Circuit Component SLAC for AXE 10

ERICSSON
REVIEW
l"
1983
Test System LPA 103 02
Development of Line Circuits for AXE 10
Line Circuit Component SLAC for AXE 10
Line Circuit Component SLIC for AXE 10
Office Communication System DIAVOX2836
Ericsson Power Systems, a New Division within RIFA ERICOM DIRECT
Sealed Lead-acid Batteries for Small Telecommunication Plants
Computer-controlled System for Road Traffic Control
Ericsson's First Automatic Telephone Exchange 100 Years
ERICSSON REVIEW
Number 4
1983
Volume 60
Responsible publisher Gosta
Editor Gosta
Lindberg
Neovius
Editorial staff Martti
Viitaniemi
Address S-126 25 S t o c k h o l m , Sweden
S u b s c r i p t i o n one year $ 12
Published in S w e d i s h , English, French and Spanish w i t h four issues per year
Copyright Telefonaktiebolaget LM Ericsson
Contents
174
Test System LPA 103 02
181
Development of Line Circuits for AXE 10
186
Line Circuit Component SLAC for AXE 10
192
Line Circuit Component
201
Office Communication System DIAVOX2836
207
Ericsson Power Systems, a New Division within RIFA
214
ERICOM DIRECT
222
Sealed Lead-acid Batteries for Small Telecommunication Plants
226
Computer-controlled System for Road Traffic Control
233
Ericsson's First Automatic Telephone Exchange 100 Years
SLIC for AXE 10
Cover
1983 was the International Communications Year,
and in connection with this Ericsson arranged an
exhibition of the 100-year old automatic telephone
exchange, which in a period setting was connected to a manual exchange and a number of
telephone sets. The function of the automatic
exchange and other equipment was demonstrated
in a short play
Test System LPA103 02
Carl-Olof Gillgren
Ericsson has developed a test system, LPA 10302, for factory testing of printed
board assemblies. It forms part of a system family which also includes a system
for testing PABX equipment at system level. All test systems in the family are
equipped with the same type of control system, which is not only advantageous
for the users but also facilitates the development of new test systems and new
common support systems.
Test system LPA 10302 has been designed for functional testing of a wide range
of printed board assemblies with the best possible overall economy. The system
permits both advanced analog measurements and high-speed testing of complex
digital functions.
UDC621 3 049:621 317
With the increased use of microprocessors and other complex LSI/VLSI
circuits the number of functions
provided by a single printed board assembly can now be very large. At the
same time dynamic components with
high clock frequencies are becoming increasingly common.
LPA 103 02 has been designed to
provide functional testing of a wide
range of printed board assemblies in
one and the same test system with the
best possible overall economy. The system permits both advanced analog
measurements and high-speed testing
of complex digital functions. Fig. 2
shows a block diagram of the system.
Fig.1
Test system LPA 103 02
The test station and the operator's terminal have
been designed to give the operator a comfortable
working position
CARL-OLOF GILLGREN
Division for Public Telecommunications
Telefonaktiebolaget LM Ericsson
The main aim has been to develop a system that is easy and simple to use. The
main features of the system are:
- short setting-up time for different test
objects
- easy connection of the test objects
- short testing times
- efficient fault-tracing aids
- good ergonomic features
- possibilities of integration with the
production process.
Efforts have also been made to limit the
amount of work required for the design
and administration of test programs in
spite of the fact that the test objects are
becoming increasingly complex. This
requires
- good integration with different design data bases
- powerful simulation aids
- a machine-independent high-level
language for the test programs.
Control system
The control system is built up around a
computer whose operating system
works in real time, fig. 3. All communica-
Fig. 2
Block diagram of test system L P A 1 0 3 0 2
POS
SEL
DMG
PWR
DIO
PSC
Position finder
Selector
Dynamic measurement and generator
Test object power
Direct communication with the test object
Power supply and control
tion with the function blocks in the test
station takes place via the internationally standardized instrument bus,
GPIB', fig. 4. In addition to the computer
the control system hardware includes
mass storage units, an operatorterminal
and a control unit.
Basically the control system functions
as an "administrative" computer, which
handles the large amounts of data for
each type of test object. The control system contains general software for entering, translating and verifying test programs that are associated with specific
objects, and for carrying out the commands from the operator during testing
and fault tracing.
Fig. 3
The control system is common for a whole family
of test systems
Basing a whole family of test systems on
the same control system means that the
users' need of training and spares is reduced. This also facilitates the development of new test systems and new common support systems. The first test system in the LPA103 family was
LPA103 01. It is used for the final testing
of
fully
equipped
cabinets
for
PABXMD110 and has been used in the
Swedish production since the spring of
1982.
The Veto language
Veto is both a programming language
and a specification language. A program written in Veto is not tied to any
Fig. 4
The instrument bus GPIB is an internationally
standardized interface (IEC-625/IEEE-488), intended primarily for the control of measuring instruments
particular automatic test system. It contains all information about how the testing is to be carried out, for example
which tests are to be made, the test connections and permitted measurement
values in each test, and the stipulated
actions in the case of measurement values that fall outside the limits. This
makes the Veto programs universal, i.e.
they can be used in test systems having
different structures and characteristics
without any modifications being required
The Veto language has unique and very
powerful instructions for digital real
time testing, and as regards analog
functions it is similar to the standard
language ATLAS2. The Veto syntax has
been inspired by PASCAL, which makes
it easier to design well-structured programs.
Digital measurement and
generating functions
All testing of digital units is done by applying digital signal patterns to the input
of the test object, after which the resultantsignal patternson theoutputsofthe
test object are sensed and compared
with the expected values. If new signal
patterns are input at a rate that is similar
to what the test object will experience in
normal operation, and in addition the
strobing of the resultant signal patterns
takes place at times that correspond to
Fig. 5
Test system L P A 1 0 3 0 1 is used for the final
testing of fully equipped cabinets for
PABXMD110
the requirements of the environment as
regards response times, the testing is
said to be in real time.
Block DMG (Dynamic Measurement and
Generator) in LPA 103 02 is designed for
advanced real time testing, and can
transmit and receive new signal patterns
at intervals of d o w n to 100 ns. DMG can
be equipped with up to 400 two-way digital channels. Both time intervals and
signal levels can be program controlled
with high resolution and flexibility.
The DMG block has been constructed
using ECL circuits in order to meet the
requirements for speed. The drivers in
the digital channels consist of custom
hybrid circuits.
Fig. 6
The drivers in the digital channels are constructed using custom hybrid circuits from the
Swedish semiconductor manufacturer R I F A A B
Fig. 7
Basic diagram for one of the 400 digital channels
Ref
1/0
Z
X
Analog reference levels (six per channel)
Digital data
High-Impedance drivers slate
Cancellation of the comparator error flag
Digital levels
The digital channels contain drivers and
comparators. The voltage levels (high
and low level respectively) of the drivers
can be set individually for each channel
t o t h e desired voltage w i t h i n t h e range 12V to + 1 2 V . The drivers can give an
output current of up to ± 2 0 mA. It is also
possible to program a current load within the interval - 1 0 m A to +10mA.
The voltage levels of the comparators
can also be set individually within the
range - 1 2 V to + 1 2 V . The input impedance is 1 0 0 k o h m s .
Digital data
Digital data are given in the test program
with the aid of the Veto instructions PUT
and GET. The test program designer
uses the PUT c o m m a n d to specify which
data are to be fed in to the inputs of the
test object. Each of the digital channels
can give high level (1) or low level (0), or
take up a high-impedance state (Z),
fig. 7. The GET c o m m a n d is used to define the expected data from the outputs
of the test object as high level (1), low
level (0) or " d o n ' t care" (X). The TEST
instruction is used to compare the expected and the obtained data.
In static tests, the PUT, GET and TEST
c o m m a n d s are repeated for each new
signal pattern, as shown in the program
example in fig.8.
177
PUT 01 ATINP1, INP2
GET 1 AT NANDOUT
TEST
PUT10ATINP1.INP2
GET 1 AT NANDOUT
TEST
PUT 11 ATINP1, INP2
GET 0 AT NANDOUT
TEST
Fig. 8
Veto instructions for static testing of a simple
NAND gate
Time control
The instructions PUT and GET define
the signal patterns in the digital test.
With real time testing it is also necessary
to specify detailed time patterns.
In the Veto language the digital real time
testing of a test object is described in
three stages. First the block PROTOCOL
specifies the times at which data are to
be changed for each input and strobed
from each output respectively. The digital signal patterns are then defined in
the block PREPARE REAL TIME.
Each real time test sequence is initiated
by the Veto instruction RUN, which
specifies the protocol in question and
the relevant signal pattern definition, as
shown in the program example in fig. 9.
Fig. 9
An example of Veto instructions for real time
testing
PROTOCOL ALPHA
FUNCTION ATE-CLOCK
PERIOD = 2 US
TIME 200 NS TO 1200NSPUT POS-PULSE AT ENABLEPIN
END FUNCTION
CYCLE WRITECYCLE
CYCLETIME = 2 US
TIME 0 NS TO 1500 NS PUT DATA AT RS0, RS1,RW, CS
TIME 0 NS TO 1500 NS PUT DATA AT DAT
TIME 0 NS TO 2000 NS PUT DATA AT RESETPIN
TIME 1900 NS GET DATA AT PA
END CYCLE
END PROTOCOL
PREPARE REAL TIME TEST1 USING ALPHA
WRITECYCLE
PUT 0 AT RESETPIN
WRITECYCLE
PUT 1 AT RESETPIN
PUT 0001 AT RS0, RS1, RW, CS
PUTH:AAATDAT
GETH:AAATPA
END PREPARE
RUN TEST1 MAX-TIME 100 MS
DMG in LPA10302 enables the test program designer to define different time
patterns with a high degree of flexibility.
The digital real time testing is divided
into test cycles. Up to 16 different cycle
types can be defined simultaneously,
and also mixed optionally in one and the
same stream of test cycles. Each type of
cycle has its own definitions of sensing
times and a cycle length of between
100 ns and 40 ns.
Block DMG contains eight program controlled pulse generators, whose sixteen
pulse edges can be used unrestrictedly
for pulse control of drivers and comparators. The pulse edge times can be
programmed with a resolution of 1 ns.
Each digital channel has 4 kbit memories for digital data. The clocking out of
the memory content towards the test object can be synchronized with the test
station clock or the test object clock. In
the latter case the clock frequency can
lie in the interval 25kHz - 50 MHz.
The microprogrammed control unit can
carry out jumps, run in loops and carry
out subroutine calls, and it can also wait
for an external condition.
Selector functions and
instruments
The function of the selector is to connect the instruments included in the system to the desired test points on the test
object. The instruments can be actual
measuring instruments, passive load
units or active generators. A very wide
range of instruments can be connected
in via the standardized instrument bus
GPIB and controlled from the test programs.
In the basic version, for testing digital
printed board assemblies, the test station is equipped with a digital multimeter and a time meter.
A wide range of selector equipment can
be used, and both single-wire and twowire selectors can be used, fig. 10.
The single-wire selector can be equipped with four selector channels per digital test channel in DMG, or it can be
equipped with up to 128 selector channels, which are then connected via a
cross-connection interface.
The two-wire selector is used mainly for
analog measurements which are not referred to common signal earth. It can be
equipped with up to 128 selector channels, which are connected via the crossconnection interface. Both the singlewire and the two-wire selector can connect in up to four instruments simultaneously, each with the aid of an instrument selector with eight instrument
connections.
A test station intended mainly for digital
testing is normally equipped with a single-wire selector, with one selector
channel per digital test channel in DMG.
This reduces to a minimum the need for
different cross-connection interfaces.
Fault tracing aids
The test station is equipped with two test
probes for fault tracing, each of which
can be used for either digital or analog
measurements. Together with the software of the control system the test
probes constitute an advanced aid for
locating faults in both digital and analog
printed board assemblies.
Test probes
For analog tests the test probes can be
used to input signals or for making
measurements. Thus all instruments
that are connected to the instrument selector can be connected to optional
points in the test object.
Fig. 10
The selector functions of the test station provide
facilities for connecting in different types of
instruments for measurements and signal generation
SEL
CSEL-A
CSEL-B,
CSEL-C
ISEL-A,
ISEL-B
DMG
Selector block
Two-wire selector
Single-wire selectors
Instrument selectors
Dynamic measurement and generator
Digitally the test probes are used for advanced real time fault tracing on digital
circuits. Each probe contains a glitch
detector, w h i c h detects any change
from high or low level to an indeterminate value. The length and position in
time of the sensing interval can be programmed in steps of 1 ns. The test probe
detector has a 4 kbit buffer memory and
can distinguish the f o l l o w i n g node
states:
- High or low level
- Neither high nor low level
- Positive or negative edge
- Positive or negative pulse
- Positive or negative deviation to an
indeterminate value
- Ditto before or after a pulse edge
- Ditto before, during or after a pulse.
Position finder for quicker fault tracing
The t w o test probes can be supplemented by a position finder, w h i c h with
the aid of step motors positions the
probes at the given coordinates on the
test object in less than a second. The
coordinates are fetched from the pattern design database, and an otherwise
time-consuming part of the fault tracing
routine is thus made quick and accurate.
Software for automatic fault location
When the test system detects a discrepancy between the expected and the detected data the operator is informed that
the printed board assembly in question
is faulty. However, before the fault can
be repaired it must be located to a faulty
component or c o n n e c t i o n .
The software for automatic fault location makes it possible to trace a fault
automatically to a certain node on the
printed board assembly. The method is
based on information c o n c e r n i n g the
expected data in each node, and how
the nodes affect each other via driving
logic circuits.
Assume that the data obtained from one
of the c o n n e c t i o n pins on the test object
deviate from the expected data. The operator is then instructed to position the
test probe on each of the inputs to the
circuit that drives the pin in question.
When an input with the wrong data is
found, the operator is instructed to
place the probe on each of the inputs to
the circuit that drives the faulty input.
The fault is thus traced backwards from
effect to cause, and the process will lead
to a node w h i c h gives faulty data in spite
of the fact that all driving inputs have the
correct data. The faulty node has then
been located.
When the position finder is used all
movements of the test probes are entirely automatic, w h i c h saves a considerable a m o u n t of time.
Designing test programs
Simulation aids
A simulation system designated VDS
(Veto Design System), fig. 11, has been
developed as an aid in designing test
data for test system LPA103 02. VDS is
intended primarily for objects that require real time testing. The system is
installed in a powerful main frame computer and can be used as an aid both for
verifying designs and for designing test
programs.
One main feature of VDS is that delay
times are analyzed statistically and the
probabilities of different sequences of
events are calculated from the mean values and standard deviations for the delay times of the various elements in the
circuit.
1Q83
179
VDS reads the circuit diagram information for the test object from the database
for pattern design. Information regarding the components is fetched from a
model library. The functions of complex
components are described in the highlevel language Modelette.
The test program designer specifies the
desired input signals using Veto. The
simulation in VDS results in Veto instructions, which define input signals
and expected output signals from a
fault-free test object, and also the data
that are required for automatic fault location. The information can be transferred directly to an LPA10302 for further processing and verification.
One advantage of simulation is that it is
possible to measure the degree of fault
coverage and the fault resolution of the
test program.
Manual test program design
The Veto program and the fault locating
information from VDS can be used directly in LPA103 02 since the same program language is used. Hence, after the
simulation the test program designer
can concentrate on the analog part of
Fig. 11
The simulation system VDS consists of a number
of interworking modules
the test program and any supplementing of the digital part that might be necessary. The program is then compiled
and verified by testing an actual printed
board assembly of the type in question.
Connecting up the test
object
The test objects are usually printed
board assemblies with components and
connectors. They are connected up to
the various functions of the test station
via a test fixture and a cross-connection
interface All printed board assemblies
having the same type of connectors and
the same mechanical dimensions can
be tested in one and the same fixture,
fig. 12. The fixture for printed board assemblies of type ROF is used for both
R0F13 and R0F16, and it has
motorized opening and closing for easier handling.
The cross-connection interface normally has direct connections between
the fixture and the test system, but special connections can be made when
necessary. The cross-connection interface can also be used to connect in
matching boards between the object
and the test system.
The test station is equipped with a special interface unit for communicating
with test objects having standardized
signalling
interfaces,
such
as
CCITTV.24.
Power supply
The test station can be equipped with up
to six power supply units, with programmable voltage and current limiting, for
the supply of power to the test objects. A
special selector for choosing the supply
pins gives greater flexibility.
In a test station equipped with 400 digital channels the DMG magazine alone
uses almost 4kW. New technology has
to a certain extent been used to solve the
problems of voltage distribution and
heat dissipation. Fan-cooled parallel
power units feed 200mm 2 copper bars,
which are screwed direct to a rear plate
of sandwich type. The connectors are of
a special type, in which the pins that are
used for power distribution are pressed
into the aluminium rear plate.
180
Fig. 12
The test station can easily be equipped with test
fixtures for widely disparate uses
The test station is air cooled by two tangential fans with a total capacity of
2500 m3/h. The fans are mounted directly above the DMG magazine. In addition the temperature is monitored in several places in the rack, and if abnormal
values are detected an alarm is sent to
the control system, which in the case of
large deviations automatically cuts off
the current to the test station.
Maintenance and calibration
The test system contains a number of
maintenance programs, which check
that all parts of the system function correctly and that all measurement and
generator functions are correctly calibrated. Some maintenance programs are
run daily, others weekly, yet others
monthly etc. The serviceman uses a
maintenance menu to initiate, for example, weekly maintenance, after which all
weekly maintenance programs are run
automatically. The serviceman is informed about any faults that are found,
and the unit likely to contain the fault is
indicated.
Summary
The main characteristics of test system
LPA103 02 are
- sophisticated functions for real time
testing of digital printed board assemblies
- flexible measuring and connection
facilities for analog printed board assemblies
- little need for separate auxiliary
equipment for each type of test object
- short setting-up times
- efficient fault tracing aids.
References
1. IEEE Standard. Digital Interface for
Programmable Instrumentation. IEEE
Std 488-1978.
2. IEEE Standard. ATLAS Test Language.
IEEE Std 416-1981.
1983
Development of Line Circuits for AXE 10
Gunnar Bjurel, Andre Dudnik and Royne Hjortendal
The line circuit is that part of a telephone exchange system where the subscriber
lines are connected in. Since the design of the circuit to a great extent affects the
characteristics of a digital exchange system, such as the space requirement,
cost, power loss and transmission quality, extensive development work has been
undertaken in order to obtain a new generation of components for the line circuit
in AXE 10.
The authors describe the work and the functions and characteristics of the line
circuit. In the two subsequent articles in this issue of Ericsson Review the two
special components, SLIC and SLAC, which have been developed for the line
circuit are described in detail. SLIC is a bipolar LSI component which contains
the high-voltage parts of the line circuit, whereas SLAC is a digital VLSI
component in MOS technique.
UDC 621.3.049
621.395.3
Fig. 1
A functional comparison between an analog and a
digital subscriber stage
Blue
Red
LIC-A
LIC-A/D
LIC-D
SS
TS
AJC/BJC
PCD
JTC
GSS
ITC/OTC
ETC
Analog functions
Digital functions
Analog line circuit
Analog/digital line circuit
Digital line circuit
Analog subscriber stage
Time switch in the digital subscriber stage
Junctor circuits for A and B-subscriber functions
Analog/digital converter for 32 channels
Link to the group selector
Digital group selector
Incoming and outgoing analog trunk circuits
Terminal equipment for PCM line systems
The introduction of the digital subscriber stage12 in AXE 10 was the second
stage in a digitalization process, which
started with the introduction of the digital group selector, fig. 1. The possibility
of remote siting of the subscriber stage
means that the advantages of digital
transmission, namely higher transmission quality and lower cost, can also be
utilized in the primary network. Since
the digital subscriber stage consists almost exclusively of electronic equipment the possibilities offered by VLSI
technique as regard space, power and
cost reductions can be exploited fully.
However, this requires extensive component development work. This applies
particularly in the case of the line circuit
components, since they contain a large
number of complex functions, which,
because the selector stage is digital,
must be individual for each subscriber
line.
The third stage in the digitalization process is shown in fig. 1 below. Here the
digital subscriber stage has been supplemented by digital line circuits for
connecting digital subscriber lines. Development work is in progress on the
first generation of components and circuits, for connecting digital subscriber
lines to the digital subscriber stage in
AXE 10.
The digital subscriber stage
The digial subscriber stage in AXE 10 is
built up in modules of 128 lines. Sixteen
such modules together form a fully built
out stage of 2048 lines. The subscriber
stage can either be connected direct to
the group selector or it can be connected remotely via digital line systems.
Subscribers connected to a remote subscriber stage have access to the same
functions and services as the other subscribers and the operating company can
make use of the same maintenance
functions for remotely and centrally
connected subscribers.
1
GUNNAR BJUREL
ANDRE DUDNIK
ROYNE HJORTENDAL
ELLEMTEL Utvecklings AB
Each line switch module, LSM, fig.2,
contains 128 lines circuits, time switch,
DTMF receivers, connections for a line
system or link to the group selector and
test equipment which permits testing of
subscriber lines and line circuits. In addition each line module contains a regional processor. The different units in
the line module are connected together
via three bus systems for speech, control and testing. The line circuits are assembled in groups of 4-8 (depending on
the type of line circuit) on a line circuit
board and are controlled by a common
microprocessor.
