion is then controlled at all frequencies in the pass- 
band. It is generally considered advantageous to apply the 
feedback round more than one stage because a greater loop 



gain can be achieved^ but it may be shown that, when 
overall bandwidths of about 30 Mc/s are involved, the 
loop cut-off frequency is about 1 20 Mc/s; the excess phase 
shifts due to transit time and the physical length of the 
loop may then cause instability. This can be overcome only 
by reducing the loop gain but it was found that the loss in 
performance is so great that more feedback can be applied 
in a single stage by omission of the decoupling capacitor 
across the cathode resistor. 

3.2 Distortion in Transistors 

Transistors have become available which have the 
required gain-bandwidth product, and some measure- 




Fig. 5 — A possible form of wide-band negative feedback 
amplifier 



ments were carried out to assess the level of distortion. 
These measurements showed that, with suitable circuit 
design, the i.p.s will be roughly equal to those of a valve 
with the same standing current and working into the same 
load impedance. Thus, in order to give a performance 
comparable with the E810F, the transistor will require to 
pass 35 mA and to have an output capacitance of about 
3 pF. With a bad impedance of 400 ohms this implies a 
maximum collector-to-emitter voltage of about 30V and a 
continuous dissipation of half a watt, while the voltage 
gain -bandwidth product will require to be greater than 
about 150 Mc/s. At present, only provisional specifica- 
tions of transistors approaching these requirements have 
been issued. It therefore seems unlikely that there will be 
any significant advantage in using transistors until further 
improvements in performance have been made. 

4. Wide-band Amplifiers 

A two-stage wide-band amplifier was constructed in 
order to obtain an idea of the practical performance 
attainable with modern valves. The frequency range was 
taken as 100 kc/s to 30 Mc/s and the maximum gain as 34 
dB with the input matched to 100 ohms and the load 
impedance equal to 100 ohms. The details are given in 
Section 4.1 and some circuits with improved performance 
are considered in Section 4.2. 

4.1 The Two-valve Cascade Amplifier 
4.1.1 The Circuit 

Fig. 6 shows the basic circuit used. The step-up ratio of 
transformer T^ is determined by the stray capacitance to 
ground at the grid of V^, R^ being chosen to provide a 
match at the input. Likewise, R^ is governed by the stray 
capacitances at the anode of V^ and the grid of V^ while 
the ratio of transformer T^ is determined by the stray 
capacitance at the anode of Fj. Cathode negative-feed- 
back is provided by R2 and R^, these being adjusted to 
meet two requirements. First, the total feedback for the 
two stages must allow the required overall gain to be 
obtained. Second, more feedback is applied to V^ than to 
Fj so that, in spite of the greater signal current in V2, the 
amount of second-order distortion is the same in each 
valve; under these circumstances there is nominal can- 
cellation of the second-order i.p.s in the complete ampli- 
fier because they are of opposite phases in the two stages. 

A full circuit diagram of an experimental amplifier using 
E8 lOF valves is given in Fig. 7. It is true that the E55L has a 




output 



Fig. 6 — Basic diagram of the cascade wide-band amplifier 



higher figure of merit (Section 3. 1 .2) but by a small margin 
only and the much greater h.t. consumption was thought 
to be undesirable. The peaking inductance in the anode 
circuit of V^ has been added to improve the performance 
at the higher frequencies. Large (decoupled) resistors are 
used in the cathode circuits to stabilize the working points 
of the valves and common screen- and control-grid po- 
tentials are provided. For practical reasons the suppressor 
grids are not at the same potential but this is unimportiint 
because the suppressor grid base is large. 

4.1.2 Gain, Bandwidth, and Noise Factor 

Fig. 8 shows the gain/frequency response of the ampli- 
fierforthe three combinations of feedback resistors shown 
in Fig. 7. When the feedback resistors were changed the 
peaking coil was readjusted in order to maintain the gain 
up to 30 Mc/s. This compensated for the different effects 
of stray capacitances at the different gains. 

The noise factor was 7 dB at frequencies below 5 Mc/s 
and rose gradually above this frequency to about 9-5 dB 
at 30 Mc/s. This performance is considered adequate. 

4.1.3 Intermodulation Products 

The non-linearity of the amplifier was tested by applying 
two inputs of equal amplitude but different frequencies, 
the resulting i.p.s being measured. 

