JPRS ID: 10159 WORLDWIDE REPORT TELECOMMUNICATIONS POLICY, RESEARCH AND DEVELOPMENT

• Document Type: CREST
• Collection: JPRS
• Document Number (FOIA) /ESDN (CREST): CIA-RDP82-00850R000400080013-5
• Release Decision: RIF
• Original Classification: U
• Document Page Count: 68
• Document Creation Date: November 1, 2016
• Sequence Number: 13
• Content Type: REPORTS

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R440400080013-5
FOR OFFICIAL USE ONLY
JPRS L/10159
3 December 1981
Worldwide Report
TELECOMMUNICATIONS POIICY,
RESEARCH AND DEVElOPMENT ~
(FOUO 17/81)
~
FBIS FOREIGN BROADCAST INFORIUTATION SERVICE
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00850R000400480013-5
NOTE
JPRS publications contain information primarily from foreign
newspapers, periodicals and books, but also from news agency
transmissions and broadcasts. Materials from foreign-language
sources are translated; those from English-language sources
are transcribed or reprinted, with the original phrasing and
other characteristics retained.
Headlines, editorial reports, and material enclosed in brackets
are supplied by JPRS. Processing indicators such as [Text]
or [Excerpt] in the first line of each item, or following the
last line of a brief, indicate how the original information was
_ processed. Where no processing indicator is given, the infor-
mation was summarized or extracted.
Unfamiliar names rendered phonetically or transliterated are
enclosed in parentheses. Words or names preceded by a ques-
tion mark and enclosed in parentheses were not clear in the
original but have been supplied as appropriate in context.
Other unattributed parenthetical notes within the body of an
item originate with the source. Times within items are as
given by source.
The contents of this publication in no way represent the poli-
cies, views or attitudes of the U.S. Government.
COPYRIGHT LAWS AND REGiJL,ATIONS GOVERNING CWNERSHIP OF
MATERIALS REPRODUCED HEREIN REQUIRE THAT DISSEMINATION
OF THIS PUBLICATION BE RESTRICTiD FOR OFFICIAL USE ONI,Y.
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400084013-5
FOR OFFICIAL USE ONLY
JPRS L/10159
3 December 1981
WORLDW IDE RE-PORT
- TELECOMMUNICATIONS POLICY, RESEARCH AND DEVELOPMENT
(FOUO 17/81)
CONTENTS
,
USSR
- Integrated Istok Analog-Digital Commtmications System Test
Results
(L. Ya, Misulovin, et al.; ELEKTROS`JYAZ�, Sep 81) 1
WEST EUROPE
ITALY
Equipment Used at Rai Mt. Venda Transtnitting Station
(Giulio Paolo Pacini; ELETTRONICA E TELECOMUNICAZIONI,
May-.7un 81) 14
Silicon Avalanche Photodetector for Optical Communications
(M. Conti, et al.; ELETTRONICA E TELECOMUNICAZIONI,
May-Jun 81) 48
_ a _ [III - WW - 140 FOUO]
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFF'ICIAL USE ONLY
USSR
UDC 621.395.345.3
INTEGRA.TED ISTOK ANALOG-DIGITAL COMMUNICATIONS SYSTEM TEST RESULTS
Moscow ELEKTROSVYAZ' in Russian No 9, Sep 81 (manuscript received l;i May 80)
pp 4-10
[Article by L.Xa. Misulovin, V.V. Makarov and Yu.A. Baklanov: "Test Results of the
Integrated Analog and Digital Communications Systent: the 'Istok' Analog and Digital
Unif ied Communications Network"]
[Text] The development of the first analog and digital communicaGions system with
centralized control within the bounds of 3 network section has been completed in
the USSR and the nations of socialist cooperatioTi, wnere this system is called the
- �1unif ied (for the USSR and GDR) analog-digital communications system", the "TStok"
YeSS ATs. The system design is the result of the cooperation of two CEMA member
nations: the USSR and GDR. The rnajor switching components, basic circuits and soff-
ware were developed by USSR specialists while the GDR specialists designed the basic
- structure, including the connectors, operational prcacess Fundamentals and designer
- documentation for the control complex.
Tao tes"" regions were set up to conduct thorough tests of the unified analog-
digital communications system equipment: in the Istrir.skiy rayon of the Moskovskaya
oblast and in Berlin. Both regions were crear_ed through the joint efforts of USSR
- and GDR enterprises. The alignment and testing of the equipment ia the Istrinskiy
- test region were performed by USSR specialists, while in the Bcrlin region, it was
done by USSR and GDR specialists.
The tests of both zones were completed at the beginning of 1980. Tte results of
the tests made it possible to correct the design documentation for the sytem and
turn it over for industrial production in the USSR and GDR, as well as work out
the system rrogram software for series prodiiction and work up the operational
documentation. But getting the "Istok" YeSS ATs in series production also poses
new problems: the creation of a single programming center and a center for train-
ing operational personnel.
The successful completion of tests of the Istrinskiy eest zone of the "Istok"
YeSS ATs is the result of the self-sacrificizig labor of equipment designers and
specialists of the Istrinskiy RUS (regional communications center).
1
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400080013-5
FOR OFFICiAL USE ONLY
Work is aow underway on thP utilization of the "Istok" YeSS ATs as a municipal ATS
[automatic telephone exchange], for which the maximum capacity of the system is
being expanded iip to 3,000 - 10,000 numbers.
The first industrial r.:odels of the "Istok" YeSS ATs will be installed in the
Ogre RUS of the Latvian SSR (Liyelvarde) and in Saratov.
The Organization of the Istrinskiy Test Zone. The structural configura-
' tion of the Istrinskiy test zone communications network of the unified analog-
digital communications system is showm in Figure 1; the zone is inscribed within
the bounds of the existing telephone network. The unified analog-digital communi-
- cations system test zone, just as all rural telephone networks (STS), is designed
on a radial junction center principle; included in it were the key exchange, OPS
(type 1) and three terminal exchanges OS1 - OS3 (type 3). The key exchange was
incorporat2d as a central exchange while the OS1 - OS3 exchanges were included as
_ terminal offices. In order Lo assure completeness of the tests, the key exchanges
of the test zone, in contrast to thz key office of a conventional rural telephone
network, has a direct output to the regional ATS's and AMTS's [automated long-
distance telephone exchanges] of Moscow.
Several junction line (SL) trunk groups were set up between the central exchange
and the key exchanges. The OS1 [terminal exchange 11 was connected to the key
excilange via physica.l junction lines (FSL), and a common control channel, OKU-FSL,
was set up to form the remote control system for OS1 via two two-wire physical
junction lines. The OS2 and OS3 terminal offices were connected to the key ex-
change via lKY [PCM, pulse code modulation] channels, where two channels were used
in both cases to check the stand-by system. Linear PCM channels were set up using
the "Zona" equipment, where the IKM-30 transmission system was used as the termi-
nal stations.
A1.1. oF the cali. routings which wera set up in the communications network of Figure
l. are indicated in Table 1. As can be seen from the table, 48 different call
_ roueings were established in all. We note that the control of all calls, as well
as the offering of additional services (DVO) and all technical operations (TER)
were accomplished under the control of the central control unit (TsUU), installed
at the key exchange. It was necessary to design program::software (PO) with a
volume of 200 Kbytes for the central control unit to realize all of the functions
enumerated above.
Suppl.emental kinds of services were made available to subscribers in the test
zone, which are shown in Table 2. The types of technical operations Xealized in
the test zone, as well as brief description of their contents, are given in Table 3.
A block diagram of the prototype exchanges for the unified analog and digital com-
munications system which were installed in the Istrinskiy zone is shown in Figure
2 with the equipment indicated, which was housed in more than 20 bays. The follow-
ing symbols are used in the drawing:
2
FOR OFFICIAL USE OPILY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00850R000400480013-5
F'OR OFFICIAL USE ONLY
-
t.AfocK6a Moscaw
J Ncmpa ~ ~ PA7C
Istra u~ �(1) .
PMTC \ ~ ,4MTC (4)
(2) o ' 
~ _ _ -k,^`j k3,PATC .
- }N
onc
6~
' / ~ . .
~ yCA0'fN6/B 0#03HQVeNlLA: Symbols:
mPOm yM PCt4 channel
Onbimnaa ~ona ~ � -fg~ ~
(7) f cc A y c.~o~rpo~cKne'~ _ _ _ _ aKy " opcA (11) .
. i OCi 1r0;1an,V Ty (12)
07cn (13)
(nvc /1aBnaBc- OC7\f10) I
Kapwo6oaa � I
I (9) i
Figure 1.
Key: 1. TsS = central office;
2. RMTS = regional long distance exchange;
3. RATS = regional automatic telephane exchange;
4. AMTS = automatic long distance telephone extahange;
5. OPS = key telephone exchange;
6. OS1 = 'Lerminal exchange 1;
7. Test zone af the analog-digital unified communications
system;
8. Pokrovskoye village, OS3 [termi.nal exchange 3];
9. Se*tlement of Pavlovskaya Sloboda;
- 10. OS2 = terminal exchange 2;
11. Common control channel for physical junction lines;
12. Voice frequency telegraphy channels;
13. Physical junction lines.
AH [P.K] - subscriber sets which serve for the subscriber line loop;
A11 [AT,j - subscriber line.;
AU,O [ATsO] - the analog-digital equipment of the IKM-30 transmission system;
E.AII [BAL] - block of subscriber lines; the switching equipment to which a sub-
scriber line is connected; intended primarily for concentrating
(compressing) the telephone load;
3
FOR OFF[CIAL USC ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
f,!-{H [BKK] - the set connection unit for connecting service sets which are busy only
in the stage of making a call connection to junction line sets, which
are occupied both 3uring the stage of making a call connection and
during the conversation;
FOI1T [BOLT] - terminal line channel unit of the "Zona" equipment; the line signal
repeater., remote power supnly for unattended repeaters and service
intercom system are installed in it;
5CII [BSLj - junction line block; the connection equipment to which the junction
line sets and service sets of various kinds are connected; it serves to
provide access to the trunk groups for various instruments;
HAT [KAT] - subscriber call fee determination set, intended for transmitting the
call fee pulsPs to meters installed at the subscriber;
HHC [KKS] - conference call unit to provide for a joint conversation with up to
_ eight conference participants;
f-iCJI [KSL] - junction line set, which powers the subscriber microphone in the case
of an external call (with respect to the exchange in which the sub-
scriber is incorporated);
HCI183 IACJV3 [ICSLV3, KSLI31 - junction line sets (incoming and outgoing three-wire
units) serve for the transmission and reception of
interaction signals with an electromechanical auto-
matic telephone exchange;
FiC-(lTH [KS-PTN] - switching system for the connection of the tonal dialing
receiver (PTN) to the service set (SK); the necessity for it is
due to the fact that the time the PTN is occupied is less than
the time that SK is occupied; t',
OKY-$CI1 [OKU-FSL] - The common control channel for physical junction lines; the
direct and back-conversion of the control signals transmitted
in ')oth directions between the key exchange and terminal
exchange 1 are accomplished in it: a series quasi-ternary code
into the line and a parallel two-level "one of four" code in
the direction of the exchange;
(1TH [PTN] - the touch tone dialing receiver which serves for receiving the touch
tone dial signals for a number from a telephone set with a touch tone
keyboard;
- f1YY [PUU] - the peripheral controller for interfaeting the high speed central
control unit to the relatively slow scaitching equipment;
CH [SKj - a service set, used in the process of making a call connection between
an exctiange and a subscriber line; it f eeds out the tone and ringing
, signal and receives the tten-step code dialing signals;
CHF [SKG] - DC service complex, which serves for transmitting and receiving DC
signals and is used in the process of establishing call connection of
an exchange with a junction line;
4
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
F'OR OF'FICIAL USE ONLY
CF{y [SKCh] - the same as the SKG, but only for frequency [AC] signals;
YH [UK] - the contr ol complex - the central controller for the entire analog-
digital un if ied communications system;
YCFS1 [USK1] - channel interface; the direct and back-conversion of "time--space"
. type digital signals are accomplished in it: time multiplexing of the
individ ual channels is provided in the direction of the hhannel end
and spatial multiplexing is provided in the section of the BSL
[junction line connection block]; in both cases the signals are
digital. Moreover, the USK1 performs the functions of a common
control channel via the PCM channel (OKU-IKM); in this case, the
control instrtiictions undergo direct and back "series-parallel" con-
version: in tile direction of the PUU [peripheral control unit], the
_ instruction is represented by a parallel code and in the direction
of the line ciiannel, by a series code;
YCF42 [USK2] - the same as USKl, but without the function of the coimnon control
channel via the PCM channel;
YCH3 [USK3] - the same as the USK1, but without the "time--space" conversion
function;
WH [ShK] - the patch cord equipment, intended for powering the microphones of the
~ calling and called subscribers in the case cf call connections within
an exchange.
The operational princ iple of this cammunications system is described in [1].
TABLE 1
~ ceT` gcc nu
~ ~Ietwork YeSS ATs
u
I cT;,,1� tIQ5 16 7
LeT, 1
~
Symbols used: are for a direct connection;
Circles: the call connection
passes through the key
exchange;
Squares: the cAll connection
passes through the central
office;
Diamonds: the call connection
passes through the key
exchange and the central
office;
Dashes: the call connections
which were not incorporated
in the test program.
C~A 2 � 0(3)
WtCT9. 0 ~
tCT6 P. E a y
NciPa L: u X ~ ~ ~ ~ ~
t
i4
eSS
4onc~
+I +I
+1
-4-I
+iol
+i+
z
~
oc
~
1
+
o
ol
o1 0 1
o~o
1
1
1
1
~
~
~
o
w
ocz ~
0
4�
0
+
0
0
0
0
D
~
r 5
] 02 I
~ O I O I-1- I
O I
0 I 01
O
O
'
'
(1)
u{,
I-F I O
I G I O I
I
I_
/
~
I
~
~
T ~
PMTC
IOI o
I� I o
� ~
s
U�~~
,
-
o
~m