Line circuit functions
Line circuit functions
Current feeding
Polarity inversion
Call detection
Detection of dial pulses
Ringing
Answer detection
Ring tripping
Line and line circuit testing
Overvoltage protection
Changeover between two-wire and four-wire
(hybrid)
Analog/digital conversion
Since the selector network in the subscriber stage is digital all analog functions must be performed in the line circuits, individually for each subscriber
line. This applies for the subscriber line
orientated high voltage functions as
well as the analog/digital conversion
and other low voltage functions in accordance with the summary of the line
circuit functions given in the table.
Two generations of line
circuits
The cost of the line circuit forms a considerable part of the total cost of the
AXE 10 system, and hence continual development work must be carried out in
order that new developments in the
fields of basic technology and components can be utilized at once. However,
in order not to have to introduce new
types of line circuit boards too often the
product development has taken place in
a number of stages, each based on a
generation of line circuit components.
Fig. 2
A line switch module, LSM, for 128 subscriber
lines
Blue
Red
LIC
LTC/LCT
KRC
ETB
JTC
TS
TSB
EMRP
EMRPB
Analog functions
Digital functions
Line circuit
Line and line circuit tester
DTMF receiver
Exchange terminal for a PCM link to the group
selector in a remote subscriber stage
Local link to the group selector when the
subscriber stage is placed in the exchange
Time switch
Time switch bus with 512 time slots
Regional processor
Regional processor bus
The first generation of line circuit components was made possible by the fact
that the complete conversion between
analog and digital representation of the
speech signal could be performed economically per subscriber with the aid of
VLSI components that were already
available on the market.
LSI technique has also been used in the
second generation for certain of the
high voltage parts of the line circuit
functions. Furthermore a special VLSI
circuit has been developed for the analog/digital conversion. This circuit includes a facility for control ling a number
of transmission parameters by means of
stored programs. This version of the line
circuit is smaller and requires less
power than the earlier version, and several other characteristics have also been
improved.
One of the prerequisites for a third generation of line circuits is that LSI technique can be used for all high voltage
functions, and the need for electromechanical relays can thus be eliminated.
First generation of line
circuits
On the first generation of line circuit
boards, fig. 3, which has been in production since 1981, the analog/digital conversion is carried out using a commercially available single-channel codec
and a filter of the switched capacitor
type. The codec is connected to the system speech bus via a special circuit in
bipolar technique. The line circuit twowire/four-wire transition takes place in a
miniaturized transformer hybrid, which
is made up of two transformers in order
to obtain optimum transmission characteristics in relation to the mechanical
size of the hybrid. The power feeding of
the subscriber line is arranged in the
conventional way through the primary
windings of the transformer hybrid via
thick film resistors. The ringing signal,
polarity inversion and line test functions
are each connected in by means of a
miniature relay.
The line equipment is protected against
overvoltages, caused by for example
lightning, by including a newly developed, highly efficient overvoltage
1983
183
Fig. 4
The first generation of line circuit boards, for four
lines per board
protection of the diode thyristor type.
Considerable effort has also been made
to separate transient currents from
other earth currents on the line circuit
board, thus greatly increasing the ability
of the line circuit to withstand overvoltages.
Cradle and dial signals and ringing answer are detected by a standard type
comparator circuit with an associated
resistance network in thick film technique.
The line circuit board, fig. 4, is equipped
with four line circuits controlled by a
common microprocessor, which also
performs a certain amount of digital signal processing. The dimensions of the
line circuit board are 222x178x17 mm.
Fig. 3
A simplified circuit diagram for the first generation of line circuit boards
MDF
LIB
LIC
POVP
SOVP
TEST
HYB
BAL
POL
DET
RING
LF
CLR
FILT
A/D
DP
I/O
Main distribution frame
Line circuit board
Line circuit components for a subscriber line
Primary overvoltage protection
Secondary overvoltage protection
Relay for connection to a test bus
Hybrid transformer
Balance network
Relay for polarity inversion
Detector function
Relay for connecting In ringing voltage
Power feeding resistance
Control of relays
PCM filter and level adjustment
PCM codec
Microprocessor
Interfaces towards the system buses
The second generation of
line circuits
The second generation of line circuits
use a number of specially developed
components and are more compact, require less power and are in the long run
cheaper than the first generation of circuits. They are also more efficient than
the first generation, and the standard
printed board design meets the requirements of most markets.
Components
The line circuit functions are carried out
by two components, SLIC (Subscriber
Line Interface Circuit) and SLAC (Subscriber Line Audio processing Circuit).
SLIC, which replaces, among other
things, the transformer hybrid of the first
generation, contains functions for
powerfeeding, receiving dial pulses and
for detecting calls and answers. The polarity inversion, which in the first generation was carried out by a relay, is now
integrated in SLIC, which also contains
drive circuits for the two remaining relays for ringing and testing.
SLAC contains the analog/digital conversion and filtering, as well as functions for stored program control of levels, the two-wire impedance and the
hybrid balance. The control signals required for SLIC are also generated in
SLAC. An analog and a digital loop can
be set up in SLAC for test purposes. The
main part of the SLAC functions are carried out using digital signal processing,
which means that the analog parts are
quite simple. The characteristics of the
circuit are therefore very stable and easy
to verify, and the possibilities of close
packing of digital functions offered by
the VLSI technique can be exploited to
the full.
Function
Generation 1
Ringing
Relay
Test connection
Relay
Power feeding
Resistances
Polarity inversion Relay
Call detection
Comparator
and film network
Ring tripping
Call detection
circuit
Two-four-wire
Transformers
conversion
Decadic impulsing,Auxiliary
e.g. towards
equipment
a PABX
Two-wire
Discrete
impedance
components
Balance
impedance
A/D conversion
Integrated
single-channel
PCM filter
Switched
capacitor
Level adjustment
Discrete
voltage
dividers
Thyristor
circuit
Overvoltage
protection
Generation 2
Relay
Relay
Chopper
LSI
LSI
LSI and
film
network
LSI
LSI
Stored
program
control
Integrated
with the
filter
Digital
Stored
program
control
Thyristor
circuit
Table 1
C o m p a r i s o n of the m e t h o d s for realizing different f u n c t i o n s in the t w o g e n e r a t i o n s of line
circuits. LSI indicates that the f u n c t i o n is integrated w i t h the p o w e r f e e d i n g
Fig. 5
A simplified circuit program for the second generation of line circuit boards
MDF
LIB
LIC
POVP
TEST
RING
SOVP
SLIC
LF
CLR
DET
SLAC
A/D
DSP
CLS
TSS
DP
I/O
Fig. 6
The second generation of line circuit boards, for
eight line circuits per printed board assembly
Main distribution frame
Line circuit board
Line circuit components for one subscriber line
Primary overvoltage protection
Relay for connection to the test bus
Relay tor connecting In ringing voltage
Secondary overvoltage protection
Line circuit component
Power feeding function
Control of relays
Detector function
Component for analog/digital conversion and for
other speech signal processing
Analog/digital conversion
Digital signal processing
Control of SLIC functions
Time slot selection
Microprocessor
Interface circuits towards the system bus
In addition to SLIC and SLAC the line
circuit board, fig. 5, contains a test relay
and a ringing connection relay per line.
It is also equipped with an overvoltage
protection of the thyristor type, to protect SLIC against any overvoltages on
the subscriber line. As in the first generation each board contains a microprocessor and a special component for
matching to the system buses.
Printed circuit board
Since SLIC and SLAC require less space
and have considerably less power loss
than the first generation of circuits, it
has been possible to place eight line circuits with the associated control and interface components, on one printed circuit board, fig. 6. A special hybrid component, CLIC (Complete Line Interface
Circuit) containing SLIC and SLAC in
chip carrier packages, and a number of
external components have been developed for this printed board assembly.
The differences between the two generations are shown in table 1.
Characteristics
Compared with the first generation, the
second generation of line circuits have a
number of advantages as regards functions and performance. With eight line
circuits on one printed board the
amount of space required for the line
circuit part of a line switch module,
LSM, is halved. This, combined with a
similar development of the control parts
of LSM, means that the space requirement has been drastically reduced.
With SLIC all known power feeding
methods, such as resistive or constant
feeding, can be arranged by programming and using simple external components. The power feeding is carried out
with the aid of a chopper, and hence the
power loss in the circuit is considerably
less than in the case of a line circuit with
conventional resistive feeding, fig.7, in
spite of the fact that SLIC imitates the
resistive power feeding with good accuracy. Another characteristic of SLIC
which helps to reduce the power consumption, for example in the case of a
short circuit on the subscriber line, is
that SLIC can be put in a mode with
extremely low feeding current towards
the line. SLIC can also handle special
signalling codes, such as decadic impulsing or single-wire calling towards
PABXs, without any peripheral equipment being required.
SLAC, with its stored program controlled amplification of thespeech level,enables the level requirements of most
markets to be satisfied with a single type
of printed board assembly. The two-wire
impedance is also stored program controlled, and all known impedances can
be achieved with good accuracy, which
in its turn means improved sidetone attenuation.
Fig. 8 shows the return loss towards a
complex impedance. The balance impedance is controllable in the same way
as the two-wire impedance, which
makes for a considerable improvement
185
Fig. 7
Comparson of the power loss as a function of the
line resistance in a conventional line circuit and
in a line circuit with simulated resistive power
feeding with a feeding resistance of 2x200 ohms
and a feeding voltage of 50 V
Conventional resistive power feeding
Simulated resistive power feeding
of the loop attenuation in the hybrid towards subscriber lines.
As in the first generation the subscriber
line and line circuits are tested with a
special test circuit, w h i c h is connected
in with the test relay. The test circuit can
be used for making measurements on
the subscriber line and for testing all
functions of the line circuit, including
the A/D conversion.
Future development
The introduction of the second generation of line circuits for AXE 10 has meant
a considerable improvement in functions and characteristics. With the introduction of a third generation, in a few
years' time, the trend of LSI technique
towards the use of higher voltages will
be exploited in order to replace, in an
economical way, the relay functions that
still remain on the line circuit board. The
further development of low voltage digital VLSI technique towards a greater degree of integration will also be exploited
in order to move other central functions
out to the line circuits.
Fig. 8
The return loss towards a complex impedance of
900 ohms in parallel with 30 nF
References:
1. Nilsson, B.-A. and Sorme, K.: AXE - A
Review. Ericsson Rev. 57(1980):4, pp.
138-148.
2. Persson, K. and Sundstrom, S.: Digital
local exchanges AXE 10. Ericsson Rev.
58(1981 ):3, pp. 102-110.
3. Norevik, B.: Remote AXE 10 Subscriber Switch in a Container. Ericsson Rev.
59 (1982):4, pp. 174-177.
4. Rydin, A. and Sundvall, J.: Line Circuit
Component SLIC for AXE 10. Ericsson
Rev. 60 (1983):4, pp. 192-200.
5. Ericsson, G. and Svensson, T.: Line
Circuit Component SLAC for AXE 10.
Ericsson Rev. 60 (1983):4, pp. 186191.
Circuit Component SLAC
for AXE 10
Gunnar Eriksson and Tommy Svensson
The development of line circuits for AXE 10 has already been described in this
issue of Ericsson Review. In this article one of the components in the line circuit
is described, namely SLAC (Subscriber Line Audio processing Circuit), whose
functions are more extensive than what is normally expected from codec and
filters. SLAC is manufactured in MOS technology using VLSI. The signal
processing in the circuit is digital and stored program controlled. The circuit has
many programmable features.
UDC 621.3.049.774.2
Fig. 1
Block diagram of the subscriber line audio processing circuit, SLAC
PFX
PFR
ADC
DAC
DF1, DF2
IF1, IF2
CFX, CFR
GCX, GCR
MFX
MFR
CMP
EXP
DOS
DIS
LSW
INS
BNS
SCL
CNTR
Analog prefilter
Analog post-filter
Analog/digital converter
Digital/analog converter
Decimation filters, in which the sampling
frequency is reduced
Interpolation filters, in which the sampling
frequency is increased
Filters for correcting the attenuation-frequency characteristics
Gain regulation units
Band pass filter 0.3-3.4 kHz
Low pass filter, 3.4 kHz
Compressor
Expander
Output stage for PCM
Input stage for PCM
Loop connection circuit
Impedance filter
Balance filter
Register for control signals to SLIC
Control unit
A prerequisite for completely digital
switching networks in a telephone exchange is that analog/digital conversion
of the telephone signals is carried out
separately for each line. The line circuit
printed board assemblies, which in addition to such traditional functions as
power feeding, ringing, sensing of the
hook condition etc. also handle analog/
digital conversion, therefore constitute
important building elements in a digital
telephone exchange. Previous generations of line circuit boards have contained single-channel codec and filters
for the analog/digital conversion. The
signal processing has then been mainly
analog.
The line circuit component SLAC (Subscriber Line Audio processing Circuit)
contains more functions than are normally found in the codec and filters. Efficient utilization of digital signal processing has reduced the number of analog functions required, and their performance data are not critical.
Digital signal processing has the following advantages:
- The circuit performance is determined by the design of the signal processing circuitry, and not by uncontrollable internal noise sources or
component variations.
- The performance can be reproduced
in one component after another.
- The performance is not affected by
temperature variations or ageing.
- It is relatively simple to incorporate
functions in the circuit which will simplify other parts of the line circuit
board.
- The above-mentioned functions can
be made programmable, which
means increased flexibility and the
possibility of improved performance
by the line circuit as a whole.
- Most of the testing of the component
can be done using digital methods,
which means a significantly lower
testing cost.
- Future cost development is considerably more favourable for circuits with
digital technology than for analog circuits.
On the line circuit board SLAC is connected to the line via SLIC (Subscriber
Line Interface Circuit). The SLIC functions can either be realized in an IC3 or
by means of discrete components.
The design of SLAC guarantees that the
transmission requirements of CCITT
Recommendation G.712 are met with
good margins.
1983
187
GUNNAR ERIKSSON
TOMMY SVENSSON
ELLEMTEL Utvecklings AB
Digital filtering
An analog filter operates on continuous signals,
usually represented by voltages and currents.
The filtering effect is obtained by means of
phase-shifting electrical components.
A digital filter, on the other hand, operates on
sequences of numbers, samples, which arrive
at a constant frequency, the sample frequency fs
The filtering effect is achieved by means of delays and arithmetic operations.
Fig. A shows one type of a simple digital filter.
The signal passes a chain of delay elements,
each giving a delay T = 1/fs The output signals
from each delay element is multiplied by a coefficient (a 0 -a 2 ) and the results are added. Each
output signal sample is thus affected by the
latest input signal sample and also by some
earlier input signal samples.
Like an analog filter, a digital filter can be
characterized by its impulse response, i.e. the
output signal that is obtained when the input
signal consists of a single sample with an amplitude equal to unity. In the filter shown in fig. A
the impulse response will be as shown in fig. B.
This type of digital filter is called an FIR filter
(Finite Impulse Response) since the duration of
the impulse response is finite (= the number of
delays).
A slightly more complicated filter is shown in
fig.C. In this example the output signal is not
only influenced by previous input signal samples but also by previous output signals. The
impulse response for such a filter is, at least
theoretically, infinitely long and the filter type is
called IIR (Infinite Impulse Response). With an
MR filter the same filter effect can be obtained as
with an FIR filter but with fewer delays.
-^-Time
Since the IIR filter contains feedback it can become unstable if the coefficients b, and b2 are
given unsuitable values. An FIR filter, however,
is always stable.
The reference articles'' 5 give a more detailed
description of digital filters.
Design
Functionally SLAC can be divided into a
number of blocks, fig. 1 The send path
comprises the blocks between the analog input and the PCM output, and the
receive path comprises the blocks between the PCM input and the analog
output. There are also blocks connected
between the t w o paths. A couple of
blocks with control f u n c t i o n s are shown
at the bottom of the diagram.
One of the main features of SLAC is the
s i m p l i c i t y o f i t s a n a l o g part. The prefilter
PFX required for the sampling process
has therefore been designed as a singlepole RC filter. In order to ensure that the
variation of the RC constant shall not
affect the transmission properties of the
circuit in the speech band to any great
extent, the filter cut off frequency is set
to 115kHz, i.e. far outside the speech
band. The analog/digital conversion
takes place in the ADC at a sampling
frequency of 512kHz. At this frequency
the attenuation of PFX is sufficient to
avoid aliasing distortion.
After ADC all signal processing in the
send path is digital. However, the sampling rate of 512 kHz is far too high for
the desired filter f u n c t i o n . The sampling
rate is therefore reduced, first to 32 kHz
by DF1 and then to 16 kHz by DF2. These
digital filters, w h i c h are of the FIR type
(see the fact panel to the left), have been
designed so that the sampling rate reduction does not cause any d i s t u r b i n g
aliasing of out-of-band signals into the
speech band.
Minor corrections of the transmission
characteristic can be made by CFX. GCX
adjusts the level of the send path output
signal to a suitable value.
The demand for high signal attenuation
for frequencies outside the speech band
is met by the band pass filter MFX. This
filter is of the IIR type.
Up to this point all digital signal processing in the send path uses linear code.
CMP therefore compresses the digital
signal samples to PCM in a c c o r d a n c e
with the A-law or |.i-law before they leave
SLAC via DOS.
The receive path contains blocks similar
to those in the send path. However, the
188
Fig. 2
Unit for digital filter calculations
signal processing takes place in the opposite order After expanding to linear
code the signal is filtered in MFR. The
level correction is carried out in GCR,
and CFR handles minor corrections of
the transmission characteristics. Up to
this point the signal processing is carried out at a sampling rate of 16kHz.
However, the sampling rate must be increased in order to obtain low complexity in the analog output stage. This is
done in the interpolating filters IF1 and
IF2. The digital/analog converter can
therefore produce analog samples at a
rate of 256 kHz. With this high rate the
output filter PFR becomes extremely
simple.
LSW is used to set up loop connections
for testing. It gives an analog and a digital loop, which are independent of each
other. The analog and the digital parts of
SLAC can therefore be tested separately.
INS is used to adjust the input impedance of SLIC, and BNS provides balancing for the two-wire/four-wire conversion.
Programmable functions
The characteristics of a digital filter are
determined by its coefficients. If these
are stored in a RAM, the filter characteristics can be changed by writing in
new coefficient values. This feature of
digital filters has been exploited in SLAC
so that several filter functions are programmable.
The parameters of the programmable filters are controlled by CNTR, which com-
Fig. 3
Block diagram of an interpolating analog/digital
converter
municates with a microprocessor that is
common to all lines on the line circuit
board. Thus the parameters of the programmable filters, and also certain
other functions, can be controlled separately for each SLAC via this processor.
The following filters can be programmed:
INS
A four-coefficient FIR filter,
with which the input impedance for SLIC is set
BNS
An eight-coefficient FIR filter, which is used as a balance network for the twowire/four-wire conversion
CFX, CFR Four-coefficient filters, with
which the transmission
characteristics can be corrected individually for the
send and receive paths.
These filters can, for example, be used to correct undesirable attenuation-frequency characteristics in
components outside SLAC
GCX, GCR The gain in the send and receive paths can be set individually over a dynamic
range of 12dB in steps of
less than 0.1 dB.
The other programmable functions include:
- Switching between active state and
idle state with reduced power consumption.
- Analog and digital test loop.
- A-law or u.-law.
- Choice of PCM clock rate: 2.048 MHz
or 4.096 MHz.
- Choice of one of 32 (2.048 MHz) or
one of 64 (4.096 MHz) time slots, independently for the send and receive directions.
- Choice of one of two PCM ports for
each of the send and receive directions.
- SLIC control: Functions associated
with SLIC require a technology that
can withstand high voltages. Such
technology is usually not suitable for
logic functions. Five SLAC outputs
have therefore been allocated for the
control of SLIC functions. Control information from the microprocessor is
decoded in CNTR and presented to
SCL as control information for SLIC,
and the digital functions in SLIC can
therefore be kept at a minimum.
Fig. 5
The left-hand part of the figure shows the whole
output signal spectrum of the analog/digital converter. The right-hand side shows the low-frequency part of the spectrum in greater detail. The
input signal has an amplitude of -2dBmO and a
frequency of 1.5 kHz
Digital filters
A digital filter requires delay elements
and units that carry out additions and
multiplications. Delays in SLAC are obtained by storing signal data in a RAM
and reading them out at suitable times.
Multiplicationsare carried out by means
of additions and shift operations. The
filter coefficients are designated so that
the number of additions is kept low.
The additions are carried out in an arithmetic logic unit, ALU. Since SLAC has a
clock frequency of 2048 kHz an ALU can
carry out 256 operations per 8 kHz sample. These operations are utilized by the
various filter blocks in fig.1 using time
multiplexing. However, two ALUs are required in order to be able to handle all
filter functions. Of the corresponding
two filter units one is used for the send
direction and BNS, and the other for the
receive direction and INS.
Fig. 2 shows the principle for the logic
implementation of a filter unit. A microprogram stored in a ROM controls the
process. The filter coefficients are obtained e i t h e r f r o m RAM or ROM depending on whether they belong to a programmable filter or not. ALU exchanges
data with a RAM, which is addressed by
the microprogram. This program also
determines which type of operation ALU
is to carry out.
The A/D converter in SLAC has been developed from an idea by J.C. Candy and
B. Wooley of Bell Laboratories 1 , which
they have called an interpolating A/D
converter.
In the interpolating A/D converter, fig. 3,
the analog input signal X(t) is compared
with a coarsely quantized signal Q A (t),
w h i c h is generated by a D/A converter.
