FINAL ENGINEERING REPORT PROJECT 100-1 PHASE I
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP78-03433A001100050001-5
Release Decision:
RIPPUB
Original Classification:
S
Document Page Count:
106
Document Creation Date:
December 27, 2016
Document Release Date:
October 18, 2012
Sequence Number:
1
Case Number:
Publication Date:
April 30, 1962
Content Type:
REPORT
File:
Attachment | Size |
---|---|
![]() | 5.04 MB |
Body:
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
CONFIDENTIAL
tta
NOT RELEASABLE TO FOREIGN NATIONALS
,I0B NO. _Z4
BOX NO. __1/
FOLDER NO. r .
TOTAL DOCS HEREIN /
FINAL ENGINEERING REPORT ? /14 .1.
Project 100-1
Phase I
Doc. No. 100-1. 5-16
Copy
April 30, 1962
DOC REV DATE ta.Y BY 467 550
OR1G COMP DPI TYPE /?1---.
ORM CLASS PAGES Jam_ REV GLASS C.--
JUST NEXT REV oa MIMI KR 10-2
This document contains in e ense
of the United States within the meaning of the Espionage Laws,
Title 18 U. S. C. , Section 793 and 794. Its transmission or the
revelation of its contents in any manner to an unauthorized person
is prohibited by law.
Approved
25X1
ThiS (ftOCUNItafA
414,1+%Se nt thsv
Eso.,orta9geA..
tutetvris4,4,1
InermeN.
pfllti 0,6
ttTE'r
s!, ?
C,
.,n
ft)* ODVOn6i
;ba mearthg .'7?
793 794.
t.%1 co,)ems
+z ta,Aitved
Approve
Project Manager
25X1
Engineering Director
CONFIDENTIAL
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
iz41 %AMIE-1UL! NI 1 It11.. ?
NOT RELEASABLE TO FOREIGN NATIONALS
TABLE OF CONTENTS
List of Illustrations
1.0 Work Statement
2.0 System Design
2. 1 System Function
2. 2 Specifications
2. 3 Design Considerations
3. 3. 1 General
a. Signal and Noise Levels
b. Bit & Character Error & False Alarm Rates
2. 3. 2 Transmitter
a. Coupling to Line
b. Audio AGC
c. Transmission Rate
d. Tone Signal Pulse Shaping
2. 3. 3 Receiver
a. Coupling
b. Audio Section
c. Threshold
2. 3. 4 Packaging
3.0 Theory of Operation
3. 1 General Design Requirements
3. 2 Transmitter
3. 3 Receiver
CONFIDENTIAL
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
a
tt7t
NOT RELEASABLE TO FOREIGN NATIONALS
Table of Contents (cont'd)
4. 0 Packaging Design
4. 1 Transmitter
4. 2 Receiver
5. 0 Performance Data
6. 0 Conclusions
Appendix: Tone Audibility Tests
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONA LS
LIST OF ILLUSTRATIONS
Figure No., Description
2.3.1a
2.3. lb
2.3.1c
2.3.2
2.3.3
3.2.2
5.1.1
5.1.2
5.1,4
5.2.4
Page
Response of Band Reject & Bandpass
Tone Filters 72
Speech Power Distribution During
Transmission 73
Probability Density Function 13
Transmitter Coupling Technique 74
Receiver Coupling Technique 75
Power Supply Schematic 76
Block Diagram, Transmitter ? 77
Block Diagram, Receiver 78
Schematic Diagram, Transmitter 79
Schematic Diagram, Receiver 80
AGC Response 81
AGC Control. 82
Error Rate vs. Line Distortion 83
Line Frequency Response 84
SECR
NOT RFT F AS A R T TO FOR F.TON NA TTON A T .S
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
1.0 WORK STATEMENT
1. 1 Conduct design and development program to convert into
hardware the fixed base station communications system design
and feasibility tested under
Task Order No. 1,
This task has been modified from
the original, requiring a more sophisticated package and a more
complex coupling technique. For details refer to Sections 2. 2. 1
and 2. 5.
1. 2 Fabricate and test one (1) each simplex system consisting of
one (1) transmitter and one (I) receiver based on the development
program conducted under Item 1. 1 above.
1. 3 Fabricate and test three (3) each transmitters and four (4)
each receivers identical to the model constructed under Item 1. 2
above.
1.4 Provide five (5) copies of each monthly letter progress report.
1. 5 Provide five (5) non-reproducible and one(1) reproducible
copies of engineering drawings describing the equipment fabricated
under items 1.2 and 1.3 above.
1. 6 Provide ten (10) copies of operation and maintenance hand-
books for the equipment fabricated under items 1. 2 and I. 3 above.
-1-
NOT RELEA'SAY&1'6 FOREIGN NATIONA TS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
25X1
25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
2.0 SYSTEM DESIGN
2. 1 System Function
The function of System 100-1 is to provide digital communica-
tion between two parties utilizing conventional public telephone
system as the transmission medium. The message is sent during
the course of a normal telephone conversation using the originator's
voice
The system concept is to convert each character of the message
into a combination of one to five audio tones. These tone combina-
tions are added to the speaker's voice and transmitted whenever the
voice level exceeds a preset threshold. Since tone levels are far
below the instantaneous voice power, the voice masks the tones,
making them inaudible to anyone listening on the line. All five
bits of each character are transmitted in parallel to eliminate
synchronization problems, a tone representing a "mark" and the
absence of a tone a "space" on the teletype code. Since each tone
is at a level considerably below peak speech levels, it is necessary
to notch out the speech spectrum around each tone frequency prior
to transmission to avoid mistaking voice components for tones at
the receiver. In the present system, the five tones are closely
spaced requiring one band reject filter to eliminate the undesired
spectral components of the voice.
-2-
N
NnT RFT F.ASART,F. TC) FC)12F,TC,N NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
25X1
25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
2. 2 Specifications
a. Transmitter
Input: (Telephone Instrument)
DC Source. Resistance: approximately 200 ohms
handset off the hook)
AC Source Impedance: 600 ohms (nominal)
Level*: -6 to -26 dbv rms
Frequency Range: 300 to 3400 cps
Output (Telephone Lines)
DC Resistance: Approximately 200 ohms (when
handset is off hook)
AC Impedance: 600 ohms (nominal)
Level*: Voice: -16 dbv L 1.5 db
Tone: Adjustable, -40 to -46 dbv
Frequency: Speech: 300 to 3400 cps notched between
1250 and 1950
Tone Characteristics: Tones: 1400, 1500, 1600, 1700,
18(X) cps. 40 millisecond
duration, isoleles triangle
envelope
Controls and Indicators
Power: OFF-ON (toggle type circuit breaker)
Standby - Transmit (toggle switch)
Power supply protection (push to reset circuit breaker)
Amber Light: Power indicator
Red Light: End of message indicator
Powe r requirements
a. 90 to 126 VAC, 45 to 60 cps, 0.6 amperes, nominal, or
b. 180 to 252 VAC, 45 to 60 cps, 0.3 amperes, nominal
*Level shown is long term average of speech in decibels above 1.0
volts rms.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
Keying Rate: 'Adjustable from 0 to 100 words/minute
during continuous speech (set at factory
for 25 wpm, nominal)
Message Tape: 11/16 inch teletype tape (5 level),
chadded. Capacity 60 feet (1200 words)
maximum
Monitor: A separate earphone is provided to permit
local monitoring of the distant speaker's voice.
Mechanical Characteristics
_Dimensions (maximum) - 15 inches wide, 9 inches
deep, 10 inches high
Weight: - 45 lbs.
Finish: Semi-gloss black enamel
Accessories
220 V motor for tape reader (1 each)
Card Extender (1 each)
Fuses - AGC - 1/4, 3 each
Environment
Temperature 00 to +50oC
Humidity: 0 to 95% relative humidity
Shock and Vibration: Air transportable
-4-
ivrrvim ID r.` T A C A Ti I T7 'Ts /1 Its f-11-1 T T T,rIA "rs T lss Ykrn
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
?
?
?
b. Receiver
Input (Telephone Line)
DC Resistance: Approximately 200 ohms
AC Impedance: 600 ohms (nominal)
Frequency Range: Voice: 300 to 1250 cps and 1950 to 3000 cps
Tones: 1400, 1500, 1600, 1700 and 1800 cps
Level: Voice: -16 to -36 (Thy (long-time average of rms voltage)
Tones: -43 to -63 dbv (peak rrns voltage)
Tone Characteristics: 40 millisecond duration, isosceles
triangle envelope
Output A (to telephone line)
DC Resistance: Approximately 200 ohms
AC Impedance: 600 ohms (nominal)
Frequency Range: Voice: 300 to 1250 cps and 1950 to 3000 cps
(No tones present)
Level: -6 to 726 dbv rms (long-time average of rms voltage)
Output B: Punched paper tape 11/16" wide, 5 level teletype,
chadded
Controls and Indicators
Power: OFF-ON (Toggle switch)
Function: Normal - Standby- Receive (rotary switch)
Power supply protection: Push to reset circuit breaker
(fuse on power supply for 220 VAC)
Gain Control: Operator sets gain of receiver using VU
meter
VU meter: Indicates level of voice signal at input to
receiver amplifier
Amber Light: Power "ON" indicator
_5-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
aux!
?
NOT RELEASABLE TO FOREIGN NATIONALS
Power Requirements
a. 90 to 12.6 VAC, 45 to 60 cps, 2 amperes nominal, or
b. 180 to 252 VAC, 45 to 60 cps, 1 ampere nominal
Information Rate: Determined by transmitter rate - 25 wpm
nominal, 250 wpm maximum
Mechanical Characteristics
Dimensions: Receiver: 21-9/16" wide, 16" deep, 13" high
Perforator: 21-9/16" wide, 17-7/8" deep, 13" high
Weight: Receiver:75 lbs.
Perforator: 79 lbs.
Finish: Hamn-iartone Gray cabinet; clear anodized front panel
Accessories
Power Adapter (1 each)
3 prong to 2 prong power adapter
Card Extender (1 each)
Fuses: 2 amp - 5 each
1/4 amp - 3 each
Microphone: High impedance lapel microphone
Environment
Temperature: 0 to +50?C
Humidity: 0 to 95% relative humidity
Shock and Vibration: Air transportable
surpET
- 6 -
L Declassified in Part - Sanitized-C-op-y-APP-Riv-ed ..f-o-rik-elese-20:i2-/i-611-8-:-CIA-1DP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release,2012/10/18 : CIA-RDP78-03433A001100050001-5
V six 414 P,rt
L
NOT RELEASABLE TO FOREIGN NATIONALS
?
?
2. 3 System Design Considerations
2. 3. 1 General
An important system consideration is the error and false alarm
rates. The error rate is a function of the number of times a trans-
mitted tone, corresponding to a one-bit, is recognized as a zero-
bit. This situation arises when sufficient noise signals (undesirable
signals) are present in the tone channel and have such a magnitude
and phase that when added to the transmitted tone, produce a signal
which is less than the established threshold in the receiving equip-
ment. The false alarm rate is concerned with the number of times
signals due to noise have sufficient magnitude to exceed the threshold
in the receiver.
A second important system consideration is the transmission
rate. This parameter is a function of the statistics of speech and
the threshold level established in the transmitter relative to the
long-time average power level of the speech. The transmission of
one bit occurs in 40 ms, corresponding to a maximum rate of 250
wpm. To provide maximum security of the system, transmissions
only occur when the instantaneous voice energy exceeds a preselected
threshold level. Whenever the speech power exceeds this threshold
and the equipment is not in the process of transmitting a bit,
transmission of a bit is initiated. At the end of the 40 ms required
for this event, transmission of the next bit depends upon the
threshold condition above being satisfied. At the time of this
writing, statistical information as to the number of times that a
given threshold level above the mean speech power is exceeded has
not been available. Further, no theoretical means has been found
to extract this information from the percentage of time that speech
power will exceed a given level referred to the average speech
power.
-7-
j4 4r0
NOT RELEASABLE TO,FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Rele.;'Se2.617/10/18 : CIA-RDP78-03433A001100050001-5
4LLY)14r.
NOT RELEASABLE TO FOREIGN NATIONALS
a. Signal and Noise Levels
This section includes a description of the noise components
considered and estimates of their relative power levels. It also
includes the various signal levels in the system which are pertinent
to determining these noise component levels.
The sources of noise considered are as follows:
A. Distortion components
B. Thermal Noise
C. Speech components which pass through the band
reject filter in the transmitter and bandpass tone
filters in the receiver.
D. Telephone line crosstalk
Using 50 db as the total useful dynamic range of the telephone
line, and 0 db as the average speech level, a maximum level of 12.5
db corresponds to peak limiting speech signal 1% of the time. This
amount of peak limiting will not noticeably affect the quality of
speech. For the power levels indicated above, line noise for a
3-kc bandwidth is at -38.5 db and line noise for a 100-cycle per
second bandwidth is at -52.5 db. A threshold level in the transmission
equipment of +10 db relative to the 0 db average speech signal was
used during the feasibility experiments and is used in the analysis.
This threshold level is the magnitude the speech signal must exceed
to initiate tone transmissions. The tone signals are established
at 26 db below the transmission threshold, or -16 db relative to
the average speech signal. It should be noted that the litie noise
is -34.5 db below the tone level (using the 100 cps bandwidth
value) and has negligible effect on error and false alarm rates.