The third-order i.p.s showed little spread when different 
valves were used and httle variation with frequency ; Fig. 9 
gives, for a typical case, the measured levels of these pro- 
ducts (relative to one of the wanted signals) as a function of 
the signal output level. The differences between the curves 
are less than the total changes in feedback for the three 
condttions. This is because the feedback has to be changed 
by more in the first valve than in the second valve to main- 
tain the condition for cancellation of the second-order 
i.p.s. The third-order i.p.s are produced mainly in the 
output stage, but the phases of these terras are such that 
the smaller contribution of the first stage is always additive. 

The second-order i.p.s showed, at low frequencies, a 
much greater variation with changes in valve characteris- 
tics as their level depends on accurate cancellation of the 
contributions from the two stages. At higher frequencies, 
these changes are masked by the deterioration caused by 
a phase shift between the stages, as mentioned in Section 
3.1.3. 

It is operationally an advantage to use unmatched valves 
in the circuit and an attempt was made to measure the 
effect of random selection of the valves. Eight valves were 
available and, since there are two possible positions for 
each valve, there was a total of fifty-six different combina- 
tions. Frequencies were chosen at which the cancellation 
ought to have been good and the level of second-order 
i.p.s was measured. Fig. 10 shows a histogram of the 
results. For the conditions of the test the cancellation 
varies from about 6 dB for the worst arrangements to 25 
dB for the best arrangements. This is quite a wide range 
but it can be seen that most of the results are below —62 
dB corresponding to a cancellation of 10 dB, which is a 



10 




520< >27 



40r 



30- 



20- 



10- 



001 



/7m 



0-001 



. V 



[001 









33 



rr\ 



;iook >Ra 



!o-5 



§ 



I , o ^ 



2:1 



10-01 



2-2H^ 2 2k< 
!o-oi "f" ' 



I 



Fig. 7 — Practical circuit of the cascade wide-band amplifier 
Overall gain dB ■ ' , * 



34 


10 


90 


28 


27 


120 


18 


82 


180 

















1 

\ 




1 


^/"^ 






I 


-i — o—c 


ill 




i 




( — ^!< j:- 


■) ; 3 


' ' 


— -;; — J<^^ 


^-^;T 




! 












1 






\'^ 




*" 






u-*^\\ 


























I 
















J 



















0-1 



1-0 
frequency, Mc/s 



10 



100 



F(|l'. 8 — Frequency response of the wide-band amplifier of Fig. 1 

O Feedback resistors \0 ohms, ^ ohms X Feedback resistors, T^ ohms, \2Q ohms 

• Feedback resistors 82 ohms, 1 80 o/ims 



11 



-M 


























- 


- 






-40 
























/ />■ 




























y/ 


\ 










-50 






















/y 


^ 






— 


- 


- 












; 








^ 


<^^ 








-60 










; 








yX 




























x^ 


1^ 


-^y 


1 






















X 












- 












y. 


^ 


y 














- 




-80 






Xy(y 


y 




















^ 


^ 


— 


— 
















'~"\ — 




-90 




■^*^ 


- 


- 
















i 


o^ 




















Kin 



























'01 1 

output voltage 

¥i^. 9 — Third-order intermodulation products in the amplifier of fig. 7 
O Gain = 34 rfS X Gain =2SdB • Gain ^\%dB 



10 



reasonable performance although it may not be adequate 
for some applications. 

4.1.4 The Ejfects of Componen t Tolerances 

Although the feedback (d.c. and a.c.) stabilizes most of 
the properties of the amplifier, the degree of cancellation is 
vulnerable to changes in certain components. Therefore, 
2 per cent tolerance resistors were specified for the a.c. 
feedback resistors and for the anode load of V^ but, be- 
cause of the power rating, the d.c. feedback resistors had 
to be wire-wound types on which the tolerance is 5 per 



12 

■^ 11 

c 10 
> ^^ 

□J :5 y 

in O ' 
■" 4-' 

o>| 5 

P "- 
L " 4 

° c 

■s7 3 




-66 -60 -64 -68 -72 -76 "60 

level ot second order intermodnlation products, dB 

Fig. 10 — Histogram showing the effect of random selection 
of valves 



cent. Simultaneously increasing the 1 -2 kn resistor (Fig. 
7) in Vi by 5 per cent and reducing both 2-2 kn resistors 
in Fj by 5 per cent changed the second-order i.p.s by a 
maximum of 3 dB when the cancellation was very good, 
and, as this represents an unlikely extreme condition, the 
effect may be neglected. 