The digital input signal to the D/A converter, Q D (nT), also constitutes the digital output signal from the A/D converter.
The A/D converter quantizing levels
have a logarithmic characteristic, w h i c h
means that the relative quantizing error
is practically independent of the signal
level. Once in each sample interval the
signal QA(t) is changed to an adjacent
quantizing level, f i g . 4 .
The error signal e(t) = X(t) - QA(t) is
integrated. Q D and its analog counterpart Q A are increased or reduced one
step in each sampling period depending
on the sign of the integrated error. This
feedback means that the output signal
will follow the input signal in such a w a y
that the low-frequency part of the output
signal accurately reproduces the lowfrequency part of the input signal.
In order to be able to exploit the advantages of digital signal processing, the A/
D converter must meet more stringent
requirements than an ordinary codec filter circuit. It must w o r k with a high sampling rate so that the analog prefilter
PFX can be made simple, and it must
have a large dynamic range since the
gain control is performed by the digital
part of the circuit, GCX.
The spectrum of the quantizing error
will therefore contain very low energy
levels at low frequencies, but higher levels at high frequencies, fig. 5. This highfrequency noise will be suppressed by
the digital filters that follow the A/D c o n verter. D e s p i t e t h e c o a r s e q u a n t i z i n g the
signal resolution in the speech band is
adequate, due to the high sampling rate
and the integration of the error, w h i c h
give an interpolation between the q u a n tizing levels.
However, the design is simplified by the
fact that it is only in the speech band that
the A/D converter has to reproduce signals with a high degree of accuracy. All
signals outside the speech band are
A digital adaptation m e t h o d has been
developed in order to meet the very
stringent demands as regards dynamics. The adaptation means that the size
of the step varies w i t h the signal level 2 .
A/D converter, ADC
Fig. 4
Input and output signal from the analog/digital
converter
suppressed by the digital filters that follow the A/D converter.
Fig. 6
The method for determining the input impedance
to SUC with the aid of the INS filter
INS filter
The digital filter INS, w h i c h couples signals from the analog input back to the
analog output, can be used to change
the input impedance of the line circuit
towards the two-wire side, fig. 6. All line
circuit boards can therefore be manufactured with the same terminating resistance RA, and the INS filter can then
be used to change the input impedance
to the value reguired for a certain application. The generated impedance
can be complex. This m e t h o d can be
used regardless of whether the c o u p l i n g
towards the two-wire side is an electronic SLIC 3 or a transformer.
Fig. 10
Equivalent diagram for the balance impedance
Fig. 8
The desired impedance and the actual value
generated with the aid of the INS filter, and the
return loss as a function of the frequency
Fig. 9
Echo suppression with the aid of the BNS filter
The best result is obtained when the delay in the feedback loop over INS is as
short as possible. The sampling rate in
INS has therefore been set to 32 kHz.
The transfer f u n c t i o n for the INS filter,
w h i c h is of the FIR type, is
H
The coefficient values Z 0 - Z 3 are input to
the circuit in serial form via the block
CNTR.
The obtained input impedance Zm can
be considered as a parallel connection
of the original impedance RA with four
part impedances, w h i c h are dependent
on the coefficient values Z 0 - Z 3 , fig.7
These part impedances are complicated
and cannot be represented by simple
RLC c o m p o n e n t s .
Fig. 8 shows an example where RA is 600
ohms and the nominal two-wire impedance is 900 o h m s in parallel with 30nF.
Receiver
coefficients,
RAM
Receiver
control
logic
Receiver,
RAM
Receiver,
ALU
Transmitter,
RAM
Transmitter,
ALU
DAC
Fig. 12
The SLAC chip, 5.7x7.0 mm
Transmitter
coefficients,
RAM
CNTR, control logic
Transmitter
control logic
ADC
BNS filter
When a signal is transmitted from the
PCM side towards the subscriber a part
of the signal will return via the send path
as an undesirable echo. If the echo path
can be imitated in the BNS filter, but with
the opposite polarity, the echo will be
suppressed. The BNS filter is of the FIR
type with eight coefficients and operates at a sampling rate of 16 kHz.
Like the INS filter the BNS can be regarded as a balance impedance consisting of Z in and eight part impedances,
each of w h i c h is proportional to one
BNS coefficient, fig. 10.
Fig. 11
The desired and actual balance impedance and
the balance return loss
Fig. 11 shows an example of how the
balance impedance of 536 ohms//100 nf
+ 330ohms can be achieved.
SLAC in LSI
A version of SLAC in LSI technology has
been developed by Advanced Micro Devices Inc., using a 4u. NMOS process.
Fig. 12 shows a photograph of the silicon chip. As can be seen from the figure, the analog parts of the circuit c o n stitute only a small part of the total surface area, less than 1 0 % . Since a digital
circuit can fairly easily be modified for
new processes using thinner
line
widths, the prospects for future reduction of the chip area are g o o d .
References
1. Candy, J.C., Ninke, W.H. and Wooley,
B.A.: A per Channel AID Converter having 15 segment n-255 Companding.
IEEE Trans. Commun., Vol. Com-24,
pp. 33-42, January 1976.
2. Eriksson, G.:An InterpolateA/D
Converter with Adaptive Quantizing Levels. IEEE 1980 National Telecommunications Conference, pp. 56.6.156.6.6.
3. Rydin, A. and Sundvall, J.: Line Circuit
Component SLIC for AXE 10. Ericsson
Rev. 60 (1983):4, pp. 192-200.
4. Shapiro, L.: The Design of Digital Filters. Electronic Engineering, July
1978, pp. 51 - 5 6 , Aug. 1978, pp. 35-39
and Sept. 1978, pp. 45-52.
5. Oppenheim, A.V. and Shafer, R.W.:
Digital Signal Processing. PrenticeHall Inc., 1975.
ine Circuit Component SLIC for AXE 10
Arne Rydin and Jackie Sundvall
The development
of the line circuit for AXE 10 has already been described in this
issue of Ericsson Review. In this article one of the components in the line circuit,
SLIC (Subscriber Line Interface Circuit), is described. It handles the analog line
circuit functions, such as power feeding and signal separation, which have
traditionally
been carried out by resistors and transformers.
The component is an
analog VLSI circuit in bipolar high-voltage, junction isolated technique. In
addition to the analog parts SLIC contains a digital interface for
communication
with the other component in the line circuit, SLAC.
monolithic circuits for high voltages.
For three years RIFA has also been manufacturing a line circuit for the Ericsson
PABX MD110. Exploiting the experience thus gained it proved to be possible to integrate most of the high-voltage
line circuit f u n c t i o n s in a circuit that
meets the requirements for public exchanges. The developed circuit, SLIC
(Subscriber Line Interface Circuit), constitutes the interface between the subscriber line and the digital line circuit
functions, SLAC 1 .
Functional description
UDC 621 3.049.774 3
Fig. 1
Block diagram of SLIC
REL
DET
AMP
DC. DC
TRDC
LO
SEP
Ringing and test relay
Detector
Output amplifier
DC'DC converter
Transversal DC regulation
Longitudinal regulation
Line sensing amplifier, separation and conversion from balanced to unbalanced a.c. and
d.c. voltages
Relay driver
Decoder for state and detector choices etc.
In a modern telephone exchange the
line circuit is responsible for a large part
of the total cost. Great efforts have
therefore been made to develop integrated circuits for this function and
hence reduce the cost. The number of
circuits required w o u l d be large, and
most semiconductor companies have
shown interest. The greatest success
has been obtained in the development
of the low-voltage codec and filter circuits, whereas the interface between the
analog subscriber line and these circuits has proved more difficult to integrate. The main difficulty has been how
to handle the voltages and currents that
occur on the line while meeting the
stringent accuracy requirements.
RIFA has for several years been w o r k i n g
on processes for manufacturing bipolar
The parts of the overall line circuit function that were considered suitable for
integration in a monolithic circuit were:
- current feeding
- transmission
- signalling, detection of line states and
relay driving.
Fig. 1 shows a block diagram of SLIC.
Current feeding
Current feeding in SLIC is carried out
with the aid of an electronic regulating
system, w h i c h senses the line voltage
and feeds out the corresponding line
current, as described in the fact panel
on current feeding on page 194. In this
way the f u n c t i o n of a conventional, resistively fed system can be imitated, but
constant current feeding can also be obtained. The desired feeding resistance
and the current can easily be arranged
by suitable choice of external resistances. However, unlike a conventional
system the current feeding is indepen-
1963
«W/J
ARNE RYDIN
JACKIE SUNDVALL
RIFA AB
dent of the actual voltage of the exchange battery, and the line current is
therefore not affected by battery voltage
variations. Nevertheless, with a long line
and low battery voltage the line voltage
could be the same as the actual battery
voltage, leaving nothing for the speech
signal. The circuit is therefore equipped
with a voltage guard, which reduces the
line current, and hence the line voltage,
to such a level that there is always voltage available for the speech signal.
In a conventional current feeding system, where the line is fed from the exchange battery via some form of physical resistors, a part of the voltage drop
will be across these resistors, with a resultant loss of power, fig. 2. In an elec-
Fig. 2
C u r r e n t f e e d i n g of a c o n v e n t i o n a l s y s t e m over
2 x 2 5 0 o h m s w i t h a t e l e p h o n e set resistance of
300 o h m s
Voltages across feeding resistors
Fig. 3
Current f e e d i n g u s i n g SLIC, w i t h a t e l e p h o n e set
r e s i s t a n c e of 300 o h m s
Voltages over the output amplifier
Voltage over the chopper regulator
193
tronic line circuit the corresponding
power loss would normally occur in the
output amplifier of the circuit. This amplifier would then have to be dimensioned for the 3 - 4 W that can be obtained if the line is short. Hitherto this
problem has usually been avoided by
using external power transistors, or by
assimilating a part of the power in external resistors. One disadvantage of this
method is the difficulty of obtaining sufficiently good performance without
making unreasonably high demands on
the accuracy of the components used.
The amount of heat produced also
meansthatthe high packing density that
would otherwise be possible with miniaturized components could not be
achieved.
Fig. B
Conventional current feeding
Current feeding
The current feeding of the line circuit is illustrated by the functional diagram in fig. A, where
the signal paths associated with current feeding are shown in red.
In a conventional system the line current is,
using the designations of fig. B,
Fig. A shows that the line voltage gives rise to a
proportional current Uab/2R, at the current output of amplifier A. A fixed current lr0, is subtracted from the proportional current, and
these two currents over the resistance give rise
to a voltage U = (|U,.b|:2R, - l,e,)R;, The amplification of the buffer amplifier can be switched
between +1 and - 1 with the aid of an external
signal, and its output voltage gives rise to a
current in to the current amplifier. After amplification with the current amplification factor K,
the resultant line current is
Consequently the terminal amplifiers in
SLIC are not fed direct from the exchange battery, but via a chopper regulator in the circuit, which with a high
degree of efficiency converts the battery
voltage to exactly the voltage that exists
across the line plus the voltage required
by the amplifiers. The circuit has unbalanced current feeding so as to avoid the
need for double choppers, i.e. thea-wire
is kept at a constant d.c. voltage whereas the b-wire is dependent on the length
of the line, fig. 3. However, from the
point of view of transmission the circuit
is balanced.
Transmission
SLIC transmits analog speech signals
between the subscriber line and SLAC.
On the line side the signal is balanced
and superposed on the d.c. feeding for
the line. Towards SLAC the signal is unbalanced, referred to earth and free
from d.c. voltages.
There are two signal paths between
SLAC and SLIC, one for the receive and
one for the send direction. The desired
impedance towards the line can be sim-
ulated by means of feedback coupling
on the SLAC side of the signal paths via
an impedance network. This network, in
its turn, can be modified in SLAC. Thus
the two-wire output impedance can be
controlled by means of software.
The requirements for suppression of unbalance signals on the subscriber line
side are extremely stringent, both as regards sending and receiving. Active balancing of the line midpoint is therefore
carried out using a special regulating
loop. The loop balances the signal transmitted from SLIC and reducesthe longitudinal impedance to such a value that
induced longitudinal spurious currents
can only give rise to low voltages, which
are further lowered by the common
mode suppression in the sensing amplifier.
Hybridizing, i.e. separating the two-way
speech signal over the subscriber line
into two speech directions, takes place
in SLAC. The hybrid balance is set by
means of software. The fact panel on
transmission describes the basic transmission principle for SLIC.
The feeding resistance can thus be varied by
means of resistor R3, which is mounted externally.
The longitudinal balance is provided by a separate feedback loop, green in fig. A. The loop
senses the line midpoint and compares it with a
longitudinal reference. The longitudinal feedback also gives the circuit low impedance with
respect to longitudinal interference signals.
This is described in more detail in the panel
"Transmission' .
Fig. A
Functional diagram of the current feeding
Longitudinal
reference
1983
195
Fig. D
The transversal Thevenin equivalent of the transmission circuit
Transmission
Fig.C shows the basic transmission principle
for SLIC. The voltage across the line terminals
U0 and U„ is assumed to have a transversal component U, and a longitudinal component U,
such that
Ua = U, + U,/2
U„ = U, - U,/2
A circuit analysis gives the transversal Thevenin
equivalent as shown in fig. D and the longitudinal impedance
Fig. D can now be supplemented with the longitudinal impedance, which gives the overall
Thevenin equivalent shown in fig. E. Choosing,
for example,
K, = 1 000
Z, = 600kohms
Z2 = 300kohms
G, = 0,05ohm-'
gives the equivalent diagram the component
dimensions shown in fig, F.
Signalling, detection and
relay driving
The various line states are monitored in
SLIC by detectors for calling, ring tripping and earth button closure.
The loop detector, which consists of a
current sensor and a comparator, detects high or low-resistance loop, i.e.
off-hook or on-hook, and dialling. The
changeover level of the detector can be
adjusted by an external resistor.
Differential
amplifier
Fig. C
Functional diagram of the transmission circuit in
SLIC
Off-hook during ringing is sensed by the
detector for ring tripping. This detector
consists of a comparator and an external resistor network, and detects the
change in the d.c. level at B-answer.
The third detector monitors ground key
closure (R-signal). It senses the longitudinal current that occurs when one
of the wires in the telephone set is
closed to earth. This detector also
makes it possible to use SLIC in PABX
applications, which often utilize this
form of signalling.
The three detectors can be connected to
a common data output. SLIC is controlled by a six-bit data bus. and two data
bits are used to select the appropriate
detector and activate the data output.
One bit determines whether the loop or
earth button detector is to be connected
to the data output at that moment. The
detector for ring tripping is always given
priority during ringing. The other data
bit activates the data output.
The otherfour data bits are used to control SLIC with respect to different line
states, for example for signalling towards the line.
In the ringing state a relay driver in SLIC
is activated and operates the ringing relay.
196
SLIC also contains a number of additional signalling functions.
Fig. 4
A n NPN t r a n s i s t o r in SLIC
0
1
2
3
4
5
6
7
8
9
Substrate
Subcollector
First isolation
Epitaxial layer
Low-resistance plug
Second isolation
Low doped base
Highly doped base
Emitter
Guard
Fig. 5
D e s i g n i n g the l a y o u t ot SLIC u s i n g a c o m p u t e r i z e d d r a w i n g s y s t e m . The different m a s k
layers of the p h o t o l i t h o g r a p h i c p r o c e s s are represented by different c o l o u r s on the d i s p l a y unit
screen
Polarity reversal is a type of signalling
which is used mainly towards PABXs
and coin box sets, but in certain markets
it is also used towards ordinary subscriber lines. The polarity reversal function is integrated in the SLIC, and hence
there is no need for any external relay for
this function. All transmission characteristics are the same in the normal state
and the reversed state. The polarity reversal is controlled from SLAC. The way
in which the current reversion has been
implemented also makes it possible to
handle induced longitudinal noise currents on the line which exceed the line
current, without the transmission characteristics being affected.
Single-wire interruption is used for
signalling towards older PABXs. This
means that SLIC feeds only over the bwire and the a-wire circuit is open (im-
pedance >50kohms). The b-wire is
grounded when a call is made from the
PABX. This is detected by the loop detector in SLIC and a changeover is made
to the normal call state.
Two-wire interruption means a break on
both the a and b-wires, i.e. no current
feeding of the line. This state can be
used for signalling towards certain
PABXs, and also for line blocking and
decadic dialling or for blocking SLIC, for
example during a line fault.
When waiting for a call from the subscriber SLIC is put in the idle state. In
this state the detectors in the circuit are
activated but the current feeding requirements are reduced. The internal
supply currents in SLIC can therefore be
reduced and hence also the power consumption.
SLIC also contains a circuit for driving a
test relay, which is controlled by one of
the data bits from SLAC.
Manufacturing process
The demands made on the manufacturing process in order to obtain the component characteristics required for
SLIC are severe and partly contradictory:
- The system requirements for high
current, low noise and an accurate,
internally generated reference voltage mean that bipolar technique is
the only possible method.
- On the subscriber line side the normal
working voltage range goes up to approximately 50V, and the residual
voltage after the secondary overvoltage protectors is momentarily approximately 120V. The circuit therefore requires a very high (for monolithic circuits) ability to withstand
high voltages. This requirement is
met in the manufacture by using a low
doped and relatively thick epitaxial
layer. The thick layer means that the
isolation diffusion round each component would be space-demanding if
it was carried out from above in the
conventional way, since the diffusion
is radial. The isolation is instead built
up in two stages from both sides of
the epitaxial layer, fig.4.
- High voltage means a risk of channels
forming between components, i.e.
that parasitic MOS transistors occur
1QR3
197
in the network. Such channels are
prevented by means of guard rings of
highly doped material round the component peripheries.
- The normal line current plus any superposed noise current can give a
working current of up to 100 mA, and
during overvoltages an instantaneous current of one ampere on the line
outputs. The transistors concerned
must therefore be able to withstand
high currents and have a low saturation voltage. The high resistivity epitaxial layer forms the collectors of the
NPN transistors. However, the real
collector, the subcollector, is situated
under the layer. The epitaxial layer
must therefore be shunted by a low
resistivity material, a plug.
- Despite the high working voltage and
the large number of amplifiers in SLIC
its own power consumption must be
low. This means that the transistors in
the amplifiers must be able to work
with very low currents (microampere)
but still have good current gain properties. Ordinary high-voltage circuits
do not have this ability. For this reason the base area of the NPN transistors in SLIC is manufactured in two
stages using a special method which
ensures very good gain at low currents.
- Both low and high resistivity resistors
are required, all having good performance and high relative accuracy.
The low resistivity resistors are manufactured in the ordinary base diffusion stage, whereas the others require an extra process stage with ion
implanting of low doped channels.
Fig. 6
Measurements on the wafer.
Each circuit is tested before the wafer is divided
into chips. The picture shows how microprobes
are used to make contact with an individual
circuit. The measurements are carried out using
an automatic measuring system
- The circuit must have very low noise.
This means extreme purity and even
quality in the manufacturing process.
- The circuit must function without any
connection pins being used for connecting external compensation capacitors, among other things for encapsulation reasons. This meant that
integrated capacitors requiring very
little space had to be produced.
Fig.4 shows a cross-section of a NPN
transistor. Ten masking stages are used
in the bipolarSLIC process. Heavily conducting n+ subcollector areas are diffused on to the p-silicon substrate. In
addition p+ areas are deposited, which
form the first isolation. A relatively thick
high resistivity (low doped) epitaxial Nlayer is then built up, followed by diffusions and ion implantations made from
the top in the usual manner.
Hence the characteristic features of the
SLIC processes can be summarized as
follows:
- The isolation is diffused from both
sides of the epitaxial layer.
- Contact is made with the subcollector
by means of a plug.
- The base area is made in two stages.
- The components are surrounded by
guards.
Design
The development of SLIC not only necessitated a new manufacturing process but has also made great demands
on the design method itself.
The first stage was to develop a function
principle which made it possible to obtain the desired performance without
making unreasonable demands on the
components used. The advantages of
the monolithic technique, such as the
good relative accuracy (matching) of the
components, had to be exploited without making excessive demands as regards the absolute accuracy. Computer
simulation was used extensively to evaluate different principles and their requirements as regards component accuracy before the actual circuit design
work started. Particular attention was
paid to the problem of achieving good
suppression of longitudinal disturbances in spite of the fact that the d.c.
feeding of the circuit was unbalanced.
198
It became clear already at an early stage
that it had to be possible to divide up this
extremely complex circuit into a number
of definable blocks, w h i c h could be designed independently. Since the layout
work was also to be based on this separation into blocks it was essential that
the number of connections between the
blocks was kept to a m i n i m u m . A plan
was prepared for the positioning of the
various blocks on the silicon chip. Consideration was then paid to the electrical
and thermal properties of the circuit.
Sensitive functions were placed away
from the chopper regulator. Low-voltage amplifiers were placed on isotherms
in order to avoid thermal feedback from
heat-dissipating components. The separation into blocks is s h o w n in the photograph of the completed silicon chip,
fig.7.
The chopper regulator, w h i c h works at
256kHz, constitutes a potential source
of interference. Balanced c o u p l i n g has
therefore been used to the greatest pos-
Fig. 7
This photograph of the silicon chip for SLIC
shows the separation into functional blocks
sible extent in order to avoid disturbances. This fact, together with a very
carefully designed layout, has made it
possible to keep undesirable transmitted signals at a very low level.
Low power c o n s u m p t i o n in the circuit
was an extremely desirable feature with
regard to packing density and heat dissipation. Great effort was therefore devoted to reducing the power. Most functional units and amplifiers in SLIC work
with currents in the order of microamperes, and wherever possible they are
fed from the + 5 V and - 5 V feeding voltages instead of the exchange battery
voltage.
