GATES

ENGINEERING
REPORT

PREVENTING FM
OVERMODULATION

GATES
 JUN 25 1975

PREVENTING FM OVERMODULATION

SUMMARY: With the advent of better broadcasting and recording systems, and the liberty
enjoyed by today's recording artists, the high frequency signals programmed into the FM trans-
mitter are often of sufficient amplitude (after pre -emphasis) to cause gross overmodulation. The
optimum placement of the peak limiting amplifier and the use of auxiliary protective devices are
discussed.

INTRODUCTION: Through the years there has been considerable controversy about the
wisdom of pre -emphasizing the signal fed into the FM transmitter. The reason for the pre -empha-
sis was that the signal-to-noise figure was enhanced by the attenuation of the high frequency
part of the audio spectrum in the receiver. Since there was appreciably less energy in the upper
section of the 30 to 15,000 cps. region than in the part below 1,000 cps., a 75 microsecond pre -
emphasis curve was adopted to take advantage of this distribution.

The standard 75 microsec-

8
and curve, shown in Fig. 1, is
14" flat below 200 cps., rises to +0.9
EMPNAS IS CUI YE t6
rile COOS 75 micaosEcars
14
db @ 1KC, +8.3 db @ 5KC, +13.8
db @ 10KC and +17 db @ 15KC.
12

10

.8
a severe curve. When it was adop-
ted, there must have been very
6
little anticipation of the kind of
4
effects that are programmed today.
.2
Also, the Broadcaster was expec-
ted to operate his FM transmitter
with rather low values of average
FREQUENCY IN CYCLES PER SECOND
POPO modulation to allow transmission
of the full dynamic range of even
FIGURE 1
classical music. Competition in
FM Broadcasting was not too
keen and programming was not stressed too highly in most areas. So, the low average modulation
concept was generally practiced. In fact, many FM Stations did not employ a peak limiting ampli-
fier, or any type of automatic levelling amplifier, due to the several reasons covered.

FM Broadcasting is becoming competitive. Programming is becoming commercial, and FM

Stations are showing a profit. This is healthy for the general public and broadcaster alike, be-
cause it will permit better coverage of many areas and a wider choice of programs in every area.

Nearly all of the practices that helped prevent FM overmodulation in the past are being
modified, in a direction that contributes to overmodulation, today. Peak limiting amplifiers are
being used extensively today. Yet, more and more FM Stations are being cited for overmodulation.

-1-
This has caused many station engineers to believe that their limiting amplifiers are defective,
or that the attack time is much too slow. However, their limiting amplifiers limited on a flat re-
sponse curve, where the high frequencies did fall below the threshold of limiting, and the ampli-
fiers were operating correctly. The problem was caused by the pre -emphasis of the signal after
it was processed through the limiting amplifier. It was typical FM overmodulation.

ANALYSIS OF PROBLEM: If the high frequency content of any part of the day's program-
ming never exceeds the curve shown in Fig. 2, the peak limiting amplifier could feed into the
FM transmitter pre -emphasis fil-
ter (assuming a fast limiter at -
110C tack time) and no overmodulation
would occur. Good microphones,
1.1Mii=
-2
tape machines, phono cartridges
-4
and recordings make this very
unlikely, however. This curve
is complementary to the standard
pre -emphasis curve, and is the
-10
one that should be used in the
12
receiver to de-emphasize the
14

STANDAID monism
3.5 signal to restore the original
16
frequency response. However,
17
MAX 18
if any of the high frequencies
FREQUENCY IN CYCLES PER SECOND
do exceed the limits of the curve
shown in Fig. 2, and go into the
shaded area - overmodulation
FIGURE 2 will result.

A good FM transmitter can tolerate extensive overmodulation without excessive distortion.

The carrier swing is extended very rapidly from the maximum allowable ±75KC, however. A pro-
gram peak of only 6 db will amplify the carrier swing to a very illegal +150KC. We have reports
of FCC citations with relatively low percentages of overmodulation. Many receivers and tuners
cannot tolerate excessive overmodulation without serious distortion because of bandwidth and/or
discriminator curve linearity restrictions. They frequently suffer from a type of base line shift
that can cause perceptible distortion in the audible frequency range. Thus, the Broadcaster has
a two -fold reason to stop overmodulating his FM transmitter.