Variations of ± 1 5 per cent in the h.t. voltage produced 
about 1 dB increase in the second-order i.p.s and so the 
circuit may be run from an unstabilized supply. A 15 per 
cent increase in heater voltage produced a similar increase 
of I dB but reduction of the heater voltage had a more 
drastic effect. A reduction of 5 per cent caused no change 
but the second-order i.p.s rose rapidly as the heater volt- 
age was further reduced, increasing by about 4 dB for a 
10 per cent reduction and by 8 dB for a 1 5 per cent reduc- 
tion. Operation of valves with these reduced voltages 
seriously affects the expected life and should therefore be 
avoided. The makers recommend that the heater voltage 
be within ±5 per cent of the nominal value. 

4.2 Push-pull Amplifier Arrangements 

The principal limitation in the performance of the am- 
plifier described in Section 4.1 is the level of the second- 
order i.p.s and, although some improvement can be ob- 
tained by using a 'balance potentiometer' in order to vary 
the control-grid potential of one of the valves, greater 
cancellation is obtained over only a limited range of fre- 
quencies and signal levels. 

On the other hand, when a push-pull stage is used, the 
cancellation should remain constant over a wide range of 
signal levels and frequencies. Fig. 11 shows the circuit of 
an amplifier with a gain of 16 dB (assuming no loss in the 
transformer) in which there is 1 2 dB of feedback. This 
push-pull amplifier may be compared with a cascade am- 



12 



plilier of similar gain in respect of second-order i.p.s as 
follows. For completeness, comparison is also made with 
a four-valve amplifier described later. 



tained by combining the cascade and push-pull circuits as 
shown in Fig, 12. In this circuit the third-order i.p.s are 
minimized, the condition for this being that the first and 



TABLE I 
Contribution to reduction of second-order i.p.s 





1 
Cascade \ Push-pull 

amplifier \ amplifier 


Four-valve 
amplifier 


Feedback 
Cancellation 

Use of push-puU circuit 
Total 


21 dB 
XidB 
OdB 
(21 ^ K,) dB 


12 dB 

3dB 

i]5 + K^)dB 


19 dB* 
KsdB 
3dB 

(22 + K^)dB 



* This figure is 4 dB greater than the first-stage feedback as explained in the text. 



Thus, to eq ual the performance of the cascade stage, the 
cancellation in the push-pull stage must be 6 dB better. 
In practice, values of A'^ of the order of 30 to 40 dB, are 
quite feasible, while for the cascade amplifier K^ may be 
only 6 to 25 dB as already mentioned. Therefore, a net 
improvement of the order of 15 dB is possible. 

Some further improvement in performance can be ob- 



final stages should produce equal relative third-order i.p.s. 
As indicated previously, they cannot be made to cancel 
and the condition simply allows the maximum benefit to 
be derived from feedback. For 16 dB gain this means that 
the feedback on the output stage is 23 dB and on the input 
stage is 1 5 dB. Considering the output stage, the feedback 
is 11 dB more than in the circuit shown in Fig. 11 but. 



2 0OV 
81mA 



>=Q 




Fig. 1 1 — Push-pull amplifier with 16 dB gain 
13 




Mk 



-VW- 






JWV- 




200 V 



/77A7 



e=< 



mm 



Fig. 11 — Four-valve distribution amplifier with 1 6 rf5 jai« 



when the equal-amplitude contribution from the input 
stage is included, the net improvement in the third-order 
i.p.s is only 5 dB. 

Because the ratio of signal currents has been arranged 
so that the third-order i.p.s are comparable in the input 
and output valves, the second-order i.p.s arise mainly in 
the input stage. The feedback for this stage is only 15 dB 
but there is a current gain of 4 dB between the input and 
output stages; the effective advantage is therefore 19 dB 
as shown in Table I above. The net improvement in the 
second-order i.p.s is seen to be 7 dB when compared with 
the simple push-pull amplifier, assuming an equal degree 
of cancellation {K^ = K^. There will, in addition, be up to 
about 6 dB of 'cascade cancellation' but this has been 
neglected to compensate for the difficulty in maintaining 
the same degree of push-pull cancellation. 