The main part of the signal processing
therefore takes place at low-voltage
level, and signals for sensing line states
and for c o n t r o l l i n g the output amplifiers
must therefore be transmitted between
different d.c. potentials, which vary according to the line length. This problem
has been solved in a comparatively sim-
199
pie way without having to use resistors
with high resistance values, by letting
the signals consist of currents instead of
voltages. Resistors would have been
voltage-dependent and would also have
required a large area on the chip.
The demands for very good balance and
high accuracy meant that the design
and layout of the circuit had to be made
with careful attention paid to matching
and uniformity of critical components,
in spite of the fact that monilithic basic
components with good properties were
available.
Another example of how the stringent
demands were satisfied is the way in
which exact resistance ratios were obtained. The resistances were built up of
uniform, similarly orientated parts in
order to achieve maximum precision in
a manner that is not dependent on the
process. The desired resistance ratios
were then obtained by means of series
or parallel connection. Particular attention was paid to the voltage dependency
Fig. 8
SLIC in a CERDIP package
of the integrated semiconductor resistors, which were placed on islands
whose potential would follow the resistor potential. Since most component
characteristics are extremely temperature dependent it has been necessary to
position and orientate matched components with respect to both static and dynamic temperature distribution over the
silicon chip.
As the individual blocks were completed
they were built up as breadboards, i.e. in
discrete form, but with components
from the monolithic process. These
blocks were then connected together to
form a complete SLIC for verification of
the design.
Performance
The quality of the subscriber line networks and the functional demands
made on the line circuit are different in
different countries. The line circuit must
therefore be sufficiently flexible so that
it can meet the requirements of different
markets. SLIC has been designed accordingly, and adaptations are made
partly by means of external components
and partly by using variants of the wiring
pattern, i.e. one of the last masks in the
manufacturing process.
One mask option gives resistive feeding
and another gives constant current
feeding. The actual values, i.e. 2x250
ohms or 2x400ohms and 20mA or
32 mA, are set by means of external resistors for each mask.
The circuit also contains a thermal protection which cuts off the current feeding if a short circuit and too high line
current are detected.
The threshold level for the loop detector
is chosen by means of an external resistor.
Other factors that vary according to the
market are line impedance, balance impedance and level regulation. These parameters are programmable partly by
means of external resistors and partly
via SLAC.
Stringent demands are also made on the
transmission characteristics of a SLIC.
The analog signal is terminated in the
200
Fig. 9
SLIC encapsulated for hybrid mounting as a part
of CLIC (Complete Line Interface Circuit)
line circuit, and in an otherwise wholly
digital network it is only in and outside
this interface that noise can arise and be
added. The demands made on SLIC and
SLAC are derived from the system requirements for a digital local exchange.
These requirements are to a great extent
determined by international recommendations, for example by CCITT. The
requirements apply for a circuit between
two subscribers, through an exchange,
and therefore refer to two SLIC and two
SLAC.
Some of the transmission characteristics of SLIC, which are necessary if the
system is to meet the recommendations
referred to above, are described below.
The level accuracy must be better than
±0.1 dB in order to meet the system requirement for stability. The frequency
dependency or attenuation distortion
and the level dependency must lie within
±0.1 dB.
The sensitivity to longitudinal signals,
i.e. the balance characteristics of the cir-
cuit, must be better than 60dB. This requirement is based on the fact that in
unbalanced conditions the longitudinal
interference can generate transversal
signals. Since such signals can give rise
to transmission currents in the speech
band or to radio interference the suppression requirements are extremely
stringent.
One aim when introducing digital exchanges is, of course, that the noise
characteristics are not impaired. This
means a maximum noise level of
- 7 5 d B m , psophometrically weighted
The power consumption is another important parameter. The chopper regulator limits the maximum power dissipation in the circuit to about 1.5W. In the
idle state the circuit develops a maximum of 100mW.
The circuit can be packaged in a 28-pin
CERDIP, fig. 8, or a so-called chip carrier
for mounting on a thick film substrate
together with SLAC. in a complete line
interface circuit, CLIC. fig.9.
References
1. Eriksson, G. and Svensson, T.: Line
Circuit Component
SLAC for AXE 10.
Ericsson Rev. 60 (1983):4, pp. 1 8 6 191.
ana
Office Communication
System DIAVOX 2836
Lars Nordstrom
DIAVOX2836 is an office communication system for up to 28 trunk lines and 36
extensions. The large number of trunk lines makes the system very suitable for
companies with a considerable amount of external traffic. The system can be
programmed for alternative operational modes.
The author describes the types of telephone sets that can be connected to the
system, the functions provided by these sets and the function, electrical design
and mechanical construction of the system.
LARS NORDSTROM
Ericsson Information Systems AB
phone attendance service for groups
and departments connected to large
PABXs.
The special sets in the system are individually connected to a central unit, to
which all trunk lines are also connected.
UDC 621.395.2
DIAVOX2836 is an electronic, stored
program controlled office communication system, fig. 1, which has been developed as a larger variant of DIAVOX 824. The latter system has previously been described in Ericsson Review1. DIAVOX 2836 can be programmed
foralternative operating modes, to serve
as
- an ordinary PABX with an operator
- a key-system without operator
- an executive-secretary system
- a multi-line system.
It is also possible to divide the system
into groups with different operational
modes. In the key-system and multi-line
system modes the incoming calls can
either be taken to all sets or distributed
cyclically to the first free telephone.
The system is convenient and easy to
operate, and it is particularly suitable for
small offices, banks, travel agencies and
airlines, and also for providing tele-
Fig. 1
Office communication system DIAVOX 2836 arranged for sveral simultaneous operational
modes
Fig. 2
The central processing unit in DIAVOX 2836
Telephone sets
The special telephone sets are connected to the central unit via six wires.
Two of the wires are used for speech
transmission, two for signalling in one
direction and two for signalling in the
other direction.
The sets are equipped with several function buttons, and a button and lamp
(LED) for each line for connection and
indication purposes, fig.3.
There are two main types of telephones,
ALPHA and BETA
- The ALPHA sets have a button and
lamp for each trunk line and indicate
the states of all lines, and they allow
the user free choice of any such line.
An extension equipped with an
ALPHA set can therefore function as
an answering position for any of the
trunk lines.
202
Fig. 3
Block diagram of a telephone set
- The BETA sets are equipped with buttons and lamps for two call facilities.
One button is used to answer calls
from a call queue or transferred calls,
depending on the operational mode
chosen for the extension. The other
button is used to reach a private line,
or alternatively a g r o u p of c o m m o n
lines.
Several variants of the two main types
are available and can be used in any
combination in both the DIAVOX824
and DIAVOX2836 systems. The only limitation is the requirement that the
ALPHA sets must have at least as many
line buttons as there are trunk lines in
the system.
Fig. 4
Telephone set ALPHA 28 Courier
ALPHA28Courier
has individual buttons and lamps for all 28 trunk lines in
DIAVOX2836. It is available with or without loudspeaker, f i g . 4 .
ALPHA 12Courier
has individual buttons and lamps for 12 trunk lines. It is
used in DIAVOX824and in DIAVOX2836
systems that are not fully built out. The
set is available with or without loudspeaker, fig. 5.
Fig. 5
Telephone set ALPHA 12 Courier
ALPHA8
has individual buttons and
lamps for eight trunk lines, fig.6. It is
mainly used in DIAVOX824 and can be
equipped with a loudspeaking attachment.
All telephone variants for normal speaking via handsets can be equipped with
auxiliary headsets.
Telephone set functions
All telephone variants contain buttons
for the f o l l o w i n g f u n c t i o n s :
Line selection (L)
These buttons are used both to initiate
o u t g o i n g calls and to answer incoming
calls. The lamp associated with each
line indicates the state of any call on the
line: i n c o m i n g calls by rapid flashing,
parked calls by slow flashing and calls in
progress by steady light.
Pushbutton set (F, G, H,
Por=,1-9,*,0
and #)
1 - 9 and 0 are used for dialling external
numbers. The number is sent to the central processing unit, and can then be
transmitted to the PABX or public exchange by means of decadic or DTMF
signalling.
The three buttons F, G and H are used as
prefixes for internal calls. Button P is
used for internal priority calls. On the
Courier sets b u t t o n P is placed in the
first vertical row.
Button = is used as a prefix for abbreviated numbers. Only the Courier sets are
equipped with this b u t t o n .
BETA Courier
has two buttons and
lamps for the connection of calls. It is
used in both DIAVOX824 and 2836, and
is available with or w i t h o u t loudspeaker
fig. 7.
Diversion (D)
Button D is used to divert incoming calls
on a private line to a predetermined extension.
BETA, w h i c h is equipped with two buttons and lamps for connecting of calls,
is used both in DIAVOX824 and 2836,
fig. 8. It can be provided with a l o u d speaking attachment.
Transfer (T)
Button T is used to transfer external
calls to other extensions. The transfer is
preceded by an internal call, during
w h i c h the external call is automatically
parked. The transfer can take place ei-
203
ther when the internal call has been
completed or before an answer is received.
Conference (C)
If, during an external call, an internal
inquiry call Is made and button C is depressed, the three parties are connected
together to a three-party conference.
Register recall (R)
Button R is used to recall the register in
the PABX to which the system is connected.
Parking (Free designation, only Courier)
The parking button is used for manual
parking of external calls.
Disconnection (Free designation, only
Courier)
The disconnection button is used for
manual disconnection of calls as an alternative to other methods when the
loudspeaking or headset function is
used.
Fig. 8
Telephone set BETA
Fig. 7
Telephone set BETA Courier
ON/OFF
These two buttons are only provided on
sets with a loudspeaker and are used to
switch the loudspeaking function on
and off. In the ON position incoming internal calls are connected through after
one ringing signal.
Secrecy (S, only in Courier)
Button S is used to switch off the microphone in telephones equipped with a
loudspeaker.
Operation
When designing DIAVOX824 and 2836
great importance was attached to the
speed and ease of operation of the systems. Efficient traffic handling is particularly important in the applications
for which the systems are intended,
such as booking and order reception.
Incoming external calls are indicated
with double ringing signals and rapid
flashing of the lamp. Calls are answered
by lifting the handset and pressing the
trunk line button. If a loudspeaking telephone is used which is set to the ON
position it is only necessary to depress
the line button.
A call is terminated by replacing the
handset, or in the case of a loudspeaking telephone by depressing the OFF or
disconnection button.
An established external call can be parked in different ways. Depressing an L
button, i.e. selecting a new trunk line, or
button T (the parking button in Courier
sets) automatically parks the external
Summary of f u n c t i o n s
Functions that inccrease the flexibility of the
system
The system can be programmed to operate as
- a PABX
- an executive-secretary system
- a queue system
- a key-system
- a system with parallel call distribution
or with a combination of up to four of these
modes in the same central processing unit.
The system can also be programmed for connection to different types of PABXs and telephone exchanges using
- decadic dialling
- DTMF signalling
- optional pause between digits
- register calling signal.
The following types of group allocations and
classifications increase the versatility of the
system:
- allocation of extensions to groups
- allocation of trunk lines to groups
- classification of certain lines as private
- classification of certain lines as barred to outgoing traffic
- classification of extensions for different
types of external call discrimination based on
the numbers dialled.
It is also possible to
- select the internal and external ringing signal
pause
- easily change the internal extension numbers.
All functions listed above can easily be reprogrammed by the customer via an ALPHA set.
The functions of the system can be changed
during non-working hours by means of a wallmounted box containing a key switch which
gives the functions
-- night blocking
- night service.
Optional extras include
- extension status display
- long-line units.
Functions that make the system more convenient and easier to handle for the user
The following functions increase the speed of
the traffic handling:
- abbreviated dialling
- group hunting
- serial internal calls
- direct speech connection to loudspeaking
sets
- no waiting for dialling tone for internal calls
- priority calls.
The convenience and usefulness of the system
are increased by
- automatic or manual parking
- automatic or manual diversion
- indication that a call is waiting
- indication of the state of trunk lines
- changeover between external calls
- choice between incoming external and internal calls
- external inquiry
- internal inquiry
- three-party conference
- transfer before or after answer
- transfer of parked calls
- variable ringing signal (volume and tone frequency)
- disconnection of the ringing signal.
call, which is indicated by slow flashing
of the lamp. Return to the parked call is
made by depressing the associated L
button.
An outgoing external call is initiated by
depressing the button for a free line. The
extension then receives dialling tone
from the public exchange or PABX and
can dial the desired number.
Abbreviated dialling of external numbers can be programmed and will then
be common for the whole system. A twodigit code (50-79) is used to select one
of a maximum of 30 external numbers.
Thisfacility isavailabletoall extensions.
An incoming internal call is indicated by
single-cadence ringing signals and is
answered by lifting the handset. If a
loudspeaking telephone is used and is
set to ON, one ringing signal will be
heard and the call will then immediately
be connected through. Intercom systems contain the same function.
An internal call isinitiated by depressing
one of the three internal prefix buttons
F, G or H, followed by a digit * or #. This
procedure gives 36 call possibilities.
The caller hears the ringing control tone
if the called extension is free, and the
busy tone if it is engaged. In the latter
case the caller can depress button P,
which gives a muted ringing signal at
the other extension in spite of the fact
that the handset is lifted. The called person then has to decide whether to continue the call in progress or finish it and
accept the priority call by depressing his
own P button.
Inquiries during established external
calls can be made both externally and
internally within the system. An external
inquiry is made by depressing button L
for a free line, which automatically results in the established external call
being parked. When the tone is received
the new number is dialled in the usual
way. Changeover between the two calls
is made by depressing the relevant L
button, which causes the other call to be
parked. In order to make an internal inquiry during an external call it is only
necessary to dial the desired internal
number. The external call is then automatically parked, and a return to this call
is made by depressing the L button. If
button C is depressed instead, a threeparty conference is set up between the
external set and the two internal extensions. The internal inquiry can also be
terminated by a transfer of the external
call by means of button T.
System structure
Fig. 9 shows the system structure. The
extensions are connected to the central
processing unit in the system by means
of three wire pairs in a star-shaped network. The maximum distance between
an extension and the central unit is 2 km.
The 28 trunk lines are also connected to
the central unit. They can come from a
public exchange or a PABX or both. The
programming, which is very simple and
is often carried out by the customer, is
done from any ALPHA set in the system.
An optional number of trunk lines can be
designated private lines. An extension
can be allocated one or two such lines
and then usually handles both the incoming and outgoing traffic. When button D is depressed all incoming calls on
a private line are diverted to a predetermined attendance position. This function can also be programmed so that the
diversion is automatically made after 20
seconds. All other trunk lines are designated common lines. All trunk lines can
be allocated to one or several of four
trunk line groups, and one or more common lines can be barred to outgoing
traffic and hence be reserved for incoming traffic.
Each extension can also be allocated to
one or more of four groups for outgoing
traffic, and to the same or different
groups for incoming traffic.
As regards outgoing traffic the extensions are also allocated to one of four
classes in what is called Trunk Call Discrimination, TCD:
- Class A has no discrimination.
- Class B is limited in accordance with a
table prepared by the customer, in
which certain numbers or number series are barred to outgoing traffic. Exceptions can be made from barred
number series. For example, if all
numbers starting with 0 are barred it
is possible to except certain parts,
e.g. by opening numbers that start
with 07 or 075.
205
Fig. 9
DIAVOX2836, system structure.
Central processing unit, star-shaped three-pair
network towards the extensions and connections
to the local exchange and/or PABX.
The maximum length of the extension lines is
2km with 0.5 mm wire. ALPHA and BETA telephones can be connected in arbitrary combinations
- Class C is similar to class B but often
with stricter limitations. It is possible
to change classes A and B to class C
by means of a key switch, in order to
prevent unauthorized long-distance
calls outside working hours, so called
night blocking.
- Class D is barred to all outgoing traffic
with the exception of the abbreviated
numbers programmed for the system.
These cannot be barred in any way
and are always available to all extensions in the system.
Electrical design
DIAVOX2836 is a stored program controlled system, fig. 10. The central processing unit contains signalling circuits, switching network, line circuits,
local junction circuits and a control system. The various blocks are controlled,
via a bus system, from the control unit,
which consists of an eight-bit microprocessor with a 16 kbyte program
store, a 2 kbyte data store, clock circuits
and a counter which generates timing
signals. It also contains a customer program store of 1 k 4-bit words of the EAROM (Electrically Alterable Read Only
Memory) type, which makes it possible
to program each system individually.
Different customer data for extensions
and trunk lines, data common to each
individual system, tables for abbreviated dialling and digit analysis for trunk
call discrimination are all entered in the
EAROM area from an ALPHA telephone
EAROM retains the stored data even in
the case of a power failure.
Fig. 10
Block diagram of DIAVOX 2836
The bus system, which is controlled by
the CPU, is used to transmit address,
data and control signals and also interrupt signals for direct memory access
(DMA). The address bus can address 64
positions with 16-bit addresses. The
data bus is two-way with a width of 8
bits. The control bus transmits signals to
specify the direction of information on
the data bus and for DMA. The bus frequency is 1 MHz.
The system is fed with - 4 8 V which is
converted to +5V for feeding the electronics, - 8 V for reed switches and
+ 12 V for relays. Internal links and electronic circuits for speech transmission
use - 4 8 V.
Each of the voltages is taken to the same
connector pin in all printed board positions in the magazine. This is done to
prevent a printed board assembly being
destroyed if it happens to be inserted in
the wrong position. The total power consumption is 120-220W depending on
the size and structure of the system.
Mechanical construction
The mechanical construction practice
BYB 901 was chosen for the central processing unit, fig.2. All components are
mounted on single or double-sided
printed wiring boards. The printed
board assemblies are plugged into a
double magazine and all connections
between the boards are made with wrapped
wiring
between
connectors
mounted at the rear of the magazine.
An equipped magazine with its case
constitutes a complete central processing unit. It can be mounted on a wall or
placed on a bench or shelf.
The case, which has air inlets at the bottom and outlets at the top, is designed
so that no extra cooling is necessary
under normal operating conditions.
The outermost printed board assemblies in the magazine contain screw terminals for connecting extensions and
trunk lines.
The switching network boards, SWN,
which are of the reed type, each contain
8x8 cross points and are electronically
controlled and supervised. This printed
Fig. 11
The modular structure of the system
Technical data
Capacity
Extensions Trunk lines Local junction
circuits
(ELC)
(BTC)
(LJC)
8
2,4,6,8,10,12
12,16,20, 12
2,4,6,8,10
24,28,32, 16
2,4,6.8
36
20
2,4,6
24
2,4
28
2
Dimensions, mm
Central unit
Telephone sets:
ALFA8, BETA
ALFA 12 Courier
ALFA 28 Courier
BETA Courier
Height
549
Width
647
Depth
290
95
95
95
95
203
310
360
280
173
242
242
242
Transmission
The handset can be equipped with a transmission circuit modified for, for example, different
power supply methods.
board assembly was
veloped for AXE 10.
Overvoltage protection
Between the speech branches and between
each branch and earth
- ^ 250 V continuously without any destruction
except the individual equipment connected
to the line
- 30 voltage pulses having an amplitude of
< 1000 V, a rise time of 10 V'ms and an interval of at least 10 seconds without any destruction.
Power supply
The central unit is fed with external 48 V d.c.
voltage for
- power feeding over the speech wires during
internal calls
- power feeding over the signalling wires for
light emitting diodes, tone ringers and signalling circuits in the telephones
- d.c. voltage conversion to secondary voltages in the central processing unit.
Nominal voltage
Tolerances
Overall power requirement
DC'DC converter
Nominal voltages
Tolerances
Maximum noise
voltage
originally
de-
The extension line circuit boards, ELC,
each contain digital signal receiver and
transmitter circuits with indicator buffer
memories for four extensions. The
boards also power the electronics (except the speech transmission) in the extensions via an electronic short-circuit
protection.
The bothway trunk circuit boards, BTC,
each hold circuits for four trunk lines.
These circuits detect ringing signals
from the superior system. They also
transmit loop signals and digit information in the form of pulses or DTMF signalling. During an external call the trunk
line is galvanically connected to the
switch. The speech circuit in the extension is thereby powered direct from the
exchange or PABX.
The local j u n c t i o n circuit boards, LJC,
each hold two internal links. During an
internal call the speech circuit in the extension is powered from LJC, The internal link also reduces the speech level so
that it is similar to that of a normal external call. The link also contains circuits
for sending the busy tone and ringing
control tone d u r i n g internal calls.
The size of the system will determine
how many of each of these four types of
printed board assemblies are included,
fig. 11. The central unit also contains
twelve other printed board assemblies
of different types.
48 V
- 42 to - 56 V
120-220W
+ 5.1 V + 12V
± 4%
±5%
50 mV
r.m.s.
Power requirement 30 W
50 mV
r.m.s.
23 W
Environmental requirements
Temperature
+ 5CC to + 40CC
Relative humidity
20% to 80%
-8V
±2,5%
50 mV
r.m.s.
8W
References
1. Jismalm, G. and Magnusson, S.: Office
Communication System DIAVOX824.
Ericsson Rev. 56 (1979):3, pp. 124129.
Ericsson Power Systems,
a New Division
within RIFA
Ake Ljungblom
Ericsson Power Systems is a newly formed division within RIFA. The Ericsson
Group has thereby concentrated
its resources for development and marketing of
power supply equipment into one division.