An extensive search of current publications failed to reveal the distribution of spectral

energy of modern recordings or of broadcast programs. After some reflection, this was readily
understood. The extensive use of program equalizers, graphic equalizers and other frequency re-
sponse shaping units, in not only the original programming - but in subsequent reproduction as
well, eliminate any chance of two stations (or even one on a regular basis) having a typical
curve. The limitations of recording media are even circumvented to a large extent by a judicious
use of levels.

-2-
 However, this study re-
quires at least an arbitrary curve
for an illustration. Various spec-
trum analyzers and wave analy-
ODD DOD zers were employed in approxi-
PEAL POWER 14RIBwr
mating the typical peak power
-"-- of MICA B Y 'R 444411 ON

01,11.0T er ICAB 1.414 4 distribution curve shown here.

-6 This is a composite curve of the
peaks observed in many types of
-10 10 programming over an appreciable
-12 12
portion of the broadcast day in a
typical FM station. A conven-
tional limiting amplifier main-
tained a constant mid -range lev-
FREQUENCY IN CYCLES PER SECOND el. The very high and very low
frequency peaks occurred much
FIGURE 3 less frequently than the inter -
mediate high and low frequency peaks. Yet, they were frequent enough to fully understand why
stations are being cited for overmodulation. It is safe to assume that nearly every FM Station
will have a peak power distribution curve that will approach or exceed that shown, unless dras-
tic preventive steps have been instituted.

Perhaps the first step taken by many stations was to simply reduce the level of modulation
to an average of 50%, effective-
ly reducing the average program
MODULATION
199%

1111112911111.1111 power to 1/4th the maximum al-

178%

+4 160%

111111 Illiii11111 lowable station power. Fig. 4

141%
2 126%

1111111111111111111111
shows the result of 50% modula-
112%
ODB 100%
tion of the low and mid frequen-
90%
111111111111011,104111.111114 80%
cies. No overmodulation occurs
-111111 70%

below 5KC on the typical peak

60%

-6 111111 111111111.
power distribution curve. It is
111111.3tVE!!!!!TEM possible, however, to overmodu-
10 11111110111111111111111111 late as much as 70% at 15KC -
30%

-12 1111 11111111

11111
which constitutes an engraved
11i11I11
-14 20%
invitation for a citation. The
-16 II 10KC point shows up to 150%
-18 modulation under the conditions
FREQUENCY IN CYCLES PER SECOND
described. Obviously, an even
FIGURE 4 lower level of mid -frequency
modulation is necessary to pre -
vent high frequency overmodulation. This is true, in spite of the fact that the mid -frequencies
are controlled with a limiting amplifier.

-3-
 After this part of the study was completed and the levels shown in Fig. 5 were establish-
ed, the manager of our engineering section responsible for FM transmitters was asked what mid -
frequency modulation level was required to prevent high frequency overmodulation with average
programming. His prompt reply was, " Around 30%". Fig. 5 shows that this study was in full a-
greement, as it resulted in a figure between 29% and 30% modulation. However, certain types of
programming could still cause
more than 100% modulation at
high frequencies, even with 29% 101IC

oou ATI coi mid -frequency modulation. Thus,

00% o
sok- even lower levels of modulation
2
liTMillairminLdul
are required to be sure that no
60- 4
overmodulation will occur.
6

.1211111712-
-8 With 30% average modula-
N --- 10
tion, the program power output
IL PEAK -7
61 TNCA 298 16/1118.1111011 LOW I
C
GMENCIES `L
12 of the station is approximately
-1- 14 1/10th of the maximum allowable
N 6 power. Many of the program peaks
18 that are causing the reduction
i It 1 1
in modulation are so high that
fREOUENCY IN CYCLES PER SECOND
some receivers will not even
FIGURE 5
pass them, and some listeners
are incapable of hearing them.
Also, they are relatively very scarce in most programming, so the wisdom of using very low mod-
ulating levels is questionable.
PRE -EMPHASIS AHEAD OF LIMITING AMPLIFIER: The second step considered by many
stations is to place the pre -emphasis before the limiting amplifier, so it can limit the high fre-
quencies as well as the mid and
low ones. If the program level
is maintained carefully at a point
1111-9ECCI10S
that is generally well below the
threshold of limiting, this method
will produce fairly acceptable
results. However, certain types
of programming will still cause
quite un-natural effects, as
shown in Fig. 6.
I I i

TYPICAL LIMITING AMPLIFIER. OPERATION WITN PRE -EMPHASIS AHEAD

A I
A conventional peak limit-
THE LIMITHIG AMPLIFIER - WITH PREDOMINANTLY RION fREOINEEIDY
CUSSI YE IIISTRINENTS OR SIMILAR PER1001C PROGRADM
ing amplifier with an attack time
of approximately one millisecond
and a 90% recovery time of a-
round two seconds will have an
FIGURE 6 output similar to that shown in