The attenuation of speech components through the band reject
filter in the transmitter and bandpass tone filter in the receiver
is shown in figure 2.3.1.a. Only the output of a tone filter
-8-
Declassified in Part - Sanitized CoPY?ATop'r7oVed?fOr-Reie-ai-e 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
adjacent to the edge of the band reject filter can contain signals of
any significance. The composite response in Figure 2.3. la shows,
however, that these components are attenuated by at least 60 db and
need not be further considered.
Cross-talk between telephone pairs is generally maintained at
about -30 db. Taking into account the 100-cycle bandwidth of the
tone filter, the cross-talk voice energy would be at about -45 db
and would not contribute to the error or false alarm rates. On the
other hand, the presence of the cross-talk of signals due to tones
used by the telephone company can be a serious problem if the
tone frequencies fall within the passband of the tone filters in the
receiving equipment. This problem can be avoided by careful
selection of tone frequencies.
The most significant noise components result from signal
distortion produced in amplifiers contained in either the transmitting
or receiving equipment or in the telephone line itself. It is assumed
that the total distortion will be kept to within 1% corresponding to
at least 40 db below the instantaneous speech power.
It thus appears that distortion components are significant
in affecting the false alarm and error rates and will now be considered
in greater detail. To accomplish this, we will establish the mean
power level of these components occurring during the transmission of
the tones as well as a statistical model for this source of noise.
From this information, we can determine the error and false alarm
rates per bit and per character. In addition we can also determine
the level of accuracies which must be maintained for a given
character error and false alarm rate and thus derive how accurate
the AGG's must be.
-97
t
TNT, V' A Pr r's ',yip -c? r?i?r 1\ A 'I" r1NT A T C
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO 1,0 REIGN NATIONALS
b. Bit and Character Error and False Alarm Rates
For 1% total harmonic and intermodulation distortion, the
total power in these components is 40 db below the speech power.
The mean speech power during transmission of the tones is
determined as follows: A transmission is initiated only when the
speech power exceeds the +10 db level. However, the speech power will
exceed this level 2.8% of the total conversation time. Assuming
that 10% of the telephone conversation time can be used in trans-
mitting the digital message, this corresponds to 28% of the trans-
mission time. If we also consider correlation existing between
signal samples, the most pessimistic conditions exist if the speech
power is always greater than the level exceeded 10% of the time.
The speech power distribution thus occurring during ttans mission
is considered to vary between 5.6 db and 12. 5 db relative to 0 db
for the average speech power. Figure 2. 3. lb shows the speech
power level during tone transmission and the percentage of time
it is exceeded varying from 100% at 6.5 db to 0% at 12.5 db. The
unprocessed speech signal will actually contain components whose
power exceed this 12. 5 db level. However, the limiter action in the
transmitter has removed these components. Taking the weighted
mean of the power as a function of the percent time it is exceeded,
an average power level of speech signal during tone transmission
of 8. 5 db is obtained.
During transmission, the mean energy in the distortion
components for 1% total distortion is 40 db below the mean speech
power. Although the statistics of the speech is not Gaussian, the
statistics of the distortion components can be considered as Gaussian
because, in general, the total distortion signal is made up of the sum
of many components. The central limit theorem states that the
statistics of a signal made up of the sum of many components tends
- 10 -
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
cArf t'P/'
, ?tt
NOT RELEASABLE T FOREIGN NATIONALS
toward Gaussian regardless of the statistics of the individual
components. The mean power level of the distortion components
is 8.5 db -40 db = -31.5 db. The tone levels are at -16 db
providing a mean signal-to-noise power ratio of 15.5 db. This
is the signal-to-noise ratio used in computing the error and
false alarm rates. It will be noted that the reduction in noise level
resulting from the use of the 100 cps tone filter is not taken into
account, and it is assumed that the distortion components will all
lie in the passband of the tone filter. It will be shown that even
with this pessimistic model, the error and false alarm rates which
can be achieved are practical. The output of the 100 cps bandpass
filter is supplied to a detector converting the Gaussian statistics
of the signals to a Rayleigh distribution with noise alone and modified
Rayleigh with signal and noise. The output of the detector is
supplied to a lowpass filter with its cutoff at 50 cps. This filter
acts as an integrator producing a signal whose st.atistics are that
which would be attained by summing independent samples at the
input to the lowpass filter in groups of two. This results in a
distribution as indicated in Figure 2.3. lc where a- is the rms noise at the
input to the detector, and where the mean value and standard
deviation of the noise alone are 2.5 a- and .928 a- respectively and
where the peak signal and standard deviation of the noise distributed
about this signal are 176- and 1.414 Cr respectively. The magnitude
of the signal for three standard deviations (corresponding to the signal
level which will be exceeded 0.1% of the time) is 5.287 and is a
case for no tones present. The magnitude of the signal corresponding
to the case for tones present and the noise subtracting from the signal
and three standard deviations is 12.76 0-- . A threshold level located
anywhere between 7.46 er and 12.28G- will produce a bit error or
false alarm rate which is less than 0.1%.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP/820-3433-A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
?
NOT RELEASABLE TO FOREIGN NATIONALS
We now wish to establish the magnitude of the threshold level
which will allow maximum gain variations in the system. Letting
L the value of this threshold level, and letting K be the variation
in voltage gain, (5. 28er K + L) corresponds to the case where the
gain is increased to the point where the false alarm rate is 0_1%.
Or?
On the other hand 12.76 = L corresponds to the case where the
voltage gain has been reduced to the point where the error rate
is 0.1%. Sdving these two equations for K, a value of 1. 55 is
obtained corresponding to - 3. 8 db. This is the voltage gain
variation which can be tolerated and still provide a bit error and
false alarm rate which is 0.1% or less. Using the value of K
indicated above, a threshold level of 8. azo--- is obtained, corresponding
to a threshold located 6. 3 db below the peak tone level. This value
of threshold level provides for a maximum variation in both in-
creasing or decreasing the gain from the optimum value.
The character error rate is obtained from the bit error rate
assuming a 0. 1% error and false alarm rate using the Bernoulli
Formula. The character error rate is determined based on having
one or more failures or errors in five bita and is given by
5 5Cs ps q5 -s
S=1
q = .9987
= 1 - 5C poq 5
= 1 - q5.
so that the character error rate is 0.64%.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A00110005nnni_
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
;
p(x)
NOT RELEASABLE TO FOREIGN NATIONALS
Threshold
Signal and Noise
S. D. = 1.4
Input S/N to detector of 15. 5 db with 100 cps bw detector
followed by low pass filter with 50 cps cutoff. Gain variations of
- 3.8 db of input signal level will result in bit error and false
alarm rates of 0. 1 percent.
Figure a. 3.1. c
PROBABILITY DENSITY FUNCTION
-13-
T ICI 9' T ACART m (-1 -c-r? D Tr`_ NT 1\ TAMT tTh 1\T A I" C
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part- Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
2. 3. 2 Transmitter (See figure 2. 3. 2)
a. Coupling to Line
The technique of using hybrids to couple the transmitter
between telephone and line was extensively explored both in theory
and laboratory experiment. The problem of high transmitter gain
(to provide good AGC control) was recognized from the beginning
of the program. Another problem discovered in laboratory tests
is the difficulty of obtaining hybrid balance with the complex telephone
instrument impedance. This is further complicated by the time
variation in resistance of the telephone (carbon microphone). Even
use of another similar telephone as the hybrid dummy load does
not yield adequate balance.
An alternate technique using time multiplexing between transmit
and receive channels was also tried. This circuit failed when connected
to the complex telephone impedance even with substantial guard
time intervals.
The final approach adopted for the Phase I equipment is to
provide a separate earphone for monitoring the incoming voice
signal at the transmitter.
b. Audio AGC
Gain control of the local speaker's voice is needed for
two reasons.
First, the level of signal fed to the line must be
kept below a value that would result in important distortion in the
telephone system but high enough to keep the tone levels well
above the ambient telephone system noise. Second, a constant
output level is desirable.from the standpoint of maintaining a
uniform information rate for a wide range of speakers and for
different telephone operating techniques, i. e. methods of holding
the handset. A practical range of levels to be expected for these
-14-
T\Tr1T DVTVACA 2 T W rT' VC\ 12. TrINT NT A 'T Tnl\T A T
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
, -50
NOT RELEASABLE TO FOREIGN NATIONALS
variations is about 20 db.
In the design of the AGC circuit, several factors in addition
to dynamic control range are important. The degree of control
necessary must be specified since this determines the information
rate versus input level characteristic. 1From the statistics of
speech, it may be observed that a very slight change in the speech
level into the threshold circuit results in a large change in
information rate. In fact, an increase in rms speech power of
only 3 db in the vicinity of the 10% point results in a doubling of the
information rate. 2Therefore, it would be desirable to limit the
range of inputs to, say, 0.1 db which would cause only a 3% change
in information rate. Assuming a linear curve in the vicinity of 10%,
however, the loop gain required to maintain the output this closely
for a 20 db change in input is quite high resulting in poor transient
3
response and complicating the loop filtering problem. A compromise
is therefore necessary in the design of the AGC circuit to achieve a
clean output signal with reasonable control. With a loop gain of
60 db, a 20 db change in input results in a 1 db output change. Using
LC filters in the loop (high pass and low pass) with 36 db per octave
rolloff, distortion can be kept below 40 db. Transient response, however,
is still poor due to the delay in AGC control, This results in a very
high output for about 100 milliseconds after the start of the input
'Another factor also affecting information rate is the talking rate
and articulation of the speaker. Compensation of this factor, however,
would require unnecessarily complex circuitry.
2See ITT Reference Data for Radio Engineers, P. 874
3
Speech components above 300 cps must be attenuated sufficiently
in the loop control signal before being applied to the gain control
element. Otherwise, distortion of the output waveform will result.
-15-
1\ TeN,T, .17% A r? A 'VI T y?-? "-NT 17,T,4"NT TT A TIT TT A V r.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
signal before the AGC takes control. The strong signal produces
an undesirable "thumping" sound each time the AGC is operated.
A technique for avoiding this effect is to employ a clipper circuit.
Distortion products in the tone channel bands are removed by the
band reject filter. The output of the audio AGC circuit is passed
through a highpass filter with a cutoff frequency of 4000 cps to
remove any undesired components that might exist outside the
useful range of the telephone system.
c. Transmission Rate
It is desirable to predict the transmission rate as a
function of threshold setting. As an upper bound, we consider the
speech as uniformly distributed; that is, the probability of
triggering the transmitter in any 300 microseconds (3 kc) interval
is uncorrelated to what has preceeded. Clearly, because of the
111 syllabic character of speech, this assumption is not valid, and the
?
transmission rate we obtain will be an optimistic one serving to
show the upper bound. In addition, since there are pauses in the
speech, there are intervals of time when no transmisions can
occur.
If we assume a trigger rate, , uniformly distributed, then
the probability of triggering in an interval of time 4 t is given by:
Pt =c),6t
(1)
Using (I), the probability of triggering in the kth interval is given
by:
= [(1 - ().?) t
^
k- I
(2)
Letting t = kt and taking the limit of (2) as it & k --Poo
so that k L t t, we get
716.,
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Releaser2012/10/18 : CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
_t,0
,e,Ate
Dividing (3) by,at yields the
P(t) = ?re_6st
The mean value of p(t) is given
00
71
tp(t)dt
probability
by
00
i t 4,.(' -
o
a, to obtain
k-11
density function
ct
(3)
(4)
(5)
(6)
= / =
t
'o
We can solve fort using (5)
4t
. ,
00'
Pt
_
t is the mean time between the end of one character and the start
of the next character. If we denote the basic character length by
then the total waveform period, T, is given by:
T= +
(7)
For our system, 7 = 40 ms. We can solve (6) fort using the
speech statistics of Figure 19 in the FTL handbook, p. 874. Pt
is the probability the threshold is exceeded, considering that d t
is 300 microseconds corresponding to a 3 kc speech bandwidth.
The results are tabulated in Table 1.
Words per minute (wpm) is given by the relation:
10
wpm= (8)
Comparing the results from Table I to the empirical data
obtained from the feasibility models, it is apparent that Table 2. 3. 2
data are optimistic by about a factor of 4.
-17-
1,k 7' rrn T rq Afl r "N e?s T1 T T.T NT A T 4"N T A T
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
;
r 4
k
?
?
NOT RELEASABLE TO FOREIGN NATIONALS
Transmission Rate
1
.0
' a
r \J
a)
)-4
E
0.,
er; - r,-;
rn N)--i
+ + +
0
o'l
r--I
.
0
i
00
i
.,
E
.1--- c",
0` r--
(NJ ,--i .--4
co
in
00
NJ
a-
,... .
?-io
H
Cr) N? CO
r..... C.; If;
vt, in in
?'"
Cr;
.o
CO.
r---
s..C.3
0
r-I
,--4 ?
in
.--1
r-I CO Cr,
NI ---4
0 0 0
Crl
--4
0
CO
N
0
0
ICI
'1'
o
o
r-i
... ......-.
r''...
(N.1
o
0
,.
A.....
(3.4
Threshold
Setting
CY, 0 ,-..i
r.--1 r--I
r?.1
r?-?-?(
Crl
e-.-4
r.-4
r?-?4
-18-
........
?'V
;;d!' ?
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
d.. Tone Signal Pulse Shaping
Any or all of the five tone generators in the transmitting
unit may be operating at any one time. Each time a tone generator
is triggered, it emits a signal burst of 40 milliseconds duration.
It is of great importance that the energy spectrum of the pulse be
controlled and confined to as narrow a region as possible. If
energy is emitted at significant levels at frequencies outside of the
desired region, such signals could cause errors in the receiving
equipment. This problem is especially severe in view of the close
proximity of the tone frequencies to each other.