AKE LJUNGBLOM
Ericsson Power Systems
RIFA AB
The a hor outlines the changes in the Group that have led to the forming of the
new division, describes the development
trends for power supply of
telecommunication
equipment and gives a summary of the division's
products.
Power electronics
Power supply equipment for telecommunication is one of the oldest products
within the Ericsson Group. At the same
time as Lars Magnus Ericsson started to
manufacture telephone sets in 1878 he
also started to manufacture batteries for
supplying the power for the telephone
circuits'.
UDC 621.311 002 6
621.395:621.311
Fig. 1
Electronic telephone exchanges are fed with d.c.
voltage, usually 48 V, from a central power plant.
The d.c. voltage is converted to electronic voltages in rack power units
Up to 1970 most of the power supply
equipment developed within the parent
company was intended for telephone
exchanges 2 . This work therefore formed
a natural part of the exchange activities.
In 1970 the sections w o r k i n g with design
and marketing of power supply equipment were separated from the telephone exchange division and were formed into a separate department for developing centralized power equipment.
Gradually all activities within the parent
company c o n c e r n i n g both centralized
and dedicated power equipment were
assembled in this department.
The introduction of electronics has resulted in different requirements and
new system designs for the power supply of t e l e c o m m u n i c a t i o n equipment.
Forexample, local converters have been
introduced 3 5 for feeding a rack, a magazine or in certain cases a single printed
board assembly, fig. 1. The development
in the c o m p o n e n t field has made possible increasingly sophisticated designs
in respect of both systems and equipment 6 ~ 8 .
By using increasingly higher frequencies it has been possible to reduce the
volume of the converters and to increase their efficiency considerably.
Power electronics with this new design
are being used not only for telecommunication equipment but for all electronic equipment that requires power
feeding.
Ericsson Power Systems
Acquisition of companies and Ericsson's extended and new activities outside the field of public t e l e c o m m u n i c a -
Central power
Distribution
Dedicated power
in the exchange
be developed and marketed to a greater
extent. The resultant increasing production volumes of the division's products
will improve Ericsson's competitiveness, not only in the field of power supply equipment itself but also as regards
all Ericsson products that contain such
equipment.
Concentration of resources
Fig. 2
The premises of Ericsson Power Systems in the
southern suburbs of Stockholm, Sweden
tions have meant a greater demand for
power supply equipment within the
Group. In connection with the recent
organizing of Ericsson into business
areas it was considered that a concentration of the power activities would
create the best conditions for rational
management of this important field. A
neutral position in the Group has also
been considered advantageous in order
to be able to meet the Groups combined
needs for different types of power supply equipment. Furthermore, in the
efforts to find markets for Ericsson's
power supply products outside the
Group the combining of the power supply and components activities to form
one business area seemed natural. Recently a number of power supply products have been developed in collaboration with RIFA. Since RIFA s sales organization can be used for marketing
power supply products it was considered logical to transfer the power supply
department of the parent company to
RIFA AB as a new division. The transfer
took place on October 1st, 1983, and the
division is called Ericsson Power Systems.
Aims of the division
The aims of the division are to provide
power supply equipment on a commercial basis for the other Ericsson business areas and to deliver power plant to
customers jointly with other business
areas. Power supply equipment intended for applications outside the Group,
in the telecommunication, data transmission and associated fields, is also to
In 1977 the power supply department
moved to new premises on the outskirts
of Stockholm (Sweden), figs.2-3. A
well-equipped laboratory was set up9,
which has facilitated the development of
advanced systems and equipment for
power feeding all types of telecommunication equipment, and also the investigation and evaluation of power
components. An extension will be completed early in 1984. It will include a new
laboratory for further development of
Ericsson's cooling systems, intended
mainly for telephone exchanges. Oneof
these cooling systems will be used to
cool the new extension building. These
premises have made it possible to
gather the whole division in one and the
same place. A nearby site also contains
a complete ERICSSON SUNWIND test
plant, a power supply system which uses
the sun and wind to generate power. The
modern and well-equipped premises offer the best possible conditions for stimulating and productive work. The very
considerable experience assembled
here also provides a good basis for the
activities.
A number of specialist functions are essential for the development of power
supply units and systems, several of
them being unique in this field. The division encompasses the whole power
supply field, from component to system
level, and collectively the staff has gained vast experience in the most important parts of these activities. A few examples are given below.
Component techniques
Good component knowledge is essential in view of the rapid development of
semiconductors, magnetic materials
and batteries for power supply equipment. Comprehensive equipment is
available to the seven engineers who are
at present working on component techniques.
963
nH. 4
The requirements of power supply in the world
209
market
The total market
Non-captive market
Expansive market
In 1979 the w o r l d market for power supply equipment, not i n c l u d i n g batteries
and standby p o w e r e q u i p m e n t , was estimated to be worth
approximately
3.5billion (109) US$. This was expected
to increase to 8 b i l l i o n US$ by 1985.
Judged by the situation in the US, the
non-captive market in 1979 was between 25 and 3 0 % , and this was expected to increase to 3 5 - 4 0 % by 1985.
Hence between 1979 and 1985 the noncaptive market should grow from approximately 1 to approximately 3 b i l l i o n
US$, c o r r e s p o n d i n g to an annual increase of about 2 0 % at current dollar
prices, f i g . 4 . The majority of the equipment in demand consists of b u i l t - i n
power units. In this category d.c./d.c.
converters
accounted
for
approximately 1 0 % in 1979 and this is expected
to increase to between 13 and 1 5 % by
1985; the rest are a.c./d.c. converters.
Power supply is required for equipment
for t e l e c o m m u n i c a t i o n , data processing, process control and instruments.
The growth rate varies for different
product fields. Word processing equipment is expected to grow by 3 2 % per
year and main frame computers by only
4%.
The division consists of two departments, one for primary power and one
for electronic power. The latter department is responsible for built-in power
supply units, power modules and elec-
Basic technology
With a large number of experienced designers it has been possible to carry out
successful basic development, w h i c h
has resulted in a number of interesting
designs. The main aims are to develop
simple, optimal designs with high efficiency, adapted to modern manufacturing methods, in order to reduce costs
and increase reliability.
System technology
In cases where the electronic equipment
is so extensive that the power has to be
supplied from a system of several units it
is important that the optimizing of the
power supply takes place at the system
level. The division has the necessary expertise for such work.
Fig. 3
One of the system test rooms on the premises of
Ericsson Power Systems. The panels to the left
are used for a.c. distribution. High direct currents
are distributed via bars in the upper part of the
picture. A small selection of the power supply
equipment developed and manufactured by Ericsson can be seen in the background
Product summary
Fig. 5
ERICSSON ENERGYMASTER is a system for
remote control and supervision of all power
equipment in the telecommunication plants of a
country or district
tricity meters. The energy plant department includes product sections and
f u n c t i o n sections for stocks and delivery planning, c o m p o n e n t and microprocessor technology, and also a section for installation methods, d o c u m e n tation and training. Extensive training of
customer personnel is carried out both
at customers' premises around the
world and at the division's training establishment in S t o c k h o l m , Sweden.
Several of the division's products have
already been described in Ericsson Review. A summary of the products is
given below, divided into eight product
fields.
Primary power plant
Telephone power plant
Telephone power plant includes
- large rectifiers
- d.c./d.c. booster converters
- inverters
- distribution racks
- distribution material
- rack and emergency lighting
- batteries.
Several
power
supply
systems
have
been developed for large exchanges as
well
as small
remote
subscriber
stages 10 14. The latest addition is a microprocessor-controlled
supervision
system, ERICSSON ENERGYMASTER,
for remote control and supervision of all
primary power e q u i p m e n t in telecommunication plants 1 5 16, fig.5.
Cooling
equipment
Approximately 9 0 % of the energy fed
into a telephone exchange via the power
supply is dissipated in the exchange
building as heat losses. The modern,
compact exchange equipment requires
efficient removal of this heat. A waterbased cooling system 1 7 has been developed for this purpose, f i g . 6. By storing cold water in tanks it is also possible
to cool an exchange during a mains
failure. The c o o l i n g plant must be equipped with a c o o l i n g reserve in order to
reach a degree of reliability comparable
with the reliability of the telecommunication equipment. 1 8 Similarly the
power plant has an energy reserve in the
form of storage batteries. Cooling systems are available to suit both large exchanges and smaller units placed in
buildings or containers.
383
211
er availability. A construction practice
has been developed which makes possible adaptation to Ericsson's standard
BYB construction practice, European
printed board standards and 19" standards. An electronic fuse in hybrid technology is being developed in collaboration with the Swedish Telecommunications Administration. Such a fuse will
give an absolute selective protection for
different electronic equipments connected to the fuse in parallel. The shortcircuit current is limited to a value just
above the rated current. The fuse can
initiate an alarm and permits remote
control. The latter facility is important if
certain loads require priority, for example in an emergency.
Fig. 7
ERICSSON SUNWIND powering a mobile radio
station in the Hardanger mountains, Norway
Fig. 6
Ericsson cooling equipment delivered to Denmark. From the left: water tank, pump units with
valves and electronic control, compressor and
evaporator
Private market products
This category comprises small rectifiers, inverters, distribution equipment
and batteries19. Systems have been developed that provide no-break direct
current, primarily for PBXs12 and for
UPS (Uninterrupted
Power Supply)
systems. UPS systems are used in computer and terminal systems, a field
which will expand rapidly as users become increasingly dependent on their
data bases, and therefore require great-
Primary power systems
Standby power units have long formed
part of telecommunication plants supplied by Ericsson20. Diesel engines and
generators have been tested in order to
find reliable equipment which is primarily suitable for feeding thyristor rectifiers. The control equipment has been
developed by Ericsson. Different methods have also been developed for generating primary power, for use on sites
without access to electricity or where
the supplies are unreliable. The low
power requirements of modern telecommunication equipment make it possible to utilize solar and wind power in
an economical way. A system, ERICSSON SUNWIND 2122 , has been developed, which uses the sun and wind, either
individually or in combination. The system always includes a battery and often
also a minidiesel with generator. The
system can be used for applications having an average power requirement of up
to approximately 500 W. The division undertakes turn-key installations, including buried plastic containers or small
buildings for the batteries, diesel power
plant, electronic control and the equipment to be fed. The first deliveries of
SUNWIND were made to the lighthouse
Blenheim belonging to the National
Swedish Administration of Shipping
and Navigation, and to the Norwegian
Telecommunications Administration's
mobile radio station in the Hardanger
mountains, fig. 7. They were put into operation
during
August-September
1983. Plant with solar panels, with or
without minidiesels, have been supplied
since 1981. By December 1983 some
thirty such systems were in operation.
Fig. 9
A comparison between two d.c./d.c. converters
for 48 V in and +5V, 20 W out. To the left a
modern flyback unit having more than 80 %
efficiency, and to the right a half bridge with an
efficiency of 69% or less
Fig. 8
A tone and ringing set BKL701 for mounting on a
wall. The set is assembled in a single aluminium
profile, which holds up to four printed board
assemblies of Ericsson's ROF type or Europa
boards
Fig. 10
A miniaturized power module for mounting on a
printed circuit board. Input +24 V, output ±12V,
25 W with 80 % efficiency
Ringing and tone signalling eguipment
Ringing and tone signalling equipment
has long been supplied together with
centralized power supply equipment23.
During recent years the development of
the equipment has been supplemented
by an analysis of the whole system for
ringing and tone signalling24. New principles have been applied which have led
to a considerable improvement in the
reliability of the system, as well as reduced costs. On the basis of these principles a new tone and ringing set,
BKL701, has been developed, fig.8.
BKL701 suits all types and makes of
public telephone exchanges, and is primarily intended to replace electromechanical tone and ringing sets. The low
capital and operating costs of the set
mean that a replacement pays off very
quickly. The division can also offer optimized system designs for signal distribution and computerized measuring
methods for system analysis.
Electronic power
Built-in power units
Rack power units for telephone exchanges and PBXs account for the
greatest part of the demand for d.c./d.c.
converters. There is a rapidly increasing
market for built-in power units (OEM
power supplies) in different types of
computer terminals, modems, radio
equipment etc. The type of conversion
usually required is from a.c. to d.c. Most
power units are custom-designed,
adapted to the equipment in which they
are to be incorporated. Circuits and
components may have been standardized but the combinations of different
requirements as regards voltage,
power, tolerances, environment and
life, together with necessary mechanical adaptations, make the variation possibilities almost unlimited. Power units
developed recently require comparatively few components and have high
efficiency, fig. 9. The division has the re-
213
sources to p r o d u c e custom-designed
power units very quickly. A prototype
can be delivered within a m o n t h .
Power
modules
A first product line of miniaturized
d.c./d.c. converters operating at a frequency of 300 kHz has been developed,
fig. 10. It comprises eight types in the
power range 2 5 - 4 0 W . These power
modules consist of hybrid circuits in
thick-film technology, designed for fully
automatic manufacture. The main advantages of the power modules are
small dimensions and low weight. They
readily provide unit designers with a
complete, noise-free and reliable power
supply.
Fig. 11
Ericsson's electronic three-phase electricity
meter for tariffs with different rates for different
parts of the day or year
Electricity
meters
A fully electronic three-phase electricity
meter has been developed for tariffs
with time-dependent rates, fig. 11. At the
beginning of 1984 a number of meters
will be delivered to a few large commercial power suppliers for field testing. Unlike conventional meters with a rotating
disc, the new meter contains a unique
current and voltage transducer, w h i c h
in several respects gives greater accuracy. The electricity meter contains a
microprocessor for storing information
and switching between the different
rates. For charging purposes the meter
is read by c o n n e c t i n g a portable terminal equipment w h i c h is afterwards connected to a computer for calculation of
charges and invoicing.
References
1. Jacobaeus, C. et a!.: Evolution of the
Technology. LM Ericsson 100 years.
Vol. Ill, 1976, pp. 223-237.
2. Ljungblom, A.: LM Ericsson Power
Supply Systems for Telecommunication Equipments. Ericsson Rev. 45
(1968):4, pp. 142-162.
3. Boije af Gennas, C : Optimization of
Power Supply Equipment for Modern
Telecommunications Systems. Ericsson Rev. 53(1976):3, pp. 142-151.
4. Viklund, B. and Assow, B.: Power Supply Units in the M5 Construction Practice. Ericsson Rev. 56 (1979):2, pp.
80-83.
5. Boije af Gennas, C. and Webrell, I.:
Power Units and Power Distribution in
the BYB Construction
Practice.
Ericsson Rev. 58 (1981 ):1. pp. 2 - 8 ,
6. Orevik, A.: DC Distribution for Power
Supply of Telecommunication Equipments. Ericsson Rev. 49(1972):1, pp.
14-28.
7. Orevik, A.: Power Supplies for Electronic Telephone Exchanges. Ericsson Rev. 51 (1974):4, pp. 120-127.
8. Wolpert, T. and Bjork, D.: Power Supply System with Booster ConvertersViewpoints after 10 Years in Operation. Ericsson Rev. 52 (1975):1, pp.
14-23.
9. Pavelsen, O.: LM Ericsson's Power
Supply Laboratory. Ericsson Rev. 56
(1979):4, pp. 164-169.
10. Michelsen, S. et a\.:A New Generation
of Power Supply Equipment. Type
BZD112. Ericsson Rev. 55 (1978):2,
pp. 46-57.
11. Hansson, A.: Uninterruptible
AC
Power Supply System. Type BZV102.
Ericsson Rev. 56 (1979):1, pp. 34-39.
12. Hansson, L. and Strickert, A.: Power
Supply System for Small Telecommunication Plants. Ericsson Rev. 56
(1979):2, pp. 57-63.
13. Hansson, L. and Santi, R.: A Rectifier
for Large Plants. Ericsson Rev. 58
(1981):2, pp. 81-87.
14. Rundkvist, K. et al.: Power Supply
Equipment for Large Telecommunication Plants. Ericsson Rev. 5S(1981):3,
pp. 111-119.
15. Ericsson, M. etal.: Computer Controlled Power Supply Equipment for Telecommunication Plants. Ericsson Rev.
60(1983):2, pp. 94-101.
16. Boije af Gennas, C : Microprocessor
Control System for Large Power
Plants. Proceedings Intelec 83.
17. Almquist, R.: A Cooling System for
Electronic
Telephone
Exchanges.
Ericsson Rev. 58 (1981):4, pp. 188195.
18. Wolpert, T.: The Reliability of Power
and Cooling Systems. Proceedings Intelec 82, pp. 181-186.
19. Andersson, H. and Bergvik, S.: Sealed
Lead Batteries for Small Telecommunication Plants. Ericsson Rev. 60
(1983):4, pp 222-225.
20. Norstrom, L.-E.: Diesel Generating Set
for Telecommunication
Equipments.
Ericsson Rev. 48 (1971 ):4, pp. 153160.
21. Akerlund, J.: ERICSSON SUNWIND.
Ericsson Rev. 59 (1982):1, pp. 40-47.
22. Eriksson, M. and Ottosson, J.: Optimizing the Power from a Wind Turbine. Ericsson Rev. 60 (1983):3, pp.
159-163.
23. Vago, F. and Persson, L: New Generation of Ringing and Tone Signalling
Equipments, BKL600. Ericsson Rev.
55 (1978):4, pp. 130-139.
24. Vago, F. and Ahl, P.: Analysis of Operation Conditions in Ringing and Tone
Systems. Proceedings Intelec 82, pp.
170-175
25. Lind, H.: Miniaturized Power Modules.
Ericsson Rev. 60 (1983):1, pp. 42-44.
ERICOM DIRECT
Lennart Skoog and Curt Sundmalm
Ericsson Security and Tele Systems AB has developed a combined intercom and
information system, ERICOM DIRECT, for applications ranging from tens of
extensions to several thousands. The development was preceded by a market
survey in order to determine what functions, in addition to the pure telephone
functions, were desirable in a new system. The most important of these new
functions is the transmission of encoded and recorded messages to or from
extensions that are temporarily unattended.
The authors describe the results of the market survey, present the functions
available in ERICOM DIRECT and finally describe the structure and operation of
the system.
UDC 621.395 2
The needs of the market for separate
systems for internal voice communication, as a complement to other telephone systems, are demonstrated by the
fact that Ericsson Security and Tele Systems AB has so far delivered approximately one million intercom lines. Modern technology makes it possible to
equip intercom systems with new and
rational facilities for information transmission. ERICOM DIRECT is such a
combined intercom and information
system.
The development of ERICOM DIRECT
was preceded by a market survey in
order to determine what functions users
of intercom systems require in addition
to the basic telephone functions. The
survey showed that:
- The intercom set must be small and
have both a loudspeaking and a non-
Fig. 1
The ERICOM DIRECT intercom set has a threeposition volume control. The set is loudspeaking
when placed on the table, but when lifted to the
ear it operates like a handset
loudspeking function, like the earlier
model ERICOM.
- The set must contain a signalling facility, for requesting the other party to
switch over to the non-loudspeaking
mode for a confidential call.
- The control devices must be well arranged and easy to handle. Onlya limited number of keys should be
provided.
• The sets must have a volume control
for adjusting the sound level to the
environment.
- It must be possible to divert calls, so
that calls to a temporarily unattended
extension can be answered at another
extension.
The system must include facilities for
encoded or recorded messages, so
that a person who leaves his room can
inform callers where he is and when
he expects to return.
- It must be possible to send a simple
coded message to a person who does
not answer a call.
It is also desirable that the caller
should be able to record a message if
the call is not answered. Telephone
operators and secretaries would then
not have to make out and deliver telephone call notes.
The caller should be informed of the
person answering the call.
It must be possible to call selected
extensions without having to dial the
whole number.
215
LENNART SKOOG
CURT SUNDMALM
Ericsson Security and Tele Systems AB
- Some of the people interviewed desired real "calling by name", i.e. it
should only be necessary to say the
name in order to get the call connected.
Manufacturing requirements
The marketing staff of Ericsson Security
and Tele Systems AB wanted a strictly
modular system, with few and always
the same building blocks regardless of
the size of the system. Two-wire connection between the extensions and the exchange was desired, instead of the fourwire connection usually required in intercom systems, in order to reduce the
installation costs and to make it possible
to use existing private networks. Tie
lines between ERICOM DIRECT exchanges and older types of Ericsson intercom systems were also demanded.
The new system should offer programmable reallocation of extension numbers. It should also be possible to connect analogue telephone sets to the digital system. A requirements specification was gradually built up that became
ERICOM DIRECT.
Performance
Loudspeaking/non-loudspeaking with
volume control
The ERICOM DIRECT intercom set,
fig. 1, can be used for either loudspeak-
Fig. 2
The keyset with 10 digit keys also has a an A-key
for, for example, accepting a call, and a B-key for
disconnecting it. The keyset is also used for
programming. In addition to the character display
and a light-emitting diode, the top panel contains
an information switch and a three-step volume
control
Fig. 3
Each ERICOM DIRECT is also an information
terminal which gives messages in the form of
displayed characters or recorded messages
ing calls or for non-loudspeaking calls
when held like a handset. Before a call is
made a tone signal can be sent in order
to request the other party to lift the set
for a confidential conversation The set
has a three-position volume control.
Information switch
The information switch, fig. 2, has three
positions. In the first position, privacy,
the extension is blocked against incoming calls and a call is accepted by depressing an acknowledgement key
With the switch in a second position the
voice of a calling person is immediately
heard in the loudspeaker. There is no
need to press any key and the call can be
answered at a distance from the set. In a
third position, information transmission, the programmed information is
sent out.