-4-
Fig. 6, with certain types of programming. This is a result of high pre -emphasized high -frequen-
cy peaks exceeding the threshold of limiting, causing the attendent gain reduction of mid -fre-
quencies. When the resultant signal is de-emphasized it sounds like the limiting amplifier could
be undergoing a blocking type of oscillation, where the gain suddenly drops for no audible rea-
son, then recovers on the normal R -C slope. This is an excellent way to gain a large group of
irate listeners, which will soon degenerate into a much smaller group of irate listeners.

The right hand portion of Fig. 6 shows the same type of programming without pre -empha-
sis ahead of the limiting amplifier. Only the upper half of the waveforms are represented in both
portions, since this a sketched representation of the scope display observed under the two con-
ditions covered.

METHODS OF CORRECTION: If the typical peak power distribution curve shown in Fig.
3 were never exceeded the problem could be eliminated by the installation of a low pass filter,
such as shown in Fig. 7. A listening test of quite a few FM stations indicate that some must be
using this method. Unfortunate-
10:14pe CC
ly, this performance is easily
surpassed by the majority of the
NOD_- TI ON AM stations with modern trans-
ODD

2
.100% MODULATION CUR
WITH PRE-00114Si I
f

F
PASO-
LTER
100%
90%
80%
mitting and programming equip-
ment.
70%
-4 --
TTPICAE PEAT POWER 1115E100 014N CURVE -"r 60%
0070 70% MODULATION 0 LOW FREQUENCIES
6 53%
A low pass filter indis-
_LOW PASS FILTER CRRVEEEN.CONUNCTICA1 I

wiTh Tutii rpvt,E.Pv: Li:31mm 1 , 40%

criminately curtails all signals
ID 30% that fall on the filter slope -
thus, degrading all programming
\
,

IS

-14 1

,
i--
.. ,,,

.
just to protect the station from
16 ,
.._ a relatively few overmodulation
18
1
1
i 1 i ,
peaks. This " shotgun" approach
FREQUENCY IN CYCLES PER SECOND
cannot be considered as an ef-
fective cure for the FM overmod-
FIGURE 7 ulation problem for many reasons.

One corrective method that does merit serious consideration is to simply repeal the de-
cision to pre -emphasize FM signals in the transmitter. There have been many advances in the
state of the art since the original decision and the original noise reduction requirement could
probably be solved with modern components. With stereophonic operation becoming more promi-
nent, with the unusually severe phase balance requirements from L+R to L-R, the injection of
pre -emphasis and de -emphasis is a big handicap. It is the opinion of many in this field that a
better overall system would result with the elimination of pre -emphasis.

Since so many existing receivers are involved in the proposed elimination of pre -emphasis,

-5-
perhaps it would be best to consider a graduated reduction of pre -emphasis. For example, after
July 1st. of this year, the pre -emphasis could be reduced from +17 db @ 15KC to +12 db @ 15KC.
Nearly all of the decent existing FM systems have tone controls that could effectively com-
pensate for the reduction of high frequency pre -emphasis.

In 1970 the 15KC peak could be reduced from +12 db to +6 db. The systems sold in the
five year interim could compensate with their tone controls. Older systems (the majority of them
would be more than ten years old) could have a simple and economical change in the de -empha-
sis circuit to make them compatible with the 6 db standard.

In 1975 the pre -emphasis could be completely eliminated. All of the FM systems sold in
the preceeding five years, and all of the older systems that had been modified to the 6 db stand-
ard, could again compensate with their tone controls. Only the sets manufactured to the 12 db
standard would need a simple change to make them fully compatible with the new 0 db standard.
Thus, in a controlled ten year program the pre -emphasis problem can be wiped out without caus-
ing anyone a great hardship. The Broadcaster should have no real objection to buying a new pre -
emphasis filter every five years. In fact, the old one could probably be modified by simply in-
stalling a factory specified shunt resistor.