While some improvement in performance is obtained 
with the circuits given in Figs. 1 1 and 12, it should be noted 
that two balanced transformers are required. The full im- 
provement may not be realized in practice unless the bal- 
ance is reasonably good. Although techniques for wide- 
band transformers are well known,* the additional cost 
and possible difficulties of manufacture of these special 
components must be borne in mind. 

5. Octave Amplifiers 

It was noted in Section 3.1.3 that if the distribution 



amplifier bandwidth is restricted to one octave the second- 
order i.p.s are eliminated. This is a convenient bandwidth 
for a simple omnidirectional aerial, and systems have been 
constructed using this principle.^ In a typical case the fre- 
quency range of interest may be • 1 Mc/s to 27 Mc/s, this 
being divided into the following bands: 
0-1 to 0-2 Mc/s 



0-2 to 0-4 Mc/s 


O'Sto 1 


Mc/s 


1 to 2 


Mc/s 


2 to 4 


Mc/s 


4 to 8 


Mc/s 


8 to 16 


Mc/s 


16 to 27 


Mc/s 



It will be observed that the highest frequency-band is, for 
practical convenience, made smaller than an octave and 
that a gap occurs between ■ 4 and • 5 Mc/s, a band which 
is not used for broadcasting and contains the receiver 
intermediate frequency. 

There are two arrangements that may be adopted in an 
octave amplifier: suitable fillers may be placed before and 
after a wide-band amplifier, or the amplifier may be so 
designed that the filters form part of the amplifier circuit. 
Both arrangements were tried and the details are given in 
the following Sections. 

5.1 The Wide-band Amplifier with External Filters 
The wide-band amplifier described in Section 4.1 was 



14 



J^^/in^^ 




L5^ ^C6 



= C7 1 



the required response upon assembly. For the 8 to 16 
Mc/s octave, the terminating capacitor in each filter was 
significantly reduced in value in order to allow for the 
stray terminal-capacitance of the amplifier, but no diffi- 
culty was experienced in constructing the filter. For the 
1 6 to 27 Mc/s filter the stray capacitance is still low enough 
to permit the required circuit values to be obtained. In 
practice the residual reactances in the ampUfier output 
circuit produced rather a humped response but this was 
not considered serious ; Fig. 14 shows the overall response. 
The values of the components of the input and output 
filters for all eight octaves are given in Appendix I: no 
allowance has been made for stray reactances or for the 
amplifier terminal impedances. 



Fig. 13 — Input (a) and output (b) filters for use with the 
wide-band amplifier 



used with the input and output filters shown in Fig. 13. 
The feedback resistors were arranged to give a gain of 34 
dB, and the peaking inductance was removed. Each filter 
was designed initially as a low-pass, maximally flat filter, 
by means of the tables given by Weinberg.' The band-pass 
filter design was then obtained by resonating each element 
of the filter at a frequency equal to the geometric mean of 
the limiting frequencies of the required pass-band. 

Filters for two octaves were constructed : 0- 5 to 1 Mc/s 
and 8 to 16 Mc/s. For the 0-5 to 1 Mc/s octave the com- 
ponent values were such that it was merely necessary to 
adjust the inductances to the required values at 1 Mc/s and 
the capacitances to the required values at 1 kc/s to obtain 



5.2 The 'Integrated' Octave Amplifier 

While it is convenient in many applications to use a 
standard wide-band amplifier with suitable input and out- 
put filters as described in Section 5.1, a further improve- 
ment in performance can be obtained by making the filters 
part of the amplifier. There are two reasons for this. First, 
a higher impedance step-up ratio is possible with a band- 
pass filter following the output valve than is possible with 
wide-band low-pass coupling for the same maximum fre- 
quency ; hence the required current swing for a given out- 
put power is lower and the i.p.s are reduced according to 
equations (4) and (5). Second, there is an increase in gain 
which may be exchanged for more feedback which also 
reduces the non-linearity. 