PATC
(lO
I
o
I o I o
I-
I-
-
I-
I
I
i-
-

/IMTC OI o I e I o
~
11
I
~
S
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
[Key to Table 1]:
1. Existing network of Istra;
2. Moscow;
3. Lang distance natwork;
4. OPS [key exchange];
5. OS1 [terminal exchange 1);
6. OS2;
7. OS3;
8. TsS [central office];
9. RMrS [regional long distance telephone exchange];
- 10. RATS [regional automatic telephone exchange];
11. AMTS [automatic long distance telephone exchange].
The tests of the "Istok" prototype were broken down into three stages, each of
which had its own program and testing procedure:
--The line tests, during which the operability of the system when making itidividual
call connections was checked and the electrical parameters of the speech channel
were measured;
--Experimental operation with the simulation of a subscriber load, which was
created by the staff workers performing the tests; the task of tl:is stage was to
check the operational capability of the test zoiie under dynamic conditions;
--Test operation with actual subscriber. In this stage, 94 telephone sets were
connected to the key telephone exchange, 52 sets to OS1 [terminal exchange 91
: 17 sets to OS2 and 17 sets to OS3; 180 telephone sets were connected in all.
Because of the incomplete utilization of the installed subscriber capacities
during trial. operation with actual subscribers, a supplemental subscriber laa.d
was produced with test telsphone sets designed so as to bring the overall load up
to six to eight calls per day per installed telephone set. The task of this
stage was to check the operational stability of the equipment with long term
exposure to the actual load.
The functioning of all components and the system as a whole was checked during the
line testing stage; the electrical measurements confirmed the high quality of the
speech channel.
Despite the positive results of the line tests, a whole series of characteristic
defects were found in the trial operational stage with the simul.ation of the sub-
scriber load, where these defects were itmnediately eliminated. The follawing are
to be numbered among them:
In the program software: the initiation of the search program cycles for trunk
routes (SP) with the overf low of the common memory field; "hanging-up" calls
because of errors in the SP search program; "hanging-up" calls when the servicing
6
FOR OFFIC'[AL USE ONGY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42109: CIA-RDP82-00850R000400080013-5
FOR ORF7CIAL USE ONLY
TABLE 2
Item No.
1
Designation of
the Service
Conference
calling
Brief Description of Serv`ice Contents
Makes it possible to hold a conference
between three to eight participants,
including the caller setting it up.
2 Call transf er
to another
_ telephone
set (TA)
Permits the subscriber to order his own
telephone to transfer a call to another
telephone when he is absent
3 Stand-by with Permits placing the calling subscriber on
return cali-up hold when the called subscriber is busy.
- After the latter is cleared, the calling
subscriber 3s first called, and after his
answer, then the called subscriber.During
' the holding time, the calling subscriber
can make outgoing and receive incoming
calls.
- 4 Abbreviated Makes it possible for the subscriber to
number dialing call subscribers of local, zonal and long-
distance networks by means of dialing an
abbreviated number.
5 Call without Makes it possible ::a call a subscriber of
number dialing a local, zonal or long distance network
without dialing the number by means of
- taking the receiver off the hook.
6 Transfer of a Makes it possible for subscriber A(or
call connection B), who is having a conversation with
to another sub- subscriber B(or A), to connect sub-
scriber subscriber B(or A) with subscriber C,
in this case, eliminating himself from the
call.
7 Obtaining inform- Permits subscriber A, who is having a
at ion during a conversation with aubscriber B, to call
conversation up eubscriber C, and after obtaining the
the inf ormation, return to the interrupted
conversation with subscriber B.
8 Inhibiting in- Permits a subscriber to temporarily
coming calls for block incoming calls from all subscribers
an indicated until the time stipulated in the order
= period of time expires.
7
FOR OFFiCIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00854R000400080013-5
FOR OFFICIAL USE ONLY
TABLE 3
Item No.
Kind of Technieal.
��:�Brie� Description of the Contents of the
Operations
Technical Operations
1
Call fee
Metering local conversations (the number
determination
of them) in centralized counters (in the
memory of the central eontrol unit) and in
meters installed at the subscriber
- 2
Statistics
Gathering data on the number of calls
which came in and the number of conversa-
tions which took place
3
Techn3cal
Monitoring the operabilitq and d3agnosing
servicing
faults in sets which parr_icipate in making
a call connection, as well as in the
switching system, peripheral controllers,
subscriber lines and the +control complex
4
Operator cammuni-
Feeding messages to the operator concern-
cations with the
ing the failure of individual devices or
system
defective situations (failures) during the
time a call is handled . Operator actions
during technical servicing are: feeding
punched tapes and out in real time, block-
ing and releasing functional devices,
analyzing the status of main frame memory
data files on the handling of calls,
f eeding out data on a change in the
category of subscriber lines.
remand register (RTO) overflows because of errors in the analysis programs;
"hanging-up" subscriber sets in the oriented check memory in the case of outgoing
ca?ls because of errors in the external outgoing traffic program3; false rpgister-
ing of failures in the PUU's [peripheral control unita] because of interference in
the scanning matrix and inadequate reliability of the programs:ifor timely monitor-
ing of the operational capability.
In the hardware: a change in the number of channels connectdd to the speech channel
(pulse code modulation), on the operation of the OKU [common control channel];
incorrect connection and disconnection of equipnent participating in establishing
call connections because of unstable operation of the PUU programmers of the OS1--
OS3 terminal exchanges; incorrect connection and disconnection of equipment partic-
ipating in mak.ing call connections because of the generation of "false" control
data for the common control channel. Failures of certain peripheral control unit
circuits of the OS1--OS3 terminal exchanges with repeated interruptions in the
primary maias voltage (220 volts).
8
FOR OFFICiAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
J
~
C"+
ti
V
S
4-4
'
~
a
a
u
u
;
~
~
.
~
aj ca
ct 00~
~
oa o
v
N
.C
41
~
U r-I GL
~ 4-1
~ G

h
ti
ti
N
N U -w (ti
v
0 4J W ai
~
ti
E
Z
E
i
~
~
cq
F G b
P
~
~
oo
.
w o~a
N~4-) G
,1 ro �ri o

b ~
u r--I
N cn
r-{ t N M
~ ~ ~ ~
~
�
~ r-i O
~
O 00 �rl
i
"
JJ Ol U bD
r
~ �rl al
41 p
co
~
3
roo5 co
~
p U O S-+
4.j U) 41 ~
cA O 0 U1
o

Q
e--I N M 1t
C
4
~
~
t�
I
I
~
o.
~G
auT7 Q .zaqTz-3sqng
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
avTI zaqT.zosqns ~uFT iay7zosqnS ,
9
FOR OFFICIAL USE ONLY

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFF'ICIAL USE ONLY
Number
~
of Calls (x 10 �j
p _ Cn
s
After the defects enumerated above were
eliminated, the number of call failures for
technical reasons dropped down to 0.3--0.5
percent.
4
� Cp
~
i ~
Days of
the Week
I ANu
T es I T , , redenu
/IOtied &W qarch trCmQ lWmn. Cyd. EacKR
Monday Wed. Fri. Sunday
Some 171,000 call connections were ma3e ~n the
trial operational stage with real subscribers;
of them, 50 to SS percent of the calls ended in
a conversation. The distribution of the over-
all number of calls Cn and the number of calls
ending in a conversation, Cp, is shown in
Figure 3 as a function of the days of th e week.
Ztao peak load hours were f ound for the key exchange
during a 24 hour day; from 9: 30 to 10; 30 AN~-and from
2:30 to 3;30 PM with the calls numbering 600 to 700.
Figure 3. During the trial operation period, which
Key: Cn = total number of lasted for several months, 28 complaints came
c alls made; in from subscribers. The reason for the com-
Cp = number of calls plaints were: 20 because of programming errors;
ending in a 3 because of damage to the cable within the
conversation. exchange; 4 because of failures in subscriber
lines; and 1 was the fault of technical
personnel. Over this same time period, there
were 30 short term interruptions (of 2 to 3 minutes) in handling calls with the
setting of the exchanges to the initial state. The reasons for their occurrence
were: 25 percenC were the fault of technical personnel; 15 percent were because of
data dropouts in the memories (ZU), since a portion of the memory capacity had no
back-up; 20 percent were because of a lack of software for the selective, facility
by facility setting of individual exchanges to the initial state; and 40 percent
were because of programming errors.
Thus, despite the fact that the first two stages of tests of the exchange in the
trial zone yielded positive results, defects continued to be found in its program
software. Moreover, there were failures in six pieces of hardware, where three of
the failed units were detected by automatic monitors, while three more were found
following test calls.
During the trial operation with an actual subscriber load, the operational quality
of the exchange equipment in the test zone was checked periodically by means of
making test calls. The results of these tests are given in Table 4. In this case,
unsuccessf ul attempts to make a call occurred for unknown reasons only in the case
of outgoing and incoming calls from telephone sets in the Istra automatic telephone
exchange.
Trial operation of the "Istok" YeSS ATs demonstrated the reliable operation of the
major system equipment. Thus, in the course of two years from the moment the
factory tests were completed until the end of trial operation, there were no faults
in 363 integrated selector switches (MIS), with the exception of those cases, the
reasons for which werc errors in the operation of other parts of the system, in
10
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00850R000400480013-5
TAALE 4
FOR OFFICIAL USE UNLY
Test Calls in the Course of an Hour
Successful
Total (Ending in a
Number of Conversation)
Attempts
Unsuccessful
For Known
Technical
Reasons
For Unknown
Technical
Reasons
1,200 1,191 5 4
1,000 990 4 6
1,200 1,188 6 g
particular, software errors. The YeFS [not further defined] connectors produced by
the GDR proved to be just as reliable. The control unit for the analog-digital
unified communications system produced by the "Robotron" comb ine during the testing
period from June to December of 1979, as well as up to the present time, did not
have a single flaw.
It should be noted that following the elimination of the bulk of the program errors,
the trial zone operated better when there were either no operational personnel at
all or the personnel were of the intermediate skill level, since the highly skilled
workers (designers), having confidence in their own kno~aledge, did not always inter-
vene in system operation with justification, something which causes additional short
term interruptions in system service which were noted above (25 percent of the
cases).
The Berlin Test Zone. To check the operat'Lonal performance of the analog-
digital unified communiaations system under actual GDR telephone network operating
conditions, a test zone for the analog-digital unified communications system was set
up in Berlin, which was practically analogous to the Istrinskaya zone (one key
exchange and three terminal exchanges). This zone was tied into the Berlin tele-
phone network, and internal, local, long distance and international calls were made
with it. In all, more than 300 subscribers were connected to the Berlin zone ex-
changes.
During the time of trial operation of the system, the overall losses in the case of
internal cotrmiunications amounted to 2.2 percent. 'I"his loss level drops off to 0.6
percent (similar to the Istrinskaya zone), if losses are excluded which are related
to program errors which were found. The overall losses for the case of external
service amounted to 1.9 percent.
Among the call connections which were made in the Berlin test zone, it is interest-
ing to note the international. ones, where the subscribers of the trial zone of the
analog-digital unified communications system in Berlin were automatically connected
via the international telephone network to the analog-digital unified communications
system test zone subscribers in Istra in the Moscow oblast.
11
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICiAL USE ONLY
The trial operation of the Berlin test zone which was carried out at the start of
1980 went successfully.
Technical Economic Indicators of the Analog-Digital Unified Communica-
tions Sy stem. The "Istok" YeSS ATs has a number of special features and unique
differences from well known foreign communications systems, which extend its capa-
b ilities and improve its economy:
--The system can operate both with autonomously controlled quasi-electronic auto-
matic telephone exchanges and with the key exchange of an iztegrated comunica-
- tions system with centralized control within the bounds of a communications net-
work section;
--It can switch both analog and digital signals without the forced conversion from
one form to the other (for switching purposes);
--Remotely controlled terminal stations can be connected to the key office both via
PCM channels and physical junction lines;
--The system has a custom designed integrated selector switch;
-It does not require air conditioning, forced ventillation or false floors;
--The system is put together using elements and technology available in the USSR
- and the GDR.
--A.s compared to a comnunications network equipped with crossbar automatic tele-
phone exchanges and transmission systems with frequency multiplexing of the
channels, the communications network organized with the "Istok" analog-digital
unified communications system has the following advantages: the overall volume
of equipment and area occupied by it are reduced; the process of technical oper-
ation has been automatdd and centralized, something which is accompanied by a
reduction of several times in the operatier.al labor intensity; the quality of
the speech channel and the carrying capacity of the system have been sharply
incr.eased; its reliability has risen sharply while the number of faults has
been reduced; subscribers are offered additional kinds of service; the possibi-
lity of transmitting d igital data at a high conf idence level is provided; the
_ prr:iucts list of equipment has been reduced, and the throughput per unit of
production area has increased.
It should be noted that all of the advantages of "Istok" YeSS ATs exchanges can
oe fully utilized in an integrated mode if a PCM tran5mission system and primary
power supplies for the terminal stations adapted to these exchanges will be
designed.
International Approval. A working exhibitinn model of the analog-digital
unified cotmnunications system [2] was constructed in a short period of time by
GDR enterprises working from corrected documentation; this prototype was success-
fully exhibited at the "Telcom-79" international exhibition in Geneva. This same
prototype was exhibited at the Leipzig spiing fair in March of 1980. In this case,
12
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R440400080013-5
F'OR OFFiCIAL USE ONLY
the results of the tests in the Istrinskiy and Berlin test zones were already taken
into account in the documentation f rom which this prototype was constructed.
A control camplex developed jointly by USSR and GDR specialists; is used as the
central control unit in the "Istok" YeSS ATs. A prototype of the control unit for
the analog-digital unif ied communications system (UK 4310) was built in the GDR
= "Robotron" combine. It was exhibited at the international exhibition devoted to
- the 30th anniversary of the CEMA, which was held in Moscow in June of 1979, after
which it was used to conduct the tests in the Istrinskiy test zone.
BIBLIOGRAPHY
1. "Integral'naya kvazielektronnaya analogo-tsifrovaya sistema svyazi - IKE ATsSS"
["An Integrated Quasielectronic Analog and Digital Communicatio:ls System, the
IKE ATsSS"], ELEKTROSVYAZ' [ELECTRICAL COMMUNICATIONS], 1975, NOs. 10, 11.
2. Tietze P., "Demonstrations-muster von Einrichtungen einer ENSAD Ortszentrale"
["Demonstration Model of the Equipment of an Analog and Digital Unified Communi-
cations System Local Central Office"], FERNMELDETECHNIK [COMMUNICATIONS
ENGINEERING], 1979, No. S.
COPYRICHT: IZDATEL'STVO "RADIO I SVYAZ "ELEKTROSVYAZ 1981.
8225
CSO: 8344/0140
13
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICiAL USE ONLY
ITALY
EQUIPMENT USED AT RAI MT. VENDA TRANSMITTING STATION
Turin ELETTRONICA E TELECOMUNICAZIONI in Italisn May-Jun 81 pp 98-114
[Article by Giulio Paolo Pacini*: "Combining Units for FM Broadcasting Transmitters
--I:quipment for the Mt Venda Transmitting Center"]
, [Text] Summary--Combining Units for FM Broadcasting Transmitters (Equipment for the
Transmitting Center of Mt Venda)--This paper deals with a general description of the
operation of a combining unit for frequency-modulated broadcasting transmitters, and
of the structure of distributed constant circuits which are the most appropriate for
the implementation of these combiners. Moreover, it presents some examples of pro-
_ totypes, designed and set up at the RAI Research Center, which have been manufac-
tured in a small mass-production for the RAI transmitting equipment. Particular at-
tention is given to one prototype implemented, as a unique model, in the Transmit-
ting Center of Mt Venda: it represents a significant example because of the particu-
lar problems its design has posed, owing to the small percent distance between ad-
jacent frequencies in the transmitters, as well as to their high power.
1. General Remarks
The FM radio programs transmitted by RAI in the $7.5-104 MHz band and radiated by a
transmitting center or by a bounce repeater are generally three and in some cases
four in number. They are r.o^r.--l?y raaiatea by a siragle wide-10s.^.d ante:.na anu not by
various separate antennas; this is because of the costs of the big transmitting an-
tennas, the power cables and installation, the space taken up in relation to the
gain and the solution of other, collateral problems.
The radiation of several programs from a single antenna requires the use of a com-
bining unit that has the task of combining the power of various transmitters in a
single antenr.a cable, with proper insulation maintained between the transmitters
connected to the unit.
In addition, the combining unit must offer good adaptation of impedance to the
ti�ansmitters; it must not introduce considerable power losses off the carrier or off
* Doctor of Engineering Giulio Paolo Pacini of the Research Center of RAI [Italian
_ Radio Broadcastinb and Television Company]-Turin.
r
Typescript received 12 March 1981.
14
FOR OFFICIaL USE ONL'Y
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42/09: CIA-RDP82-00850R000400080013-5
rOR OFFIC[AL USE ONLY
the side bands, and it must not introduce into the signal in transit considerable
distortions of phase and amplitude, that would be found in the demodulated low-band
stereophonic signal in the form of harmonic distortion, intermodulation, linear and
nonlinear diaphony, improper amplitude/frequency response.
2. Description of Several Cambining Cirucits for FM Transmitters
The circuits that make it possible to combine two or more FM transmitters can be
constructed in several ways, depending on the characteristics of the installation,
the number and power of the transmitters, and especially, the gap between two con-
tiguous carriers. They break down fundamentally into two groups: star circuit, and
ring circuit with hybrids.
2.1. Star Circuit
The basic diagram is as indicated in Figure 1 for the case of three transmitters.
These feed three lines that converge at a common point S(the center of the star),
to whcih the antenna cable is also connected. From each of the three lines is
shunted, at a quarter-wave from point S, a filtering network F with passband charac-
teristic: it presents high impedance at the frequencies of the transit channel, and
very low impedance--and therefore high attenuation--at the frequencies of the other
two channels.
P'igure 1. Diagrum of FM combining unit constructed with star circuit
If one considers, Eor example, the TX1 transmitter with carrier at fxequency fl, the
corresponding signal can transit on its own line through the F1 network but cannot
- reach the other two transmitters Uecause of the strong attenuation (ideally, a short
circui.t) introduced at the same frequency by the F2 and F3 networks: at points 2 and
- 3 there is an almost total reflection that transfers, through a line section one
quarter-wavelength long, the low impedance that exists at these points into high im-
pedance at antenna-entrance point S; the TX1 therefore sees only the antenna imped-
ance. 7'he situation is analogous for the other two transmitters. The structure of
a combiner constructed with star circuit can be different for the type of network
involved, which can be of the passband type or of the blocking-filters type.
In the former type, the required characteristic is achieved with a passband filter,
which can be obtained with distributed oonstants with two resonant elements in co-
15
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFFiCiAL USE ONLY
axial cauity connected with flux linkage or capacitive coupling. Figure 2 presents
one of the various solutions for constructing the passband network, obtained by
- means of two quarter-wuve resonutors and connected, with flux linkage, by means of a
line of one quarter-wave electrical length; the network's response is indicated
qualitatively in the same figure.
90* ElET1AiC1 y ( 1) NS~T ~
I 4' ~
I I ~4
<
I j
(2)
i riuMi a .wco, ~Y .
~2)
EMlA2 ,nIW,.: (3)
I sTi.
aa �
0 0
i
-~o -10 i
I '
I -70f -10 i
_3p I i
-~04 -~0 I II
_f0l _9C
I
L
I fi ~7 f~ II ~7 ~7
eno I~ ~m ~
Figure 2. Filtering network of passband Figure 3. Network with blocking filters
Key: type and related response Key: and related response
1. Electric 2. Attenuation 1. Compensator 2. B].ocking fil-
- ters. .s. . Attenuation
In the second type, blocking of the frequenciea to be attenuated is achieved with
series-resonant liTies shunted off the transit line, which makes for higher isola-
tion values, while compensation at the frequency of the carrier in tranait is
achieved with a line that resonates parallel with the sum of the ausceptances pre-
sented by the blocking fil.ters at the passing frequency.
Figure 3 stiows a construction of the F1 network: the blocking filters are obtained
wi.th lines a half-wave long at the frequencies f2 and f3 and are short-circuited at
their ends, while the compensator for the lowest frequency fl must be inductive and
is achieved with a line open at the end and of length greater than a quarter-wave.
The network's response in this case presents twa zeros of strong attenuation at the
frequencies f2 and f3 iFigure 3).
_ The figures that follow represent several examples of combining units built in the
RAI Research Center, which from the beginning of FM broadcasting has been involved
in the design and construction of various prototypes, different in their structure
16
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R004400084013-5
FOR OFFICIAL USE ONLY
and in the number and power of the transmitters and mass-produced, for RAI's trans-
mitr.iiig iiiyLallations, by 1tal.ian firms specializing in mechanical construction.
� � ~ � � a o s s e~ � /
J"
. � . . . . . ~
. _ . . ~
~R3i ~rnrno r~ i.nw-nro erui .-.~.n?iry+~~~p~y~.,~'~ _ .
~i. .
a)
~
n.- n, n�.,, ;,i �
~
Figure 4. FM combining unit for low powers, built with star circuits. a) Front
view with tuning controls; b) rear view with the cavities and the con-
nections to the star center by coaxial cable.
Figure 4 shows the smallest specimen--as regards both power and dimensions--built
for Che combining of three lU-watt transmitters. It is of the star type, with pass-
band filters each comnosed of two resonating cavities strongly charged capacitively
so as to reduce its length and thus permit movnting it in the same section as the
transmitting equipment.
Figure 5 shows a combining unit for four 3-kW transmi.tters; it too is of the star
type witli passband filters obtained with Eour pairs of cavities of X/4.
The conL-rul circuits, in addition to furnishing the readings, provide for the mini-
murn-power and maximum-reflection security, with automatic action, at the preestab-
lished threshold, to trip off the o of the transmitters involved in the anomaly.
'The quarter-wave cavities that constitute the filtering groups are of the flux-link-
- age type, have natural cooling, and are frequency-stabilized as regards temperature
by means of a bar of. Invar. They are of the type shown schematically in Figure 2.
FiKi,re h stiows anor.tier combining unit, for three 1-kW transmitters. This too is of
the star type, but is made with blocking filters. For each o� the three sections,
each of which corresponds to a transmitter, the two half-wave blocking filters are
17
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFIC[AL USE ONLY
visible at the sides and the compensation lines for the channel in transit are vis-
ible at the center, per the diagram indicated in Figure 3. At the bottom of the
figure are the directional probes that go to the control and security circuits.
~
R'
1 , 1 Y ~
I
�

-
i.
`18~ Figure 5. 4 X 3-kW FM combining unit built with star circuit. a) Front view show-
ing the instruments for measuring power and reflection and for tuning the
cavities; b) rear view showing the cavities and the star center, with the
coaxial antenna line leading to it (toward the bottom).
2.2. Ring Circuit with Hybrids
When the frequenci.es of the transmitters to be combined are very close to one an-
other (1.5 percent or less), star circuits are no longer suitable; in such cases it
is necessary to use circui.ts in which the isolation between the transmitters is
achieved not by the characteristic of the filtering elements but by the structure of
a bridge circuit. Indeed, it is not possible to obtain high attenuation values at
small percentage distances from the resonance with a parallel resonator, except at
the cost of heavy losses and dangerously high gradients.
= The circuits of this second group too can be differentiated by the type of filtering
network (passband or band-stop) and by the type of hybrid (directional coupling at
3 dB, or a 180� hybrid--for example, a diplexer).
2.2.1. 1lybrid
It will be recalled that a hybrid junction is a circuit with four gates joined two
by two: the joined gates are isolated from one another. Power applied at a terminal
does not appear at its conjugate but divides in equal parts between the other two
18
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
terminals; an ideal hybrid circuit is characterized by perfect adaptation of the im-
_ pedances and by infinite isolation between the conjugate gates.
a�a
s s , s �
9
Figure 6. 3 X 1-lcW FM combining unit of star type with blocking filters
In practice, a hybrid can be constructed in various forms that essentially differ as
regards the phase difference between the outgoing signals, which can be 90� or 180�.
The first type is constructed by coupling two parallel lines to a length of a quar-
ter-wave (or uneven mvltiples) to the central frequency of its operating band; this
- is a:s-dB directional coupler for TiM transmission mode (Figure 7). In resonance
condi.tion (A = 90�) and wi.Ch the other gates charged on their own characteristic im-
pedance Rc, the power applied at input gate 1 subdivides in equal parts between out-
put gate 3 and coupled gate 4, while on conjugate gate 2 it is nil. The signal on
gate 3 is delayed 90�, while that on gate 4 is in phase with the input signal:
E, i E~ i?;)Oo� Ea = i El i~i Ez = 0(v. � 3).
y~ ~ I V
2.2.2. Descripr_ion o� Ring Circuit
The ring circuit with 3-dB hybrids and filters of passband type is described. It is
composed of two hybrids and two filtering networks F identical with one another
(figure 8). The power of the TX1 transmi[ter at frequency fl enters gate 1 of the
_ first hybrid I1, and half of it exits at gate 3 and half at gate 4; on the upper
19
FOR OFFICIAL U5E ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
~

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00850R000400480013-5
FOR OFFICIAL USE ONLY
line 3-3', the phase of the signal is 90� behind the signal going through the lower
line 4-4'. The two signals can pass through the filtering networks F because they
are passband filters tuned to frequency fl; joined at gates 3' and 4' of the second
hybrid, given the phase relations between the gates af that hybrid, the two signals
recombine in phase at output gate 2', ivhile they cancel one another at gate 1' be-
cause they are in counterphase on account of a futher 90� delay of the signal of the
upper line. In this way, gates 1 and 1' of the circuit are isolated from one an-
other. The second transmitter at frequency f2 is placed at gate 1', achieving
strong isolation between the transmitters independently of the characteristics of
the filtering network.
RC
E3
0
El E2
00 R
c
E.
G)
RCc
Figure 7. 3-dB directional coupler; 0= electrical length of coupling
Figure 8. Diagram of FM combining unit conatructed by means of ring circuit with
two 3-dI3 couplers and two identical passband filters
'flie signals partially reflected by the filtera F at the lateral frequencies of the
fl transmitter travel back again behind the two lines A-3 and B-4, going back into
phase at gate 2(absorption load) of t:,e first tiybrid, not being able to return to
inptit gate 1, where they cancel one another becauae they ar.e in counterphase. This
means that that circuit is at constant input impedance, being the ref.lection of the
filters E ciissipated in absorption load Rc.
The power of the TX2 transmitter at frequency f2 applied at gate 1' of the secorid
hybrid I2 exits, half at gate 3' and half at gate 4', being still, in this case, the
- signal on the upper line, phase-delayed 90� vis-a-vis the carresponding signal of
the lower line. At points A and B, these twa signals undergo almost total reflec-
20
F'OR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
tiori because E,) is in the attenuated band of the filrera and reenter the same hy-
_ brid, going back into phase at output gate 2', while they cancel nne another, being
in counCerphase, at entrance gate 1' of transmitter f2; gate 1' is also at constant
input impedance.
Figure 9. General diagram of 4 X 10-kW FM combining unit of Mt Venda, with switch-
ing circuits
Key:
1. Ring
- Since the filters' attenuation at frequency f2 is not infinite, a small fraction of
power transits beyond the filters, reentering at gates 3 and 4 of the first hybrid
- and recombining in phase at gate 2 on the absorption load, while it cannot reenter
at l: this ensures isolation of the transmitters at frequency f2 also.
One notes that while the fl channel is narrow-band--that is, suitable for tr.ansit of
a single trancmitter corresponding to the passband of the filters--the f2 channel is
wide-band, because it corresQonds to the attenuated band of the filters. This makes
it possible to apply to that channel two or more transmitters already combined with
one another and in any case well-removed from one another in the FM band. In this
way it is possible (Figure 9), with successive rings in cascade, to combine a large
number of transmitters with one another (n transmitters by means of n-1 rings).
Another circuit, entirely similar, can be obtained by the use of filtering networks
of dual type. '
21
FOR OFFiCIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
3. Mt Venda Installation
RAI, with its program to restructure a sizable part of its FM installations, set the
objective of going ahead with renewal of the old equipment while at the same time
making the installations suitable for stereophonic traasmissions; within this frame-
work, nearly universal introduction of circular polarization in place of horizontal
polarization has been planned, in view of the considerable use advantages offered.
Figure 10. Overal] view of the 4 X 10-kW combining unit. At left, the instrument
panel; at ri.ght, the switching panel.
Within the program for renewal of the FM installations, the Mt Venda Transmitting
Center presented several clifficulties, occasioned essentially by the extreme close-
ness between the channelized frequencies, which are only 0.9 MHz apart, as against
the distance of 2 MHz or more in most of RAI's FM installations.
In the old arrangement, the installation had two superimposed radiating systems, one
of which radiated the combined power of the two transmitters farthest apart in fre-
- quency, while the other radiated the power of the transmitter at the central fre-
- quency; this was because of the lack, at the time, of a combining unit capable of
handling three frequencies so cl.ose to one another.
At the time of the restructuring of the installation, in
polarization it was seen to be necessary not to decrease
power horizontally, which was possible only by taking up
on t}ie metal lattice by means of a single antenna of hig
combining unit suitable for simultaneous broadcasting by
single antenna was posed again in this way.
n 22
FOR OFFICIAL USE ONLY
the transition to circular
the equivalent radiated
the entire space available
ner gain. The problem of a
all the transmitters on a
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
The combining unit for the Mt Venda transmitting installation was built for four
10-kW FM transmitters at the frequencies of 88.1-89.0-89.9-101.5 MHz by the combin-
ing of tliree rings with 3-dB hybrids and passband filters of the type described;
this choice was due to the very small distance between the first three frequencies.
Figiire 9 is a general diagram of the combining unit, comprising the manual switching
~ system, which is integrated with it and which in normal operating :,onditions has the
contacts disposed as indicated in the figure. It permits both switching of the
transmitters from the combining unit dirQCtly to antenna lines A and B and section-
- ing of the unit itself in case of breakdown, because of the fact that each of the
three rings of which i.t is composed can function autonomously. The A(preferential)
and B lines go to the two semiantennas into which the entire antenna can be sec-
tioned in case of breakdown of a part of it. Under normal operating conditions, the
two semiantennas are connected in parallel on line A.
Figure 11. Rear side view of the 4 X 10-kW combining unit
Figure 10 shows a view of the unit. At left front is the control-instrument panel;
on the right is the switching and sectioning frame. In the upper part the hybrids
are visible, and below, the passband cavities. Figure 11 shows a rear side view.
In the fullowing sections is described the structure of several constituent parts of
a combining unit for FM transmitters, with particular attention to several design
elements and witti special reference to the unit built for Mt Venda.
4. Structure of the Hybrid
In section 2.2.1. ttiere was defined a hybrid in the form of a directional coupler
with a single quarter-wave section. When the power applied is high, the coupled
lines in practice are normally brought back to the form of bars of rectangular or
circular section, though trie latter are not suitable for high coupling values.
23
FqR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLV
t
t2" .
F~---d---{
Figure 12. Cross-section of a semicircular-bar hybrid of the type used in the
4 X 10-kW unit
A less common structure is that which uses coupled lines of semicircular section
placed between two parallel planes (Bibliography 3). A direct derivation of this
- type of structure has been used for the coupliers installed in the Mt Venda combin-
ing unit (Fi.gure 12), The advantages of this configuration were seen to lie in the
large coupling area, together with reduced cross-section. For a coupler of quarter-
wave section and characteristic impedance of 50 ohms, the dimensional ratios af the
section indicated in Figure 12 prove to be: t/d = 0.3449 and s/d = 0.05342.
5. Relation between the Voltages of a Hybrid and of a Rir.g of Hybrids
5.1. Voltages at the Outputs of a 3-dB Coupler
A coupler is considered that is compoaed of two parallel transmission lines coupled
uniformly for an e.lectrical length 6, with power supplied at gate 1 and closed at
the other gates on characteristic impedance Rc (Figure 7).
It is demonstrated (Bibliography 1, 2) that the voltages at the other gates are:
E3 = Ei U1-k2
V 1- ka � cOS a& +j sin o&
(output gate)
121 Ea = Ei jk sin &
V1-k2 - cos$-I-jsin5
(coupled gate)
in which k is the coupling factor that corresponds to the maximum value of the ratio
E4/E1 for 0= 90� (centerband of the hybrid).
From (1] and [2) it results that for any value of 6--that is, with the variation of
ttie frequency of the input signal--IE3I~ + IE4I2 - IE1I2; that is, all the input
- power is collected at gates 3 and 4, and therefore at the conjugate gate E2 = 0. In
addition, the input impedance is constant and equal to Rc at all frequencies.
In the particular case of a 3-dB coupler that has to be E3 = E4 for 6= 90�, it re-
sults from [1] and (2] that k= 1/Y/'-2, There�ore, for the 3-dB coupler the preced-
ing relations become:
24
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFiCIAL USE ONLY
[31 Ea j El a wi th
2
1
a=
sin cos ~
_ (output gate) V T
[4] EQ _ El b with b= sin,&
1
l"2 sin 3- j cos 9V2
(coupled gate)
a and b are two complex operators that have the same phase angle and that can be re-
presented with two parallel vectors whoae ratio b/a = sin 6. It follows that with
variation of frequency, E3 and E4, even though rotating phase, always remain in
quadrature with one another. At the central frequency fo(6 = 90�), a= b= 1. This
type of coupler can be used in a frequency octave (between fo/r-2 and fo/--2-) with
a rather limited variation of the coupling, as results from [3] and [4].
_ 5.2. Voltages at the Gates of a Ring of Hybrids
A ring is considered that is composed of two 3-dB couplers, 11 and 12, with two fil-
ters placed at points A and B of the connection lines (Figure 8). Analysis of the
ring alone is done, for the sake of simplicity, by supposing that the filters have
an ideal transfer function with the value of one at passing frequency fl, and zero
- at attenuated fre(luency fZ--that is, as if the filters did not exist at frequency fl
and presente3 a short circuit at frequency f2. The ring's behavior is examined sep-
arately at the two frequencies.
5.2.1. Signal in Transit at Frequency fl
When gates 1' and 2' of the second hybrid are charged with impedance Rc, input gates
3' and 4' present a constant impedance equal to Rc at all the operating frequencies.
Therefore the voltages E3 and E4 at the output gates of the first hybrid are still
the same as expressed by [3] and [4] and are also equal to the voltages at the input
gates of the second hybrid E'3 and E'4 except for a phase constant that can be ex-
pressed by the operator
c= c-fGI (E's = Eae-iN; E'a = Ea e-fX)
which hereinafter, Lor simplicity of written expression, will be understood (1 is
the length of lines 3- 3' and 4- 4').
We now obtain the expressions of the voltages E'2 and E'1 at the output gates of tlie
second tiybrid to which are applied the voltages E'3 = E3 and E'4 = E4, respectively,
at input gates 3' and 4' (Figure 13). Considering this configuration as the super-
imposition of two situations analogous to that indicated in Figure 7, expressions
analogous to [3J and [4], respectively, are applied individually to the input gates
for ttie corresponding output gate and coupled gate, and adding tugether, at the out-
puts, the contributions of the two input signals, one has:
1 (6E3 - jaEa) = -jEla � b
[51 E` V~77
E'i = l~ ~ (bE'4 - jaEa) = - 2 (a2 - b~)�
25
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL tJSE UNLY
It is recognized immediately that if A= 90� at frequency fl, with a= b= 1 the
signal is collected entirely at gate 2', because in such case E'1 = 0; and this is
because the two signals at gate 1' are in counterphase. For 60 90�, E'1 # 0, and
therefore a fraction of signal reaches gate 1' also, thus decreasing the isalation
between the two generators at frequency fl.
Figure 13. Voltages at the gates of the second hybrid
On the basis of [3], [4) and [5], the moduli of the voltages at the various points
of the ring at frequency fl are:
IE9I _ 1 1
a
E~ l; z I I V T 1 CosQ ~
EQ I_ 1 I b I_ 1 sin a ~
I E1 V' cos2 ~
I y
sin 9
I Ei l =Ia'bl - 1
~
I
1 - y cosz .9.
, 1 cosz 3
~Ell_l~az_bzl - . -
~ 1
i El ~ 1- 2 cosz
Attention is drawn to the fact that the isolation IE'1/Ell between the two transmit-
ters at frequency fl is entrusted exclusively to the equilibrium of the ring and
that its value under balancing conditions is theoretically infinite at the hybrid's
resonance frequency (0 = 90�; a= b= 1) in relation to the vectorial combination,
at gate 1', of two magnitudes of equal amplitude and in counterphase. This requires
that in addition to having perfect symmetry in the mechancial structure of the two
branches of the ring, the two filters be identical and remain so in time: a small
variation in the Cuning of one filter vis-a-vis the other, because of inechanical or
temperature factors, can cause a considerable loss of isolation (as can be verified
_ with the second of the exnressions of [5] by multiplying E3 and E4 by the character-
istic of the filters). This requires the use of cavities that are very stable in
frequency.
26
FOR OFFICIAL +15E ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00854R000400080013-5
FOR OFF[CIAL USE ONLY
5.2.2. Signal Reflected at Frequency f2
Analysis of the fL signal is identical in form to analysis of the fl signal.
A circuit section downstream from points A and B(Figure 8) is considered, lceeping
in mind the description given, in section 2.2.2., in relation to the TX2. If, ini-
tially, one ignores the wave that travels on the two lines toward points A and B and
one supposes that fihe wave reflected at those points is, instead, a trave2ing-wave
coming from two imaginary generators at A and B, the conditions already examined at
frequency fl are found again: the signal on the upper line phase-delays 90� vis-a-
vis the corresponding signal of the lower line, and the situation is still repre-
sented by Figure 13, the generators having in this case the amplitudes:
E'aR - E,1 alPI Blly- =Pn