Programming
Calls can be directed to another extension by means of the information switch.
Display information can be programmed with the aid of the keyset,
fig.3, and given to anybody who calls.
The type of information that can be
given is:
- the time of return during the same day
- the day and month of return
- where the user is, in accordance with
optional company codes.
Fig. 4
There are two types of messages. Either the
character display shows a request to call a
person, or a recorded message is heard from the
loudspeaker
A user can also record a personal message, which is repeated to anyone who
calls, fig. 4.
corded names, PRINA will automatically
call the extension that corresponds to
the name.
When a call is not answered, the caller
can leave a display message including
his extension number and requesting a
return call. Alternatively the caller can
record a spoken message. In both cases
the light emitting diode on the called set
flashes, and the character display shows
02. The message is obtained by dialling
the displayed number 02.
Standard facilities
The standard facilities offered by the
system are:
- loudspeaker paging to all extensions
- calls to groups of extensions
- camp on busy
- enquiry call
- call transfer
- group hunting with the call being set
up to the first free extension in a
group.
In the idle position the character display
shows the extension's own number.
When a call comes in, the display shows
the caller's number. When a call is made
from the set, the display shows the dialled digits, fig. 5.
Two methods of calling without using
extension numbers
Each user can have ten personally programmed abbreviated numbers, which
are called by depressing a name-marked
key and the key A.
Fig. 5
The character display shows the following types
of information:
- Message waiting (02)
- Who is calling
- Who is answering
- When the called person will return
- Where the called person is
- The number of the extension when idle
ERICOM DIRECT can be provided with
additional equipment for personal voice
recognition (PRINA) with voice-operated calling by name, fig.6. Each user
can program some twenty name calls
just by repeating each name a few times.
When thereafter the user depresses a
start key and pronounces one of the re-
Optional facilities
The optional facilities offered by the system are
- paging by radio to pocket receivers
- conference calls
- communication with mobile radio
systems
- music distribution
- tie lines between ERICOM DIRECT exchanges
- tie lines to Ericsson's previous intercom systems and PAXs
- recorded messages.
Technical description
ERICOM DIRECT is a system for rapid
and efficient internal communication,
and is able to provide information even
when the person sought is absent. The
system has a modular structure, and
with one type of shelf and only a few
types of printed circuit board assemblies (PCBs), it can satisfy the needs of a
small office requiring only a dozen extensions, as well as an organization
needing several thousands of extensions. The system consists of a central
unit to which different intercom sets are
connected over a two-wire star shaped
network.
ERICOM DIRECT uses pulse code modulation between the extensions and the
exchange.
Intercom sets
The ERICOM DIRECT intercom set comprises a keyset, character display, light
emitting diode, volume switch and information switch. The set is normally loudspeaking, but automatically changes
over to the non-loudspeaking mode
when it is lifted.
217
lines emphasizes the need for restricting the software.
Fig. 6
Each extension has ten personally programmed
abbreviated call numbers. When making a call
one of the name marked keys and the A-key are
pressed.
With the PRINA equipment for voice-operated
dialling, a call is initiated by just saying the name
of the desired person
The loudspeaking function can be disconnected by the clipping of a diode in
the set, so that it operates as a digital
telephone for non-loudspeaking calls
only.
Several other types of intercom sets are
available. It is also possible to connect
ordinary analogue telephone sets
System structure
The main aim of the system development work has been to create a modern,
modular, flexible and progressive system, which is easy to install and put into
service, while still retaining the possibility of programming and adapting
the system to meet the individual requirements of the customers.
Fig. 7
A shelf for 14 PCBs
Today the use of sophisticated technology with processors, memories etc. is
natural, but the demand for economy for
systems containing only some tens of
It was soon clear that the central unit in a
digitalized PCM system, where the
favourable cost trend for semiconductor components is exploited, could be
made considerably cheaper than the
equivalent unit in a partly analogue system From a financial point of view it was
important that the PCBs were designed
as complete functional units and that
they could be placed in a shelf for
mounting on a wall or in a rack.
The chosen printed board size,
344x178mm (TVF115), holds either a
complete central processor (EXCP), or
four speech connection circuits including voice switching (CCDX), or equipment for connecting ten digital lines
(LIDI). These three types of PCBs are
sufficient for a functional system.
Auxiliary PCBs are available for
- four analogue telephone sets (LIAI)
- three junction lines to other exchanges via private lines (ADIO)
- two junction lines to other exchanges
via leased lines (TLDT)
- four 4-second voice
memories
(VMEM).
There are two shelf sizes, one for up to
seven PCBs (max. 50 lines, weight
11 kg), and one for a maximum of fourteen PCBs (max. 120 lines, weight 13 kg
including the PCBs). Three shelves can
be connected together via a 34-.pole
band cable to form an exchange with 42
PCB modules for approximately 300
lines. Each such basic exchange can be
equipped with a CPU PCB (EXCP) and
optionally 1-4 speech connection
PCBs (CCDX). The other modules in the
shelf can be equipped with line PCBs
(LIDI), fig.8, in optional order, together
with auxiliary PCBs.
A homogeneous exchange system for
up to 4800 lines can be obtained by connecting together several basic exchanges. The connection is made with
an optical fibre cable, which is run in a
loop between the exchanges, fig. 9. In
the event of a break in the loop, each
basic exchange will operate autonomously. An exchange that develops a
fault will be by-passed by means of an
elctro-optic switch.
Fig. 9
Optical link for 16Mbit/s between basic exchanges
OCL
HWLO
Optical transmission unit
Interlace between the exchange and OCL
Voice switching
In a loudspeaking circuit the speech direction must be controlled by a voice
switching unit, with speech amplification and speech attenuation in the respective speech directions. If this method is to work properly, it is necessary to
filter out environmental noise from the
sound channel before the sound is fed
into the control equipment.
Fig. 8
A PCB for 10 digital lines, LIDI
Since all speech information is digital,
digital filters are used and changes in
amplification are obtained by addressing memories with the actual amplitude
and the desired amplification. The amplitude value required for each such
combination can then be read out from
the memories.
The voice switching equipment forms
part of the CCDX PCB. The voice transmitter and receiver cannot be connected together since speech always
undergoes amplification changes in
loudspeaking systems, and different
time slots have to be chosen for different
parts of the transmission.
ductor for each individual bit. This
makes it possible to keep the bit rate on
the PCM bus as low as 1 Mbit/s.
The line interface PCBs have an internal
bus network of their own between the
line equipment and the regional processor, and between the line equipment
and the connection to the PCM bus
Central time channels and time slots
The central time division comprises 32
channels, each with two time slots. The
time slots of the different channels are
not tied to each other or to any function
or unit, but are used as and when required, for example for a speech connection. Two time slots are necessary
for each loudspeaking intercom set, one
for information to the set and one for
information from it. Four time slots are
thus required for a complete loudspeaking connection between two sets,
fig.11.
A non-loudspeaking call with both sets
lifted needs only two time slots, one for
each speech direction.
One of the central time slots is designed
as a carrier of a tone signal and another
for, for example, music.
Fig. 10
The PCB can be placed in an optional position in
the shelf where there is access to the necessary
buses. Only EXCP and HWLO require all buses
Bus system for the PCBNs
The central processor board (EXCP) is
built up around the processor 6809,
which via the MP bus communicates
with, for example, the optical fibre
equipment (HWLO), fig. 10. Communication with other boards takes place via
the control bus from the communication
processor 68120. All the other PCBs
contain a regional processor 6801 of
their own. In addition each voice switching equipment contains a processor,
and thus each CCDX contains an additional four processors.
OCL
HWLO
EXCP
CCDX
LIDI
LIAI
ADIO
The transmission of speech requires a
high bit rate, and it was therefore considered most suitable to design the PCM
bus as a special parallel bus with a con-
Each such complete transmission, including pauses, takes place 16000times
per second, which gives a time frame of
62.5 LIS and a frequency range up to a
TLDT
VMEM
Optical transmission unit
Interface between the exchange and OCL
Central processing unit
PCB for tour speech connection circuits
Line PCB for ten digital extensions
Line PCB for four analogue extensions
Universal PCB for analogue and digital transmlsson to and from the exchange
PCB for tie line traffic
Voice memory PCB
Signalling over the intercom wires
The connection to each extension consists of one twisted pair, over which the
current feeding also takes place.
The digital transmisssion is carried out
by means of burst signalling, i.e. the digital signals to and from the extension are
sent in groups, bursts, at different times
and separated by pauses, fig. 12.
1983
219
Fig. 11
A loudspeaking system requires voice switching
that determines the amplification in the two
speech directions. Four central time slots (1A, 1B,
2A and 2B) are used for a loudspeaking circuit.
The central system has 32 channels and 64 time
slots
maximum of 8000Hz. The process is
started from the exchange side by the
line unit sending two synchronization
bits, which are followed by PCM speech
samples and two data bits.
With the aid of the synchronization bits,
the intercom set sorts out the PCM samples and the data bits and acknowledges
by sending similar information, i.e. synchronization bits, PCM samples and
data bits, back to the exchange equipment.
The two data bits consist of one bit for
actual data (X), corresponding to a
16 kbit/s information flow, and an end of
frame bit (E), which is a one in every
thirty-second burst and a zero in the
other thirty-one bursts. In this way a
multi-frame is obtained which contains
32 bits and is repeated every other millisecond, fig. 12. These 32 bits can carry a
large amount of information. The information from the keyset is allocated six
bits in the multiframe, which with binary
representation permits 64 digit combinations. A keyset with twelve keys,
each with one make contact, has been
designed which generates such active
information. Standard keysets can be
coupled in cascade to give 24,36,48 etc.
functions. Individual keys can also be
added.
Fig. 12
Information to and from the extension is transmitted in the form of bursts having a duration of ten
time slots. The bursts in the two directions are
separated by pauses. The total time frame for a
burst in each direction is 32 time slots and
corresponds to 62.5 \>s. The data bits (X,E) are
collected from 32 such bursts. The collection is
completed when bit E (End of frame) is a one. The
multi-frame thus formed is 2000 us
The principle used for sensing the keyset also, when no active key is depressed, allows the transmission of information concerning states of a longer-
lasting passive type, such as cradleswitch states, privacy mode and indication of origin. The indication of origin
provides information as to which type of
intercom set is connected. This means
that different types of sets can be mixed
as desired, and each type can be treated
individually.
A large number of the 32 bits in the multiframe are unused and available for future functions.
The whole transmission circuit
An extension which is not connected up
for a call is either connected to a time
slot for music or to a silent channel. In
both cases burst signalling takes place
continuously between the set and the
line unit. A call attempt is therefore indicated immediately, and the associated
regional processor (RP) calls the central
processor (CPU). RP also calls CPU as
soon as there is a change in the line
information. When necessary, CPU allocates a time slot on the PCM bus to the
line in question, selects a voice switching unit and informs the unit which time
slot applies. In a similar way CPU allocates the time slots on the PCM bus that
are needed for a complete connection.
Information about the allocated time
slot is sent to RP, which thereafter tells
the line equipment in question which
time slot is to be used for sending information to or fetching information from
the PCM bus. Since CPU sends out resetting and stepping information, the
time slots for the respective line can be
determined and burst transmission
starts in the allocated time slot. The
whole system is thereby synchronized,
fig.13.
Each RP controls all lines on its own
PCB. The synchronization bits are not
included in the transmission of information to and from the PCM bus, and hence
the bus only carries data and speech
bits.
Speech encoding
Memories storing speech information
are often designed for storing words of
eight bits. There are thus reasons for
restricting the information on the PCM
bus to just eight bits. Since the data part
comprises two bits, there are six bits left
for speech. Hence the PCM bus consists
of two wires for data and six wires for
speech, a total of eight wires.
220
The encoding principle commonly used
in telephony is based on the A-law. This
law uses eight bits, of which three are
used to specify the larger amplitude
range (segment) within which an amplitude lies, and the accuracy (step) with
which the amplitude is given in the segment. Of the other bits one is used to
specify polarity and four to indicate
which of the 16 steps within the segment
applies.
The two lowest segments cover amplitude ranges of equal size The other segments differ in size by a factor of two.
The A-law is designed so that a certain
amplitude value can only be given in one
segment and one step, fig. 14.
Fig. 13
For the whole transmission circuit, zero instant is
given by the central processor, EXCP, which
sends a common resetting pulse (T 0 E ) to the
channels. Zero instant for the line circuit in the
exchange, DEC, is given by the allocated channel
(T O L ); for the line circuit in the intercom set, DIC, it
is given by the time when synchronization pulses
are obtained (T o a )
EXCP
DEC
DIC
TOE
T0L
T0A
Central processor
L i n e c i r c u i t in t h e e x c h a n g e , w h i c h is a l l o c a t e d
c h a n n e l 14, T h e l i n e c i r c u i t c o u n t e r is r e s e t at t h e
t i m e of c h a n n e l 14
L i n e c i r c u i t in t h e i n t e r c o m set, w h o s e c o u n t e r is
reset by t h e s y n c h r o n i z a t i o n p u l s e t r a n s i t i o n
Central resetting
R e s e t t i n g of t h e line
R e s e t t i n g of t h e i n t e r c o m set
In the code converter developed by
Ericsson Security and Tele Systems AB,
the number of steps in each segment is
doubled, so that in each segment the
amplitude values for all lower segments
can also be given, but with a lower degree of accuracy. This means that, if regard is paid to the mean value of the
speech amplitude, a large number of
successive encodings can be given
without the segment having to be
changed. The segment can be retained
during a comparatively long time (tens
of ms) and only six bits have to be transmitted. When the segment has to be
changed a message is sent from the
transmitter to the receiver, which works
synchronously. A code converter of the
new type uses a simple six-bit linear encoder plus a variable amplifier/attenuator.
Thus the information regarding segments and steps obtained on the receiving side is identical with the corresponding levels obtained according to the Alaw.
Special circuits
The new code allows simple circuit designs and Ericsson Security and Tele
Systems AB has designed a special
CMOS circuit (DIC) for the intercom set.
The circuit contains all the functions required for the set except the microphone amplifier and the output stage.
The special circuit (40 pin) contains
code converter, oscillator, synchronization, biphase control functions for controlling the light emitting diode and the
character display, transmitter, receiver
and line matching.
Furthermore, Ericsson Security and
Tele Systems AB has designed a character display module, which is connected
directly to the intercom set (DIC) via
three wires. The module (DIM) has five 7segment characters and can display digits, certain letters and a minus sign.
The company has also designed a special circuit in NMOS. This circuit (DEC)
is connected to the exchange side of the
extension line and contains all the functions necessary for all the digital connections between the set and the rest of
the system. DEC communicates with the
PCM bus via bus drive circuits, and with
the regional processor on the line board
via an internal bus. DEC is connected to
the PCM bus in a time slot which is determined by EXCP via the regional processor, and hence no selector as such is
required, the connection being made by
means of time division.
DEC is transparent to transmission of
speech and data to and from the PCM
bus. In the case of transmission of data
to EXCP, the information is sifted by the
regional processor so that only permanent changes necessitate calling in the
central processor.
Each DEC contains functions for handling two lines; towards the line the cir-
Fig. 14
With speech encoding for code conversion in
accordance with the A-law the speech amplitude
is fitted Into a table arranged in segments, with
each segment divided into steps. The segments,
000, 001, ,.., are arranged sequentially as specified in the A-law. Amplitude A in the figure falls in
segment 011, which is divided into sixteen steps.
Amplitude A is therefore defined as step 8 in
segment 011. With the code converter developed
by Ericsson Security and Tele Systems AB the
segments are placed side by side and each
segment consists of 32 steps. Amplitude A can
therefore be defined in a number of segments,
and the specified segment can be retained during
several coding processes. The applicable segment is thus changed relatively seldom and
hence the number of transmitted bits can be
reduced to six, as against the eight bits required
for the A-law
cuit automatically chooses between two
different sensitivity levels. The maximum line length is 800m.
Start-up and programming of
the system
All exchanges are delivered with standard programming. Local alterations
and modifications are made by means of
strapping.
When the power is switched on, EXCP
automatically starts the exchange by
reading off all strappings, mapping the
physical positions of the PCBs and registering the types of auxiliary PCBs inserted.
All connected extensions are automatically allocated numbers in a consecutive number series, and the character display on each intercom set shows
the allocated number.
The system can thus be put into operation immediately, and will then work in
accordance with the PCBs and the
strappings that have been made in order
to modify the factory programming.
If the customer desires, special programming can now be carried out: for
example free number allocation, allocation for group hunting etc.
All essential programs and data are
stored in memories whose contents are
preserved if a power failure should occur.
Number transfers or number changes
can be programmed from one of the extensions using an ERICOM DIRECT with
a character display. In larger systems
the programming can be simplified with
the aid of a portable computer with a
special program, supplied by Ericsson
Security and Tele Systems AB.
Summary
ERICOM DIRECT is a technically advanced system which uses new technology and processors. The system is unconventional insofar as it requires practically no programming for small or medium-sized systems. Moreover, very
large systems can be built up from exactly the same equipment as small ones,
but using more extensive programming.
Sealed Lead-acid Batteries for Small
Telecommunication Plants
Hans Andersson and Sven Bergvik
Ericsson has developed a battery unit with sealed lead-acid batteries for 24 and
48 V. It is designed for mounting in a 19" rack and can therefore be placed in the
same rack as the other power or telecommunication equipment. Individual
voltage regulators ensure that each cell in the battery has the same voltage. This
gives high reliability and maximum battery life.
The authors discuss the advantages of sealed batteries as regards maintenance
and application, specify the types of batteries available in the market and
describe the design and characteristics of the newly developed battery unit.
UDC621 355
621 395:621.311
Battery maintenance is expensive. It
consists of regular checking of voltage
and density, topping up with water,
cleaning, greasing etc. If the site is remote the travelling costs will be high.
Sealed (gas-tight), maintenance-free
batteries would reduce the maintenance costs considerably. The saving
justifies the use of such batteries in spite
of the fact that the purchase price is
higher than for conventional batteries.
Another application for sealed batteries
is in plant where the battery and telecommunication equipment are mounted in the same rack and placed in an
office. Batteries that give off gases are
not acceptable for such applications.
Different types of
maintenance-free lead-acid
batteries
Fig. 1
Expected life (MTTF) for battery cells with trickle
charging and without temperature compensation
of the charging voltage
Fig. 2
Individual voltage regulators for each battery cell
Maintenance-free lead-acid batteries
with totally sealed cells can be divided
into three main types:
- Gel batteries, whose cells contain
conventional upright plates with calcium alloy lead in the grids. The electrolyte is gelatinized, with considerable surplus quantity. A safety valve
opens at a pressure of 0.5-1 kp. The
batteries are usually supplied in
blocks of 6V or 12V. The block containers consist of an ABS plastic
which allows the diffusion of gas. This
type of battery is also made with tubular positive plates for capacities of up
to 1350Ah.
- Batteries with round cells, in which
the generated gases are recombined
into water. The plates are wound spirals and contain pure lead in the grids
The amount of electrolyte is small and
is absorbed in the plates and separators. A safety valve opens at a pressure of approximately 4kp. The batteries are supplied as single cells or as
blocks. The cell container consists of
polypropylene, which prevents gas
diffusion. A metal sheath surrounds
the container.
- Batteries with rectangular cells, in
which the generated gases are recombined into water. The plates contain calcium alloy lead in the grids.
The amount of electrolyte is small and
is absorbed in the plates and separators. A safety valve opens at a pressure of 0.5-1 kp. The batteries are
usually supplied as blocks of 6V or
12V, but single cells are also available. The cell containers are normally
made of ABS plastics.
Characteristics of battery
cells
If all cells in a battery were identical they
would all get exactly the same voltage by
the trickle charging. However, there are
differences between individual cells,
caused by manufacturing tolerances,
ageing etc. Certain cells can be kept
fully charged by a fairly low current,
whereas others require a considerably
higher current. This shows up as a voltage discrepancy between the cells. The
voltage of a battery cell falls if the cell
does not receive sufficient current. If the
voltage becomes too low the plates become sulphated and the cell loses its
capacity. This phenomenon is particularly marked in maintenance-free batteries with low self-discharge.
223
HANS ANDERSSON
SVEN BERGVIK
Ericsson Power Systems
RIFA AB
Nominal voltage, V
Final voltage, V
Number of cells
Operating time
Load, W
100
200
300
500
1 000
2 000
4 000
Methods for equalizing the
cell voltages in a battery
48
43
23
h min
48
43
24
h min
24
21,5
12
h min
10
4 30
2 45
1 30
30
10
11
5 30
3 20
1 45
38
13
3
5 30
2 30
1
38
13
3
Table 1
Expected operating time as a function of the load
Fig. 3
Battery unit for 48 V, 25 Ah
Regular charging at raised voltage is
usually recommended for equalizing
voltage differences between cells in a
battery. However, battery manufacturers do not recommend periodic
equalizing charging for maintenancefree batteries. For these a relatively high
floating voltage is recommended instead, varying between 2.30 and 2.38V
per cell when the number of cells in the
battery is 12 or more. The current will
then be so high that even the worst cells
receive sufficient current for full charging. In the case of batteries with 24 cells
or more the manufacturers suggest a
division into groups of 6 or 12 cells,
which are charged independently12.
Disadvantages of high cell voltage
High cell voltage has the following disadvantages:
- The overall voltage is high. For example, with 23 cells and 2.38V per cell
the total voltage is 54.7 V, which is too
high for many applications. 2.25 V per
cell would give 51.8 V, which is an ideal voltage for most applications.