An octave amplifier for the 8 to 16 Mc/s octave was 
constructed, the design of the input and output filters be- 
ing based on those described in Section 5.1, but arranged 
to include the necessary impedance transformation. The 



T3 



30 
25 
20 
15 
10 
5 

-5 
-10 
-15 
20 
25 
30 
35 

.in 






i 




i J 


r* 


^ 


r""'^ 


























/ 






^ 




















i 


i 


/ 






I 






















1 












\ 




































\ 




































\ 
































1 


























5 




1 






























/ 




1 






























/ 




1 






























/ 




































/ 












\ ^ 




















/ 














\ 1 


















/ 
























i 








/ 





























10 
frequency, Mc/s 



100 



Fig. 14 — Insertion gain of wide-band amplifier plus filters for the 8-16 Mc/s octave 

15 




017CV S2mA 



C12 L11 
211 p O-^lfi 

001 ' Tl"!^ 

LlOg j£llXci3 

75 6p 




Fig. ] 5 — TTie circuit of the ^integrated' octave amplifier. All capacitances are the total 
required values and make no allowance for stray capacitances 



inter-stage coupling was a simple three-section network 
designed to give a maximally flat bandwidth of 12 Mc/s 
which is more than adequate. The design does not repre- 
sent the best possible as the design effort was limited by the 
time available. It is, however, sufficiently near the opti- 
mnm to show the advantages to be gained by this type of 
circuit. 

Fig. 1 5 shows the circuit of the amplifier constructed for 
the 8 to 16 Mc/s octave and Fig. 16 shows a comparison of 
the third-order i.p.s for the two types of octave ampliiier. 
Also plotted in Fig. 16 is the curve for one of the existing 



octave amplifiers at Caversham.^ It can be seen that the 
wide-band amplifier (plus filters) is about 14 dB better 
than the existing Caversham amplifiers but is up to 25 dB 
short of the performance attained with the 'integrated' 
octave amplifier. There was, however, great diificulty in 
measuring the integrated amplifier because the levels 
approached the limit set by the measuring apparatus ; it is 
thought that the true curve would show the expected im- 
provement (26 dB) over the whole range. It should also be 
noted that the overload level is about 4 or 5 dB higher in 
the integrated amplifier and therefore use of the other 




10 
output voltage 

Fig. 16 — Measured third-order i.p.s in octave amplifiers 

Integrated octave amplifier X Wide-band amplifier plus fitters 

O Octave amplifier (Caversham) 

16 



arrangement involves a considerable sacrifice in perform- 
ance. 

The bandwidth to be amplified depends on the values of 
the inductances and associated capacitances, those for the 
8 to 1 6 Mc/s band being given in Fig. 15. The values for all 
octaves are given in Appendix II, together with the modi- 
fied circuits for the 16 to 27 Mc/s band. 

6. Conclusions 

Because of the world-wide increase in transmitter powers 
and the large number of new transmitting stations being 
built, the principal difficulty in the construction of distri- 
bution amplifiers arises from the need to avoid inter- 
modulation between the various signals. Passive distribu- 
tion systems avoid this possibility and in many situations 
the degradation in overall noise factor is well worth the 
freedom from intermodulation products, provided that the 
overall sensitivity is not Umited by lack of gain. Where it is 
not permissible to use a passive system, narrow-band 
amplifiers should be used if possible and the octave system 
is particularly valuable in this respect because it has the 
greatest bandwidth which may be used if the second-order 
i.p.s are to be eliminated. 

Where there is no alternative to a wide-band amplifier, 
the design may be based on the principles given in Section 
3, In general the second-order i.p.s are the most difficult 
to eliminate. The levels of these may, however, be pre- 
dicted with reasonable accuracy from the circuit diagram 
of the amplifier. 

In practice, the signals from the aerials will arise from a 
very large number of transmissions, but for test purposes 
it is sufficient to assume that there are only two interfering 
signals. As an example, the performance of the wide-band 
amplifiers has been computed when two 60 mV signals are 
applied. It is assumed that each amphfiier has a gain of 
16 dB and that the distribution loss following each ampli- 
fier is also 16 dB. Assuming that both of the interfering 
signals and the i.p.s of interest fall within the pass-band 
of the amplifier, Table II below shows the magnitude of 
the principal i.p.s at the input terminals of the monitoring 
receivers. Some spread about these values will occur in 
practice but the degree of variation should be about the 
same for each amplifier. 



A similar table has been drawn up for the octave amph- 
fiers based on measured results. As before, two 60 mV 
signals have been assumed to be present at the input and 
the amplifier gain has been taken as 30 dB (with a distri- 
bution loss of 30 dB). In this case only the third-order 
i.p.s are of interest; 

TABLE III 



Type of octave 
amplifier 


Fig. 