In the case under consideration, the tone frequencies are
located at 100 cps intervals between 1400 cps and 1800 cps.
Clearly, a minimum of energy should be generated in the spectrum
more than 50 cps removed from a given tone frequency; indeed, it is
essential that the cross-channel interference level (defined as the
unwanted energies in any one tone channel due to tones in all others)
be at least 20 db below the wanted tone level. Two approaches exist
to achieve the desired result:
1. Narrow band filters can be used to limit the energy
spectrum to the required extent, or,
2. The energy spectrum can be limited by shaping of
the pulse amplitude.
Approach No. 1 is undesirable for these reasons:
a. It is difficult to maintain center frequency and bandwidth
accurately over the temperature range;
b. Filters are large, heavy, and expensive;
c. Five filters are required for the five tones.
-19-
irrir
47,1 ?
T T7 A C. A T'YI rn /". Tni-N V` 1e-4 'TT NT A f-VAT A T C'
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
fy
NOT RELEASABLE TO FOREIGN NATIONALS
Approach No. 2 makes use of one wave shape generator and five
modulators. It has been shown in theory and practice that this
approach is practicable.
There exist two types of modulation envelopes for which
the energies in the undesired frequency spectrum are particularly
low; these are the "cosine-squared" and theuisosceles-triangle"
envelopes. Approximate amplitudes of the frequency functions of
these pulses can be gleaned from graphs in the I. T. & T. Reference
Data for Engineers, page 1013 and page 1014. It is seen that the
amplitudes of the two frequency functions are nearly alike. Thus,
it is reasonable that a preference for one or the other of the two
pulses can be based upon other considerations. A very good reason
for selecting the "isosceles-triangle" pulse shape is the fact that
such a pulse can be easily generated with good stability of amplitude,
symmetry, and period. The total unwanted energy of an isosceles-
triangle modulated pulse has been calculated to be 22 db below the
proper tone level; and the peak amplitude of an interfering signal in
the adjacent channel is calculated to be 28. 6 db below the proper level.
Measurements taken to date have consistently indicated a maximum
interfering signal level in the adjacent channel (due to one oscillator)
of 30 db below the desired signal. This result is in good agreement
with the theory. It has not been resolved whether the remaining
difference of 1.4 db can be ascribed to calibration errors, or whether
the slight amount of rounding at the beginning of the triangular ramp
is responsible for the improved performance. The interfering signal
level in one channel due to tone signals in all others should reduce
the margin between the interfering signal and the proper signal to
approximately 25 db. This latter figure has been checked experimentally
and is in agreement with the theory.
7 207
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NA TIONALS
2, 3. 3 Receiver
a. Coupling
The receiver is coupled directly to the telephone instrument
rather than between telephone and line, and consequently, is some-
what simpler than the transmitter coupling unit Figure 2. 3.3 illustrates
the receiver coupling technique. A notch filter identical to the trans-
mitter filter is inserted between the handset microphone and the telephone
hybrid. The rejection of the telephone hybrid is rather poor (around
15 db) and cannot be relied upon to isolate the local speaker's voice
from the receiver. A better hybrid is furnished with the receiver, but
variations in line impedance preclude any possibility of maintaining a
high degree of balance. The notch filter assures that the signal leaking
past the hybrid does not cause false alarm errors in the received
message.
b. Audio Section
Distortion is the most important factor in the design of the audio
unit. Distortion is held below 40 db using heavy negative feedback in all
amplifiers preceeding the bandpass filters. With the tones 30 db below
the transmitter keying threshold, it can be shown that speech distortion
becomes an important factor if the dynamic range of input levels is too
great (more than - 3.8 db). Since the rate of input levels is 20 db or
more, it appears desirable to incorporate a coarse manual gain control
in the receiver which the operator adjusts with the aid of a meter at the
beginning of each call. Figure 2. 3. 3b is a block diagram of the receiver.
The noise plus tone signal from the hybrid is fed to a manual gain control.
The control is set on voice peaks with the aid of a VU meter. An
emitter follower isolates the bandpass filter from the hybrid (and the
variable line impedance). This filter removes the voice from the
voice-tone combination and sends the multitone signal to a 60 db amplifier.
Separate emitter followers on the output of this amplifier drive the five
-21-
`r4
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
---
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
telye,
NOT RELEASABLE TO FOREIGN NATIONALS
?
?
tone channel filters. The filters each have a 3 db bandwidth of
100 cps and center frequencies at 1400, 1500, 1600, 1700, and 1800 cps.
The individual tones are delivered to the logic circuit which contains
an automatic threshold.
c. Threshold
A threshold must be provided to determine whether a mark
or space was sent at the transmitter. The original approach was to
provide a closed loop AGC on the tones after speech components are
removed by filtering. A loop was designed and built but variable phase
shift across the 100 cps bandwidth of the filters proved too severe to
allow a reasonable closed loop device.
The final circuit adopted uses a simple open loop servo which
rectifies the output of each channel filter and applies a common d-c
threshold to each channel detector equal to some percentage of the
strongest channel output, e. g. 6 db below the peak of highest level
channel. Since this d-c level is always less than the rectified input,
highly stable emitter followers may be utilized to provide reliable
and precise threshold action.
2. 3. 4 Packaging
Z. 3. 4. 1 Scope
This section outlines the detailed design plan for packaging of
the transmitter and receiver units.
2. 3. 4. 2 General Requirements
These units are to withstand a temperature range of 00C to +500 C.
The construction is to be of good commercial practice.
2. 3. 4. 3 Transmitter Unit
The following specifications were considered in the packaging of
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
this unit. Size: 9" wide by 15" long and 10" high; these are the
maximum outside dimensions. Weight: approximately 40 pounds.
Finish: Semi-gloss black enamel. A ten-foot external cable is
required. Duty Cycle: 1 hour on and 5 hours off.
Transmitter sub-units are as follows: Tally reader, Model 424C
modified to ACI drawing 100-1-03-080, card basket and EECO "T"
module chassis. ACI 100-1-03-078, filter chassis, ACI 100-1-03-079,
and power supply chassis, ACI specification No. 100-1-10-006.
ACI drawing 100-1-03-008 shows the over-all transmitter package.
The external dimensions of the box are 9" wide, 15" long by 10" high;
weight is 6 pounds.
The case for the transmitter is fabricated from 2 Zero Manufacturing
Company .063" thick 9" x 9" x 8" deep, drawn aluminum boxes. One
side of each box is removed and the two remaining halves are butt-joined,
welded and are reinforced by an open frame member at the joint to form
the case. A 1/8" thick aluminum panel is welded to the open end of the
case to form a rigid enclosed box. The panel has cutouts into which
mount the Tally reader and the card basket chassis. The power supply
is mounted to the case below the Tally reader. Two removable plates
are provided on the side of the case for access to the connectors located
on the filter chassis and for access to the power supply terminals.
The top of the transmitter is enclosed by a cover. The cover is held
down to the case panel by screws. A hinged cover over the Tally reader
permits access to the reader mechanism for replacing tapes and also
provides a storage area for the extension cable. The cover is fabricated
from two Zero aluminum boxes .063" thick, 9" x 9" x 2" with one side
of each removed. Two handles, one at each end of the case are provided
for carrying the unit.
-23-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
The blower in the Tally reader is used to exhaust the cooling
air from this unit. The cooling air enters through vents at the
plug-in chassis side of the transmitter, flows over the EECO "T"
modules, cards, filter chassis, power supplies, through the Tally
reader and out the opposite end of the case. Excluding the Tally
reader, approximately 10 watts of power is dissipated.
9II
The use of four deep-drawn aluminum boxes of the same 9" x
size with edge and corner radii gives the transmitter case a clean
external contour as well as structural strength. This method was
chosen for this design over more costly techniques such as all-welded
or one piece deep-drawn construction.
2.4.4 Receiver Units
The following specifications were considered in the packaging
of this unit:
Mounting to be compatible with standard 19" relay rack
Duty Cycle: one hour on, five hours off
The two major sub-units of the receiver, the Tally perforator and
the receiver control chassis, are packaged separately in standard
relay, desk-size steel cabinets approximately 14" wide x 15" deep x 23"
high (Bud No. CR1727/Wyco C2119) painted black. The cabinet weight
is approximately 40 lbs.
-24-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
3.0 THEORY OF OPERATION
3.1 General Design Requirements
No environmental requirements have been specified by the
customer. However, to assure reliability, a tentative temperature
range of 0oC to +50oC on the equipment is considered reasonable.
Allowing for temperature rise inside the package, circuits are
designed and tested for a range of 0?C to 65?C.
Transistors and semiconductor diodes are used in all circuit
designs to improve reliability and reduce size. All semiconductors
are derated according to their individual specifications for 650
operation. Resistor dissipation is derated by a factor of 5.
The equipment will he connected to a standard telephone system
using 600-ohm impedances and levels of approximately -17 dbrn.
3.2 Transmitter
3.2.1 General
The theory of operation of the transmitter is explained by
pursuing the flow of a signal through the system and observing the
operation of the individual circuits.
Input and output circuits, as well as the operation of the power
supply, are discussed first.
-25-
NOT RFT EASABLE TO FOREIGN NATIONALS
;
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
4t1;i1E/
NOT RELEASABLE TO FOREIGN NATIONALS
3. 2..2 Detailed Theory
3. 2. 2. 1 Power Supply
The transmitter power supply is an integral unit which operates
off 115 V of 230 V line voltage.
Theory of Operation of Power Supply - The EM1140B Power Supply
converts 115 volt, 60 cycle power to plus and minus 12 volts well
regulated DC power and -24 volts unregulated DC power. An
electrical schematic of the circuit is shown in the attached drawing.
One power transformer is used for all outputs to minimize weight
and size.
Since the two regulated sections are identical, only one will be
considered for this discussion. Input power is fed to the transformer
where the line voltage is stepped down. The reduced voltage is
rectified in a full wave bridge circuit and filtered by a choke input
filter.
The regulator section is. made up of a compari son circuit, error
signal amplifier, bias supply and series regulating transistor. The
comparison circuit is a bridge consisting of two resistors, two zener
diodes and a potentiometer as shown below.
a-.
Vt\
-
D
-26-
0
Output voltage
1\Ir-vr D'F'T 7ACA1kT 7 Tr- fl 1:27TrN NTATTnNA1,S
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
"t1.21t:
NOT RELEASABLE TO FOREIGN NATIONALS
The output voltage is applied across the arms CAD and CBD to
generate an error signal. The potential of point A with respect
to D is the output voltage minus the fixed zener voltage, and is
a potential that varies considerably for changes in output voltage.
The potential of point B with respect to D is approximately equal
to the fixed zener voltage and is fairly constant. Therefore, the
error signal which is the difference of potential between points A
and B will vary directly with output voltage. If the output voltage
increases, the error signal will increase and vice versa.
Reference is now made to the power supply schematic, Figure 3. 2. 2.
The error signal is fed to a transistor amplifier which drives the series
regulating transistor. In order to operate properly, the base of the series
regulating transistor must be biased negatively with respect to the emitter.
The bias is obtained from the -12 volt supply for the -1-12 volt regulator
and from a separate bias supply for the -12 volt regulator. When the
output voltage increases, the error signal changes so that the base current
of the series regulating transistor decreases. When the base current
decreases, the collector current decreases and the voltage across the
transistor increases. This decreases the output voltage and compensates
for the original rise in output voltage.
The third output is unregulated and consists of a bridge rectifier and
filter section to provide a low ripple DC voltage. The primary of the
transformer is made up of two sections, which are wired in parallel for
115 volt operation and in series for 230 volt operation. Overload
protection is provided by a 0.5 ampere circuit breaker when the power
supply is wired for 115 V AC operation; when it is wired for 230 V AC
operation, a fuze inside the power supply provides the protection.
-27-
-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
. ,
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
- OFT
NOT RELEASABLE TO FOREIGN NATIONALS
?
?
3.2.3 Input and Output Circuits
In operation, the transmitter is inserted in the telephone line
between the transmitter telephone and the line.
3. Z. 3.1 Input Circuits
The input signal to the transmitter is obtained from the mouth-
piece of the transmitter telephone whenever (a) operating power is
applied to the transmitter, and (b) the transmitter telephone is "off
the hook."
The operation of this circuit is as follows (Refer to Transmitter
Schematic Diagram 100-1-03-064):
Assume that operating power has been applied to the transmitter.
Lifting the phone off the hook completes the biasing circuit which
enables the microphone in the mouthpiece of the transmitter telephone.
The biasing circuit consists of resistors R1 and R4; capacitors C4,
C9 and C10; inductor Li; relay coil K1; and the carbon microphone of
the telephone.
With the exception of the carbon microphone (which is outside
of the transmitter) and the inductor L1 (mounted on the filter chassis),
all parts of the biasing network are mounted on TBi.
Resistors R1 and R4 determine the microphone bias current of
approximately 50 ma . Capacitors C9and C10 in conjunction with
the resistors R1 and R4 form a filter network for the bias current.
Inductor L1 represents the load impedance for the transmitter
telephone. The signal return path is provided by C4. Bridged
across Load L1 is the first amplifier stage of the transmitter (02,
Card No. 10).
3.2.3.2 Output Circuits
After the input signal has been processed by the transmitter, it
appears as the output signal of transformer T1, and is impressed
through 'capacitors C2 and C3 across L2 and R2* When the phone is off
;"
-28.-
L.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
^Ft- zY,11
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
?
the hook (and K1 thus is closed) L2 and R2 also serve to terminate the
incoming telephone line. It is seen that under this condition any signal
present across L2 and R2 is transmitted.