(81 VZ
E'ett = E/1 bI PI eicv-:3+)
VZ
in which the subscript R is introduced to recall that two signals reflected at fre-
quency f2 are involved. In addition, there is E'1 instead of E1, in conformity with
the amplitudes at the gates of the respective generators; Ipl and ~ are the modulus
and the phase of Che coefficient of reflection of the filtering networks at points A
and [3, and s is the distance between the latter and the hybrids.
In the case hypothesized for the transfer function: lpl = 1; ~ _7, and except for a
phase constant which, not being essential, is understood, the expressions of [8] are
identical in form to the expressions indicated in Figure 13, corresponding to. [3]
and [4]. In practice, lpl is very close to unity, and therefore the simplifying hy-
pothesis does not alter the conclusions.
With this premised, the signals at frequency f2--E'2R/E'1 and E'1R/E'1, exiting at
gates 2' and 1', respectively--are expressible by means of the same expressions as
in [5] and [7], in which, in this case, 6 is calculated at f2. One notes that when
the ratio E'1/E1 is calculated at �1, it represents the isolation of the ring be-
tween the two generators at fl, and when it is calculated at f2, it represents the
coefficient of reflection at gate 1'.
In Figure 14, the magnitudes examined in function of the electrical length 6 of the
coupler and of normalized frequency f/fo (f/fo = 0/90�) are represented in dB.
The first graph indicates, in accordance with [6],
output and coupled gates of the first hybrid with
second graph represents the ring's output at gate
for the signal at f2, in accordance with the first
graph represents both the isolation of the ring at
gate 1' at frequency f2 on the basis of the second
the course
voltage app
2' both for
expression
F1 and the
expression
of the voltages at the
lied at gate 1. The
the signal at fl and
of [7]. The third
return losses at
of [7].
6. Filtering Networks
6.1. ConsideraCions on the Transfer Function of the Filter
Analy�r.inK the requirements for the filtering networks, one easily recognizes the
factors that introduce losses and distortions.
27
FOR OF'FICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42109: CIA-RDP82-00850R000400080013-5
IFOR OFFICIAL USE ONLY

ae
E
;
~
,
4s
IE I
-
I
F
o
a
t I
-o~
T
~
F
iE
I
i
2o
dB
~
-
' '
�

,
e
~
_
a
i
6
0
~
-
I
~
~
-t0
60' 70' ~0' 90' 100' 110' 1 0'
0,7 0,1 0,9 ~ fo
Figure 14. Top: course of the voltages at the output gate and at the coupled gate
of the first hybrid in relation to the input volrage. Middle: course of voltage at
the output gate of the second hybrid in relation to the input signal at the two dif-
ferent frequencies. Bottom: isolation at frequericy fl and return losses at fre-
quency f,,..
Yassing ctiannel 1, encountering an amplitude chaY-acteristic that is not perfectly
flat and a phase characteristic not perfectly linear, undergoes an alteration in the
amplitude and the phase of the spectral spectral lines that constitute the modulated
signal entering, with consequent distortiona and power loss in the outgoing signal:
a loss from dissipation at the center of the channel, and predominantly by reflec- tion at thP extremes.
Keflected channel 2 encounters the characteristic at one aide at frequency f2.
Keeping it in mind that the useful signal, in this case, is the reflected one, and
that the characteristic does not present an infinite attenuation value, part o� the
_.signal passes and dissipates on the absorption ch.arge with loss of power, while the
28.
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42/09: CIA-RDP82-00850R000400080013-5
FOR UFFICIAL USE ONLY
useful signal undergoes a dissymmetrical treatment of the side bands, being re-
flected on a side of the characteristic that presents an attenuation and therefore a
coefficient of reflection variable with the frequency within the framework of the
channel; and this gives rise to distortions.
The designing of the filtering network can
transfer function H of the passband at the
function HK for reflected channel 2, which
cient of reflection p at the input of the
nel 2--a coefficient that must prove close
the framework of the channel.
therefore be premised on stLdy of the
frequencies of channel 1 and of transfer
in this case is represented by coeffi-
same network at the frequencies of chan-
to unity and variable very little within
6.1.1. Width of the FM Channel
= The band width necessary for transmission of the FM channel is conventionally as-
sumed to be + 100 kHz; but in cantrast to the television channel, a radiofrequency
tolerance mask is not required. What is required under the heading of technical
speciFications, though, is all the characteristics and tolerances for the demodu-
lated signal in both multiplex and monophonic operation (amplitude-frequency char-
acteristic, nonlinear distortion, AM synchronous modulation, linear and nonlinear
diaphony between the A and B channels).
This �act allows a certain freedom in case-by-case definition of the band character-
_ istics, permitting simplifications with reduced losses and distortions in the case
of adjacent channels farther away from one another; but it requires careful study of
the transfer function of the filters in the opposite case.
_ The band width necessary can be calculated with the well-known approximate expres-
sion called Carson's rule; but more rigorously, it is possible to put the channel
width into relation with the percentage of power transmitted by it, referring to
[38] and [41] in the Appendix, by means of which Table 1 was calculated.
6.2. Elementary Component of the Filter
The filter is constructed on a distributed-constants basis with passband elements.
T'he simplest element of this type would be composed of a line of length s equal to a
quarter-wave at freyuency fl and circuited at the extremity, shunted off the trans-
mission line (Figure lSa), which, as is known, behaves, around the first resonance,
like a parallel resonant circuit. (The representation as a two-wire line is by way
- of example.)
In Figure 15b is s}iown the equivalent circuit with the values of the elements L, C,
- Itp expressed in function of the characteristics of the distributed-constants resona-
tor (see Tab:.e l., I3ibliography 6); 1, c, a are, respectively, the inductance, the
capacity and the constant of attenuation per unit of length.
- It should be said at once that an element of this type is not suitable for solving
the problem under consideration. Indeed, the filter's characteristic must go from
negligible attenuation values at frequency fl to quite high values at frequency f2
in a percentage interval of frequency Af/f of about 1.1 percent. This requires a
varial-ion of the input admittance Ye of the elenent, in the vicinity of the very
higli fl resonance; Ye can be represented (within the limits of validity expressed in
Table 1, Bibliography 6) wi_th the expression:
29
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
Table 1
Sf = * 75 kHz
Channel
Width
Fraction of
Total Power
- t10
0 kHz
Q > 99.99%
Modulating
Channel
Power
Channel
Power
frequencies
used
Fraction
used
Fraction
(kHz)
(kHz)
Transmitted
(kHz)
Transmitted
1
+
82
0.99990
5
+
100
0.99998
10
+ 100
0.99948
+
110
0.99994
15
+ 105
0.99926
+
120
0.99993
10 + 15*
+ 105
0.99978
+
120
0.99999
Stereo**
+ 106
0.99564
+
159
0.99991
* Two tones of half-amplitude in relation to the single tone,
which varies by + 75 kHz.
Lines fl = 23 kHz and f2 = 53 kHz, obtained by modulating the
right and left channels with two tones in counterphase of
frequency fm = 15 kHz, each of one-half amplitude.
i
Rcc S
RC
N _7RC
a)
Rc
r� L ~t Rp RC
b) I
R
L_~ Is ; C= 2 5 ; RP=a I
Figure 15. a) distributed-constarts resonator shunted off transmission line of
characteriatic impedance Rc; b) equivalent circuit in vicinity of
first resonance. 1, c, cx are t}ie inductance, capacity, and attenuation
constant per unit of length of the distributed-constants resonator.
[9l
1 r 1 ~f
p` 4(~l � Roe 2 keo f 1
30
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00854R000400080013-5
FOR OFFICiAL USE ONLY
with Llf = f- fl; Rcc, characteristic impedance; and Q, coefficient of quality of
the resonator. In other words, Q being practically constant in the vicinity of pf,
the derivative IdYe/dfl =Tt/2f1Rcc has to be large and thereforF the characteristic
impedance Rcc has to be small. The order of magnitude of Rcc can be deduced from a
practical case relative to the X ring of the Mt Venda unit (fl = 89 MHz; f2 = 89.9
MHz). With a value of lpl= 0.99 placed on f2, Rcc � 0.13 ohm would result (see
also [221). The practical impossibility of physically achieving such low values of
the characteristic impedance of line s is obvious.
It is nevertheless possible to reduce the characteristic impedance of a parallel re-
sonant circuit by coupling the resonator to the external circuit in such a way as to
reduce the impedance level to the value desired. This is possible by means of de-
vices of a magnetic or electrical nature. The former case can in theory be achieved
by means of an impedance transformer.
L.- asi-Lsin2 aTT Lacast-L
g:X/4 ~
jr-a s
~ L C L' C'
R~~ jn,. I^s.
Rp RP
idealQ
~
f
-
L'=L sin2 a
sin a~:1 C'= C/sin2 a.TTF
RPsRp sir12 e -~f
Figure 16. Transformation of characteristic impedance by means of a parallel dis-
tributed-constants resonant circuit; equivalent circuits L, C, Rp have
the same values as indicated in the preceding figure.
Figure 16 sh:ws a simple transformation method capable of reducing the degree of
(a < 1) coupl,ing nf the resonator to the line to which it is connected.
a) I
Rc
ti L: !EC R P Rc
b)
Figure 17. a) distributed-constants resonator with intermediate input connected
to the transmission line; b) equivalent circuit.
31
FOR OFFIC[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
The equivalent circuit, valid in the vicinity of the first resonance (s =X1/4) and
indicated in the same figure, shows the same values L, C, Rp valid for the quarter-
wave resonator when it is connected at its open end to the transmission line, and
- takes into account the change of impedance with an ideal transformer that has a sin
ratio sin(a /2):1, which is the ratio between the voltage at the intermediate input
and the voltage at the open end. The new configuration is represented in Figure 17
together with its equivalent circuit.
One notes important differences from the preceding case:
1) The level of the impedances is lower; in fact, the impedance of each element is
sin2(a7/2) times the corresponding impedance for connection to the open terminal.
In addition, while on the one hand the level of characteristic impedance from the
value Rcc to the value R'cc = RC4 sin2(an/2) goes down, as is desired, this entails,
at equal Q, an increase in the losses, with the resistance Rp, which represents the
equivalent of the losses in the resonator, dropping in the same ratio.
2) Ttie voltage at the open end increasea in the ratio 1/sin(a /2) vis-a-vis the in-
_ put voltage, and with it, all the electrical stresses in the resonator, including
the reactive power.
3) The coupling to an autotransformer introduces a coupling reactance w�La, in
which La = asl - L sin2(a7/2) that cannot always be ignored.
The effect of the coupling reactance is to introduce several modifications into the
course of the transfer function, with regard to the L'C' circuit only. Indeed, this
last-named circuit resonates parallel to frequency fl and is capacitive at frequen-
cies higher than fl. Consequently there exists a frequency fs > fl at which that
element is in series resonance with La, and the resonator's curve of reactance
therefore contains a pole at frequency fl and a zero at a higher frequency fs. The
zero gets closer to the pole, on the axis of the w's, as the coupling reactance in-
creases with respect to resonance rear.tance wiL', or the higher the transformation
ratio is; at the same time, the asymmetry of the characteristic increases with re-
spect to central frequency fl, and therefore the attenuation of the circuit for a
given distance pf from fl vis-a-vis the attenuation at -Af.
In the distributed-constants circuit, the parallel reaonance is determined by the
entire length s=X1/4 of the reaonator, and the series resonance by the line sec-
tion (1 - a)s open at the end when, a*_ frequency fs, it bPcomPS in turn a quartex-
wave long.
The value of the admittance at the resonator input in the vicinity of the resonance
can be represented by the approximate expression:
[101 y= ~ ~l+'2R �-f~
r fo
4QR� sin" - 1a 2
(see Table 1, Bibliography 6) in which 4f = f- fo, with fo with fo = resonance fre-
quency, Q= coefficient of quality of the coaxial line charged with the losses in-
trinsic to the resonator only (see section 8), and Rcc = characteristic impedance of
the resonator. It is valid in the vicinity of the first resonance under the condi-
tions IQf/fol � 1 and
32
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42109: CIA-RDP82-00850R000400080013-5
FOR OFFTCIAL USE ONLY
[11)
I .
Af
f o tg
�I /
- [10] was obtained by means of series developments applied to the equations of the
- lines with losses, and condition [11] results from the first term that was ignored
in the development.
4-", dp t�-DCXa xb i XC
I
� ,
~
I
t
i ,
i
i ~Ld+~- C-Ir+--
r--- D-0-+I j~- 0-.4
a) b) c) 8,24
Figure 18. Equivalent structures with autotransformer and transformer. a) direct
coupling; b), c) turn-of-winding coupling. .
The autotransformer structure so far considered (Figure 18a) is in practice used for
high degrees of Soupling. In addition, it representa an easy term af transition, by
means of equivalence systems, in the calculation of transformer atructures with
turn-of-winding coupling (Figure 1$b, c); these latter are prefereable for high
transformation ratios. The procedure consists in determining the dimensions of the
turn that produce the same concatenated flux as the sutotransformer. Tn the case of
Figure 18b), for example, by making the flux of the turn-of-winding equal to that
relative to the autotransformer ofFigure 18a), one obtains the equivalence condi-
tion: [12] sin (3Xb�log D/Db = sin (3Xa�log D/d [turn-of-winding b]
in which (3 = 2Tr/ and the other magnitudes are indicated in Figure 18.
Analogously, in the case of Figure 18c) one obtains:
[13) sin (3Xc�log Dc/d = sin (3X8�log D/d [turn-of-winding c]
= The turn-of-winding resonator thus obtained retains, in the vicinity of the reson-
ance, the same characteriatics as the autotransformer resonator--in particular, the
same ratio of transformation and the same electrical stresses--except for the value
of the coupling reactance, which, in order to be in relation with the inductance of
the turn-of-winding, can take on different values. The input admittance of the
turn-of-winding coupled resonator is still represented by equation [10].
33
FOR OFFICIAL USE OIeILY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OEF[CIAL USE ONLY
6.3. Filtering Network of the Combining Unit
Figure 19. Cavities of passband type coupled inductively for the Mt Venda filter
Key:
1. Air-inlet tube 3. Air filter
2. Tuning mechanism 4. Holes for air outlet
The selectivity characteristics necessary for the networks of the combining unit for
Mt Venda are obtained by means of two identical elements of the type described,
turn-of-winding coupled by means of a quarter-wave line (Figure 19). The equivalent
circuit of the filter thus obtained is represented in Figure 20: it expresses a
passband characteristic; its transfer function in the vicinity of the resonance
(Ipf/fol � 1), with the limitations considered, is expressed in modulus and phase
by
[1lJ 1 HY1 _
2 ra
{[4 (r+ 1) Q Aflfo]l -i- [r= -I- (r -f- 1)2 - (2Q Aflfo)2]213:
[ 151 0: = artg (2Q Aflfo)Y - r2 - (r -f- 1)2
4 (r + 1) Q Oflfa
in which the normalized impedance values r= R/Rc and x= X/Rc are, for [10]
_ [16] r _ Rcc ~ Q sin2(aTr/2),
c
Rcc 2 sin2(a~r/2)
[17] x = Rc pf/fo
34
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
from which there results:
[181
a circuit of this type therefore proves to be completely defined by three para-
meters, thus chosen: fo, Q, r.
R r- --R FllJ. R ~
ll ~
E ~ R" X R= X i Rc
~ i