ACTION WHILE AWAITING THE DOCTOR: It is somewhat doubtful that the proposal out-
enough to immediately solve the FM overmodulation problem.
There surely must be a method of controlling the high frequency peaks, yet permitting decent
levels of modulation with full quality. This problem is not unique in the FM broadcasting in-
dustry. Disc recorders have been faced with an almost identical problem, and a product :has
been marketed that certainly has the correct approach to the problem. It essentially pre -empha-
sizes the signal along a slope
that corresponds to the problem
Ee- IOU area of the device that it event-
008
ually feeds, wipes out any peak
2
that exceeds this slope, de -em -
100% M0001AT1
4
phasizes in a complementary
WISH -NC
slope to give an overall flat re -
6
C106 ST sponse. Fig. 8 shows the adver-
-8
tised curves in solid lines, plot-
10
ted in the same scale as all of
-12
the previous figures in this
5' 14
4- study. This graph is advertised
-16
for use with a 75 microsecond
18
pre -emphasis curve in FM trans-
FREOUENCY IN CYCLES PER SECOND
mitters. The standard 75 micro-
second pre -emphasis curve is
shown in the dotted line on the
FIGURE 8 graph. One of the curves falls

-6-
remarkably close to the standard 75 microsecond curve and would eliminate FM overmodulation
with much of the standard programming. Selecting one of the two more severe curves would sure-
ly eliminate the problem. However, they would cause unnecessary reduction of all frequencies
above 600 to 900 cps. with some resultant degradation of signal.

.4 TION
Fig. 9 shows the amount
MI% of overmodulation possible with
.2
112%
100%
the curve that falls closest to
004

-2
.0%
80%
the 75 microsecond curve. With
-4
100%1110011 TIN
70%
the low frequencies held to ap-
60%

5c4
proximately 90% modulation by
6 --sELETTurimE4mrsuris-corir
40%
the peak limiting amplifier
8

(which should preceed this unit

10
30%
for best control), it is possible
12

20%
to get a small amount of overmod-
14
ulation in the 1100 to 1200 cps.
14
area. Then, it is possible to get
-18 lox
a serious amount above 6500
MO

FREQUENCY IN CYCLES PER SECOND

cps, with up to 130% @ 15KC.
Another apparent deficiency oc-
FIGURE 9
curs in the 2KC to 5KC region,
where the Networks and larger stations generally install circuitry to give the system a gentle
hump for " presence effect". The curves shown will give an almost opposite effect.

The system that would apparently give the least degradation of signal, yet yield full pro -
tection against FM overmodulation is shown in Fig.. 10. The signal is pre -emphasized with a
standard 75 microsecond curve
01 1.1111114121
and any random program peak
.18
00ope
that exceeds this curve is clip-
CC

16 owl ped off. Then, the signal is fed

14 SEcTio
08-11111111414
through a standard 75 microsec-
011110009-15 mietesecen
.12 ond de -emphasis curve to give
10 an overall flat response.
1.0
.8
ovERAu. tesposte+
000
-1.0 Actually, the three curves
10
shown are carefully plotted from
+4 tia
the test data on the new FM Top -
2. SBA T5 MICROS low
14
Level Unit that was developed
OBS
specifically to eliminate FM
overmodulation. The pre -empha-
sis filter measured within 0.1 db
FREQUENCY IN CYCLES PER SECOND
of the standard curve. The over-
all response measured 40.2 db
FIGURE 10 @ 10KC and -0.2 db @ 15KC, it

-7-
was essentially flat below these frequencies. The M6467 FM Top -Level has two identical sections
with essentially complete separation. Thus, it may be used for Left and Right stereo processing,
main channel and sub channel processing, etc. It may even be used for FM in one channel and
TV on the other.

The same problem that has been discussed for FM does exist on the aural portion of TV
Stations. It needs the same method of correction.

Fig. 11 shows the block

CLIPPER
diagram of the FM Top -Level.
PRE DE -
After the variable input gain
INPUT X1STOR I ISTOR OUTPUT
11 AKP. ANP.
control a transistor amplifier
1
.18 DBM TO -10 Dell
OUTPUT LEVEL boosts the gain to the proper
CLIPPING
THRESHOLD processing level. This is fol-
-10 DBM TO
.24 DUN IN-
DISABLING 20 V
POWER lowed by a precision pre -empha-
PUT LEVEL. SUPPLY diki.)"4
I I7V.
sis filter, a balanced series clip-
CLIPPING 50/00mme.
THRESHOLD
per with its output matching
.18 DUN TO -10 DON
OUTPUT LEVEL section. This feeds into another
transistor amplifier to recover
/2 Y
L
ISTOR
AMP.