- The power consumption is high. With
54.7Vtheconsumption is 18% higher
than with 51.8 V.
- The difference between the working
and discharge voltages is large and
hence the number of cells in the battery has to be limited. This means that
the cells cannot be discharged sufficiently to ensure full utilization of
their capacity.
- The battery life is greatly reduced.
With 2.40V per cell its life is only half
the life of a cell at 2.25V, see fig. 1.
(This applies to, for example, cylindrical cells from Gates Energy Products2.)
Regulator for individual control of the
voltage of each cell in a battery
A radical solution to the problem of uneven cell voltages and high working
voltage is to have a regulator for controlling the voltage of each individual cell in
a battery. Ericsson has developed such
a regulator, and its function is illustrated
in fig.2. The regulator is placed directly
over the cell and has no outgoing connections. Its function is thus entirely independent of the number of cells in the
battery.
The regulator works as follows: Normally a current of approximately 150 mA
passes through a shunt consisting of a
resistor and a transistor. For a cell that
requires more current than a normal cell
the current through the shunt decreases
correspondingly. Foracell that requires
less current than a normal cell the current through the shunt increases. Each
cell thus takes exactly the amount of
current it requires to stay fully charged.
Extensive tests have shown that 2.25 V is
ideal and gives the maximum cell life.
At 2.25 V a fully charged cell of 25 Ah requires approximately 10 mA, and since
the shunt accepts current variations up
to 150 mA between the cells it is clear
that the system tolerates cells that are
radically different.
The transistor in the shunt circuit is controlled by an individual control circuit,
which senses the voltage across the eel I.
The power consumption in the shunt circuit is small compared with the power
saved as a result of the reduction in voltage obtained when a regulator is used.
Fig. 4
Printed board assembly with voltage regulators
and alarm circuits
Battery unit 24 V, 48 V, 25 Ah
The battery unit contains battery cells
and cell voltage equalizers with cell connections and undervoltage guards. It is
mounted in a rack in direct connection
with the units being supplied with
power. Installation is merely a question
of connecting the plus and minus cables
from a rectifier and cabling for any outgoing alarms.
Battery
The battery cells are manufactured by
Gates Energy Products and have spirally
wound plates. They work on the principle of recombining oxygen and hydrogen into water. The cells are gas-tight
with a safety valve. The self-discharge is
low. The nominal capacity is 25Ah. The
cells have very low internal resistance,
permitting a relatively high drain capacity even with a high discharge current.
Fig. 5
Cell voltages measured on batteries with 23 cells
for 25 Ah which have been in service for a
considerable period. The uniformity of the cell
voltages is a feature of the system
Highest cell voltage
Average voltage
Lowest cell voltage
Structure
The structure of the battery unit is
shown in fig. 3. The unit is available for
24V (12 cells) and 48V (23 or 24 cells).
The dimensions of a 48V unit suitable
for mounting in a 19" rack are
length 420 mm
width
220 mm
height 380 mm
weight 47 kg.
A printed board assembly placed above
the cells, fig. 4, contains all voltage regulators (one per cell), cell connections
and alarm circuits. In this way loose wiring is avoided.
Fuse and alarm
The battery unit includes an automatic
fuse and two undervoltage guards,
which in a 48V system work as follows:
- Guard no. 1 gives an alarm at 43 V and
lights an LED lamp. Aterminal is available for an outgoing alarm.
- Guard no. 2 disconnects the battery
by means of the automatic fuse at
38 V. The fuse is reset manually. The
alarm lights an LED lamp. A terminal
is available for an outgoing alarm.
Voltage levels, charging times
During operation the voltage should be
2.25 V per cell, which gives the following
overall voltages:
with 12 cells 12x2,25 V = 27,0 V
with 23 cells 23x2,25 V = 51,8 V
with 24 cells 24x2,25 V = 54,0 V
Raised voltage should not be used. With
2.25V full charge is reached within 24
hours. 8 0 - 9 0 % charging is reached in
one or a few hours depending on the
available rectifier capacity.
Storage
The battery unit can be stored for 6-12
months at 20-30° C without any damage
to the cells. The internal consumption of
the regulators is only about 250|iA,
which means a loss of capacity of approximately 2Ah per year. This is negligible compared with the self-discharge
of the cell.
Service, tests
The battery unit with its sealed cells requires no maintenance. However, in
plants of vital importance it may be necessary to check that the battery retains
its capacity. The most reliable method
would be to carry out discharge tests,
but such tests are expensive and can
interfere with the operation.
A good idea of the condition of the cells
can be obtained from the cell voltages.
The battery unit is equipped with an
easily accessible jack for reading off all
cell voltages. The unit is thus prepared
for output to a measurement processor.
Fig. 6
The cell voltage as a function of time, with a
discharge current of 12.5 A for a battery of 23
cells, of 25 Ah capacity
If all cell voltages lie in the range 2.242.26 V the cells can be considered as
new. If a cell shows a voltage as low as
2.20-2.21 V this means that it draws abnormally high current (10-15 times the
usual value). Tests have proved that
225
Fig. 7
The cell voltage as a function of time when
charging a battery with 23 cells of 25 Ah with the
current limited to 5 A
such a cell still has full capacity, but the
cell should be changed on the first suitable occasion, since there is a risk of further deterioration.
When a new cell is inserted in a battery
unit it does not necessarily have to be
fully charged. The cell will be charged
after a time in the battery unit.
Figs. 5, 6 and 7 show curves for a battery
unit obtained using a measurement processor.
Fig. 8
The battery unit mounted in a rack with rectifiers
Applications
Voltage equalization gives the battery
unit remarkably high operational reliability. The battery can w o r k at temperatures d o w n to - 4 0 ° C , and can w i t h stand occasional temperatures up to
4 0 - 4 5 ° C w i t h o u t noticeable shortening
of its life. The battery life is estimated to
be 8 - 1 0 years at 2 0 - 2 5 ° C , and approximately 4 years at 35°C continuously.
Two or more battery units can be connected in parallel. The installation is
easy, the unit being m o u n t e d in the
same rack as the electronic equipment,
fig.8. The battery cells are classified by
IATA (International
Air
Transportation
Association) as " d r y " , w h i c h means that
they can be sent by air w i t h o u t restriction.
These characteristics give the battery
unit a wide field of application, and it is
particularly suitable for small rural exchanges, repeaters on long-distance cables, radio relay link, PBXs, military applications etc.
Operating time
The operating times of different units,
with different loads, are s h o w n in tablei.
References
1. Accumulatorenfabrik
Sonnenschein
GmbH, Prospekt Nr. 7023, p. 10.
2. Harrison, M. R., Gates Energy Products,
Denver,
Colorado,
USA:
Considerations in the Application of
Sealed, Recombining Recharge Leadacid Batteries to Remote Installations
and UPS Systems on Telecommunications Networks. International Telecommunications Energy Conference, Oct.
3, 4, 5, 6, 1982, Washington DC. USA,
pp. 424-428.
Computer-controlled System for
Road Traffic Control
Knud J. Larsen and Hans Jorn Nielsen
Dansk Signal Industri AS is the member of the Ericsson Group responsible for
developing road traffic control equipment. In this article a co-ordinated traffic
control system is described, with emphasis on the centre that carries out the
control and supervision of the traffic signals throughout a town or city. The main
characteristic of Ericsson's co-ordinated traffic control system is the extensive
use of computers, and the authors describe the possibilities and advantages of
this technology.
UDC 625 746 5:681 3
The intensive use of the road network in
and around large towns and cities results in demands for traffic control systems to handle traffic efficiently. The requirements for such a traffic control system are stringent, both as regards the
actual traffic control and the possibilities of efficient maintenance of the
system. Ericssons co-ordinated traffic
control system is divided into three levels, each with its own tasks as regards
safety, control and supervision, fig. 1. At
the local level the control is carried out
by a local controller, usually controlling
a single intersection. At the middle or
area level master controllers coordinate
up to 32 local controllers, and at the top
level a centre can supervise and control
a large number of controllers, covering
a whole town. This centre can coordinate the traffic at the intersections in a
larger or smaller area depending on the
amount of traffic or the time of day.
The basic philosophy behind the division of the traffic control system into
levels is that functions that ensure the
safety of individual road users are
placed as close as possible to the signal
device, i.e. at the local level. Functions
for optimizing the traffic handling, on
the other hand, can be placed at any
level, depending on the size of the area
to be optimized. The main advantage of
such a structure is that the safety of the
traffic handling is retained, even if parts
of the system are out of operation. Furthermore, the use of processors at all
levels means that even if a fault occurs at
a superior level the largest possible part
of the system will operate optimally on
its own. At certain times of the day it is
also best to optimize the traffic at the
local level. On the other hand, superior
levels, particularly the centre at the top
level, afford the best facilities for efficient maintenance. For example, faults
can be reported efficiently and modifications of the system functions can be
carried out from the centre.
Fig. 1
Ericsson's co-ordinated traffic control system
with three control levels. The levels are:
1. Control of individual intersections, using local
controllers
2. Control of small areas by master controllers
3. Control and supervision of a whole town or city
from a centre
1983
227
KNUD J. LARSEN
HANSJORN NIELSEN
Dansk Signal Industri A/'S
Hvidovre, Denmark
Fig. 2
The detector sites in a typical intersection are
shown in blue. The geometrical layout of each
detector site determines which vehicles will be
detected, and whether it is the passage or presence of the vehicle that is detected
Local control
The purpose of the local controllers at
the individual intersections is to ensure
that the traffic is managed with complete safety, and that the road users
have the shortest possible waiting time
at the intersection. Among other things,
the local controller can adjust green periods according to the number of vehicles in the different directions For this
purpose a number of electronic vehicle
detectors are embedded in the road surface. Fig.2 shows a typical intersection.
It is obvious that the traffic handling depends on the detectors (which can be
placed up to 200 m from the controller)
functioning correctly, and hence the
controller automatically monitors their
operation. The signal lights are also
monitored, and thisfunction can also be
used as a safety measure, so that green
light can only be given when the red
light is actually on in the conflicting directions. The system contains several
other supervision functions, and if a serious fault is detected the signals are set
to the state that is the safest possible for
the road users, and an alarm is sent to
the superior level.
Local controller JCF 150, fig.3, includes
a microprocessor, which contains all
control logic. This enables the control-
Fig. 3
Local controller type JCF 150. The microprocessor is of type Intel 8080A, with a 44 kbytes
EPROM and an 8 kbytes RAM
ler to meet the many and varied demands made on each individual unit,
both during installation and any subsequent modification. Moreover, adaptation to the requirements of other countries can easily be implemented.
Coordination within an area
In order to obtain efficient traffic handling throughout a large area it is necessary to coordinate the changeover times
in a number of local controllers. This
creates green waves. This function is
carried out by master controllers, each
of which can coordinate up to 32 local
controllers. The master controller can
select different timing plans, i.e. plans
for signal changeover times in each local controller, and thereby adapt to the
traffic flow pattern. The choice of timing
plan can be made on the basis of actual
traffic density (based on detector readings) or time (time of day plus day of the
week).
The master controller is connected to
the local controllers via transmission
lines over which synchronization signals are sent to the local controllers and
information is received regarding the
status of the controllers and each individual signal.
Fig. 5
Operator's position in a centre with a graphic
colour display unit and a printer. The display unit
shows traffic density measurements at different
points in the controlled area
The master controller, JCC 150, fig. 4, is
bmit up around a microprocessor which
contains all the available timing plans
and the logic necessary for choosing
timing plans for the area. This gives the
master controller flexibility to adapt to
different requirements, just as the local
controller JCF150.
Control centre
The top level in Ericsson's co-ordinated
traffic control system consists of a control and supervision centre. There are
several versions of this centre, series
JCC16X, and the choice is dependent
on the traffic regulation algorithms to be
used. The centre, which is computerbased, communicates with a number of
master and local controllers, so that the
whole system can be controlled and supervised from one centre.
The three main tasks in a large traffic
control system are to:
- optimize traffic handling
- maintain the equipment
- establish new controllers.
The main advantage of a superior control and supervision centre is that the
execution and supervision of the above
tasks can be carried out centrally, since
the centre collects all the information in
the system and presents it in a clear and
comprehensible way. It must be noted
that even if the system contains only one
central computer the presentation of information can take place simultaneously in many places, for example
Fig. 4
Master controller type JCC 150. The microprocessor is of type Intel 8080A, with a 3 kbytes
EPROM and a 17 kbytes RAM
Fig. 6
A display which continuously shows the situation
in an area. The name of the area and the selected
area timing plan are shown at the top, together
with the current timing plan step. The picture
shows a plan of the area, and for each intersecion
the status of the local controller and the indication being given by the most important signals
are also shown. The input commands are shown
at the bottom of the display
1983
Fig. 7
A display picture which continuously shows the
situation at an intersection. The display is divided
up in the same way as in fig. 6. The state of each
individual signal is shown, and also whether the
detectors are occupied or not
for the service staff, police, traffic engineers etc., since video display units, either monochrome or colour, are used
for the presentation, fig. 5.
The following functions are performed
by the centre:
- System supervision
- Presentation of information
- System control
- Data recording
- System modifications.
System supervision
The supervision entails automatic collection of information concerning the
status of all units and devices in the system, and selecting information that indicates changes in status, or particularly
important information, for example fault
Fig. 8
A display picture which shows the current situation in a small area in which the signals are
coordinated in a green wave. The picture is
divided up in the same way as in fig. 6. On the lefthand side a stylized picture of the road is shown,
the distances between the intersections being
proportional to the real distances. The display
then shows, level with each intersection, what the
most important signals have indicated during the
last 120 seconds. Signals in the direction of the
green wave are shown in green, and signals in the
opposite direction in violet. The speed of the
green wave is easily read off as the gradient of a
line that passes through the front edge of each
green line
reports or messages concerning the
traffic in the parts of the road network
with the highest load.
Presentation of information
The collected information is presented
to the operators in an easily understood
manner. Video display units are used,
with the picture layouts designed so that
the information concerning the state of
the system is clear, for example by using
flashing to indicate particularly important information, such as status
changes. All texts are in clear language
as decided by the customer, and therefore use the expressions and designations which the customer uses daily. The
customer can also choose the formulation of commands, which makes the system very adaptable to the user's needs.
230
Fig. 9
The transmission system divided into three levels. Data for all controllers in a certain area are
transmitted between the master controller or
concentrator and the centre. Up to 50 master
controllers or concentrators can be connected to
a centre. Concentrators are used where no microprocessor-based master controllers are installed,
or where no coordination is required. Up to 8
local controllers can be connected to each of the
four transmission lines to a master controller or
concentrator. Microprocessor-based controllers,
such as JCF150, can be mixed with older controllers connected via JCA151
The picture layout and operating functions are the same irrespective of
whether m o n o c h r o m e or colour display
units are used. Graphic display units can
show exact plans of the road network
and individual intersections, figs. 6 and
7. Colour is used to emphasize traffic
signals, with the current signal states
shown in the actual colours. Detectors
and traffic measurements are also emphasized by colour displays. Fig. 8
shows an important use of graphics and
colour with the signal indications in a
number of intersections s h o w n graphically. This makes it possible to plan and
control green waves, since distances
and times can easily be read off from the
display.
times in the controllers in the group. The
timing plan to be used is selected on the
basis of a c o m b i n a t i o n of traffic flow
measurements made at various strategic places in an area. Timetables can
also be selected with respect to the time
of day and day of the week, or for a
particular type of day, such as a holiday
that falls on a weekday. It is also possible
to use a display unit to select a timing
plan. The c o o r d i n a t i o n of different
groups is arranged with the aid of a special
function-linking-whereby
the
choice of a t i m i n g plan for a group can
affect the choice in othergroups.sothat
overall the system continuously adapts
to the traffic. The decision criteria described here can be supplemented by
custom-made algorithms.
System control
It is particularly important, especially
when optimizing the traffic handling,
that modifications of the controllers in
the network can immediately be initiated from the display unit. This is usually
only possible in microprocessor-based
controllers. In older controllers the control from the centre is limited to changing the operating mode, for example setting all signals operated by a controller
to amber flashing. Centres of type
JCC162 include traffic regulation algorithms. The connected local controllers can be grouped arbitrarily, and for
each group timing plans are prepared
that coordinate the signal changeover
Data recording
One important task of the traffic control
system is to record information for later
use, for example for statistics or for analyzing a specific situation. The centre is
equipped with a disc store for storing
such i n f o r m a t i o n , and a magnetic tape
station can also be included for longterm storage of large quantities of data.
The stored information can be recalled
for statistical purposes and for presentation f u n c t i o n s , w h i c h can show signal
changes and detector information as if
they were taking place at that moment.
The speed of such presentation can vary
between frame freezing and four times
the original speed.
Centre
Master controller
Concentrator
Local controller
Older local
controller
1983
231
User functions
Network manager
Protocol administration
Line driver
Physical link
Fig. 10
The different communication levels in the transmission system. The advantage is that the upper
levels, particularly the user functions, need not be
changed even if the lower levels have to be
modified, for example, to adapt to another transmission protocol used in a part of a network, or
another type of physical signalling
System modifications
Any traffic control system will have to be
modified continually, and it is necessary
to ensure that such modifications are
made in the most efficient way. The central computer is used for this purpose. A
complete set of data, which describes
each individual unit and device, is
stored in the computer, and it is thus
possible to follow up every physical
change in the system with a modification of the data. In the microprocessorcontrolled units the updated information must be input in the processor, and
this is done by sending it over the transmission line. The new controller is then
in operation.
System configurations
The various units in Ericsson's co-ordinated traffic control system have been
described above, but the main nerve of
the system is the transmission system
that connects these units together,
fig. 9. The transmission system, which is
designed on the basis of experience
gained regarding datatransmission networks, is built up in several levels, each
with its own function in the communication process, fig. 10. The great advantage of such a structure is that the lower
levels can easily be modified individually for special physical conditions, such
as connection to a direct cable or to a
leased line via a modem. This structure
also makes it possible for a whole network section to communicate by an entirely different transmission protocol
since the lower levels can be exchanged
for others having the necessary special
functions without the operation of the
network being affected. The method of
transmission used in the network means
that the data are divided into packets
with added redundancy for fault detection purposes. This method is in accordance with CCITT Recommendation
X.25. When a fault is detected in a received packet a request for retransmission is sent to the transmitter, therefore
the probability of faults not being detected by the transmission system is
very small. Ericssons new microprocessor-based controllers can of
course be connected to the transmission system direct. Older controllers
(electronic or with relays) must be connected via a special transmission adapter, JCA151, fig. 11, which can be adapted to each individual controller. The
transmission adapter is also controlled
by a microprocessor (Intel 8085) and is
therefore extremely flexible.
System installations
The first centre, type JCC161, will be
installed in Stockholm, Sweden, in 1984.
The computer is a Digital PDP-11/44,
equipped with three graphic colour display units for the operators. Two of the
display units will be installed in a separate alarm centre, situated 8 km from the
main centre.
Fig. 11
Transmission adapter JCA151. This unit can be
plugged into a local controller. It has 32 relay
inputs and 32 relay outputs, which are connected
to the controller via a cable. The small printed
board assembly mounted on the large board
contains devices for adaptation to the physical
transmission line, and it can be changed if
another line has to be used, for example a leased
telephone line. The microprocessor is an Intel
8085, with an 8 kbytes EPROM and a 2.25 kbytes
RAM. The same printed board assembly is used
for concentrators, but it is then mounted in a
shelf
The centre will be equipped with a disc
store with large capacity (122 Mbyte),
and a magnetic tape station for recording purposes. Printers will also be included for printout of changes in the
operational status of the controllers,
fault reports and other recorded data.
In the first stage the centre will be connected to approximately 200 controllers, fig. 12, but it can be extended to
approximately 600 controllers. Since
most controllers are of older types,
Fig. 12
Map of central Stockholm showing the individual
controllers. The colours indicate the three stages
ot connection to the centre:
Pilot system
First stage
Second stage
Other controllers
transmission adapters of type JCA151
will be used to connect them into the
transmission network. An existing network of direct cables from the centre to
the master controllers and on to the local controllers will be used for the transmission.
Summary
The use of computers in traffic control
systems has greatly simplified the operation of such systems. This in its turn
leads to improved traffic flow and reduces fuel consumption and road building costs.
Ericsson's co-ordinated traffic control
system comprises:
- Local controllers JCF150 and
JCF151
- Master controllers JCC 150
- Centre, series JCC 16X
- Transmission adapters JCA151
- Detectors JCD191
- Signalling equipment.
All functions in the system are performed by computers and microprocessors. The system is therefore extremely flexible and can be adapted to
suit the requirements of different countries.
1983
Ericsson's First
Automatic Telephone
Exchange 100 Years
Gosta Thames
It is now 100 years since Lars Magnus Ericsson and Henrik Tore Cedergren
obtained a patent for an automatic telephone exchange. The author, who
recently retired from Ericsson, is now researching the technical history of the
company, and after studies in museums and archives he has compiled a
description of the hundred year old telephone exchange.
UDC 621.395.34
In 1883 Lars Magnus Ericsson and
Henrik Tore Cedergren were granted a
patent for an automatic telephone exchange, fig. 1, and an automatic digit
pulse transmitter. Comments in contemporary literature were very complimentary about the exchange:
"... dieser ausserst sinnreiche Apparat
"1
"... contains many points of great ingenuity ..." 2
"... une des inventions les plus ingenieuses parmi celles ..." 3
"... they are the best of their kind ...""