Transformer T1 and inductor L2 are mounted on the filter
chassis. R2, C2, and C3 are located on TB,.
3. 2.3. 3 Audio and AGC Circuits
(a) First Audio Amplifier (Part of Card No. 10) - The input signal
from the transmitter telephone appears at connector P101 pin 20 after
having passed through the input circuit (see 3, 2, 3. 1). From here, it
is routed to P10 - 17 on audio card No. 10.
P10 - 17 is the input terminal of the first audio amplifier. This
circuit consists of transistor Q2 connected in common emitter con-
figuration. The gain of this stage is 10 db V. and the maximum peak-to-
peak input signal occurring at the base of Q2 is 4.0 volts. The output
signal of the amplifier is capacitively coupled via P10 - 20 to P9 - 22
on the AGC card (card No. 9).
(b) AGC Loop (Card No. 9 and Part of Card 8) - The AGC card
contains most of the circuits that form the AGC loop. Two amplifiers,
which are also a part of the AGC loop, are located on the audio card
(Card No. 8).
The AGC loop is designed to maintain the average voice power per
syllable at a relatively constant level, regardless of fluctuation oF the
incoming signal level. This effect is achieved by providing a "fast-
attack-slow release" gain control characteristic in the loop. The
loop gain is sufficient to maintain the output signal level within L. 1 db
for input signal level fluctuations of - 10 db.
Gain Control is achieved by means of a variable voltage divider
- 2 9 -
NOT RELEASABLE TO FOREIGN NATIONALS
2,.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
RC
NOT RELEASABLE TO FOREIGN NATIONALS
consisting of R1 and R20. R20 is a "Varistor," i. e. , a current-
sensitive resistor. Its characteristic is such that an increase in
the varistor bias current causes the varistor AC resistance to de-
crease. It is apparent that, if the varistor bias current is varied
as a function of the voice power, gain control can be exercised
over the incoming signal.
To explain the operation of the AGC loop, assume initially
that no signal is present at P9-22. In this condition Q1 is reverse-
biased and only a small current flows through the resistor network
of R2, R20 and R3. This current causes the varistor R20 to
assume a better defined value of resistance than it can without a
bias current. Capacitor C2 acts as a filter to isolate the varistor
from the line, and simultaneously, provides the signal ground.
Assume now that a signal is introduced at P9 - 22, small enough
not to exceed the AGC threshold. This signal appears across the
varistor R20 which is typically 1 K ohm under the given minimum
bias condition. Transformer T1 is bridged across R20. It serves
two purposes: (a) it provides a 2:1 step-up of the signal voltage,
and (b) it allows the introduction of the DC bias current to Q without
2
loading effects or I problems.
co
Q2 operates as an emitter follower amplifier providing a high
termination impedance for transformer T1, and a low driving
impedance for the following filter. FL2 is a high-pass filter having
a cutoff frequency of 300 cps; its input and output impedances are
600 ohms; and its function is to prevent low frequency oscillations
within the AGC loop, as well as to reject undesired low frequency
signals which are not part of speech.
-30-
NC)T R1.1 FASATIT.F. Tn FnRF.TC,N NATTCWA IS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
IvREF
NOT RELEASABLE TO FOREIGN NATIONALS
Q3 is an amplifier which is biased to achieve a gain of 26 db
with a large dynamic operating range. The output of Q3 is
transformer coupled to the clipper circuit and to emitter follower
Q5. (Both circuits are on Card No. 8). The secondary winding
which
of?T2 is returned to ground in order to obtain a signal swing
is symmetrical with respect to ground. This characteristic
necessary for the operation of the clipper circuit.
is
The signal in the AGC loop next flows into amplifiers 0 and
Q6. Q5 is*an emitter follower which performs two functions; (a)
it provides the high impedance necessary to prevent loading of T2,
and (b) it provides isolation between the clippers and amplifier Q6'
This isolation is required to prevent signal distortion at T2 prior
to AGC attack when very large signal voltages (up to 20 volts peak-
to-peak) can occur at T2. Q6 provides the final voltage amplification
of the AGC signal prior to its application to the detector circuits.
Transformer T1 (on card No. 8) couples the AGC signal to the
detector driver Q (on the AGC card). This transistor is biased so
4
that its current output capability is limited only by its own DC current
amplification factor and the drive available at the base. Since the
voltage applied to the base of Q4 can become much larger than - 12 V
with respect to ground, diode CR2 is installed to limit the base voltage
to -(12 volts. The negative swing is limited to -12 volts by 0 since
its collector-base junction becomes forward biased at larger voltages.
The output of 04 is direct-coupled to detector diode CR4.
The entire detector circuit, including its output current amplifiers
05 and 06' is DC coupled. This approach eliminates the time-constant
problem inherent in AC coupled detectors. To achieve good AGC
-31-
r\ ID T Tr A C A Ta T L"' 'T' rl t-nDL-TC:1\T KT A 'T TrINT A T
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
at wit
?
?
NOT RELEASABLE TO FOREIGN NATIONALS
threshold stability over the temperature range from 0oC to 65oC,
compensating techniques were applied to the circuits. For instance,
to insure that the AGC attack and release time constants are stable
with temperature, transistor Q4 obtains its bias voltage from
Diode CR3. Since diode CR3 and the emitter-base diode of Q4 are
similar, their temperature characteristics are similar, and the
DC bias voltage applied to CR4 is very close to zero. This action
insures that the quiescent DC voltage on the memory capacitor C11
is always near zero volts.
The DC signal present on C11 is applied to two complementary
emitter followers in tandem. The use of complementary PNP-NPN
followers reduces the DC shift through the amplifier pair to a very
small amount. Transistor 0 is biased to present a very high impedance
-5
to the time constant network C11' R thus making the time constant
essentially independent of the load.
16'
At this point it should be noted that values of R15' R16' and C11
have been determined through listening tests. R15 and C11 control
the AGC attack time and the network C11, R16 controls the release
time. Either time constant can readily be changed by varying R15
or R16.
The output voltage of Q6 is approximately zero volts in the
quiescent state. In order to establish a definite AGC threshold,
a 5.0 volt zener diode is inserted in the output of Q6. The diode is
held in the break-down condition by a current through R19. Thus,
in the quiescent condition a voltage of +5.0 V is applied via low-pass
filter FL 1 to the base of the varistor control transistor Q1, holding
it off. The selection of a 5-volt zener diode as the AGC threshold
is prompted by two considertions: (a) a 5-V zener diode has an
-32-
NOT RELEASABLE TO FOREIGN NATTnNA r.s
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
'It
NOT RELEASABLE TO FOREIGN NATIONALS
extremely low temperature coefficient, and (b) operation with
high level signal makes any changes with temperature in the junction
voltage of Q1 insignificant as far as threshold stability is concerned.
The low-pass filter FL 1 has a ZOO cps cut-off frequency. Its
purpose is to insure stable operation of the AGC loop.
The
-
The AGC loop is designed to reduce fluctuations of 2- 10 dip
in the average voice level to less than ? 1.0 db. The average RMS
Voice level at P 9-22 is 0.5 volts. The AGC threshold is exceeded
with 0.1 volts average RMS.
As mentioned before, the attack and release time constants of
the AGC system were determined in listening tests. The results indicated
that an attack time constant in the order of ZOO milliseconds is desirable.
Because of this relatively slow attack characteristic, it is clear that
initial bursts of voice energy far in excess of the desirable levels are
applied to Card No. 8. In order to avoid overloading of the succeeding
stages prior to AGC attack, amplitude limiters are provided on Card
No. 8. The release time constant of the AGC is several seconds.
(c) Audio Card (Card No. 8) - The input signal is applied to Card No. 8
at P8 - 1. R1' a high-resistance ia the order of 20K ohms, feeds the
signal to the clipper diodes CR and CR2' The clipping level is
adjustable by means of potentiometers R and R4' with C1 and C2
providing the signal paths to ground.
The value of R1 is selected in test; this is a convenient means
of adjusting the gain of the system between the signal input and the
clipper diodes to the desired value.
The output of the amplitude limiter circuit is applied to an emitter
follower which in turn supplies the signal to the 4000 cps low-pass
filter FL 1. This filter operates between 600 ohm impedances and
-33-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
rdit
NOT RELEASABLE TO FOREIGN NATIONALS
limits the signal baridwidth to 4000 cps.
Having been passed through the filter, the signal is amplified by
transistor Q2. This stage has a gain of 20 db and can handle a large
signal swing.
Emitter follower Q3 is used to supply the impedance match and
power required to drive the band reject filter FL 3.
FL 3 operates between 600 ohm impedances; its reject band lies
between 1350 and 1850 cps; and its insertion loss is in the order of
7 db.
FL 3 is capacitively coupled to the base of emitter follower Q7.
This stage provides the driving power for output amplifier Q4 as
well as for low pass filter FL 4. The tone bursts are combined with
the transmitted voice power at the base of Q4. Bias network CR 4,
R26.
and C14 establishes the operating of Q4 in the most linear
region. Linear operation of Q4 is necessary in order to avoid the
generation of frequencies between 1350 and 1850 cps which have just
been removed from the signal spectrum by the preceding filter FL 3.
Signals in the reject band, generated by non- linearities of Q4, are
approximately 70 db below the operating signal level. Aside from
providing gain, Q4 establishes a constant, resistive system output
impedance of 600 ohms.
The output signal from Q4 is applied to hybrid transformer T1
on the filLer chassis. Most of the signal appears on the output
winding of Ti and is impressed across L2, R2 of the output circuit
as described in 6. 2.2. A small amount of the output signal appears
across the third winding of the hybrid transformer and is available
at J 10 - 21. Also present at J 10 - 21 are signals which may be
ctrri?rf,r
abit
-34- ?
Nryr P7T VAcART 7 Tr) .p-r-11:27Tr.NT T\TATInNT4 T
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part- Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
?
NOT RELEASABLE TO FOREIGN NATIONALS
arriving from the telephone line. In order to monitor these incoming
signals, an amplifier is provided on Card No. 10 which drives an
earphone.
It is obviously desirable to balance hybrid transformer T1
sufficiently well so that the level of the signal transmitted from 04
and appearing at J 10 - 21 is small compared with the signals
received from the telephone line. Unfortunately, it is not generally
possible to achieve a good degree of balance with the hybrid
transformer. This is due to the fact that the hybrid transformer
balance is a function of the impedance presented by the telephone
line to the transformer. SiI,ce nearly every telephone line presents
a different impedance, no single balance network will yield satisfactory
results. In the given circuit, a balance of approximately 12 db is
obtained iTy means of ba.lancing potentiometer R1. The amplifiers
on Card No. 10 are designed to prevent "blasting" the transmitter
operator's ear with his own voice. (d) Audio Card No. 10 - Card
No, 10 contai:is the audio amplifiers which are required to enable
the transmitter operator to listen to the incoming call.
The incoming telephone signal (along with the leakage of the
outgoing signal) appears at P 10 - 21 of Card No. 10. Audio amplifier
Q3 is isolated from the hybrid transformer by emitter follower Ql.
This insures that no distortion will be introduced into the line
regardless of the drive level applied to the amplifier.
A set of amplitude limiter diodes is connected across the
output of the emitter follower. This technique 'insures that the
signal level applied to Q3 will never exceed 300 triv peak-to-peak
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
9flL
NOT RELEASABLE TO FOREIGN NATIONALS
and, consequently, Q3 cannot be blocked by excessive incoming
signal levels. The gain of Q3, when the amplifier is loaded with
a 1000-ohm impedance earpiece, is approximately 15 dbv. This
gain is sufficient to raise the incoming signal to a comfortable
listening level.
It was indicated previously that under most operating conditions,
a significant level of the transmitter operator's voice can be present
at P10-21. To limit the signal level applied to the earpiece under this
condition, limiter diodes are connected across the earpiece. The
distortion introduced h.r the amplitude limiter diodes is not objectionable
as was borne out by listening tests.
3.2.3.4 Logic and Tone Generator Circuits
(a) Detector Card (Card No. 7) - The detector card contains
circuits which monitor the instantaneous signal level of the transmitted
? voice present at 38-22. After amplification and fullwave rectification,
?
the signal is applied to an adjustable threshold circuit which generates
an output pulse every time the signal exceeds the threshold. In flowing
from P8-22 to the detector card, the signal passes through filter F1-4
mounted on TB1 which is a part of the filter chassis. F14 is a low-pass
filter with a cutoff frequency of 750 cps. Its purpose is twofold:
First, it prevents noise spikes from triggering a tone burst when there
is no voice energy present to mask the tone; and secondly, it increases
the probability of masking a tone by triggering only on low-frequency
-36-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
JEW%
NOT RFI. EASABLE TO FOREIGN NATIONALS
components of the voice which exceed the thresholO. This
latter fact is based on evidence that the rate of decay of a syllable
composed of high frequencies is much faster than the rate of
decay of a low-frequency syllable. The output of the 750 cps low-
pass filter is applied to P7-1. Amplifier Q6 accepts this signal
and provides a voltage gain of approximately 28 db. Emitter
follower Q7 is direct-coupled to the collector of Q6' and provides
the pew er gain required to drive step-up transformer T1. The
center tapped secondary winding of T1 feeds two diodes connected
as a fullwave rectifier. In order to achieve good rectification
efficiency, the diodes are held in the conducting state during
quiescent signal conditions by a small current caused by a voltage
developed across Ri and CR3. This current of approximately 20
micro-amperes per diode is relatively independent of temperature.