r
- = 2Q Qf/fo ;
x
Figure 20. Equivalent circuit for filter obtained with two identical cavities
coupled with quarter-wave line.
Key:
1. Filter
Figure 21 illustrates the characteristics of the system adopted. In the upper part
of the figure is drawn the curve A2 of insertion attenuar_ion derived from [14]:
[191 A2 = 10 log10IH I-y[dB]
with the values stabilized for ring X: fo = 89 MHz; Q= 14,650; r= 51. The other
magnitudes that appear in the preceding expressions are (still for ring X): charac-
teristic impedance of circuit Rc = 50 ohms; characteristic impedance of the cavities
Rcc = 76.77 ohms; degree of coupling a= 2.69�10-2, and therefore, transformation
ratio 1/sin (a 7/2) =.23.7 In ehe same figure, insertion attenuation A2 of a single
cavity is shown in a broken line, together with the values measured.
tn the lower part of Figure 21 are drawn the phase curves and the group delay.
One notes several characteristic values of *_he curves relative to ring X:
Attenuation at carrier fl - 0.17 dB
Attenuation at + 100 kHz - 0.38 dB
Attenuation at carrier f2 - 24.57 dB
Band width at -3 dB 520 kHz
Group delay in channel + 100 kHz 151 ns
6.4. Coefficient of Reflection
It has already been noted (section 6.1.) that coefficient of reflection p at the in-
put of the filtering network represents the transfer function Hr for the reflected
channel when it is calculated in the band of channel 2. In addition, if calculated
_ at the frequencies of channel 1 it furnishes the value of the power lost through re-
flection by transmitter 1 in transit and dissipated in the absorption loAd.
- Calculation of requires knowledge of the,input impedance at section e of the fil-
tering network (Figure 20). With simple calculations it is possible to transform
35
FOR OIFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400084413-5
FOR OFFICIAL USE ONLY
all the elements of the network at aection e into the two elementa in parallel Re
and X. (rigure 22). Impedance Ze at the filter input is then:
[20] 1 _ 1 + ' 1
Ze Re jXe
in which the active and reactive components normalized at impedance Rc are:
[21J 1--1 /r
(1 + 1/r)' (lfx)a ,
1 1 1. f a~
Xo _ x - (1 + llr)a + Tllx)a .
~
d8
0
I
I
'
~
I
-

I
- ia
I
'it
A
I
-i~
 E9 MFIZ
}
~
I
o
_to
'
r' S~
~
I
-77
Q � 14650
i
~
jB 89.5 -'�o""t8
9'o�""'
89.5 4 9i
MHz
n I
qr1idi
itoo
47 ~
� iooo
~
I i
�~s~
1
i
f
 89 MHZ
o
i
~
r , Sl
U e 14650
coo
'~10~
~
j I
!
I
i ~
- 770'
i
100
~ I
~
I i
es,s
r,
Figure 21.
Key:
1
86 883 "99`"1 8919�`"' E9,5 Ti Bo
I MHZ I
Top: course af insertion attenuation for two-cavity filter (curve A2)
and one-cavity filter (curve A1); bottom: course of phase and group
delay for the two-cavity filter. The small circles indicate the values
measured.
Degrees
36
FOR OFFICIAL USE ONLY
1
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R040400080013-5
FOR OFFICIAL USE ONLY
e
_ Rc I
i
R X' _ ~e
Figure 22. Transformation at network-elements input indicated in Figure 20.
The modulus and the phase of the coefficient of reflection at input p= Ze'Rc/Ze+Rc
are as:
[ 221 Ipi = 1 -
1 1 2 -f- 1 (r� + 1/r�) + r.f (4x3.)
�Jx,
art
~ 23 ~ ~ ~ 1 - 1/rt. - 1/~ca, '
I,
I
I
i 141
~t
~
~
~
I ~
�
�
.
p..aito a nco.no
74t reti filtrantl
.
s
I
! I
I 1
~
~
.
~
~
~
pordlto d
i Ntorno
2
f
 E9 MN2
0
p-~
e
entrtta anetlo X
r. 51
1
~
44-1,
4.
6-4
,
w~
G= 1G6S0
I i~
~ ~I
I
I
1 -
- -
B8 ee,s - f:e9~m. ea,s t:E9,9 90
,
MH=
f
CoeMkent� di ritlex.
I41 (3)
ntl tNtrantl
P"to di ritorno
ac.tow9~ (1)
reti Nltrantl
~ -100 KMZ
0.9968
_ Sp
0.9864
TX, f~.89.9 MNx
0.9970
� 60 1W=
0.9975
.
� 100
0.9979
- roo Kr,:
14.00 cis
- su
.
25.03
TX~ f, . 89 MNz
716.89