No. 


Level of third- 
order i.p.s 


Caversham octave 
Cascade plus 

filters 
Integrated 


Ref. 5 

7,13 
15 


105 

21 

3 



The signal levels in the last column of Table III above 
are much higher than those in the fourth column of 
Table Tl because the amplifier gain assumed is 14 dB 
greater. Distortion products at other signal levels may be 
estimated from the simple power laws given in equations 
(4) and (5), provided that the output voltage is less than 
about half the maximum possible output voltage. Above 
this level the distortion products rise more rapidly as can 
be seen, for example, iu Fig. 16. 

7. Acknowledgments 

The valuable help of Mr C. J. W. Hill, lately Engineer- 
in-charge of the monitoring station at Caversham, and of 
the present Engineer-in-charge, Mr F. Masterman, in the 
practical assessment of the performance of various opera- 
tional and experimental amplifiers is gratefully acknow- 
ledged. 

8. References 

1. Revision of Atmospheric Radio Noise Data, CCIR Report No. 322, 
International Teiecommimication Union, Geneva, 1963. 

2. Parker, P., Electronics, Edward Arnold (Publishers) Ltd, 1950. 

3. Thomason, J. G., Linear Feedbacic Analysis, Pergamon Press Ltd, 
1955. 



TABLE II 



Type of wide-band 
amplifier 


Fig. 
No. 


Level of second- 
order i.p.s 


Level of third- 
order i.p.s 


Degree of 

cancellation 

assumed 

dB 


Cascade 

Push-pull 

Four-valve 


1 
11 
12 


3-4 

M 
0'48 


0-34 
0-30 
017 


14 
30 
30 



Note: The levels of second and third harmonics generated from one 60 mV signal are obtained 
by dividing the voltages given for second- and third-order i.p.s by two and three respectively. 

17 



4. Maurice, D., and Minns, R. H., Very-Wide Band Radio-Fre~ 
quency Transformers, Wireless Engineer, 24, June 1947,285, p. 168, 
andJuly 1947, 286, p. 209. 

5. Hill, C. J. W., and Bishop, H. S., The Engineering Facilities of tlie 



BBC Monitorii^ Service, BBC Engineering Division Monograph 

No. 22, January 1959. 

6. Weinberg, L., Network Analysis and Synthesis, McGraw-Hill 
Book Co. Inc., 1962. 



APPENDIX I 

Component Values for the Octave Filters {Fig. 13) 

The values shown are the total values required : that is, no allowance has been made for stray reactances or 
the amplifier terminal impedances. 



Input Filter 



Frequency band 


L^ 


Cx 


L, 


c. 


L, 


c. 


L, 


c. 


Mcls 


^H 


PF 


f.H 


pF 


I.H 


pF 


^H 


pF 


O'lto 0-2 


108 


12100 


50' 1 


26100 


261 


5010 


121 


10800 


0'2to 0-4 


54-1 


6050 


25'1 


13100 


131 


2510 


60-5 


5410 


0-5to 1-0 


21-6 


2420 


100 


5230 


52'3 


1000 


24-2 


2170 


1-Oto 20 


10-8 


1210 


5-01 


2610 


26-1 


501 


121 


1080 


2-Oto 40 


5-4 


605 


2-51 


1310 


13-1 


251 


6-05 


541 


4-0 to 80 


2-70 


303 


1-25 


653 


6-53 


125 


3-02 


270 


8-0tol6-0 


1-35 


151 


0-626 


327 


3-27 


62-6 


1-51 


135 


16-0 to 27-0 


0-98 


60-7 


0-251 


238 


2-38 


25-1 


0-607 


98 



Output Filter 



Frequency band 

Mcjs 



0-1 to 
0-2 to 
0-5 to 

-Oto 

-Oto 

-Oto 
8-0 to 16-0 
16-0 to 27 



1- 
2- 
4- 



8-0 



L, 


c. 