This effect is achieved by deriving the detector biasing current from
a voltage source (CR3 and R1) which exhibits a temperature character-
istic similar to that of detector diodes CR1 and CR2..
Potentiometer R3 is required to permit balancing of the two
detector legs which otherwise would differ in their outputs because
of inherent differences in diode and winding impedances.
The detected output is developed across R4.
It is in order here to consider the reason why a fullwave rectifier
was inserted instead of applying the AC signal directly to the threshoki.
Fully/aye rectification of the audio signal provides two significant
improvements over the use of unprocessed audio: (a) it causes a higher
NcYi RFT,FASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
?
?
NOT RELEASABLE TO FOREIGN NATIONALS
tone transmission rate, and (b) it increases the probability of
masking the tone by voice energy.
To explain the mechanism of achieving higher tone trans-
mission rates by using fullwave rectification, note that through
fullwave rectification, use is made of the positive and the negative
signal excursions; and that twice as many signals are sampled by
the threshold.
The use of positive and negative signal excursions speeds up
the transmission rate by providing the possibility of exceedin g the
threshold several milliseconds sooner than in case of no rectification.
Thus, there is an increased probability that two, three, or even more
tone bursts can be transmitted per syllable, simply because of the
fact that threshold conditions are detected sooner, tone transmission
is initiated sooner, and less dead time occurs between tone bursts.
The fact that threshold conditions are detected without unnecessary
delay permits the initiation of the tone burst at the earliest possible
moment, thus increasing the probability of masking the tone burst even
in the case of very short syllables.
Returning to the description of the operation of the detector
circuits, it is remembered that the signal had been traced to R4.
Emitter follower Q1 employs a type 2N1420 transistor which features
a very high current amplification factor. The intention is to prevent
any undue loading of the detector circuit by developing a very high
input impedance to the base of Ql.
E mitter follower Q1 does not operate as a linear amplifier over
-38-
Declassified in in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
itn.006AiraffrtegtsrViv
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Ors,,VM
NOT RELEASABLE TO FOREIGN NATIONALS
the full dynamic range of the input signal. To understand its
operation, it is necessary to first explore the operating
characteristic of the threshold stage Qz.
The emitter voltage applied to Q2 is deriVed from voltage
divider R12 R13 and is adjustable between +12 and +6 volts.
Under quiescent conditions and until the signal reaches threshold
level, Q2 is held in saturation by a voltage negative with respect
to the emitter. This negative voltage is supplied through R9.
It becomes obvious at this point that follower Q1 need not operate
at all until the base voltage at Q1 approaches +6 volts. During
quiescent conditions when the base voltage of Qi is near zero volts,
R9 will continue to supply base current to Q2 as if Q1 were non -
existent. Indeed, under such conditions, Qi is reverse-biased
by the following mechanism. Assume that R12 is adjusted to +8
volts. Q2 is now saturated, and the base current of Q2 is
approximately 1.7 ma. Because of voltage divider action, the
emitter of Q1 will be at approximately +7 volts, indicating that Q1
is reverse-biased. Certainly, Q1 could be designed to operate
as a linear amplifier by decreasing R9 considerably in value.
However, this approach generates two problems: (a) it tends to
reduce the input impedance to Q1, thus loading the detector, and
(b) it causes excessive currents to flow through the base of Q
thus abusing Q2 and also causing unnecessary loading of the
divider R12 - R13' It is seen, therefore, that non-linear operation
of Q1 is desirable.
-39-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
g
?
NOT RELEASABLE TO FOREIGN NATIONALS
Suppose, now, that the input signal at the base of Q1 exceeds
+8 volts. Now 08 goes into the conducting state and its emitter will
'
follow the base. It is apparent that any further positive increase in
signal voltage at the base of Qi will cause Q2 to be turned off, and a
negative-going pulse to be generated at the collector of Q2. Its duration
is equal to the length of time that the threshold is exceeded. .
A threshold stability of better than +- 1 db is achieved through the
use of complementary transistors, i. e. , Q1 is an NPN, and Q2 is a PNP
silicon transistor.
The normal operating level (long-time average of RMS voice
voltage) at the base of Q6 is approximately 50 my rms. The wiper
of R12 is typically set to +8 VDC.
(b) EECO Rack - The EECO Rack contains most of the digital
logic elements of the transmitter unit, namely, a Schmitt Squaring
Amplifier and three One Multivibrators.
The input to the EECO rack consists of a negative pulse which is
derived from Q2 on the detector card and which is available at P7-8.
The pulse is capacitively coupled to the squaring amplifier Z1. Its
output is applied to a gate circuit consisting of RI, R2, and CRl?
If no pulse has been applied to this gate during the preceeding 40
milliseconds, the gate is enabled, and the output pulse from Z1 is
passed to one-shot Z2. This circuit provides a 20-millisecond time
delay. The normal (positive-going) output pulse of Z2 is passed
through an emitter follower (part of Z3) and is fed to P6-1 to be used
in the control of the ramp generator. The trailing edge of this pulse.
-40-
NOT RELF,ASART,F. Tn Fr1 R Vir.NF NT A l'InNT A T C
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for?Re"le?as; 2012/10/18: CIA-RDP78-03433A001100050001-5
fa IAA 1
NOT RELEASABLE TO FOREIGN NATIONALS
serves as the trigger for the second 20 millisecond one-shot Z4.
The inverted (negative-going) output of Z2 is applied to Pg-21
to drive emitter follower Q1 on Card No. 6.
At the moment when the timing interval of ?Z2 ends, Z4 is
fired which also operates as a 20 millisecond delay circuit. Z4
performs several functions. Immediately upon firing it delivers
a negative pulse to trigger the 4 rn sec one-shot Z5. Simultaneously,
the normal (positive-going) pulse is applied., via an emitter follower
(part of Z 3 ) to the gate which controls Z2. This positive pulse
inhibits the gate and prevents retriggering of Z2 until Z4 has timed
out. The very same pulse also clamps (through CR2) to zero the
trailing edge of the negative pulse of Z2. This technique yields a
fast fall time for Z .
The inverted (negative-going) output of Z4 is applied to P6-9
to be used for control of the tone generators. At the end of the
20 millisecond interval of Z4, the input to the gate is removed and
4,2 can be retrigge red.
(c) Ramp Generator - The ramp generator comprises
transistors Q2 and 03. The control signal for the generator appears
at P6-1. It consists of a positive pulse (-11 V to -3 V) of 20 milli-
second duration. In order to achieve good linearity of the trangular
waveform, the circuit operates between -12 V and +12 V.
The ramp generator operates as follows:
The -11 to -3 volt signal at P6-1 is shifted by a 15 volt
zener diode (CR1) to obtain a +4 to 4-12 volt swing at the base of Q2.
In the quiescent condition, the base of Q2 is at +4 volts. In this
condition, Q is saturated, its collector is at +4 volts, and Q
2 3
is cut off since its emitter is clamped to ground by diode CR2.
_41-
,11. Icr
, e
NOT RFI, FASABLF, TO FORFIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
.4tura,
NOT RELEASABLE TO FOREIGN NATIONALS
During this period, there is a constant +200 my on the emitter
capacitor of Q3.
Assume, now, that a pulse arrives which elevates the base of
02 to +11 volts. 02 now turns off, since its emitter is returned
to +10 volts. With Q2 off, the base of Q3 is returned to -12 volts
which causes 03 to go into saturation. The emitter of Q3, which
initially was clamped to ground, now begins to shift negatively,
trying to follow the base toward -12 volts. The time constant is
dimensioned so that the emitter capacitor of Q3 changes to - 1. 2
volts in 20 milliseconds. Good linearity of the ramp voltage is
achieved by limiting the emitter swing to -1.2 volts.
After 20 milliseconds of charging, the base of Q2 returns
to +4 volts. At this moment Q2 again goes into saturation and Q3
is cut off. The emitter capacitor of Q3 now charges from -1.3
volts toward +12 volts through the emitter resistor of Q3. This
charge time is so dimensioned that after 20 milliseconds, the
voltage on the capacitor is +200 millivolts. At this moment, diode
CR2 conducts, and the ramp is completed.
The output of the ramp generator is amplified by emitter
follower 05 and applied to all five of the tone frequency modulators.
Note that the ramp signal will be fed to the tone modulators every
time that the logic circuits are triggered. This is true regardless
of the tape reader input. The ramp signal appears at P6-8.
(d) Tone Oscillator and Modulator (Card No. 1)
There are five tone oscillators and five modulators in the
transmitter. The modulators are identical, and the tone oscillators
-42-
RET
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for R-ereiie-267-2710/18 : CIA-RDP78-03433A001100050001-5
4E:WM 1
?
NOT RELEASABLE TO FOREIGN NATIONALS
differ only in the inductance values used in the frequency determining
networks. Consequently, only one oscillator and one'modulator
will be discussed.
The tone oscillator is a conventional two-transistor type with
the frequency determining elements connected between the emitters
of the transistors. This design makes the oscillator frequency
essentially independent of the load, provides a relatively low output
impedance, a large voltage swing, and good starting characteristics.
The oscillator is controlled by a transistor switch (Q3) which
normally clamps the collector of Q4 to ground, inhibiting operation.
To start the oscillator, Q3 is reverse-biased. This allows the
collector of Q4 to swing positive and oscillation begins. The tone
oscillator signal is coupled to the modulator and impressed across
diodes CR and CR . The tone modulator is a very simple circuit.
3
It operates as follows:
Assume that initially no input is present at the audio channel.
Thus, the ramp generator is in its quiescent state. Remembering
that the quiescent voltage at J6-8 is slightly positive, it is seen that
the emitter of QI tends to be above ground by about 400 mv due to
the junction drop of emitter follower Q1. This voltage insures that
diodes CR2 and CR3 of the modulator are conducting, and that no
output can exist from the modulator.
Now assume that at t=0, a signal is received which calls for the
generation of a ramp, and that, simultaneously, the input from
the tape reader requires the generation of a certain tone.
The following events now take place:
-43-
Declassified in Part- Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Sttlit 1
NOT RELEASABLE TO FOREIGN NATIONALS
(1) The proper tone oscillator is turned on and its output
is fed into the modulator; (2) the emitter of Q1 follows the negative-
going ramp, and an increasing reverse bias is established across
diode CR the oscillator signal is therefore always clipped in its
positive swing, ,but it will build up to the equivalent of the
instantaneous reverse bias of diode CR2 in its negative swing.
Signals above this threshold are short-circuited by the large
emitter capacitor of Q1. Since the amplitude of the tone is larger
than that of the ramp, a 100% modulated wave results.
It is desired to multiplex the outputs of the five modulators.
This cannot be accomplished passively for reasons of isolation
and impedance matching. Amplifier Q5 is therefore inserted after
the modulator. This stage is AC coupled and has unity voltage gain.
In order not to load the modulator, a high input impedance must be
achieved for the amplifier. The ensuing requirement for a high
impedance bias circuitry calls for the use of a silicon transistor
as the amplifier. There is one additional consideration which
requires attention; namely, the fact that the modulator output
consists of a negative-going pulse only. Thus, the average output
of the modulator is a DC voltage, the magnitude of which is a
function of the repetition rate of the pulse. If the amplifier were
to follow the average output to any degree, the linearity of the
amplification would be lost, and the output wave shape would no
longer be triangular. It is necessary, therefore, that the time con-
stant of the coupling network be of the order of one cycle of the tone
oscillator. The implication is that now the waveform of the carrier
tone suffers some distortion. However, since the unwanted
frequencies are far removed from the desired spectrum, it is easy
SECI ET
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
iaLL;
NOT RELEASABLE TO FOREIGN NATIONALS
to reject them. Low-pass filter C7' C8' L1 on Card No. 6, having
a cutoff frequency of 2000 cps, serves this purpose.
(e) Tone Control Circuits -
The tone control circuits are located on Cards No. 1 and 6.
In conjunction with the tape reader, these circuits initiate and terminate
the tone bursts at the proper instants. Throughout the following
paragraphs, reference will be made to the timing diagram of Figure
3. 2. 2, 2. 4.
To explain the operation of the tone control circuits, assume
first that no audio signal is present at the transmitter input.
Consequently, the system is in its quiescent condition, and the
following DC voltages are present at the control circuits:
Transistor Q3 on Card No. 1 is in saturation, causing the tone
generator to be "OFF". Q3 is held in the "ON" condition by a
voltage of +4 volts which is derived through a zener diode from
emitter follower Q6 on the Ramp Generator Card No. 6. If the
tape reader senses a hole in the tape, the t4 volt potential is
present also at diode CRi. This diode is "ON" because of the
return of R2 to +5 volts. Clearly, with +4 volts applied to the base
circuit of Q3, the transistor will be in saturation.
Consider the second possibility, namely, that the tape reader
does not sense a hole in the tape. In this case, no DC continuity
exists between Q6 on Card No. 6 and diode CR on Card No. 1.
However, Q3 on Card No. 1 is "ON" whenever there is no DC
continuity through the tape reader, 1. e., whenever the reader does
not sense a hole in the tape.
-45-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
? --
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
tiziumorn
NOT RELEASABLE TO FOREIGN NATIONALS
Consider, now, the final possibility, namely, that an audio
signal has exceeded the threshold, and that the first 20 millisecond
one-shot Z2 has fired. Call this time t .