GATES RA010 COMPANY

_ ^
ii
CLIPPER
SISTOR
AMP.
OUTPUT
2

IN -TOP LEVEL
the circuit losses of the pre-
QUINCY, ILLINOIS BLOCK DIAGRAM
ceeding components; then into
a precision de -emphasis filter,
isolation transformer, variable
FIGURE 11
output gain control, and finally
into a 6 db isolation pad.

The unit has 28 db gain, guaranteed +1 db frequency response, 0.5% maximum distortion
(below the clipping point), 75 db or better noise, instantaneous attack and release time, an in-
put level range from -10 dbm to +24 dbm, and an output level range from +18 dbm to -10 dbm
after the 6 db line isolation pad.
The unit does not cause deterioration of the programming when properly used because it
does not act until the offending program peak tries to exceed 100% modulation. Then it neatly
and instantaneously clips the excessive portion of the peak only, without affecting any of the
associated signal. The harmonics generated by this clipping are greatly attenuated by the fol-
lowing de -emphasis filter so the action of the unit is extremely difficult to detect by critical
listening tests.

IN SUMMARY: Pre -emphasized FM overmodulation presents a very complex problem that

requires custom designed equipment to correct. With properly installed and operating corrective
equipment, the FM system is actually enhanced since gross overmodulation and associated dis-
tortion are completely eliminated.

W. J. Kabrick
Advance Development Engineer
Gates Radio Company

-8-
 r -.
4.. ;0,....
.
' Clirrl'"- - s !;`, 94.
0- '
t,
..
,

-
. .
, . !, .44. _00.. _, , ..
.
It2... _ -y,?.
'lit...to *..; ..,
..... , ..
-r !..-zi -ii,'Itr4!-- _.....:1,1 ._.. , ir.
.
' , e *7..
,_
1677,.,-. 16.4..F. .- iroru,!..44,-;;44--`
7' :
''' '% '-±--...-'..,
-T,./. ;.;--..1-.-_ -
;
l'' ri..;,_-, 1..,- 4;7_ :.. ,,i0 -,4-T, - -
7.- --dier"."
.. 1.-- 0 '
--
.
7, - - 47.--.! '_:_'
--
7
11, ,
-.0- ..'," c...,
,,-
_
-;%

, ;:ti,t-.- --7 ,- c -
-. -
.
' "`.-1 - '
't'.0
'
..-'' - -.'-'; -
. '
,:, .
,
reTi'
- ,-
". 4 ' -
- . -.4, 'Or,
- -'''. .` --- ',- 11:- ....v.t,4,,111,
.1 317e
-
t' jr
4? "
"at .
't.'- .4.- a. _:.-
.. -
't.,1.. t-i:
,,,,,f13,..
aa.,04
r
11 7 .---4412
'- ; , :31* " - .
'
.
a..r*.'
' 4, : 'pw i
_
It' 40 1:11; .;$., 11110PINI 4r a-:.-..' .
.: f t *-.
-4.104 r0.24 taried. -4.1.
Z:r%.+4tif.
. "
-
; ip?",.
. 4. . . .. , s,), ..,
,
.1
A, ' .: ..
.
, 1, ..1' ,,,.e,,,,,,, .
.:
#1
.1i1;
,
.1 , .-.,
-
--, _ 2144:-1.,,,-,,:: - ,...,,,
, a 1
1-; TQV.--,,,--:'*,
-
.....
44 . ,-, . -. :I,
-
,..,t, Iv , .s
go -a
I. ,-,...ak- .
..1...
'
,
1.1.

. .
'.lit.,,ri '
14 ' .
i- , -,,- .
-
, A.
- 7 4 ' '
it t. N.._ 1
. ''''`. r""': .J51:-._.
.
.
. ., _ .. es 'L, .r.
, -
, ,
;-;
.'
.-t '
4' ' Fr'
410, - - 4, v
'
i4
.....
v. " . i* IN" ..- ,
.11*e.
_ 1.',- a4
I,
..... . ..../It , _ ,,,,....,- -- tftf- ...K"'--: 2 . L- ....... .... . .. ,2.1-,?". .
.
V. 3r.,,,_
,
__
' 3 ''' ' -1*; " ' . if-
c I"-
'
, ,. sl.:k. .: f -.) 7 ,.
- 4_,-,.: .
'''
4:
. -
."- V 'P. :srtlr. -
-
sil:"tW-
7 ',. rf- - ;,
P
-414%. t ... r k ,Pc! fi A ,-! ,5)- "'' T l'''' -
.
_r S
,. 1
..
-
#
ci,f.'', ----,14e7:,,it ../ .
' :116 n:p.
" - iir'., * -
-' -
-1:-,' ...,,,,O :' .5- ' -45Ar 0' '
'11- - -.0 ...* ,
-f , .. lop ,.. ,_, mr.
,
-106
''''ir -,...I.%
'',,, %- .'-}. 1444 ih--, .
-
44 .. ,
' .-
.4., th" rr, 7. 'A111116.::::ibii _
J 71sp :11006
i ,,. :-
alt.
.
r, ..2.7.'.
--
"I? - '1'4 -
. .'..-7
,.41 -
:

d
.
,
4:: 44' 11,14,4,
- -
.114.4,:"i
.* 1.,
.
3/.1t . :
Y.4'1*
- kWe'
*-'
ivar, ;tor -It
.- .
-.1)-
fr -
k -Tv
;;41, ?..;
#r- _ 10
tit or.07;!.

;
.
,
7

.
v. t*.
-Ilia. - .
. 1,1
-
-,
7
a
:1 r ,..,!spe C. 174
I..11_!. - .'"' . 4
_,-4-:t .....r .

.
,....,,,.4.-%

,4 '
;
-' .4".`
ra
. T.

-
..
iW1,&-(;_. 4 if '
41 '415' - , .1
t,
'116.00 , i
, *IA&
" r- skiirA d
-
D t) -9N- ?kw.
. t -
,4
fp a
"4"!
*,e
,
.,
$41
40-4111'
..
4r1" 't
74: ,

:7
I
val.c4111.
....10.111 ,±17iirk
7. Con. -
.."-it*/44-r, '.i I
,,1,6' *It ,
L
404, ' . 17 14.,..i:4.1*4,7411174....
.
. '4;
-gLe.a V
kipioe'' 44 ri.011,4 .
S iar-
SW- .0.'1; =-..
-1 ;011"11 *re.,
Itv :: ..-.. --4. -'
-- f" - ve-, sr. -
-. ; -,
-,:_.4 :_ ..... - .4, .-
_
,-t,
.
:
94....
" -
Alta....,ti-,,'
jo
' .
-
4-h.,
It..,' vpst- -.. '..`"- 4 --.
' :
,,.'
A -
i -i ,41'
-
,c.-*._ .e:. t.AP's .) :,4: :...,--- ,- - .
-;
.:-
'
';"'t--:-:
,:.-
....v- e,-,..0, , '-
e
-
;......'",r...,. ,,,----
- . 4,-.7r.,; -., _
...
.. 1
,
'''''''',1,!-7.L.,t*,:e.' _
.
" .
' "
.1- .4- t-,
.4.n: '4; 1 : . , ZT "; '.z. 7.4 ...i;:.,
4.. .s Lig.;
.
., .,. it.,.." .s 1,4! :: :A. . 0. '4-.
'",:.
... .',,
.r.
,
,,. .,...31, , ,._
Ait .0 t i
'
. e ,, . : .. -- '1'....:. ,t, ,
!,, ,
4r,:' -4,Fi. t
'-
,
t...41,.. -c:' ,,n 4 ." 4e KA,
-
..:14.1.
4,
-
- 1'4.1^ '

1:,
. ,

.
.
'vs
:
-
1., .,...
C.

.,,"164,
-
,, . -,...
".
;,?,,, -
'

-_
'f< k .
-. +.
qpi
74.1
.311
. _
;,...4-4
'-141, 47. -.." ..-- .
I', - . ' 4).
,:#,ri
' L ,---,
, ,.
.
-.0.*
r ....
-
-,
,C, . ..,,,
., 1
..- .-. -:
,:i ,..1..- 3;-:7..z..:,4
" 4: ' i; Cr
e
*.: -
'
sH.'.. -1'
1...k
I ' ''' . . "a;' ' '''
--::.::.7:4,:: -1
-1.,.. ,- ,_ .a, ,.__, _ ' . 7 ita
-rtifiV34-
...,L=..
....4,,,,,,;_. :_tr:i0:711:1:::ror.7.4.10.
- -- . ' ..4:`,-; -
',--,1k-"a --1.-.. *.kg t..Ler '-;
. ,
- 141,0
." " 1;?,
7
cr; 4F14A
 HARRIS
INFTEFIlrlfFPE
CORPORATION
GATES
GATES RADIO COMPANY
A Subsidiary of Offices. New York, Houston, Los Angeles. Washington, D C
Intertype Corporation Export: Roche International Corporation. New York City
QUINCY, ILLINOIS 62302 In Canada Canadian Marconi Company. Montreal