I � 50 KHt
25.e3
l � 100 �
:
14,00 :
- Figure 23. Modulus of coefficient of reflection p(curve 1); return losses at fil-
- ter input (curve 2); return loeaea at ring input (curve 3)
Key:
1. Return losses, filtering networka 3. Coefficient of reflection,
2. Return losses, ring-X input filtering networks
37
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
The course of the modulus Ipl`at the network input as expressed by [22j is drawn in
Figure 23 (curve 1) for the values of fo, r and Q achieved for ring X. The values
for the channel of the TX2 are given in the table that accompanies Figure 23.
In the same figure, graph 2 relative to the network's return losses is drawn, and
these losses are presented in a table for channel 1. Finally, in curve 3 are given,
for comparison with curve 2, the return losses measured at the ring-X input. For
transit channel 1, these losses are highzr than 40 dB; this means, as already noted,
that the reflection of the networks is almost entirely sent into the absorption
load.
Still for ring X, the group delay in channel 2 was calculated by using [23]:
[24)
1 _
TR - dt
2Tt df
and proved to be 30.8 ns within the framework of the channel's + 100 kHz as against
- the 151 ns relative to transit channel 1.
Before concluding on this subject, it is noted that the quarter-wave line that
couples the cavities (Figure 20) was considered to be of constant electrical length
with variation of frequency. It is pointed out that all the formulas were derived
also without introducing this approximation, and it was verified that at least in
the field of freRuencies considered, the errors inzroduced with this simplification
prove lower than 0.5 percent for all the magnitudes calculated, while the formulas
not simplified are formally somewhat more complex; and therefore this appror.imation
was accepted.
7. Distribution of the Powers of the Transmitters in the Elements of the Unit
With the losses in the connection lines and in the hybrids considered as negligible
(the values measured are on the order of hundredths of a dB), the power drawn by two
transmitters of a ring is distributed into six elements (Figure 24). The part PRC
reaches the useful load Rc that can be the antenna or a following ring; a second
part PR1 dissipates on the absorption load; and the other four parts, 2PR1 and 2PR2,
are absorbed by the four cavities and dissipated in the elements diagrammed with the
equivalent resietances at losses RZ and R2 in Figure 20.
(1)
-
, C I riueo C i
I
~ pai Pei ~
L _ _ _ _ _ _ _ _ _ J
F I PpL P2 j
~ CMCO (2)
AfSORG11i TXt r--------~ TX2 Pnc
~ I
i
~ PR, P.~ cuuco
j ~ uTiLc R~
` C3 rLrRo . 3)
Figure 24. Distribution of the input powers in the elements of the ring
Key:
1. Filter 2. Absorption load 3. Useful load
38
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
It is useful to know the distribution of the power of the transmitters, both for
evaluating the losses and for the dimensioning and disposal of the dissipated heat.
7.1. Nonmodulated Transmitters
A single branch of the ring is initially considered: it is presumed to apply power
to the two-cavity filter examined (Figure 20) with a sinusuidal generator of avail-
able power P1. Part of the incident power P+ = P1 is reflected by the filter be-
cause at the input (section e), the coefficient of reflection is different from
zero. The reflected power P' _ 1012P1 constitutes the return loss expressed in
watts. The part of power Pe that effectively enters the filter is the difference
_ of the preceding ones: Pe =(1 - IPI2)P1� This is distributed in the elements Rc
_ (useful load), R1 and R2 (cavities). The active power Pe = IVeI2/Re (Figure 22)
dissipated in the imagi.lary resistance Re is in reality dissipated in the resistive
elements of the filter per a distribution that can be derived from analysis of the
network (Figure 20) and that is:
_ [25) PRC - Pe'(X, PR1 � Pe'S, PR2 - Pe'Y,
in which the quantities a, (3, y indicate the fractions of power Pe dissipated, re-
spectively, in RC) R1 and Rz; their values are expressed by:
r
a = ~1 , r) + (1 + 1/r)E 1 (1l~)2 ,
1 + ~ )2+ ~ z
~261 ~ - , ,
r) T IIr}1 + ~1Ix~1
1
Y (1 -f- r) + + 1/r)2 -f- (1/X)2 '
in which r and x are again furnished by [16) and [17]; one has a+ S+ Y= 1.
The extension of these considerations to the complete ring (Figure 24) is immediate.
ror transmitter 1, one has:
[27]
PRc
Pl = (1 -I PI 2)
pRl R - ~ P ~Y) � Q r
PR2 1
pl = l (1 P ~E) �Y,
PRl =I
P PI`,
i
in which, if the values of Ipl and af the parameters a, Q, y are calculated at car-
rier frequency fl, the expressions of [27) express the power distribution of non-
modulated transmitter 1.
Analogously for transmitter 2:
39
F'OR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-40850R040400084013-5
[ 281
FOR OFFiCIAL USE ONLY
PR� =I PI4,
P2
PZl = 2(1- I P I Z) � Y~
Pzz 2 (1 -I P1z)'p
PRL
Pz =(1 - I P I2) � a
with Ipl, p,, Q, Y calculated at carrier frequency f2 of transmitter 2.
+ The factors 1/2 that appear in the preceding formulas take account of the subdivi-
~ sion of power of each transmitter on the two branches of the ri.ng.
7.2. Modulated Transmitters
When the transmitters are modulated, the power distribution varies. With a non-
modulated carrier, all the power of a transmitter is associated with the carrier;
with a modulated carrier, in accordance with Par.seval's equation, the power associ-
ated with the modulated signal is the sum of the powers carried by the individual
lines of the spectrum. Referring to Appendix A, formulas [37j and [38], and modu-
lating with a sinusoidal tone of frequency fm, the expressions of [27] relative to
transmitter 1 modify into the following:
PRC pk OCk � Jk (m)lk- 0 'f'
Pi
w
-E- d E cl - 1 Pk Iz 'J"t c'n~ ,
k-l
PRi 1
Pl = ~ [(1 - ~ Pk ~2) � Pk (r+)]r-o -I-
w
+ pk ~a) � Px ' Jk (m) r
_ [291 k'1
PR2
Pl I Pk I' � "(k ~ (Yll)]k_ 0
~
-I- I Pk IZ) � Yk ' Ji (m) ,
k~l
PRL = LI Pk Ji ~'n+))x-o -F-
~
+2Z 1 Pk 11 'Jk(m)r
k-l
in which Jk(m) is the first-type Bessel's function, o� order k and subject m(index
of modulation). The first term represents the power relative to the carrier and the
sums of the terms of the series refer to the lateral lines. Coefficient 2 presup-
poses an identical contribution of the two linea of each pair of order k, and this
derives from the fact that for a small Ipflfl, the modulus of the filter's transfer
function is symmekrical vis-a-vis the tuning frequency. In the expressions of [29),
the subscripts k of the functions Ipl, a, S, Y i.ndicate that they are to calculated
by putting into expression [17] for normalized reactance, of which all of them are a
function, the value:
[30) (Af)k = kfm�10-3 (with fm in kHz, fo in MHz),
40
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
in which k ralces on, successively, all the whole-number values between 0 and n.
The sum of the power ratios [29] is equal to 1 when sums of the terms of the series
are extended to number n of the pairs of lines that make up the signal (in theory,
n=-); in practice, the calculation was done by putting into the computer program a
"test" that interrupts the sums of the terms of the series fcr that value of n by
which there results:
~
PRC/P1 + 2PRl/P1 + 2PR2/P1 + PRL/P1 ? 4,
Q having been fixed as 0.9999; in this way, the 2n + 1 lines of the spectrum thst
_ contribute 95.99 percent to production of total power are taken into consideration.
- One notes that by putting into the expressions of (29] m= 0, the expressions of
[27] are obtained again, keeping it in mind that Jk(0) = 1 for k= 0 and Jk(0) = 0
for k # 0.
Ttie upper part of Table 2 presents the values of the powers calculated with both
_ nonmodulated and modulated carrie.r, by menas of [27] and [29), for a frequency devi-
ation af 75 kHz and for a transmitter power P1 = 10 kWatt at the frequency of
89 M}iz (ring X).
In an analogous manner, by modulating transmitter 2 one can obtain the expressions
- �f PRC/P2, PRl/P2, PR2/P2 amd PRL/P2. In this case, though, the sums of the terms
of the series relative to the higher lines and to the lower lines with regard to the
carrier must appear separately, because frequency f2 of TX2 is on a side of the
_ characteristic and the lines of the spectrum are treated in a dissymmetrical manner.
The lower part of Table 2 presents the values calculated for transmitter 2 at 89.9
of ring X with P2 = 10 kW.
The total power dissipated in each cavity and in the absorption load is obtained as
the sum of the corresponding powers dissipated by the two transmitters, and the sum
of the voltages must be assumed as the maximum voltages.
It s}iould be noted that for the TX2 one has both a negligible difference in the dis-
tribution of the powers as between the conditions of nonmodulated transmitter and
modulated transmitter, and somewhat reduced loases of useful power (see section 9).
Once tne value of the power for each cavity is known, it is easy to go back to the
maximum currents and voltages on the internal elements, knowing the transformation
~ ratio. For example, the voltage at the open end of the cavities C1 and C2 (Figure
- 24), relative to nonmodulated TX1 only, is:
r.A Pai � r� Rc _ 1.1847 volt.
- sin (a ;:f'l)
8. Resonant Cavities of the Filter
Once the structure of the cavities and the Q necessary for achieving the desired
radioelectric characteristics are established, one goes on to determine their dia-
meters. This subject was dealt with in sections 6 and 7(Bibliography 6), and
therefore only the essential part of it is reviewed here.
41
FOR OFFICIAL U5E ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400084013-5
FOR OFFIC[AL USE ONLY
_
Table 2
-
Nonmodulated
Modulated
car.ri.er f.l
TX1
carrier fl
fm = 1 kHz
fm = 15 kHz
('9 MHz)
(W)
(W)
(W)
PRC
9,615.46
9,546.32
9,544.55
PR1
98.00
108.22
108.17
PR2
94.27
93.60
93.57
PRL
0.36�10-3
49.07
51.30
Loss of
useful
- 0.17
- 0.20
- 0.20
power (dB)
Nonmodulated
Modulated
carrier f2
TX2
carrier f2
fm = 1 kHz
fm = 15 kHz
(89.9 MHz)
(41)
(W)
(W)
PRC
9,940.60
9,938.12
9,938.45
PR1
0.34
0 35
0.35
PR2
11.90
12.03
12.03
PRL
34.90
36.12
36.11
Loss of
- useful
- 0.026
- 0.027
- 0.027
power (dB)
the coefficient of quality for the coaxial resonator is:
[31] Q= 4.17/ _f/pr �D�~(D/d) (with f in MHz, D in mm)
in which D= inside diameter of the external conductor, d= diameter of the internal
conductor, and pr = resistivity as referred to that of copper.
The function ~(D/d) presents a maximum, of unitary value, for D/d = 3.6, corresporid-
ing to a characteristic impedance Rcc = 77 ohms.
The value furnished by (31] is the maximum obtainable with a coaxial line; in prac-
tice, lower values are obtained.
When the energy stored in the resonator's electromagnetic field remains constant,
independently of the number i of dissipative elements, the real Qo is obtained as a
- parallel of the Q's that there would be from inserting the i sources of loss one at
a time:
42
C'OR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFdCIAL USE aNLY
[32] Qo = KQ with K= 1/(1 + PI/P + Pi/P),
P being ttle power dissipated in the coaxial and Pi the powers dissipated by the i
sources of added losses.
In the case being considered, it is possible to individuate principally two causes
of loss (Figure 19): the short-circuit disc, which closes the cavity, into which all
the maximum current Io goes, an3 the strip contacts necessar.y for tuning the cavity.
The expressions of the loss ratios (formulas [25] and [26] of [as published] Biblio-
graptiy 6) are referred to again; for the short-circuit disc:
[33] P1/P = 3.71�10-6�f�D�~(D/d)
and for the strip contacts:
[34] PZ/P = 26.66�10-6 f,L,cos26o
(f in MHz, D in mm)
(f in MHz, L in mm)
in w}iich Cp is the distance in degrees hetween the selection field of the strip con-
tacts and the shorshort-circuit plane and L is an equivalent line length that would
give rise to the same losses as those due to the "fingers."
For the cavities of ring X one has: 6o = 55.4� and L= 355 mm; ~(D/d) = 1.
9. Distortions Introduced by the Combining tlnit
In section 6 the necessity was pointed out of limiting the baseband distortions of
the frequency-modulated aignal transiting in the combining unit, especially in
multiplex operation.
The problem of the distortions causes all the more concern the narrower the band of
the circuits is, because of very close channeling, as is the case with the Mt Venda
installation.
The KF signal outgoing from tYie filter is affected by
stantaneous amplitude and phase. The amplitude, which
in synchronism wi.th ttie modulation of frequency by the
width, giving rise to synchronous AM modulation, which
tween the di.fference and the sum of the maximum values
the KF-signal envelope--that is:
iistortions that alter its in-
is no longer constant, varies
effect of the limited channel
is defined as the ratio be-
VM and minimum values Vm of
[ 351 AP1 = VM ^ Vn'` .
pM + Vm
In reception, this distortion is eliminated by the limiters, while the phase distor-
tions are transferred by the demodulator inko the baseband signal.
For checicing the distortions in the stereophonic signal, an original calculation
p:ocedure was used; it is described in Bibliography 8, and is applicable to any lin-
ear quaciripole network whose transfer function is known. It calculates the distor-
tions present in the A and B channels (left and right) after the RF signal modulated
by a multiplex has transited through the filter.
43
FOR OF'F'IC'[AL U5E ONY.Y
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R400400080013-5
F'OR OFFICIAL USE ONLY
ln summary: the spectrum of the modulated signal, calculated by Fourier's transform,
is multiplied by transfer function H(w) of the filter, in the sense that each line
' of the spectrum is altered in amplitude and phase in relation to the modulus and
phase of HU at its frequency. At output fram the filter, the antitransformed sig-
nal presents a variable instantaneous-amplitude envelope from which the synchronous
- AM is obtained, wh.ile the distortions are calculated or_ the basis of the demodulated
signal, obtained by me3ns of derivation of the instantaneous phase.
Several of the values calculated--for example, for channel 1(89 MHz) in transit in
ring X, applying power to channel A only with modulating frequency of 1 kHz and
~ df = + 75 kHz--are:
synchronous AM 0.93%
Linear diaphony - 56.5 dB
Nonlinear diaphony - 72.1 dB
Harmonic distortions - 71.6 d3 ,
which values are to be considered good.
Another procedure is mentioned, for evaluation of synchronous AM (Bibliography 9),
that uses an approximate method but has the virtue of simplicity. It consists in
assuming for VM and Vm in [35] the maximum and minimum values taken on by the trans-
fer function with variation of the �requency from the value fo of the carrier to the
values of the deflection peak fo � df. This procedure, called quasistationary ap-
proximation, furnishes synchronous-AM values approximated by defect. It has been
observed experimentally that the approximation can be improved if the values
fo �(df + fm) are considered as frequency extremes.
As regards reflected transmitter 2, the problem of distortions proves less important
if the dimensioning of the circuit is appropriate.
Several differences between the characteristics of the TX1 in transit and those of
the reElected TX2 of the same ring have already been noted (sections 7 and 6). For
the latter, they are: lower losses, smaller differences in power distributioe be-
tween modulated and nonmodulated carriers, less group cielay in the channel. Syn-
- chronous AM is also lower, being 0.12 percent for TX2 of ring X.
10. Auxiliary Circuits For Control of the Unit
The unit described is provided with a system of electronic circuits for the func-
tions of monitoring, signaling,equipment protection and personnel safety.
10.1. rlonitors
The monitoring circuits, situated in the instrument panel, furnishes readings of the
direct and reflected power at the unit inputs for each transformer and at the output
from the various rings, on lines A and B and on the artificial load (indicated by
Ra in Fibtire 9).
In the same section there is monitoring of the reciprocal isolation of the trans-
mitters between the gates 1- 1' (Figure 8) for each ring, and switchable reading of
the maximum temperature of the cavities.
44
FOR OFFICIAL USE" 4NLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFLCfA.L USE ONLY
MoniLoring of the cavity-tuning conditions is furnished by reading of the power re-
flected by the cavities that goes back onto the absorption loads and by reading of
the synchronous AM, whi.ch must be mini.mal in conditions of perfect tuning. These
measurements too are switchable to each ring.
10.2. Signaling
Sigraling is done with a number of L�ED's that constantly indicate the conditions of
the circuit and signal any breakdowns. They are located partly in a group under the
instrument panel and partly on the manual-switching frame. The fornier signal: power
on in the transmitters; the position of the manual switches; presence of the protec-
tion systems. 'nce lights on the manual-switch panel, though, permit switching only
if the transmitters are not under power.
10.3 Equipment Protection
This consists in automatic action to shut down
anomalies occur in tlie cirruits Powered by *_he
parameters monitorzd by the protection systems
lines A and 13 (shtit-off af transmitters at the
power on artificial load Ra; maximum reentry o
temperature.
10.4. Personnel Safety
one or more transmitters whenever
unit, including the antenna. The
are: maximum reflection on antenna
preestablished threshold); maximum
nto absorption loads; maximum cavity
In addition to the usual safety provisions prescribed for every piece of equipment
under tension, the unit and the manual-switching frame have been provided with a
circuit for intervention in case of wrong or dangerous maneuver. This circuit re-
produces, in direct current, all the possible runs of the RF power; only when all
the runs of the RF are closed are the relays operated that enable the various trans-
mitters ro stay under power. Any dangerous maneuver is prevented, a little ahead of
time, by the shut-off of the trar.smitter or transmitters transiting by way of the
circuit section in question.
Aclcnowledgements
We cite: the essential contribution of Mr Giuseppe Novaira, who competently and
skilfully carried out the entire cycle of ineasurements and the development of the
enti.re radio circuit of the un.it �or. Mt Venda, through its installation; Dr Frances-
co Rossi Doria, wiio efficiently supervised the construction and detailed installa-
tion oF the auxiliary monitoring and safety circuits; Engineer Renato Orta, who on
the occasion of thi.s project did an original and rigorous study of the distortions
oI the stereophonic signal (Bibliography 8), filling a void in the technical litera-
ttire on tile subject; and the Radiof.requency Laboratory for its notable contribution
to the workinb-oul and construction of the protection and safety circuits and re-
lated electronics.
APPENDIX
A) The relations thar linic a sinusoidal magnitude v(t) of pulsation wo (carrier)
wi.th a generic fu�ction of time x(t), modulating, are considered to be known. It is
recalled that in frequency modul.ation, a biunivocal correspondence is established
45
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
oecween x(t) and the instantaneous separation of the frequencies of v(t) in relation
to that of the nonmodulated sinusoid. For the instantaeous pulsation, oae has:
wi(t) = Wo + Kx(t).
Keepinb it in mind that the instantaneous pulsation is the derivative, with regard
to time, of the instantaneous phase, in the particular case of sinusoidal modulation
--that is, x(t) = cos wmt--one has:
[361 [36] v (t) _ ,A sin [wo t K j x (t) dt =
- A sin [wo t+m sin wm t]
in which m= df/fm is the index of modulation with Sf frequency deviation corres-
ponding to the peak value of the separation of the instantaneous frequency, and fm
is the modulating frequency. Expanding in [1] the sin and cos te nns of the subject
(m�sin wt), one has also:
[371 v (t) = d {�To (m) sin wot
t
~Jk (m) [sin (wo 1- kWn.) t l)k. y1D (WO - kWm) t]}
k~l
in which Jk(m) is the first-type Bessel's function, of order k and subject m.
The total power associated with the signal v(t) is given by the sum of the powers
associated with the individual lines.
The fraction of power PQ transmitted in a channel with limited band BQ = 2pfm is ex-
pressed by:
[381 Po =AZI Jk(m) =AZ�Q
te v
with Q< 1 and p is the number of pairs of lines considered around the carrier; for
_ P = w, Q = 1 results.
li) If the modulating signal is composed of the sum of two sinusoidal tones
x(t) = cos wlt + cos w2t, one obtains analogously:
[39] v(t) = A sin[wot + mi sin wlt + 1712 sin w2t]
witti ml = Sf/fl and m2 = 6f/f2, which, expanded, can be written (Bibliography 7) as:
1401 (t) = 3 7 1 Jn (mi) �
� Jx (ms) � sin (coo ~ hwi k(as) t.
The Craction of power YQ transmitted in a channel of limited band BQ is:
+ n a
411 FU - A= ~ Z LJn (mi) � Jk (ms)l= = -422
L--r k --v
with the condition -B2/2 < + hwl + kw2 BQ/2 and in which p and q are the maximum
orders ot the Bessel's functions corresponding to the lines contained in the band
considered BQ = 2Mw2 (with (1)2 > wl and M positive integer). For p= q=�O, Q= 1
results.
46
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R400400080013-5
FOR OFFIC[AL USE ONLY
BIBLIOGRAPHY
1. Young, L., "The Analyticat Equivalence of TEM-Mode Directional Couplers and
Transmission-Line Stepped-Impedance Filters," PROC IEE, 110, February 1963,
PP 275-281.
2. Matthaei, G., Young, L., and Jones, E., "Microwave Filters, Impedance-Matching
Networks, and CoupJ.ing Structures," McGraw-Hill.
3. Chuck Y. Pon, "A Wide-Band 3-dB Hybrid Using Semi-Circular Coupled Cross-Sec-
tion," THE MICROWAVE JOURNAL, October 1969, pp 81-85.
4. Pacini, G.P., "25-kW UHF Combining and Vestigial Filter," ELETTRONICA, No 1,
1963, pp 2-9.
5. I'acini, G.P., "Au[omatic Frequency Stabilization for Distributed-Constants Re-
sonating Circuits by Means of a Mechanical-Hydraulic Device," ELETTRONICA E
TELECOMUNICAZIONI, No 6, 1969, pp 210-212.
- 6. Pacini, G.P., "Design Method for Audio-Video Combining Filters," ELETTRONICA E
TELECOMUNICAZIONI, Part l: No 2, 1970, pp 54-64; Part 2: No 3, 1970, pp 106-114.
7. Black, H., "Modulation Theory," D. Van Nostrand Company, Inc.
8. Orta, R., "Distortion of Stereophonic Signal in Frequency-Modulation Transmis-
- sions," Technical report No 80-22-1, November 1980. RAI, Research Center,
Turin.
9. Boccazi, F., and Luzzatto, G., "The Synchronous AM Problem in FM TV Transmit-
ters," IEEE TRANSACTIONS ON BROADCASTING, Vol 13C-15, No 3, September 1969.
COPYRIGHT: 1974 by ERI-EDIZIONI RAI RADIOTELEVISIONE ITALIANA
11267
CSO: 5500/2309
47
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R440400080013-5
FOR OFFICIAL USE ONLY
ITALY
SILICON AVALANCHE PHOTODETECTOR FOR OPTICAL COMMUNICATIONS
Turin ELETTRONICA E TELECOMUNICAZIONI in Italian May-Jun 81 pp 116-124
[Article by M. Conti, M. De Padova and A. Modelli*]
[Text] Summary--Silicon avaianche photodetector for optical-fiber communications.
This paper describes the implementation of a reach-through avalanche photodetector
_ for optical-fiber communication systems with 0.8 to 1 um wavelengths. This photo-
detector has an n+pTrp+ silicon structure and has been developed using the planar
technique. Particular attention has been devoted to dopant profile, diffusion tech-
nique, geometry and other factors for use optimization. Technological features
(guard ring, channel stop, field plate, getter, passivation) made it possible to
achieve very low dark current and noise and even lower than those typical of avail-
able photodetectors. Two different systems have been implemented: diffusions on top
by side [as published] of the incident light and overturn position. This avalanche
photodetector operates with voltages from 200 to 300 V and variable avalanche gain
up to a value 100. The responsivity is between 0.55 and 0.65 A/W for the wavelength
range from 0.8 to 0.9 um. Besides, low capacitance allows this photodetector to be
an excellent device for 34 Mbit/s communication systems.
1. InCroduction
In optical-fiber communication systems, the incoming optical signal is converted in-
to an electrical signal by a photodetector that must have adequate characteristics--
that is, it must not introduce appreciable distortion in the incoming signal and
must generate the least noise possible. The greater the noise, the greater the sig-
nal level needed to ensure satisfactory communication, and consequently, the shorter
is the communication section achievable at equal transmission power.
The maximum performance characteristics in the 0.7-0.9 um band are obtainablle with
a silicon avalance photodetector (Bibliography 1). Siich a photodetector has an in-
ternal gain, and in this way the noise input of the succeeding amplifier can be made
negligible. In addition, because of the physical properties of silicon, the ava-
lanche noise is intrinsically low. In this way, sensitivities even 10 dB higher
than those obtained with a p-i-n photodetector can be achieved.
* Doctor of Engineering Mario Conti of the SC'S-ATES (expansion unknown); Dr Matteo
De Padova of the CSELT [Telecommunications Research and Study Center]; Dr Alberto
Modelli of the SGS-ATF.S.
Typescript received 2 March 1981.
48
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R400400080013-5
FOR OFFICIAL USE ONLY
I'arti.cularly interestinb is the RAPD (Reach-through Avalanche Photo-Detector) struc-
ture, which can be developed with planar technology, gives responsivities only a
little lower than the theoretical limit even with supply voltages of only 150-350 V
--thus far lower than for the step-junction detectors (Bibliography 1)--and finelly,
has low dark current and high reliability. For these reasons, the avalanche photo-
- detector designed and built at the SGS/ATES by the joint SGS/ATES-CSELT group is of
this type, and is styled by the acronym OCPDA (Optical-Communication Photo-Detector-
Avalanche). Below are reported the optimal-design and fabrication conditions for
this device.
2. Operating Principles and Simulation
The cross section of the photodetector is represented in Figure 1. The active zone
is the one between the two broken lines, and it is an n+p7p+ junction typical of the
"reach-through" structure.
coa(aQ st 02~ ro(at) , 360~m _r.j fP'AI) taa(at)
I 1 0 1 r.
_1ZLu^"�~~�
J / i n� ~
xpN ; xj p~ ANC4UARllOWAt 3~ ` CN4NNEl y
~ STOPPCP X70
5` ~ xn n
D- ~ p' ~ I
X
70N4_ 5YU07A7A ~ ?ONAATiIWI I ?pNJ~ ; Vl1DT4TA
/
~ x \Z% llJ
Key:
Figure 1. Structure of the OCPDA photodiode
1. Emptied zone 2. Active zone 3. Guard ring
Figure 2 represents: above, the profile of net concentration Nd - Na per cm3 (Nd =
concentration of donors, Na = concentration of acceptors) along the lateral axis x
(indicated also in Figure 1), and below, the profile of the electrical field E in
i the condition of voltage sufficiently high to empty the TT region completely (Reach-
Through situation: RT),.
One notes that the maximum value of the electrical field ^2.5 - 3.105 V/cm is local-
ized at the juuction sJ, and that the multiplication of the carriers generated by
the incident light is therefore concentrated there. There is also a far broader re-
gion, in which the electrical field is considerably lower and rather uniform, which
functions as an absorption region for the incident photona and for collection of the
charges generated by them.
All the regions present on the front are achieved with planar technology--that is,
by obtaining, by photolithographic technique, adequate windows in a silicon-oxide
state increased by heat oxidation, and by diffusing or implanting the appropriate
doping agent through t-hem (Figure 1).
Since the n+ region is very thin, it is necessary to use an n+ guard ring (deeper
diffusion) that avoids problems of premature breakdowns and of inetalization at the
edge.
The "channel stopper" external p+ ring servea to limit the emptied region laterally.
49
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00854R000400080013-5
FOR OFFICIAL USE ONLY
INd-Nal~m a~
10 i8 n.
i06 p
P''
to"
TT
~Z
x
~ x;
;X
I
I
x.p.
~ob 1
b)
l '
Icm
,c~ ~
~
~
104 1 .
I
x
-=-r-~---.
0 vi ,xpn
~ X
�
~
~
nD