At time t , the emitter of Q6 on Card No. 6 swings from - 3 V
to -11 V; and, due to zener diode CR3, the voltage at P 6-22
swings from +4 V to -4V. Assuming that the tape reader senses
a hole, DC continuity exists between P 6-22 and P 1-1. Since Q6
on Card No. 6 represents a very low source impedance for negative-
going signals, it is capable of discharging capacitor C1 on
Card No. I very quickly, thus reverse biasing Q3 and permitting
the tone generator to start. The above voltage conditions persist
until Z2 times out after 20 milliseconds. Let this time be called
t1. At t1, the voltage at P 1-1 swings to +4 volts, reverse biasing
CR1.
Remembering that a tone burst lasts for 40 milliseconds, it is
apparent that the tone generator must continue to run for 20 milli-
seconds past time t1. This operation is effected by dimensioning the
time constant R2 x CI> 20 milliseconds. Thus, the tone generator
tends to run for longer than tI + 20 milliseconds. The exact end
of the 40 millisecond tone burst at time t2 is effected through
transistor Q2. This circuit operates as follows: Let t2 be the
instant when the second 20 millisecond one-shot times out. At
time t2' the voltage at P6-9 swings from -11 V to -3 V and transistor
Q4 is turned "ON" through the resistive voltage shift network R2
and R20. Switching of Q4 causes a sharp negative pulse to appear
at the base of Q2 on Card No. 1. The effect is that Q2 conducts
hard for a brief instant. Its current gain is sufficient to remove
quickly the residual negative charge from C1 (which held Q2 "OFF"
- 4 6--
ECik,
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NiaLl
NOT RELEASABLE TO FOREIGN NATIONALS
and the tone generator "ON"), and to replace it by a positive charge,
which causes Q3 to conduct and squelch the output of the tone
generator. Thereafter, Q3 is held "ON" through R2, and Q2 is
held "OFF" through R14 on Card No. 6.
Since the tape reader participates in the tone control operation
only during the first 20 milliseconds, (from to to t1 the succeeding
20 millisecond interval (t1 to t2) can be used to advance the paper
tape reader by one step. Thus, ample time is available for the
advance of the mechanism, and for the settling of the tape reader
contacts.
_47-
\r?
4.4.010
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
JLU
NOT RELEASABLE TO FOREIGN NATIONALS
3.3 Receiver
3.3.1 General
An over-all block diagram of the receiver is shown in
Figure 3.2 and a detailed schematic diagram in Figure 100-1-04-034.
Input signals consisting of notched speech plus information tones
are applied to the receiver via a public telephone system. The
message information detected by the receiver is contained within the
1300 to 1900 cps frequency band. Message signals are transmitted
to the receiver at low amplitude, about 27 db below the rms level
of the 300 to 3000 cps voice signal.
The received message consists of alpha-numerical characters
in standard five-baud teletype code. The five bauds comprising
each character are transmitted simultaneously in the form of
40-millisecond pul ses having a triangular shape. A sEparate
carrier frequency is used for each of the five pulses. The five
carrier frequencies are:
1400 cps
1500 cps
1600 cps
1700 cps
1800 cps
Bandpass filters are used in separate amplifier channels to
separate the individual code pulses from the composite signal.
The detected and amplified pulses are supplied to logic circuits
which drive a Tally Paper Tape Perforator. The receiver is
capable of processing input code groups at any rate up to 25
-48-
MrIT PTT 1=k T TC n1:2 -P"' 1 YMT NT TnIcr T
Declassified in Part- Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
v.
4;
NOT RELEASABLE TO FOREIGN NATIONALS
characters per second, or approximately 250 words per minute.
The receiver output is a punched paper tape suitable for use on any
standard teletype printer.
The receiver utilizes a special sliding-threshold agc circuit
which provides a constant output from each of the five detector channels
over a dynamic range of 10 db. A manual gain control circuit and
VU meter provide for adjustment of the input signal amplitude to
compensate for the wide variations between local and long-distance
transmissions.
An Engineered Magnetics Model EM 1140B regulated power
supply provides operating voltages for the receiver.
3. 3. 2 Detailed Description
3. 3. 2. 1 Input-Output Circuit
The input-output circuit of the receiver consists of the outpu
amplifier and band-rejection filter which notches the voice spectrum
of the receiving station operator to prevent errors in the received
message; the input circuit for the receiver; and coupling circuits
which permit the use of a single-pair transmission line for the
two-way conversation.
The voice signal born the receiver telephone is applied to
pin A of J3 in the receiver. This signal is generated' in a high-
impedance crystal microphone which develops a peak-to-peak
voltage of about 200 millivolts across a load impedance of 75 K ohms
presented by R9 and the input impedance of Ql, both of which are
located on TB2. The conventional carbon microphone normally
used in the telephone is replaced to minimize distortion and to
reduce the possibility of false alarm errors in the received message,
GRE
-49-
NOT RFTF AS A 'P, T TC) FC)RFTC;N N A TTC)NA
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18 : CIA-RDP78-03433A001100050001-5
E 411C, 1
NOT REL EASABLE TO FOREIGN NATIONALS
The need for low distortion in this application is obtained from the
following considerations. The spectrum of normal speech is highly
peaked in the region below 700 cps and decreases rapidly at frequencies
above 1000 cps. When the speech is distorted (distortion in a carbon
microphone is as much as 30%), however, the spectrum tends to flatten
and additional energy appears above 1000 cps. Since the voice signal,
due to the local speaker, is already 20 db stronger than that of the
distant speaker, and since the notch filter (FL1) has finite attenuation
in the message bandwidth, distortion in the local speaker's voice
can cause a significant number of errors to occur whenever the
receiving operator speaks. The use of a linear crystal microphone
minimizes this effect.
Additional isolation of the local speaker's voice from the receiver
input is provided by the hybrid transformer T2 which is described
in detail in the later part of this section.
Returning to the input circuit, the voice signal fro m the
microphone passes through emitter follower Q1 on TB2. This
signal is applied to the base of Q2 where it is amplified to a level
of about 850 millivolts peak-to-peak. After further.Amplification
in 01 (audio card No. 1), the speech signal passes through the
band reject filter FL1 which attenuates speech energy in the 1300 to
1900 cps message bandwidth when the function switch is in the
RECEIVE OR STANDBY position. In the normal position the
un-notched signal at the base of Q1 by-passes the notch filter
via Sl-Al and is supplied directly to Q2 base on audio card No. 1.
Q2 is an pmitter follower that isolates the band-,reject filter from
line impedance variations. The output of this stage is fed through
a 330-ohm resistor (R9) to hybrid transformer T2. R9 plus the
-50-
NOT RELEASA.RLF, TO PnR-RIT',N NATTC)NAT
Declassified in Part- Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT R.ELEASABLE TO FOREIGN NATIONALS
source impedance of emitter follower 02 add to yield a resistance.
of about 350 ohms, which is just equal to the resistance of gain
control R1 (front panel). Hybrid transformer T2 is exactly
balanced when R2 (chassis, Figure 8.1) plus R6 (TB2) just equals
the line resistance. When this occurs, the voltage drop across
R6 - R2 is exactly equal to and 1800 out-of-phase with the voltage
across pins 7 and 9 of T2 Consequently, no current flows in R1
(gain control) due to a signal applied to pin 4 of T2. (Since the
-line impedance varies for different telephone circuits, the balance
is not perfect and some feed-through occurs.) Instead, the signal
is coupled through T2 to T1, where it is delivered to the telephone
line for transmission to the transmitter station.
The incoming voice signal arriving from the transmitter station
is applied to the telephone earpiece through terminal D of J3. In
the "NORM" mode, this voice signal is applied directly from
terminal 9 of the transformer T2 to J3-D, through contacts of
Sl -A2 and Sl-AS. In the "STANDBY" and "RECEIVE" modes,
emitter followers 01 and 02 on terminal board TB1 are placed
in series with the voice signal output at T2-9. In these modes, the
received signal is also applied to GAIN CONTROL R1 in the
receiver input channel. The emitter followers provide isolation
between the receiver input channel and the earpiece to minimize
distortion and voice feed-through. 1
Note 1: The earpiece in the receiver telephone acts like a low-
efficiency microphone without isolation; unnotched speech
can appear at the input to the information channel and can
cause false alarm errors.
-51-
T r \ 13 T L' A C A T.) T L' mr tr\rD T T T NT A 9' T CVNT A T C
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
0-11
?
IA.,
NOT RELEASABLE TO FOREIGN NATIONALS
In the "STANDBY" and "RECEIVE" modes, voice input signals
received over the external telephone transmission lines at J3-E
and -F are coupled through transformers T1 and T2 to the emitter
followers and the receiver GAIN CONTROL, RI.
3. 3. 2, 2 Receiver Input Channel
The receiver input channel consists of a gain control, a signal-
level monitoring circuit, and a preamplifier. GAIN CONTROL R1
provides a means of adjusting the input signal level to compensate
for the variation between local and remote transmissions, For
local signals, the maximum peak-to-peak amplitude of voice
input signals may be as high as 3 volts; for input signals from remote
transmitters, the input may be as much as 20 db below this level,
or about 300 millivolts peak-to-peak.
To permit proper adjustment of GAIN CONTROL RI, a signal-
level monitoring circuit is provided. The monitoring circuit consists
of emitter follower 05, amplifiers 06 and 04, a peak clipper circuit,
and VU meter.Ml. The transistor stages are mounted on audio
card No. l, the clipper and meter circuits on terminal board 'TB].?
The voice input signal at the arm of GAIN CONTROL RI is applied
to the monitoring circuit through emitter follower Q3, which also
drives the preamplifier circuit.
Emitter follower Q5 at the input to the monitoring circuit drives
a two-stage cascade amplifier, 04 and 06, the output of which is
applied to METER ADJUST potentiometer R14. This potentiometer
provides a means of calibrating the VU meter. The amplified output
at the arm of R14? is coupled by C6 and C5 to VU meter Ml.
Positive and negative peak clippers are connected across the
. 1/4
-52-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
output to protect the meter against high-amplitude peak signals.
The clipper diodes, CR1 and CR2 on terminal board TB1, are
back-biased by voltage divider resistors R4 through R7, so that
only high amplitude peaks of the signal cause the diodes to conduct.
The dial of VU meter M1 is marked to indicate the optimum
signal level, and GAIN CONTROL R1 is adjusted so that the meter
reading coincides with this mark. Since the ratio between average
level of the voice signal and the low-level message signal is nearly
constant, this adjustment brings the input signal amplitude within
the limits of the dynamic AGC range of the receiver.
The preamplifier circuit consists of bandpass filter FL2, and
a three-stage amplifier and emitter follower located on audio card
No. 2. The bandpass filter passes the low-level message pulses
within the 1300 to 1900 cps band, and rejects all components of the
voice signal.
The three-stage preamplifier, consisting of transistors Ql,
Q2 and Q3 on audio card No. 2, provides approximately 60 db of
gain for the 1300 to 1900 cps signal components. An over-all gain
of approximately 100 db is required for the receiver; however, the
preamplifier gain is limited to 60 db in order to maintain linearity
and to prevent intermodulation distortion. Intermodulation distortion
might occur if two tones were received simultaneously, or if an
exceptionally strong signal outside the message band were not
attenuated sufficiently by the bandpass filter. This could cause a
false output from one or more of the five detector channels, and in
turn cause an incorrect character to be punched by the Tally
perforator.
-53-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18:dA:kiff8-0-343.3A001 100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NAM ONALS
The input signals to the preamplifier consist of triangular pulses
of 40-milliseconds duration, at five different carrier frequencies,
spaced 100 cps apart, from 1400 to 1800 cps. Transformer coupling
is used between the three amplifier stages and between the third
stage and output emitter follower Q4. Thermistor RT1 in the emitter
of Q3 provides compensation for gain variations with temperature
changes. The preamplifier is disabled by the operating mode selector
switch, except in the "RECEIVE" mode. When Si is set to either
"NORM" or "STANDBY", the emitter of amplifier Q4 is shorted to
ground through contacts of S1-B5. In "RECEIVE" mode, the short
is removed allowing Q4 to conduct.
The output of the preamplifier circuit is applied from the
emitter of Q4, through J6-22 to the inputs of the five audio detector
channels.
L31,3 Audio Detector Channels
The five audio detector channels provide further amplification
of the message pulses. All five detector cards are identical except
for the bandpass filter at the input of each card which selects only
one of the five modulated frequencies. Each bandpass filter has
a 3-db bandwidth of 100cps. Since the five detector cards are
identical, only detector card No. 1 will be described.
The input signal to the detector card consists of the amplified
message signal from emitter follower Q4 in audio card No. 2.
This signal is applied through J7-1 to emitter follower Ql, which
provides an impedance match between the preamplifier output and
bandpass filter FL3, and isolates the input circuits of the five
detector cards.
ft, eft ? ir .??r _
Declassified in Part - Sanitized Copy Approved for Release-2012/10/18 : dA-RDP78-03433A001100050001-5
tt.t.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
The bandpass filter FL3 has a bandwidth of 100 cps, centered
at 1400 cps, so that only the 1400-cps message pulses are coupled
to the first amplifier stage Q8. Amplifiers Q8, Q2, and Q3 in
cascade provide 40 to 50 db of amplification of the 1400-cps signal.
Potentiometer R3 at the input to transistor Q2 provides a gain
adjustment for the amplifier circuit. The narrow bandwidth of the
amplifier eliminates all intermodulation distortion, since no two
signals within the 100-cps band can generate a cross-product that
might be interpreted as a valid signal.