U 


Q 


L, 


c, 


L, 


c. 


i.H 


pF 


i.H 


pF 


^H 


pF 


ixH 


pF 


62-5 


21000 


228 


5730 


81-9 


16000 


57-0 


23000 


31-2 


10500 


114 


2870 


41-0 


7990 


28-5 


11500 


12-5 


4190 


45-7 


1150 


16-4 


3200 


11-4 


4590 


625 


2100 


22-8 


573 


8-19 


1600 


5-70 


2300 


3-12 


1050 


11-4 


287 


4-10 


800 


2-85 


1150 


1-56 


524 


5-71 


143 


2-05 


400 


1-43 


574 


0-78 


262 


2-86 


71-6 


1-02 


200 


0-71 


287 


0-31 


191 


2-08 


28-8 


0-41 


145 


0-52 


115 



18 



APPENDIX II 

filter Component Values for the Integrated Octave Amplifier (Fig. 15) 
The values shown below are the total values required: no allowance has been made for stray reactances. 



input Circuit 



Frequency band 


L, 


Ci 


L, 


c. 


c. 


L, 


L, 


c. 


Mcjs 


^H 


pF 


,.H 


pF 


pF 


iJ,H 


t^H 


pF 


0-lto 0-2 


108 


18100 


50-1 


26100 


5010 


383 


827 


1080 


0-2to 0-4 


54- 1 


6050 


25-1 


13100 


2510 


191 


414 


541 


0-5to 1-0 


21-6 


2420 


10-0 


5230 


1000 


76-5 


165 


217 


lOto 20 


10-8 


1210 


5-0 


2610 


501 


38-3 


82>7 


108 


20 to 4-0 


5-4 


605 


2'51 


1307 


251 


191 


41-4 


54- 1 


40to 80 


2-70 


303 


1-25 


653 


125 


9-57 


20-7 


27-0 


80tol60 


1-35 


151 


0-626 


327 


62-6 


4-78 


10-3 


13-5 


16'0to27-0 


0-98 


61 


0-251 


224 


39-1 


* 


4-84 


9-84 



* This elemen: in this filter consists of i ■ 24 \j.H in parailei witb 21 ■ ^pF. 



Interstage Filler 



Frequency band 


c. 


L, 


L, Q C, 


L, 


Cb 


Mcjs 


pF 


P-H 


/ii/ pF pF 


,.H 


pF 


0-lto 0-2 


779 


1160 


762 677 2580 


1344 


581 


0-2to 0'4 


389 


580 


381 338 1290 


672 


290 


0-5to 1-0 


156 


232 


152 135 517 


269 


116 


1-Oto 20 


77'9 


116 


76-2 67-7 258 


134 


58 1 


2-Oto 4'0 


28'9 


58 


38-1 33-8 129 


67-2 


29 


4-Oto 8-0 


19'5 


29 


19-0 16-9 64-7 


33-6 


14-5 


8-Otol6-0 


974 


14'5 


9-52 8-46 32-3 


16-8 


7-26 


16-0to27-0 






SfP V\n 17 (■q^l 

















Output Circuit 



Frequency band 


C, 


L, 


L, 


^10 


^10 


Cn 


Cn 


Cl3 


ill 


Mcjs 


pF 


fj,H 


iiH 


pF 


l^H 


pF 


pF 


pF 


fiH 


0-lto 0-2 


524 


1960 


537 


3110 


151 


4220 


16900 


6050 


57 


0'2to 0-4 


262 


981 


268 


1550 


75-5 


2110 


8452 


3020 


28-5 


0-5 to 1-0 


105 


392 


107 


622 


30-2 


843 


3380 


1210 


114 


1-Oto 2-0 


52-4 


196 


53-7 


311 


15-1 


422 


1690 


605 


5-70 


2-Oto 4-0 


26-2 


98-1 


26-8 


155 


7-55 


211 


845 


302 


2-85 


4-Oto 8-0 


131 


49-1 


13-4 


77-7 


3-78 


105 


423 


151 


1-43 


8-0tol6-0 


6-55 


24-5 


6-71 


38-9 


1-89 


52-7 


211 


75-6 


0-71 


16-0to27-0 








See 


Fig. 17 (b 


) 
















1 









19 



6 8|j 



18iJ 




(b) 



Fig. \1 — The interstage and output fillers for the 16 to 27 Mc/s 
'integrated' octave amplifier 



(C) THE BRITISH BROADCASTING CORPORATION 1964 



Published by theHritishBroadcosI'mg Corporation, j; Marykbone High Street, London, W, i. Frsmsd jn England an Ba^ingv^erk 
Parchment in Timc-i Atw /Ionian bj' The Sroadwatsr Prcsi iid, Wel^yn Garden Cily, Hetts. Nff. 58(7