5~
Figure 2. Profile of concentration a) and profile of electricat field b) along
axis X(see Figure 1) in the active zone of the photodiode
w
P I
REGIONE SVU0TA7A(1)--- Ij Q+
~I REGIONE01 MOLTIPLIGAZIONE(2) ~
1
~ - 1
~
0

I ~ O~ I
~
1
hJ ~ I
I
I
I j ~ I
x
�
i
0 xj lpyl i x*p X
IOMZ2AZIONE(3) COPPIA
A VALANGA PRIMARIA
�7S1
_ Figure 3. Diagram of the operating principle of the "Reach-through Avalanche Photo-
Detector" (RAPD) photodi.ode
hey:
1. Emptied region 3. Avalanche ionizatian
2. Multiplication region 4. Primary pair
Metalization, in the classic planar structure with field electrode (FP = field Plate,
Figure 1) and equipotenti.al ring EQR (Figure 1),is useful in stabilizing the situa-
tion of the electric:al. field at the surface. In thia way, surface breakdown phenom-
ena are avoided and the leakage current ia kept low and stable in time.
50
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2047102109: CIA-RDP82-00850R400404080013-5
Figure 4.
FOR OFFICIAL USE ONLY
Concentration profiles used in digital simulation of the diffusion pro-
cess (logarithmic scale fo: cor.centration). 1: profile of n+ diffusion
with surface concentration Cgn+; 2: profile of p diffusion with surface
concentration CSP; 3: constant concentration Cg in the n zone; 4: profile
of p+ difEusion. ~
(V]
SoC
~
a pc
a
Cp
Q
b WC
u
~
n
~ 2�
r'1
1~
na
0
LD � LUNGHCIIA
`
pffuSqNC
1/
v
G`SiAATO p
~
F
ti
0
Y' 'n o
N
J
n a
)
~ so
3 es
2,80,
'
Nm
715
2
yT
1 2D~~ S 7 10 1 2 3 S 7 10
(4)TCMPO D~ 1FUS1DME A 920�Cj0(r) TcWo a arry"c Avio-c (oac) (4)
Figure 5. Simulation of the n+ diffusion process: a) parameter of each curve diffu-
- sion length Lp [as published] of the p layer; b) parameter surface con-
KeY: centration CSp [as published] of the p layer.
1. LP = diffusion length, p layer 4. Diffusion time at 920�C (hours)
2. CSP = surface concentration of boron 5. atoms per cm3
3. Breakdown voltage
The sensitive area of the device is circular, with diameter 360 um, and its size was
determined so as to limit the device's capacity as much as possible while permitting
easy coupling to the optical telecammunications fiber, which, as is known, has an
outside diameter of 120-200 Um.
51
FOR OFFICIAL USE ONLY
C,p � COrKCNTR
StDE Af C WL C
/ ~DI BORG
\y
b)
~
f
r;
D
1
r
z
�
~
~
t
f,t
�tomy/un~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R400404080013-5
When an electron-gap pair is generated by absorption of a photon, the electron, by
effect of the electrical field, moves toward the n+ region and passes through the
high-field region siruated at the ju.-Lction xj (Figure 3). In this region, the elec-
trical field is such as to give it sufficient energy to ionize by impact--that isy
to create an electron-gap pair, which in turn can produce another pair, and so on.
Consequently, the current relative to a single photon becomes multiplied by a factor
that depends on the electrical field and therefore on the voltage applied to the
photodetector. The electrons are then collected by the n+ region, while the gaps
move toward the p+ region and are collected there. 'I`his thin p+ layer, in the rear
of the chip (Figures 1 and 3), keeps the emptied region from reaching the rear sur-
face and also makes good ohmic contact possible.
_ The structure of Figure 2, which has been studied in depth, has the particular char-
acteristic that the avalanche multiplier effect, and therefore the gain of the pho-
- todiode, grow in a relatively slow manner with increase of the voltage applied.
_ The profile of reference impurities is shown in Figure 4.
The profile of the electrical field E(x) is calculated with Poisson's equation:
d2v _ p
dx7 e
in which v= electrostatic potential, e= dielectric constant, p= spatial charge
_ assumed equal to the net concentration of doping agent--that is:
[21
p = q(Nd - Na)
with q= charge of the electron.
Equation [1] is solved in a numerical manner by imposing the corresponding contour
conditions at an applied reverse voltage v= VR. A complication arises from the
fact that the thickness xd o�- the emptied region is noC known beforehand. The calcu-
lation therefore proceeds by beginning with a zero emptying thickness and increasing
= it gradually until the required voltage VR is reached.
The capacity per unit of surface of the photodiode is given by:
[3l
c = e/xd.
The multiplication factor M, defined as the ratio between carriers collected for
each photon absorbed, is given by the expression:
_ cZP ~+~(a. _ aD) dx]
u _ o
i_ Jo �n exr. f o(�. -(xp) ax'j ax
in which:
151 an - anoo exp Ln1E)f aY = a9oo eZP by/E).
The values of the coefficients valid for silicon (Bibliography 2) at ambient
temperature are:
52
FOR UFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400084013-5
Cln(� = 3.8� lU6 cm-1,
bn = 1.75�106 V/cm,
FOR OFFiCIAL USE OIVLY
cxpoo = 2.25�107 cm^1,
bp = 3.26�106 V/cm.
Since an and aP according to [5) depend on E, M proves to be an increasing function
of VK, and for a certain value VR = VB tends to infinity (for VR>Vg, it becomes neg-
ative). Vg obviously represents the breakdown voltage of the photodiode, which is
therefore defined by the expression M(VB)
It is of particular interest for dasign purposes to analyze the influence on VB of
the net doping profile. It is strongly determined by the n+ diffuse region, which
must be regulated in such a way as to neutralize the correct quantity of acceptors
introduced with the p diffusion. As will be explained later, this is achieved by
depositing an appropriate quantity of n+ doping agent (phosphorus) on the surface
and raising the chip to 900-1,000 �C for a time tn, so that the phosphorus diffuses
in rhe chip with Gaussian profile whose characteristic length is Ln = Dntn in
whicli Dn is the diffusivity of the phosphorus under the operating canditions. The
final profile can be approximated by a distribution of acceptors with practically
irtunobile Gaussian distribution from which one subtracts the distribution of donors,
which depends on the heat treatment described.
With these premises, the voltage Vg was calculat2d in function of diffusion time tn.
The case of diffusion at 920 �C is represented in Figure 5a)b).
One notes that the curves of breakdown voltage in funr_tion of diffusion time are
composed of two straight lines. In the first line, the breakdown voltage is lower
than the voltage necessary for emptying the p zone and the breakdown voltage in-
creases slowly. As soon as the n+ diffusion has sufficiently compensated for the p
diffusion, the breakdown voltage begins to increase far more rapidly with diffusion
time.
For the purposes of study of the process, it has been interesting to study how the
form of the diffusion curves varies with the variation of two initial parameters,
which are the length L of diffusion of the p layer and its total dose. The curves
of Figure 5a) were calculated for a consCant total boron dose Qp by varying its dif-
Lusion length Lp that appears as a parameter. Figure 5b), on the other hand, corre-
sponds to the case of a p diffusion with constant diffusion length Lp but with vari-
able total dose Qp; in this case, CSp is the surface concentration of boron, assumed
as a j>arameter. The curves indicated by 1 represent the breakdown voltage VB, and
t}iose indicated by 2 represent the voltage at which emptying of the 7 region begins.
_a
Comparing Figures Sa) and b), one notes that the process is far moze critical caith
regard to the predepositing of boron rather than with regard to the diffusion of
boron. Indeed, going from a boron surface concentration of 1.1�1016 to 1.2�1016
atoms per cm2, a good 1 hr 30' shift of the rediffusion curves at 920� is obtained.
This shows that very high uniformity is necessary in the boron dose deposited in
order to avoid localized breakdowns.
3. Fabrication
As the starting material for making these avalanche photodiodes (APD), Float Zone
silicon chips of 2" diameter, with resistivity of 2,000-4,000 ohms�cm and orienta-
tion (111) are used. The high rQSistivity, obtainable only with material nf "detec-
51
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R440400080013-5
FOR OFFICIAL USE ONLY
.:j
tor" class, is indispensable for achieving empty regions more than a hundred microns
thick with breakdown voltages lower than 350 V.
After the initial polishing and oxidation, the first two processes are carried outc
the making of the ny guard ring and of the p+ "channel stopper," achieved with plan-
ar dif.fusion in an atmosphere of, respectively, phosphorus (from POC13) and boron
(from BN).
As regards the p deposition, after many tests carried out with the traditional depo-
sition process (predepositing with BN chips followed by a doubie "steam leach"), the
necessity of having greater uniformity emerged and it was decided to go over to de-
- position by ionic implantation. This method does indeed make it possible to regu-
late the dose deposited--which in our case is about 4�1012 atoms per cm2--with great
_ precision.
A measuring process that makes it possible to check accurately the dose deposited
has also been developed. It uses the four-points method (Bibliography 3) and makes
it possible to have a direct check on the dose implanted. It is essential for the
dose to fall within very narrow limits; indeed, if there is a deficiency of boron,
the chips go into breakdown at excessively high voltages, and if it is too abundant,
it is necessary to rediffuse the n+ for too long a time, which reduces quantic effi-
ciency in case of illumination from the front.
After implantation of boron, the chip is raised to 1,200� C in a controlled atmo-
sphere for several hours so as to achieve rather deep diffusion, with characteristic
length Dptp = 1.8 um.
The chip is then brought to a thickness of 120 um by means of lapping and chemical
attack on the rear face. An implantation of about 1014 atoms/cm2 of boron on the
rear constitutes the rear p+ region.
Finally, the chip is treated in a phosphorus (POC13) atmosphere at 920� for a suit-
able time so as to cause penetration of the phosphorus previously deposited on the
front and to achieve the required profile of acceptors in the p region. The treat-
_ ment takes place in a phosphorus atmosphere, and in this way a"gettering" effect
(Bibliography 4) is obtained wtiich is useful for eliminating microdefects in the ma-
terial and consequent causes of premature breakdowns. The cooling cycle, which has
to be a very slow one, is also very important.
4. Metalization
The two types of arrangement that have been made and evaluated are represented in
Figure 6. In the first, a), light falls on the photodetector from the n+ diffused
region, or from [he front, while in the second case, b), it falls from the p+ re-
gion, or from the rear.
The first arrangement is of a type conventional for integrated-circuits technology.
The metalization on the front (FP field electrode of Figure 1 and equipotential ring
EQR of Figure 1) is of aluminum, while that on the rear it.is of nickel-chromiiim. An
antireflectant layer of Si3N4 or Si0 is then deposited, as described later on.
In the second arrangement (Figure 6b), it is necessary to make a metallic "cushion"
of sufficient thickness to keep the alloy from making contact with the equipotential
5r+
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2047102109: CIA-RDP82-00850R400404080013-5
FOR OFFIC[AL USE UNLY
11
ring during the operation of soldering the plate to the bottom, causing a short cir-
cuit of the diode.
Figure 6. Section of two OCPDA photodiodes with di�ferent metalization: a) for
front illumination; b) for rear illumination.

(wl
20
t0
'o
- ~ S
N
~
~ e
W ?
Q
N ~
~
OS
o2
o+
0
9
,
o,e
0,7
0,6
0,9
0.94
o,e
o uoo aoo ioo ( v) o ioo n0 1 v )
(1) 1CH3I01[ iNYLRSA (1) iCNVONC WVCASA
Figure 7. Responsiviry in function of polarization voltage for various values of
incident wavelength: priotodetector illuminated from the side of a) p+
diffusion; b) n+ diffusion.
_ Key:
1. Reverse voltage 2. Responsivity
55 ,
FOR OFFICiAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
For this purpose, a"cushion" of gold, about 50 um thick, is made from galvanic
growth, starting from a layer of gold-chromium evaporated over the aluminum metaliz-
- ation and delineated by photolithography. At the same time, the metalizatinn on the
p{' side is done, aligned with the preceding and achieved by evaporation of aluminum.
In this case also, an antireflectant layer of suitable material and thickeess is de-
posited.
The plate is then soldered onto a type TO-5 container with ceramic mount. In this
way, both cathode and anode are separated from the metal case, and the diode's satel-
lite capacitances can therefore be reduced or eliminated.
Closed photodiodes Are then made, with cap with transparent window of glass or with
optical fiber incorporated. This last-named version is of particular interest for
optical-fiber communications, since in such case the connection to the line fiber is
by a simple aligned auto joint and does not require an additional operation.
5. Performance Characteristics of the Avalanche Photodetector
5.1 Responsivity
The responsivity of an APD detector is defined as the ratio between the detected
current and the incident photonic power at a given gain and wavelength.
In the case of a detector with given gain M, it can be expressed as the result of the
following processes: reflection at the surface, absorption in the frontal region,
conversion of phontons into electron-gap pairs, and internal multiplication.
a) Reflection
Reflection of light at the surface of the diode can be reduced if it is covered with
a layer of antireflectant material. It has been demonstrated (Bibliography 5) that
if this antireflectant layer with refraction index n2 has the thickness d=X/4n2
(a = wavelength of the incident radiation), the reflectivity has the value:
.R ( 713 ti 1 - n z:
~ 61 113 � 1 Nzz 1
in which nl is the r.efract:ion index of the external space and n~ is that of the sili-
con. As emerges from [6], reflection R can be minimized if n2 is the geometric mean
between nl and n3.
For nl = 1(sir, vacuum), and since for a= 0.7 - 1 um for silicon ene has n3 =
3.5 - 3.7, then in order to obtain R=-0, n2 - 1.9 is necessary. Silicon nitride
(Si3N4) and silicon oxide (Si0) have a refraction index close to that value.
b) Absorptiort in the Frontal Region
On the surface of the photodetector there is always a semiconductor layer (for ex-
ample, n+, Figures 1 and 3) that absorbs photons without contributing to responsiv-
ity. Given oe as this layer's coefficient of absorption and ws as its thickness, the
fraction of luminous energy transmitted is:
exp(-pr,ws ) .
56
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/42109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE O1NLY
It is obviousl.y necessary to reduce ws to the minimum, given that q, depends on the
material used and is a function of the wavelength a employed.
c) Conversion
The photons that.,reach the emptied region of thickness w(Figure 3) are converted in-
to electron-gap pairs. If the materi2l is very pure, as is the "float zone" silicon
generally used, the parasite absorption phenomena are negligible and the number of
pairs created equals that of the photons absorbed.
Given Po as the photon power entering into the emptied region, the generation of
pairs is given by:
X d.P (x~ aa
9'x) � hc dx I hc Po esp c~)
in which h= I'lanck's constant and c= speed of light.
The current I produced by photon power Po therefore has the value:
I= q f o 9(x) 4�~ [i - eap aw)l
- in which q= electron charge.
In this expression, the component of photons reflected by the rear surface has not
been taken into consideration.
The total respoonsivity is the product of the factors considered and therefore has
the value:
[ 71 (R, (,lf, a) = ~(1 - R) 3I exp aw.)
[1 - eap ( - aw)]
in wtiich M is given by [4].
~ [7] is valid for a photodetector, such as the one considered, made with a homojunc-
tion; it makes it possible to study the influence of pe--which, as was said, depends
on X--on responsivity bq.
In silicon, pc is large at short wavelengths, and uq is therefore limited by the term
exp(- wS). With increase of a, a decreases rapidly, and this term therefore tends
toward 1; but t.he bracketed term, which, for a tending toward zero, also tends toward
zero, becorties decisive. It is therefore important to use the biggest possible col-
lection ttiicknesses w. This has a negative effect on response time, as will be shown
_ farther on. -
In Figures 7a) and b), the curves of responsivity 4? are presented in function of the
reverse voltage applied for various wavelengths, for APD's illuminated both from the
p+ side and from the n+ side.
Measurement was done by illuminating the detector with monochromatic light of known
intensity, calibration having f-irst been done with a rated EG & G radiometer. The
current generated by the APD is amplified by an operational amplifier and recorded by
_ a logaritlimic recorder in function of the voltage applied. Gain M is evaluated as
_ the ratio between the light current at the voltage involved and that of a p-i-n de-
tector built with the same technology.
57
FOR OFFICIAL USE QNLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2047/02/09: CIA-RDP82-00850R400404080013-5
FOR OFFICIAL USE ONLY
one notes that R (Figure 7a) is an increasing function of the voltage applied and
repeats the course of M, which in fact appears in expression [7].
Ah abrupt increase occurs for V_ 80-90 V--the voltage at which the reach-through
situation appears. For lower voltages, & is that of a p-i-n diode, while with high-
er voltages it is much greater.
'fhe case of illumination from the n+ side (Figure 7b) is similar to the preceding,
except that one notes that the falloff of 67 at the short wavelengths is greater.
Comparison is given in Figure 8, in which the high-gain responsivities of both the p+
APD and the n+ APD are presented. It can be noted that for high wavelengths
(0.9 - 1 m), the OCPDA devices have better responsivity than the best devices avail-
able on the market. This is explained by their having a rather thick intrinsic zone
120 um). At the short wavelengths, the responsivity of the n+ 1 detector is lower
ttian that of the 9+ type; this is because of the fact that a part of the radiation is
absorbed by the front region n+. The n+ 2 specimen, though, in which the n+ diffu-
sion is less heavy, has high responsivity even at thia wavelength.
gL
(w)
30
20
Figure 8
~
~
b
0,6 0,7 OB a(Nm) 1
Spectral response of several photodiodes illuminated from the p+ or n+
side for a multiplication value M= 60.
If the results presented in Figure 8 are compared with those calculated with [7k, it
is verified that experimental iR is only slightly less than the theoretical value in
the field 0.7 - 0.9 um--that is, in one of the areas of greatest interest for optical
communications.
5.2 Capacity
Ttie capacity associ.ated with the detector has negative effects on both speed of re-
sponsP and noise, contributing to an increase in amplifier noise (see below). It is
[herefore important to limit it as much as possible.
It consists of three components:
--capacity associated witti the central active region and with the n+ guard-ring re-
gion; it is that of a flat condenser, and for high voltages there�ore tends toward
vacuum value (Bibliography 3);
--capacity relative to the lateral cylindrical region and to the field electrode FP;
at operating voltage, though, these are negligible;
58
FOR OFF[CIAL USE ONLY
~
'.\3
xo,