The amplified 1400-cps signal at the collector of Q3 is applied
in parallel to emitter followers Q4 and Q7. Emitter follower Q4
drives the diode detector, and Q7 supplies one of the five input
signals to the AGC circuit. (See paragraph3; 5) Emitter follower
Q4 and transformer Ti provide a low-impedance charging source
for the RC network at the output of the detector. Detector diode CR2
is back-biased by the AGC circuit, through the secondary winding
of Ti. Capacitor CI in AGC card No. 2 provides an AC ground
return for the secondary of Tl. The AGC circuit establishes both
a minimum and a dynamic threshold bias level on the detector.
Capacitors C4 and C7, connected across the detector output,
are charged very rapidly to the peak value of the input signal, through
the low-impedance source of Ti and Q4. A delay is introduced,
however, by the RC time constant of resistor R18 and series
capacitors C5 and C8. This delay provides sufficient time for the
AGC circuit to respond to the input signal, and also prevents any
stray noise pulses of short duration from exceeding the trigger
threshold. Thermistor RT1 forms a part of the voltage divider which
establishes the bias level for diode detector CR2 and switching
-55-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
t'S -14 r
tit sf
NOT RELEASABLE TO FOREIGN NATIONALS
transistor Q5. The thermistor acts to maintain this bias at a
constant level over a wide range of temperature variation.
Switching transistor Q5 is normally conducting. When an
incoming signal pulse exceeds the AGC threshold, CR2 conducts
and a positive-going signal is coupled to the base of Q5, cutting
off the transistor. This produces a negative-going pulse at the
collector of Q5 and at the output of emitter follower Q6. The
negative-going output of Q6 is coupled via P103-22 to a Schmitt
trigger at the input to the control logic circuits.
During any 40-millisecond interval, a negative-going output
pulse from any of the five detector cards represents a binary 1,
or true signal; conversely, the absence of a pulse represents a
binary 0, or false signal. The combination of the five outputs
represents the teletype code for a particular alphabetical or
numerical character in the incoming message.
33d4 Control Logic Circuits
The control logic circuits respond to the outputs of the five
detector channels to provide simultaneous 1-millisecond output
pulses to the perforator. The logic circuits include a 10-millisecond
delay which allows sufficient time for detection of signals in all
five channels, and a 1-millisecond one-shot multivibrator which
synchronizes the outputs of the five channels.
Assuming that a signal above threshold level is detected in
channel 1, a negative-going output signal will be coupled from the
emitter of Q6 in detector card No. 1, through P103-22 to the
channel-1 Schmitt trigger Zi. Because of the delay in the detector
circuit, the transition from the false to true state is relatively
-56-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
RE I
NOT RELEASABLE TO FOREIGN NATIONALS
slow: the Schmitt trigger, however, provides an output pulse with
a very fast rise time. This output pulse is coupled to the channel-1
20-millisecond one-shot multivibrator, Z6. Identical Schmitt
trigger circuits, Z2 through Z5, and 20-millisecond one-shots,
Z7 through Z10, are provided for the other four detector channels.
The output of the channel-1 20-millisecond one-shot, designated
Z6, is combined in an AND gate with the Z14 term from a timing
circuit. The Z6 term becomes true when the 20-millisecond one-
shot fires; the Z14 term becomes true 10 milliseconds later, for a
period of 1 millisecond. During the 1-millisecond period when
both Z6 and Z14 are true, an output pulse is coupled to the perforator
by emitter follower Z11-A, At the end of the 20-millisecond period
of Z6, the circuit returns to the quiescent, or off, state, ready to
respond to the next incoming signal. In the same manner, 1-milli-
second output pulses are delivered simultaneously to the perforator
from all other channels where a signal was detected.
The timing circuit, which synchronizes the five channels,
consists of an OR gate, a 10-millisecond delay one-shot, a 1-milli-
second one-shot, and associated emitter followers. The output
signals from the 20-millisecond one-shots in the five signal channels
are combined in the OR gate, comprising CR1 through CR5 in the
logic card. If any one of the inputs to the OR gate is true, the
10-millisecond one-shot Z13 is triggered. This one-shot provides a
10-millisecond delay to allow time for all five detector channels
to respond to an input signal. Normally, all five bets for each
character in the message code are transmitted simultaneously;
however, the transmission time for each of the five carrier
frequencies may vary several milliseconds when the signals are
transmitted over long distances.
-57-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Gii,7
NOT RELEASABLE TO FOREIGN NATIONALS
At the end of the 10-millisecond delay, the trailing edge of the
one-shot output triggers the 1-millisecond one-shot Z14. The
1-millisecond output pulse from Z14 is applied to emitter followers
01 and 02 in cascade, on the logic card. Emitter follower 02
provides the D1, D2, D3, and D4 and D5 terms for the output AND
gate in each signal channel, as described above. The output of 02
is also coupled by capacitor Cl, through P110-8 and J1-40, to the
Tally perforator. This signal is applied through a delay circuit
to the sprocket drive in the perforator to advance the tape one step
after the appropriate code has been punched.
13.45 AGC Circuit
The AGC circuit in the receiver provides a sliding-threshold
voltage for each detector channel, to provide positive triggering of
the control logic circuits over a 10-db dynamic range of input
signal variations.
As described in paragraph 132.3, the amplified signal at the
collector of Q3 in each detector card is applied to emitter follower
Q7. The output of 07 is coupled to the AGC circuit through
transformer T2 and diode CR3, which comprises one element
of an OR gate. The combined signals from all five detector circuits
are applied through the OR gate to the base of emitter follower Qi
on AGC card No. 1.. The level of AGC voltage developed by the
circuit is determined by the input signal having the highest amplitude.
Two emitter followers, 01 and Q2 on AGC card No. 1, are
connected in cascade to provide isolation of the input circuit and a
low-impedance source for the AGC detector. The response of the
AGC circuit is sufficiently rapid to follow the linear rise of the
-58-
T L' I L"' AC ART rl t--e--? T r" NT NT A '7' T 1\T A T C
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
triangular modulation envelope of the input signal.
Diodes CR1, CR2 and CR3 provide temperature compensation
for emitter followers 01 and Q2, and AGC detector diode CR4.
Diode CR4 rectifies the input signal, and the negative AGC output
voltage is developed across a filter network comprising C3 and R6.
Two additional emitter followers, Q3 and Q4, provide sufficient
power to drive the five parallel output stages. Emitter followers
Q5 through 09 provide a separate AGC output circuit for each of
the five signal channels. A potentiometer in the emitter circuit of
each output stage permits separate adjustment of the AGC voltage
supplied to each channel, to provide compensation for variations
in gain between channels, and for differences in input signal levels
at the different modulation frequencies. Diodes CR5 and CR6 in
series between Q3 and Q4 provide temperature compensation for the
output stages.
The five outputs from AGC card No. 1 are coupled to AGC card
No. 2, where each is combined with a fixed bias circuit. The fixed
bias circuit establishes a minimum threshold voltage for the
associated detector channel. For detector channel 1, the minimum
threshold is established by the setting of potentiometer RI, which
forms part of a voltage divider connected across the -12 volt supply.
The minimum threshold is adjusted to optimum level and applied
through isolation diode CR2 to transformer Tl on detector card No. 1.
When the AGC voltage from Q9 and potentiometer R19 in AGC card
No. I is less than the fixed bias level, the AGC volt age has no effect.
However, when the AGC voltage exceeds the fixed bias or minimum
threshold, CR2 is biased off and the AGC voltage is applied to
,59-
;v1:5..3,1"
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
the channel 1 detector. Diode CR1 provides temperature compen-
sation for CR2, and acts to maintain a constant level of fixed bias.
The bias circuits for channels 2 through 4 are identical to the
channel 1 circuit.
The action of the AGC circuit is such that signals having a
peak amplitude less than the minimum threshold established by
the fixed bias circuit cannot produce a detector output signal.
Signals having a peak amplitude greater than the minimum threshold
produce a detector output signal which is held nearly constant over
an input range of 10 db.
-60-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for 2-0-1-2/10/18 : CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
4.0 PACKAGING
4.1 Design of Transmitter Sub-Units
4.1.1 Reader
The Tally reader purchased per ACI specification drawing
No. 100-1-10-018 is 7-7/8" x 8-1/2" x 6-1/4" high; weight,
approximately 13 lbs. This unit is flange-mounted and secured
to the case panel by 6 No. 8-32 screws. The reader unit panel
and cover are fabricated from rigidamp material supplied by
Barry Controls of Glendale, California to quiet the operation of
this unit. To complete this unit, a control box, consisting of a
toggle switch (standby and transmit positions), two indicator
lights, one for power, the other for end of message, and two circuit
breakers are moulted on the Tally reader panel. A small aluminum
box is used, only requiring a hole through the Tally panel for the
cables and for the mounting screws, with all the control components
mounting in the box.
A supply bin for 30' d 11/16" teletype paper tape is mounted
to the read head mechanism to handle the pre-punched tape.
4.1.2 Card Basket
Card Basket and EECO "T" Module chassis size is approximately
8-1/2" x 6-1/2" x 6"; weight, approximately 6 pounds. This chassis
is fabricated from aluminum in the shape of an open box with 10 taper
pin Kennedy printed circuit connectors mounted at the bottom. Ten
card guide grooves are machined in the sides for the cards. Along
one side of the chassis is mounted a bracket holding the Seven EECO
"`P' modules (2 spares). Flanges are located at each end for mounting
this chassis to the case panel by 4 screws. A cable and plug connect
-61-
NOT PF,T,FASABLF. To FORFTnN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
OLORL
NOT RELEASABLE' TO FOREIGN NATIONALS
this unit to the filter chassis. The plug-in cards have components
mounted between terminals and use point-to-point wiring. EECO
"T" modules are purchased items. Refer to ACI drawing 100-1-03-078
for details.
4.1.3 Filter Chassis
This chassis is frabricated from aluminum plate. The trans-
former and filters are mounted to this chassis by a hold-down plate.
A small component board and three connectors are also mounted on this
chassis. The connectors form the junction box interconnecting the
sub- chas sis .
The approximate size of this unit is 6-1/2" x 8-1/2"x 4";
weight, 9 pounds. Refer to ACI drawing 100-1-03-079 for details.
4.1.4 Power Supply
The power supply is purchased per ACI specification 100-1-10-006.
For details refer to this specification.
4.2 Design of Receiver Units
4.2.1 Tally Perforator Unit
The Tally perforator 420 and drive package 1424 is packaged per
ACI Drawing 100-1-04-074 on a 10-1/2" x 19" panel. The perforator
unit is enclosed in a sound-proof box with a hinged cover over the
punch head and chad box for access to the tape supply. The tape
supply reel is mounted to the right of the perforator on the front of
the panel. For details, refer to the ACI drawing. The perforator
and drive package unit is ordered from Tally Register Corporation per ACI
Specification 100-1-10-019. Approximate weight of this unit is 35 pounds.
4.2.2 Receiver Control Unit
The receiver control is packaged in a 10-1/2" x 19" panel x 14"
deep chassis and mounted in a standard 10-1/2" cabinet modified to
-62-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
.1aELT. 1
NOT RELEASABLE TO FOREIGN NATIONALS
ACI drawing 100-1-04-056.
Mounted on the chassis are one card basket for 10 plug-in
cards, one power supply (ACI 100-1-10-006), 2 component boards,
the EECO plug-in modules, 7 filters, 2 transformers, connectors,
and the necessary inner component cabling.
Switches, meters and controls are mounted on the front panel
of the unit. Refer to ACI drawing 100-1-04-056 for details.
-63-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Ott-
NOT RELEASAB LE TO FOREIGN NATIONALS
5.0 PERFORMANCE DATA
Several tests were conducted on the breadboard and model to
determine performance of the system. No precise performance
specifications have been established by the customer but the following
basic requirements appear reasonable from a systems viewpoint
based on results obtained in the feasibility program:
1. Information rate: 25 to 50 words per minute
2. Voice quality: telephone toll quality
3. Tone level: set below threshold of audibility
4. Error rate: less than 1 percent
5. Ambient temperature: 0 to 50?C
Some of these parameters are quite subjective and difficult to
measure. For example, voice quality and tone audibility are very
? hard to measure in quantitative terms. The error rate is dependent
?
on a number of factors - tone level, line distortion, pulses and
crosstalk on the line, etc. The approach followed was to first
determine a reasonable tone level based on listening tests by several
observers and then to make error rate measurements at and around
this tone level for different values of simulated line distortion. In
addition, tests were made over several local telephone circuits with
measured performance characteristics.
5.1 Breadboard Tests
5.1.1 Output Level
This test was performed to determine the AGC control chb_racteristics
using CW input. Figure 5.1.1 shows that the dynamic range of the AGC
is 20 db and the variation in output level for this range of inputs is 1.0 db.
-64-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
ffirmsnrii"
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
5. 2 Keying Rate for Various Speakers
Keying rate as a function of input level was measured using
several different speakers. In each test, the input level was varied
to determine the center of the AGC control range. This data is
plotted in Figure 5.1.2.
5.1. 3 Tone Audibility
In this test, each aural observer listens to the transmitter output
(using headphones) which contains both a recorded voice and the
information tones. The highest tone level at which each observer
can no longer hear the tones is noted. Each observer listens to
several different types of voices including both male and female.
The same number of tones is keyed each time the threshold is exceeded,
a much more severe condition than actually exists in normal operation.
Table 5.1. 3 below lists the audibility thresholds for the observers used
in the test. It is believed that slightly higher tone levels than those
indicated would be safe because the observers had a priori information
on the characteristics of the tones.
TABLE 5. 1. 3
Tone Audibility
Threshold of Audibility -db below rms speech power
Observer One Tone
Two Tones Three Tones
-30 -30 -27
-30 -33 -36
not audible -27 not audible
1- 27
-27
not audible
-27
65-
-27 -33
-30 -30
-27 -30
-24 -27
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
4LIAL
NOT RELEASABLE TO FOREIGN NATIONALS
5. 1. 4 Error Rate
A test was performed to determine the effect of line distortion
on system error rate for several tone levels around -27 db. Four
one-minute recordings of different speakers were used to make the
statistics invariant. Line distortion was simulated using an amplifier
whose peak distortion could be set using a switch. Figure 5. 1.4
summarizes the data obtained in this test. The error rate for the
breadboard is slightly poorer than that obtained using the model.
This is probably due to the improper rejection characteristic of the
filter used in the transmitter breadboard.
5. 2 Model Tests
5. 2. 1 Keying Rate vs. Line Voltage
The effect of variations in line voltage on information rate was
measured using the transmitter model and a tape recorded input.
This data is tabulated below:
TABLE 5. 2. 1
Keying Rate vs. Line Voltage
Keying Rate
117 VAC
- Char. /Minute
105 VAC
128 VAC
290
287
303
210
212
232
200
220
225
240
235
245
5. 2. 2 Keying Rate Vs. Temperature
Information rate was checked for the transmitter model over a
temperature range of 0 to +50oC. Table 5. 2. 2 summarized the
-66-
A Ptl e`s ./Nfl n T e?-? TT TT A TT /1 TTA T
25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5 .
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
o
NOT RELEASABLE TO FOREIGN NATIONALS
results obtained. A tape recorded input signal was used.
TABLE 5.2.2
Keying Rate Vs. Temperature
Speaker +25?C
+55oC
0?C
Max. Change
From 25?C
Value
251
242
257
3.6%
195
183
193
6.2%
162
146
153
9.9%
222
197
220
2.3%
5.2.3 Error Rate Vs. Distortion
The error rate tests performed on the breadboard (5.1.4) were
repeated using the model and a tape recorder. With the tones set 27 db
below rms speech power and a peak line distortion of 1/2% (-46 db),
the error rate is 0.3%. Data for other distortion values is listed below
in table 5.2.3.
TABLE 5.2.3
Model Error Rate Measurements
Peak Distortion
(db below peak speech)
Total No. of
Characters
Transmitted
Total No.
Of Errors
Character Error
Rate - %
-62
1143
11
0.1
-54
1162
0
0
-40
1203
30
2.5
Note 1: This error occurred at the end of one message and may be due
to transient from switching to "Standby."
-67-
Declassified in Part - Sanitized copy-kp-prO;eific;rieTe-ase-20-1-271-0/16-: CIAr2R131';8--03433A001100050001-5
?,.
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18 : CIA-RDP78-03433A001100050001-5
F
NOT RELEASABLE TO FOREIGN NATIONALS
5.2.4 Local Line Tests
Error rate was also measured with the model transmitter and
receiver connected through the telephone exchange. Two different
configurations were tried
1. Diamond-to-Diamond and
2. Diamond - to- Triangle
Before these tests were run, frequency response, distortion
and insertion loss were measured for both lines. Frequency response
data is plotted in Figure 5.2.4. Neither line exhibited any measurable
distortion (output distortion was identical to input distortion which was
less than 0.1% or -60 db.) Insertion loss for Di-Di is approximately
10 db while the Di-Tri is 20 db. Each line was dialed five times and a
five-minute passage transmitted each time. The results are tabulated
below.
TABLE 5.2.4
Error Rate over Local Circuits
Total No. of Total No. of
Line Characters Sent Received Errors Error Rate
Diamond/Diamond 860
0%
Diamond/Triangle 1092 57 5.2%
Errors in the Di-Tri test were apparently due to clicks from dialing
at other locations which are either inductively or capacitively coupled to
the line used in the test. Error-free copy was obtained on several occasions
when the line was quiet.
5.2.5 Error Rate Vs. Receiver Gain Control Setting
In all of the previous error rate tests, the receiver gain control
was accurately adjusted to establish proper tone levels at the input to
-68-
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
4J
?Fkf
NOT RELEASABLE TO FOREIGN NATIONALS
the detectors. Since this control is set on voice peaks using the
VU meter, it is subject to some error which depends on the care
exercised and skill of the receiving operator. This test was made
to establish both the resettability error using several operators and
the variation in error rate due to improper setting of the gain
control. Table 5. 2. 5a lists the reset errors for five operators using
two different recorded voices.
TABLE 5. 2. 5a
Deviation in Gain Control from Optimum Setting Using
Inexperienced Operators
Operator
Average Gain Error
(Three measurements on each voice)
Voice A Voice B
-1. 2 db
?0.8 db
-1.0 db
+1. 8 db
?0.6 db
-1. 8 db
+0.4 db
-0. 1 db
+2.0 db
-1. 1 db
With a keying threshold 6 db below the tone peak and a minimum
keying threshold 8 db below nominal, the receiver AGC range is
- 4 db.
In every instance, operators were able to set the gain well within
these limits as Table 5. 2. 5a shows. However, as a check on
Receiver AGC performance, error rate was measured at these
extreme ends of AGC control. Table 5. 2. 5b is a compilation of error
rate vs. gain control setting with a simulated line distortion of 1%.
-69-
NOT RELEASABLE TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
TABLE 5.2.5b
Error Rate vs. Receiver Gain Control Setting
(1% simulated line distortion)
No. of Characters
Gain Setting Sent
Optimum 900
4 db high 828'
'4 db low 860
-70-
No. of Errors
Error Rate
Received
- %
21
2.3
32
3.8
13
1.5
NT (Yr P T LiZ A Ft T f-) TrINT 1\T A MTCINT A T C
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release- 2012/10/18: CIA-RDP78-03433A001100050001-5
41:1111,L
NOT RELEASABLE TO FOREIGN NATIONALS
? 6.0 CONCLUSIONS
?
?
All tests conducted on the transistorized equipment developed
under this contract confirm the feasibility, demonstrated previously
with a vacuum tube version, of using audio tones to carry covert
information via standard public telephone installations. A significant
reduction in the transmitter size was accomplished by packaging and
by grouping the five tones in a single wide slot between 1300 and 1900
cps as opposed to the five individual slots spread between 700 and
2400 cps employed in the feasibility model. In addition to size
reduction, improved quality in the notched speech was obtained.
As a consequence of the extreme importance of system security,
tone levels approximately 10 db lower than those used in the feasibility
transmitter are presently being used. For this reason, receiver gain
is 10 db higher making it more susceptable to noise and line distortion.
A method for obtaining the same degree of security at higher tone
levels (6 to 8 db) through use of a magnetic tape delay technique
(ACI proposal No. P-027) has been proposed. It is believed that this
modification is essential to assure reliable long-range operation.
-71-
Ncyr R F. T AS A B T ,F, TO FOREIGN NATIONALS
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
.4-/a?/-*/VC Y,
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
_
?
?/'7
?20
?30
-to
?--ss-c)
6C)
-ga
?90
?
in 1-- mum so l4 0 um m m0 4111114111161 I'll
mom a mu ono s m MOM MO mama
1111111111111111 I III 11 Op 11O lin ni Oulloolmmou mem
n so nauseous f m mum m ? mom Immo nirmummiammumum
mammommumm m lik MEIN In mul mini imi 111641.14:.-IMETEMIMMOMM
1111111. 1.1 II M"On111 m Mon allm 1111?1111mrelimIllogla
imam iii 1111110111 imp AIM OA :120411111161gii
I
annum meg Ns MIMI 16 sopposmonommogns nous ? mornmon
SOMMEms MEM 111 MUM o i OM MOOMMMOMMMOIMOMff MIBMINOMMO EMMIIMM
SMMEM snmansi mumaillen sim mmimmummarminummommi OIMMOMM
ma......_IMMO InflOOKI IBM ME OW MIIIIIM MN NomMoMOMOOMMUMMUM
Wig MOW ? Mani 1.1 MI I
el 111011111r1=
NM' IsillIlln Ile III I s Inillnam mmo.mmum.muminsinum
Mlle Islimmil imonsmanum sum m A mm mum mmummummom mum
mlim i 1111111 111111111A I III IMI II! 1111111 1111111111111111111
mommososomm?mmommummmomommonms moms
11111 MIMI MUM MO ? I II MEW OM MO ? MEMO ME I MMUMMMOMMEM
111 sili:memismIlmill mineillmo islusilir nomomommInsun
II ? f MONEMME MI ? NI MI TIMM Mff ? II ? MOMMIMMII MINN
"11.11.11111 1411111 MOM
OM
MEMO 11111I MOOMMULMI 1111?111111111 111111111141111111111111111111
ill lel "111111111111111,11 1 II 111111 III 1111 11111111
Miff WM MMIBIrf MOOS w OMNI ? OrnmomilIME MMONNAMMEMMOIMM
Milfoolf10 PM MOM MEMEMMIIMM DOMEMME
FIN 1.1111111111111 11
MMEMITEMONOMOMMO MUM MIMMOMEMMI
iOM IMBIMMOMOIN meow . ppm ?
onssonnlarymIsnmom AmEginnAloil mmlimmulls1111666 mil
ns 666666' I 111466611 11111411111 milml 11110 Ingillm 1111
ins "61711 I m .111 nor ., ..11 MO ? MO III ..111.. I 116.111MMOMMOUR
III ,p .111.1111,116.111111111 ill 11111...111111 1111.11111111
Ar"""I'llii" r''' ilitiri Ilia lalliiiii1114111221111 mg
111111111ogill ? PI III di 4" 1iiiiii* ?
^ 101116.4.1.0 ?
mIllInisll Illsol =AI I m mnimps
Ilps les mg_ nub im las =MG e
11111111111111111111:1111111111"1"11111:11 1101111111 "m
maim NON i ? MI mil MIIM Milla II f MOM OM
^ am ma i IIIIIIII 1111 111111111111111111111 111111111111111111
11 ?
OM
6 mon
mom
61111111 III A 11 IP 1 AIII1MM Oil Mill AI AA
I MO
II MO OM Imm nr s ma
4111111111 11111 41111'11111 IIIII 1 I
g ? ? b
mil 111111111 km al . m ?
1111141111 611111 lill lam
MORM 111 II 11. II Map
I MI MR I Mal I IIMEM OM I NM Ile
1 ION MIL II ? OM ?
rli IBM ? If
IIM WOE
II 1111 i
1 I J _ II IIM
? 11 1 MIN II dir
liwr ,, Emu 11.1
1,we 1 ..-+,-,11 311
I o
mIIIIIIIIII III II L 1 ,_
? s
11411 miss ? ? II
.c.p-val m qm_11 MI
I B If ? ij g I oda rim =
1
.IIHWillill"411111"le
1 ml111k-' ii-iiJ
1
1m mil
mIlms III 1 imm.4.
1 4..
? ms?
IIINIIIIIIIIL
mi m mml m
'? N. 4:1011112ilie 4 4 4:1
Ajli'' ''afrbl g milipaic 4110-. li III
?imaiiillilimilm mom
in
?
? ?
?
mum ??m mum Ems ?
i/i Z31
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
tl
J-11 mom
immil I IIIIIIIIII III
I: CI .1--.17r sup ?
?? ? terns's= mu ? i I
mu -Al MOINNIONN MO MO M
IMMOMA-L:AllailIMMMUMMIMMIMO M
10?11111ffirrillannin?Min
ivliria...m Iiirmemy
?lila.11. la rieni IlMilthd
l___S
I 1
.:j__!.
II
? MN __,III
gill III ill II
IR
e4111 1 IU 111111111 ? milli
1 I
rinal Idlle?P ?
MOM .1110 m
IMMO 111 MOM ? ? II M 11
I
MMEMOMIIRMII lir OR
11111111:" le I
InIM III
aisa
ME
III ? 41 III .
All= 111 '
E.....Lsa 1
nun
.. 1
80
40
?
2
I 11111.1.1111111111 1..
?II. "1"919111111111111.11
111111111111111111111111111111
1-114.1.111115111111111..111
?11111141.111.111111111111111
IIII 11111111.111111111111111111
.. .. ..1........1. ... muss
g mism m maga ? mummummommEE
1111111111 111111 JIM? 11
il
Essmung..
1111111111 musiggres
mi molgo EE
ms Elm 11 ?
Ilm
lil IP i !Film
1111 1111 Mil
Iffl
1 1.1111ma
um ..
I
? a ? ?
IL I . I II I ? M
MO Et OM ql
1
/0
?
Z, .1
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
" rflir
r
VkL.
NOT RELEASABLE TO FOREIGN NATIONALS
C.G.,)P t\',
- T
Declassified in Part - Sanitized Copy Approved - - - -
fOrReiea-;e261271071-8 : CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
NOT RELEASABLE TO FOREIGN NATIONALS
co
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
Declassified in Part - Sanitized Copy Approved for Release 2012/10/18: CIA-RDP78-03433A001100050001-5
cn
0
>
0
LC1
.
r-
-! I
P4 ?
r=c' 'y. 1 C4 I
i Q.)
. c4 1
1 1
1 i
6, _l_._
- si 1,, , 1 --------! i ,
.
? 1 . ,24 NJ ! ?
) a e?-.1 'El 0
1...
(V
o i
1 _.,
? c ._)11 _
,
1
1 (?) T
1
;