~
1 RC
A
i
2 OCPOA p
3 OCPOA 1� n'
4 OCPDA 20 n�
4.
~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R400400080013-5
FUR OHFiC1A1. USE UNLY
--capacity due to the bottom; this capacity is minimized by mounting the detector on
an insulating plate in such a way that no terminal of it is in electrical:contact
with the bottom; with this mounting, the parasite capaciky can be reduced to about
- ha1F what it would be with direct mounting on the bottom (0.3 as against 0.6 pF).
- The capacity-voltage curve is given in Figure 9. At the voltage of 70-90 V, one
notes an abrupt dropoff of capacity, which is due to arriving at the reach-through
situation. For lower voltages one observes a second "hump" that is due to stabiliz-
ation of the emptying situation in the semiconductor under the field electrode.
u
c
(pF)
10
OL
0
50 100 150 ( Y ) 200
rigure. 9. Characteristic of capacity in function of the polarization voltage of an
OCPDA pYiotodiode
In the last analysis, there is liLtle difference in capacity in operating condi-
tions in the two situations, n+ and p+; its value is 1.5 and 1.8 pF in connection
with insulated bottom and with grounded bottom.
5.3 Speed of Response
'The photogenerated char.ges pass through the w region at a speed v that deper.ds on
the electrical field present in it. As is known, with silicon it is proportionate,
with low fields, to the electrical field in accordance with a coefficient of propor-
tionality p called mobility. But when the electrical field becomes high, on the or-
der of 104 V/cm and above, the speed tends toward a saturation value vs on the order
of 107 cm/s. There is also a second cause of delay: the time needed by the ava-
lanche to stabilize (Hibliography 1). But it is important only in devices that are
very fast and not thi.ck, such as the present ones, in which high responsivity is re-
quired for X = 0.8 - 0.9 um.
Transit time can be calculated with the expression tt = 0.8 w/v; it conditions the
photodetector's intrinsic cut-off frequency, given by the expression:
f3dB - 0.4 v/w = 0.5/tt.
In order to keep t}ie cut-off frequency high, a high v must be achieved, in view of
the fact that w cannot be too small, which would reduce responsivity. This is done
by keeping the electrical field high in the w region, and therefore a part of the
- voltage applied has the purpose of achieving this electrical field.
I'he pulse response of lliis pho*_odetector is presented in Figure 10.
59
FOR OFFIC[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5
F()R UNF'I('IA1, t15H: ()NLY
C... :i~ d"M~
EME!06
p~~W13
rFI_Z ~'it~n
M-7P~C' 0
..ft._
_ Figure 10. Oscillograms of the response of OCPDA photodiodes subjected to a light
pulse. The response of a fast p-i-n diode is given for comparison.
Axis of the abscissas: 1 ns/div.
0
ia
~"Q)
t07
1
?
10
3
4
1
011--
p 100 200 300 ( v) 400
Figure 11. Inverse characteristics in dark of four OCPDA diodes
4 10 m
A 1
(~rl
z
1C i
~f~
rp
-
-
I
-
-
) o
,
,

,

v
~
O ~
-
-
,
i
7
5
416 11
n
d
0
20 40 70 100 M 200
Figure 12. Spectral density of noise current in function of multiplication fackor
Key M(the broken line shows, Eor comparison, a course proportionate to M).
:
l. Side
'1'he light pulse is Yenerated by a type-C 30025 laser and has a width t* that is
measured by s fast p-i-n detector with a rise time ts < 0.35 ns.
60
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE OIYLY
}tesponse time with impulse excitation is obtained by subtracting time t* from the
measilred value tm with the expression:
t = e tm2 - t'i .
One thus obtains values of about 2 ns and 3 ns, respectively, for detectors of type
p+ and n+. The differei.ce, and also the rather different form of the pulse de-
tected, are due to the fact that in the first type, multiplication is rather uniform
for all the photogenerated carriers, while such is not the case with the n+,
The times given are entirely satisfactory for 34-Mb/s PCM systems and also for high-
er frequencies. They can be further reduced by reducing the thickness of the de-
tector.
5.4 Noise
As was said in the introduction, it is important for the detector-ampZifier system
to introduce very low noise, so as not to deteriorate the incoming information. The
lower the noise, the lower the optical power necessary to ensure detection of pre-
establi.shed quality (signal-to-noise ratio, error rate, etc), and therefore, the
longer the section that can be constructed for a given transmission pawer. Let us
review the various causes of noise introduced by the detector and evaluate their
contribution to [otal noise.
a) Quantic Noise
With an optical power Po there is associated a flux of n electrons (that flow inde-
pendently) in whic}i n= pPo/hv) in which n quantic efficiency of the detector, h=
Planck's constant, v= frequency of the radiation. The fluctuation of these elec-
trons gives rise to a"shot" noise current whose mean quadratic value is given by:
(81 is? = 2qBIp with Ip = tlqn = photogenerated current for M= 1, and B= pass band.
b) Avalanche Noise
In the avalanche process, the individual light pulses are multiplied in accordance
with a stochastic process, and the noise is therefore greater than what would be
calculable if only the indi.vidual packets Ip�M were considered; this is taken into
account by means of ttie coefficient F, that is called the "excess noise factor."
The noise current is thereFore given by:
191 in? = 2qFBIpM2
in whicti:
[101 F_M 1- L) (M -1)z ~
M
x
with k = ap/an
in whi.cticxp and an are, respectively, the coefficients of ionization for the gaps
_ and for the electrons. F therefore proves to be an increasing function of multipli-
cation factor M. rtoreover, the lower k is, the smaller F is; it is therefore good
61
FOR OFF1C[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400080013-5
- FOR OFFICIAL USE 01VLY
fur the primary cliarges (electrons in silicon) to be those with the highest coeffi-
cient of ionizatior.. This explains the choice of the n+p-ffp+ structure for this
photodetector.
From the expressions [S] of a and cxP it is deduced that ks smaller as the avalanche
electrical field is lower. However, low fields require considerable avalanche
_ thicknesses and therefore high polarization voltages.
In the case in question, a value of k= 0.02 was obtained, which entails a rather
small value of F in operational multiplications.
c) Dark Current
The detector's reverse dark current also involves a fluctuation and therefore noise:
it is composed of a surface term IS and a"bulk" or mass term Ib. They are given
by:
- IS = qnivrAS, Ib = 2 q~ Aj w
in which: ni = intrinsic concentration, vr = speed of surface recombination, T=
average life in vacuum in silicon, and AS and A� are the surface area not involved
in the multiplication and the surface area of tge junction (Bibliography 3).
The darlc noise current is calculated with the Shot-effect formula, and is:
[11) -1-7 = 2qB (IS + IbMZF).
In Figure 11 are given the reverse-dark-current curves for several OCPDA diodes in
function of applied voltage.
d) Noise Due to Amplifier
The photodetector is closed on a resistance RL which is the amplifier's input resis-
tance: a thermal noise is associated with it. In addition, the amplifier is the
site of noise that can be represented by increasing the noise produced by RL in ae-
cordance witti the coefficient FA (noise factor of the amplifier). The noise current
of tlie amplifier therefore proves to be:
z 202
[ 121 - 4 ~TH 11 FA (1 R3 ~
1
in which K= Boltzman's constant, T= absolute temperature, C= total capacity at
amplifier input, and u> = 27rf is the angular frequency of the signal.
1 q~. a
Since the signal power associated with Po is given by 2~lw POM), the oyerall ra-
,
tio S/N between the signal and the noise is given by:
[13] S
4
2 (r~ Po~f s
q~'o
2qB [1. (Ib -f-7) hv M2F + 4KTI~yI3 1+Fd 1+ c02X2 L C=
3
62
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007102109: CIA-RDP82-00850R000400080013-5
FOR OFFICIAL USE ONLY
6
F
5
4
3
2
1
10 20 50 100 M 200
.
�
>
- TEOR
CA (k c002)
�OGPDA - p'
� � � OCPDA - n`
*
+
~
Figure 13. Course cf excess noise factor F with increasp of M: diode illumin-
ated from the p+ diffusion side; diode i luminated from the n+ side
_ Key: V
1. Theoretical \
s�1o
NEP 4
I 1 4 OGPDA
1 HZ ~ 3
2 I I I I I I
20 50 100 M 200
Figure 14. Noise Equivalent Power in function of M
[131 expresses quantitatively the dependence of the signal-to-noise ratio on the
various parameters of the detector and of the amplifier. We find the fact that with
the presence as denominator of a term not dependent on M, S/N initially increases
with the increase of M. Beyond a certain value, though, the tendency reverses,
since the first term--in which, because of the presence of F, dependence on M is
greater than in the numerator--becomes predominant as denominator.
Ttius there exists an optimal value of M that optimizes the S/N ratio; to ir corre-
sponds the situation in which the two quantities to be added L-o the denominator are
about equal. In these conditions, the maximuR S/N value depends on the input-signal
level Po, on 1/RI,, on F(factor intrinsic to the detector) and on the amplifier
noise factor FA, as well as on the total capacity in parallel to the detector C.
hleasurement of noise was done by sending to the photodetector an optical power sig-
- nal equal to 40 nW, obtained from an LED HRED 956 L emitting at 0.9 um. The re-
sponse was registered by a transimpedance amplifier with very low noise and was read
with a voltmeter of true eEfective value.
With variation of the photodiode polarization voltage and therefore of multiplica-
tion M, the noise voltage was read, both with illuminated di.ode (VL) and with diode
63
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400080013-5
FOR OFFIC(AL USE ONLY
in dark (V4). Subtracting the effective values, one obtains the value Vn related to
_ the photonic noise--that is:
Vn= VL-Vg ;
This value, divided by the amplifier's reaction resistance, provides the value of
noise current in given by [9).
The course of in in function of M is represented in Figure 12; it can be noted that
in increases faster than M. One can then deduce also the "excess noise" factor F
given by [10].
In Figure 13 are given the F(M) curves for devices illuminated both from the p+ di.f-
suion side and from the n+ diffusion side. The experimental results coincide very
well with the thecretical curve deduced from [10] and drawn in the figure for
k= 0.02. This value is among the lowest reported in the literature, and it has
- been possible to actiieve it only with appropriate geometry and careful technology.
The commercially available devices have slightly higher k values: typically,
k = '-0.025.
Another notable datum useful for expressing the good quality of the detector is the
NEP (Noise Equivalent Power), defined as the incident-optical-power value necessary
in order to have a unitary signal-to-noise ratio (S/N = 1); it proves to be:
NEP ln
vL  D
'The value of NEP is derived from the preceding measurements of in given by [9] and
of a given by [7]. The typical course of OPCDA photodetectcrs is given in Figure
14; it presents a minimum for values of M around 60 - 70. This va2ue of the multi-
plication factor P1 is therefore the one that optimizes the S/N ratio of the detector
system.
6. Conclusions
The optimization criteria, the fabrication process, and the electro-optical perform-
ance characteristics of the UCPDA avalanche photodiode designed to be used in optic-
al fiber systems with wavelength between 0.8 and 1 um are described.
The n+p7p+ planar structure with channel stopper and other technological expedients
- has made it possi.ble to obtain devices with high multiplication and very low dark
current.
Two possible arrangements of the diode were constructed and studied: diffusion from
the incident-light side and i.n the reversed position. For the wavelength involved,
0.8 - 0.9 pm, a small difference in responsivity values was observed; as regards
noise also, it was seen that it decreases only slightly with illumination of the di-
ode from the p+ side. The structure with iliumination from the p+ side (Figure 5b)
therefore does not seem so mucti supeiior to the other as to justify the necessary
teclinological complications, except for very advanced applications.
In conclusi.on, the OCPDA device made by the SGS/ATES-CSEL'T group presents character-
istics comparable to those of the competition, and in some respects superior, as re-
- gards dark currents and noise.
64
_ FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5

APPROVED FOR RELEASE: 2407102109: CIA-RDP82-00850R000400480013-5
Acknowledgements
FOR OFFICIAL USE ONLY
The authors wish to thank Dr G. Randone for the discussions with him and Messrs
L. Gandolfi, G Vento and A. Destro for their contributions to the construction of
the devices.
BIBLIOGRAPHY
- 1. Webb, P., McIntyre, R.J., and Conradi, J., RCA REVIEW, 35, 1974, p 235.
2. Sze, S.M., and Gibbons, G., APFL PHYS LETT, 8, 1966, p 111.
3. Grove, A.S., "Fisica e Tecnologia dei Dispositivi a Semiconduttore" [Physics and
Technology of Semiconductor Devices], F. Angeli Ed., Milano, 1978, p 90.
4. ].bid, p 249.
- 5. I3orn, M., and Wolf, E., "Principles of Optics," Pergamon Press, 1975, p 61.
6. Philipp, H.R., and Taft, E.A., PHYS REV, 120, 1960, p 37.
- 7. PicIntyre, R.J., IEEE TRANS EL DEVICES, ED-13, 1966, p 164.
COPYRIGHT: 1974 by ERI-EDIZIONI RAI RADIOTELEVISIONE ITALIANA
11267
CSO: 5500/2310
END
65k
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400080013-5