Fig. 1
Ericsson's first automatic exchange, for ten lines
GOSTA THAMES
Telefonaktiebolaget LM Ericsson
Stockholm
As far as is known the EricssonCedergren exchange was the first commercially available automatic telephone
exchange in the world. It was widely distributed, and probably more than 300
units were manufactured.
The main features of the exchanges
were
- possibilities for both internal and external calls
- complete secrecy during calling as
well as conversation
- busy indication
- remote control by the operators at the
central exchange
- one main line and up to ten local lines
- identical telephone instruments, handled in the same manner, for both direct and automatic lines.
In 1880 the Stockholm Bell Telephone
Company (Stockholms Bell Telefon-Aktiebolag) was formed. H. T. Cedergren
decided to form a competing telephone
company, which was named Stockholms Allmanna Telefon A.-B. (SAT).
Fig. 2
Automatic digit pulse transmitter intended tor
exchanges of version 2
Naturally enough Cedergren did not expect the Bell Group to supply him with
telephone equipment, and instead the
29-year old engineer tried to convince
Lars Magnus Ericsson that they could
compete with the powerful International
Bell Telephone Co.
If Cedergren and SAT were to succeed,
telephone tariffs would have to be set
much lower than those of the Bell Company. One way was to let several subscribers share a common line.
The first automatic
exchanges
Fig. 3
Automatic exchange, version 2, for five lines
Three different automatic exchanges
were displayed at the Exposition Internationale d'Electricite in Paris in 1881.
Maybe they gave Cedergren some ideas.
In any case Ericsson worked on an automatic exchange in 18826, and on February 10th, 1883, Cedergren and Ericsson
applied for a patent, which was granted
during the same year". The exchange
had the disadvantage that the subscribers connected to it could not talk to each
other, only to the subscribers at the central exchange. An extra, common conductor was also required which connected the telephones to the automatic
exchange.
These problems were quickly solved,
however, and already in July Cedergren-Ericsson were able to submit a
new application, which resulted in a new
patent. Subscribers connected to the
new exchange were now also able to
speak to each other, and the common
conductor was no longer required,
fig.3. 712
The improved exchange was ready to be
shown at the Vienna Exhibition, which
opened in August, 1883s. A.L. Paul acquired the English rights, and as early as
November the English magazine The
Telegraphic Journal reported that "Mr.
A.L. Paul, of The W.T. Henley's Telegraph Works Company, Limited, has a
set of apparatus at work in the office of
the company, at 8, Draper's Gardens,
and can give every information on the
subject." 2
Before long (in 1884) further imporvements had been made, and it is this third
version that achieved such wide distribution fig.4. 3 9 1 0 1 3
In addition there was a two-party line
relay which was sometimes called an exchange. However, this relay will not be
covered in this article.
A misunderstanding
The article mentioned above states that
Paul is the owner of the patent for this
country.2 In Paris the English article was
considered so interesting that within a
fortnight it was reported in La Lumiere
Electrique under the headline "Telephone multiplex de M. A.L. Paul"'7 and
later Elektrotech. Zeitschrift mentioned
the exchange "patented by A.L. Paul...
in England" 18 . It was clearly these references that misled Dr. Rothen of the
Swiss Telegraph Administration when
he described the latest version of the
exchange as follows: "One finds, in Mr
Ericsson's and Mr Cedergren's equipment, all the details that are in A.L.
Paul's equipment, plus additional ones,
which makes the former one of the most
brilliant inventions ..." 3 . That the "Paul
exchange" was in reality an Ericssonexchange is confirmed by J.E. Kingsbury,
among others. He describes "... a system the English rights of which had
been acquired by Mr. A.L. Paul. This,
according to my recollection, was the
production of Ericsson of Stockholm
"19
The range of application of
the exchanges
In May 1886 SAT reported that the company had more than 150 exchanges in
operation in the Stockholm area, a number which later rose to approximately
17514. Switzerland had 45 exchanges15,
and the exchange was also in use in
Norway16. The exchanges were certainly
in service in other countries as well, and
the number produced was most likely in
excess of 300.
1983
235
Fig. 5
50-line switchboard without multiple jacks. The
first SAT telephone exchange, 1883-1887
Fig. 4
Automatic exchanges, version 3, for five, seven
and ten lines
As has already been mentioned, the
main reason for designing the exchanges was to reduce line costs.
However, as the number of subscribers
increased and the telephone network
grew, new problems arose, such as:
- limited capacity of the central exchanges
- crosstalk
- concentration of overhead lines at the
central exchanges.
Increasing the capacity of the central
exchanges
Each switchboard usually had 50 lines.
Since the multiple principle had not yet
been developed, the connection to a
subscriber in another switchboard in
the exchange was made through a link.
To put through a call, an operator had to
shout to a colleague and announce the
number of the link and that of the subscriber being called. In order that the
distance between operators should not
be too great, the switchboards were arranged in a square and the number of
boards was limited to 16-20 114 , fig. 5.
Despite this the noise level was very
high.
Asthenumberof subscribers increased,
the telephone companies were forced to
build more exchanges. Connecting subscribers via automatic exchanges was a
feasible alternative. For example, the capacity of an 800-line central exchange
was increased by 200 if 50 five-line automatic exchanges were connected as
satellites.
Fig. 6
Winter view of the largest telephone tower in
Stockholm. It was taken into service in 1887
Crosstalk over "long-distance" lines
The network technology applied in the
early days of telephony was a carry over
from telegraphy. The subscriber line
consisted of a single conductor, with
earth as the return circuit. The pairing
and twisting of lines were unknown concepts. When networks were extended
the problem of crosstalk arose.
All telephone companies had the same
experience, and in its annual report for
1883 the Swiss Telegraph Administration says: "Since it has not yet been possible to use two or several telephone
wires that run close to each other over
long distances because speech is transmitted from one wire to another, only
one telephone wire can be run between
two communities" 15 .
The lines did not have to be very long
before crosstalk occurred, as the following extract shows: "If two single telephone lines are mounted on common
poles over a distance of barely a couple
of kilometres ... Even if the lines are run
on opposite sides of a broad highway
and continue farenough-10 kilometres
or more-similar induction is noted" 20 .
Before the two-wire system (metallic circuit) was introduced, subscribers in remote communities could only obtain a
"non-party-line" connection to a central
exchange through a line to a satellite
exchange served by an operator. When
there were only a few subscribers, operating costs were very high, and night
service was out of the question. With an
automatic exchange both problems disappeared, at least in the Stockholm
area, where the central exchanges
provided 24-hour service from the middle of 1884.
The first automatic exchanges developed by Bell System were the result
of the same problem of service and cost
of telephone exchanges in small communities21.
Reducing the number of lines
The first multiple exchange produced by
Ericsson was taken into service in the
middle of 1884. It was so successful that
SAT started engineering a large exchange for Stockholm in the same year.
This exchange was fully operative in
July 1887 and was at the time the largest
in the world, with 4000 connected lines
and an ultimate capacity of 7000.
Larger telephone exchanges have been
built since then, but never a larger tower
for telephone wires, fig. 6". In time, developments in the cable field made the
telephone tower superfluous. Until this
,1983
237
occurred, however, it was a good thing
that the number of wires could be reduced with the help of automatic exchanges.
Setting up calls
The same telephone instrument was
used irrespective of whether the subscriber was connected to the central exchange via a direct line or an automatic
exchange. Nor was there any difference
in the way calls were handled, or in the
telephone numbers. A call to the central
exchange was made by means of the
signal generator in the telephone instrument, which also caused the bell on the
own telephone to ring. The bell did not
ring if the automatic exchange was engaged. When the operator had connected up the desired number she announced: "Ready!" The subscriber then
had to ring again. This was done to reduce the work of the telephone operators, since there were no powered signal
generators in those days. The subscriber had to give a strong second signal.
There was no point in giving an extra
signal in order to attract the attention of
the called party, because the third signal
always serve as the disconnect signal,
which meant the end of the call.
Fig. 7
These telephone sets from the early 1880s were
used with the automatic exchanges
In the central exchange number register
ten numbers were reserved for each automatic exchange, and a corresponding
jack in the switchboard was marked with
a multiple of ten. For example, when
connecting up a call to subscriber 2387,
who was connected to an automatic exchange, the operator connected up to
the jack marked 2380 and used a digit
transmitter to send out seven impulses
before she announced: "Ready!" When
the disconnect signal was received from
an automatic line the cords were disconnected in the ordinary way, but the operator also had to press a button to restore
the selector to the zero position, unless
the central exchange was equipped with
a resetting device, fig. 15, that did this
automatically.
Contemporary exchange
designs
In his report for 1882 the chief engineer
of the Swedish Telegraph Administration wrote as follows: "The problem of
replacing the personnel required in a
central exchange by means of a self-
functioning switching instrument was
solved theoretically as early as 1880"6.
He is referring to the two Connelly brothers and McTighe, who patented the first
automatic telephone exchange in 1879,
closely followed by Westinghouse Jr.
Fourother Americansand two Belgians,
Bartelous and Leduc, applied for patents before Ericsson and Cedergren did.
After them, many others sought new
ways to solve the problem of automatic
connection. "... many of the fundamental ideas came from inventors who were
without technical training or practical
telephone experience, and whose mechanical arrangements for embodying
their ideas were apt to be impracticable
or unworkable" 22 . With the manufacturing technology then available, the difficulties of achieving adequate precision
were probably substantial. Here Lars
Magnus Ericsson, as a manufacturer,
had a great advantage because of his
wide experience in precision mechanics, combined with brilliant design
skills.
The automatic exchange is included in
the first three editions of the Ericsson
catalogue (1886-1892), but does not
appear in the fourth (1897) edition. As
has already been mentioned, most of the
exchanges were in operation in the SAT
network. When this was converted to the
two-wire system (in 1895) the exchanges
could no longer be used and were removed. In other places the exchanges
are said to have been in operation well
into the 20th century23.
A corresponding automatic exchange
for two-wire networks was patented by
Ericsson, but not until 189924. It was only
used to a limited extent, which confirms
the view that the automatic exchanges
were used primarily to solve the
crosstalk problem.
The design of the automatic
exchange
As has already been mentioned the exchange was modified successively.
Version 1, fig. 1, did not contain line relays, and hence the subscribers connected to the exchange could not talk to
each other. The design of the selector
permitted the connection of 15 lines, but
only 10-line exchanges are still in exis-
Fig. 8
Overvoltage protection
tence. Exchanges with 15 as well as 25
lines were mentioned, but cannot be
verified in contemporary documentation.
Version 2, fig. 3, had individual line relays but the circuit diagram was such
that each internal call combination required a selector position of its own.
With n lines the required number of sem +1
Thus the
lector positions was n
15 selector positions only permitted the
connection of five lines.
Fig. 9
Selector for seven lines
Version 3, fig. 4 also had individual line
relays, but the design had been improved so that internal calls no longer
required special selector positions. In
the home position of the selector all subscriber lines were connected together
via individual connectors. For mechanical reasons this limited the number of
lines to ten Variants with five and seven
lines were also made.
The components common to all versions were
- a selector
- a line relay (galvanometer) for the
main line.
Fig. 10
Galvanometer relay with the oscillation brake to
the right and the correction magnet at the bottom
All equipment was enclosed in a hardware cabinet with a glass door. The lines
to the central exchange, subscribers
and earth were connected at the top of
the cabinet. Some exchanges were
equipped with a lightning protector,
fig.8. The semicircular upper part was
then covered with a metal strip connected to the earth screw. All line screws
had springs which exerted pressure on
the earth strip. A ribbed tape provided
the insulation between the strip and the
line screws.
Selector
The selector was a step-by-step mechanism, fig. 9. It had two electromagnets,
one with a driving latch and one with a
locking latch, both operating against a
ratchet wheel. When the selector was
stepped, tension was created in a return
spring, so that the selector returned to
its original position when the locking
magnet was energized. This made for a
highly reliable selector, since it always
started from the some position. The
principle was later used in the selector
systems in railway telephony.
Galvanometer relay
The galvanometer relay consisted of a
coil, at the middle of which was a magnet system capable of a to-and-fro
movement, fig. 10. The latter had three
parallel magnetic bars, one inside the
coil and the other two along the outer
sides of the coil. Another magnet was
fastened parallel to and below the magnet system. By turning this magnet the
effect of the external magnetic field on
the magnet system could be compensated.
The relay functioned as a three-position
polarized relay, which returned to the
central position when the coil was not
energized. In each of the outer positions
a contact was made. The contacts were
designed so that they were made during
the greater part of the turning movement of the magnet system. A brush attached to the upper bar magnet braked
the movement of the magnet system
when it returned to the central position.
The brush passed through a device similar to a metal comb, thereby preventing
the magnet system from swinging back
and forth.
The relay was always connected in series with the line. This meant that, in
addition to the positive and negative
control pulses, the speech current and
ringing signal also passed through the
relay. The effect of the relay on the
speech and ringing signals was negligible, however, since the relay resistance
was very low, 35 S.U. (One Siemens unit,
1 S.U. corresponds to—-rrohms). Fur1 .UD
Fig. 11
Line relays with balancing weights in the version 2 automatic exchange
thermore the relay did not react to the
low-frequency alternating current of the
ringing signal.
1983
i-ig. i *
Double key pulse generator
Line relays
As has already been mentioned, version
1 of the automatic exchange did not
contain line relays.11
Fig. 13
Impulse spring with twin contacts in the automatic digit pulse transmitter, from 1883
In version 2, a line relay was connected
in series with each line, fig. 11. The relay
was included to prevent the bells of
other telephones connected to the exchange from ringing when a call was
made to the central exchange. The line
of the calling set was short-circuited via
the relay contacts so as not to attenuate
the ringing signal. The relay armature
was equipped with an adjustable counterweight in order to prevent it acting as
a buzzer with an automatic interrupter.
The relays had to be adjusted individually to the connected line.
The line relays in version 3 were also
connected in series with the line, and
their purpose was the same as in version
2. The operation was quite different,
however. When the caller's line relay operated it was held mechanically, which
meant that the line bypassed its own relay, and in all other relays the connections to the telephone sets were broken.
The resetting relay released all line relays when the selector was reset after a
call had been completed.10
Fig. 14
Ericsson's first rotary dial, from 1895
Fig. 15
Resetting device with the cover removed
Pulse generators
The simplest pulse generator used was
the so-called double key, a telegraph instrument with two keys, one for positive
pulses and the other for negative, fig. 12.
When the positive key was depressed
the negative pole of the battery was connected to earth. Depressing the negative
key gave earthing of the positive pole.10
It may be noted that Strowger also used
telegraph keys in his first exchange.
An automatic digit transmitter, fig. 2,
was included in the first patent.11 It had
many features in common with the telephone dial of today: rotary pulse cam,
speed regulator, release block and a spiral spring as the drive mechanism. Apart
from the fact that it had no contact for
short-circuiting the speech circuit this
digit pulse transmitter could be used in
present-day automatic exchanges.
The high remanence of the relay iron
then available meant that the release
times were longer than was desirable.
This problem was solved by short-cir-
239
cuiting the selector stepping magnet
after each application in order to remove the stored energy quickly, or, as it
was expressed in the patent, "with a
view to discharging the secondary current". This was achieved by equipping
the digit pulse transmitter with two
pulse wheels, one of which gave pulses
with a ratio of 40/60. This ratio is still
used by Ericsson. The cams on the second pulse wheel were displaced in relation to the first wheel and provided very
short pulses of opposite polarity. The
above-mentioned telegraph key works
in accordance with the same principle.
The pulse springs of the digit pulse
transmitter had twin contacts, fig. 13.
This appears to be the first time twin
contacts were used in telephony. In this
case the contacts were made of pure
platinum, but other contacts consisted
of a silver alloy with 10% copper, an
alloy which is still used today.
Each digit (1-10) had a pawl that could
be raised. A number was dialled by raising the pawl for the desired digit and
depressing a button at the lower edge of
the box. This released the pointer arm,
the sending of pulses started and continued until the arm stopped at the raised pawl. Before a new digit was selected
the arm was turned back to the starting
position and the pawl was turned down.
The knob on the top of the box was used
to reset the exchanges after a call. A
version with 15 positions, 1,2,3,4,5,1-2,
1-3 etc. was produced for version 2 of
the exchange.
A rotary dial was produced in 1895. Except that only the digits from 1 to 5 could
be dialled it had all the functions of present-day dials. It was also equipped with a
locking device which held the dial in the
set position. The lock was released electrically. The method of operation was
probably as follows: When a caller requested a number, the operator dialled
the number and the dial was locked.
When the operator then plugged in the
B-line the lock was released and the
pulses were transmitted automatically.
It is not known how common this dial
was. Only one dial appears to have been
preserved.
Resetting device
When the operator received a disconnect signal she had to remove the two
Fig. 16
A contemporary diagram of an automatic exchange, version 31
plugs. If the line went to an automatic
exchange she also had to send a negative pulse, so that the exchange selector
and relays could be reset.
Exchanges with many automatic satellite exchanges were equipped with a resetting device. Probably its purpose was
to facilitate the work of the operators by
making the disconnection procedure
the same irrespective of whether a direct
or an automatic line was used. It has not
been possible to find a connection diagram, so the exact function of the resetting device cannot be determined.
Each automatic exchange was connected to an individual contact bank in
the resetting device, and a maximum of
100 exchanges could be connected,
fig. 15. The contact arm, which was connected to negative potential, rotated
continuously over the contact banks
and a negative pulse was thus sent to
each bank. Contemporary relay technology was sufficiently advanced to control
the pulses so that they were only sent
out on lines with disconnect indications,
and only one pulse was sent.
Fig. 17
A Leclanch6 unit
Three resetting devices have been preserved, together with Lars Magnus
Ericsson's assembly drawing, in his own
hand. They were probably produced
around 1884-1885.
With a base of pearwood and its mechanical parts made of ornamented
brass, this device was typical of the
pains Ericsson personally took in designing his products to make them attractive-long before industrial design
became an accepted concept.
Power supply
The automatic exchanges received their
current from batteries situated at the
central exchange, fig. 17. The battery
consisted of 35-60 Lechlanche units,
which together gave 50-90V. The voltage used was determined by the line resistance10. The automatic exchanges
drew current only when calls were being
connected or disconnected. Each telephone set had its own microphone battery, as was the case with all carbon microphone instruments at the time.
References
1. Maier, J. och Preece, W. H.: Das Telephon
und dessert praktische
Verwendung.
Stuttgart 1889. pp. 186, 215, 219, 231-232
303-307.
2. The Telegraphic Journal and Electrical Review. Vol. XIII (1883), pp. 382-384.
3. Rothen, T.: Etude sur la tel£phonie. Plusieurs stations sur un seul HI. Journal Telegraphique Vol.XI (1887), pp. 193-195.
4. Bennet, A. R.: The Telephone Systems of
the Continent of Europe. Longmans, Green
and Co., London 1895; Reprint by Arno
Press, New York 1974, pp. 294, 354,358 and
382.
5. Chapuis, R. J.: 700 Years of Telephone
Switching (1878-1978). Part 1, North-Holland Publishing Company, Amsterdam-New
York-Oxford 1982. pp. 58.
6. Wennman, M.: Till Kongl. Telegrafstyrelsen
afgifven berattelse ofver framstegen inom
telegraftekniken under ar 1882. (Report to
the Royal Swedish Telegraph Administration on progress made in telegraphic technology in 1882.) Stockholm, March 31,
1883, pp. 16-19.
7. Ofveringenidrens
till Kongl. Telegrafstyrelsen
afgifna
berattelse
diver
framstegen inom telegraftekniken under aV
1883. (Chief Engineer's Report to the Royal
Swedish Telegraph Administration on progress made in telegraphic technology in
1883.) Stockholm, March 26, 1884, pp. 2329.
8. Ibid., appendix Nystrom, C.A.: Berattelse
om internationella Elektricitetsutstallningen i Wien Sir 1883. (Report on the International Electrical Exhibition in Vienna, 1883.)
Stockholm. October 31, 1883, pp. 15.
9. Ibid., 1884, Stockholm. March 25,1885, pp.
20-22.
10. Ibid., 1887, appendix: Automatisk Telefonvexel. (Automatic Telephone Exchange.)
Stockholm 1887, pp. 1-9.
11. Swedish patent no. 208/1883. Application
dated February 10, 1883. Issued June 23,
1883. English patents No. 2025 and 5008,
A.D. 1883. French patent No. 154.885. German patent D.R.P. 27703.
12. Swedish patent no. 398/1883. Application
dated July 13, 1883. Issued November 2,
1883.
13. Nystrom,
C.
A.
och
Wennmann.:
Description d'un "distributeur automatique", invente par MM. L. M. Ericsson,
mecanicien, et H. Cedergren. ingenieur
civil. Journal Telegraphique Vol.X (1886),
pp. 145-147.
14. Johansson. H.: Telefonaktiebolaget LM.
Ericsson. Vol. I. From 1876 to 1918. LM
Ericsson, Stockholm. 1953, pp. 48-51, 60,
64-65.
15. Hundert Jahre elektrisches Nachrichtenwesen in der Schweiz 1852-1952. Vol. II,
Generaldirektion PTT, Bern 1959, pp. 182184, 711.
16. Lenaes: Privattelefonen i Norge. (Private
telephony in Norway.) 1966, p. 17.
17. La Lumiere Electrique, Vol. X (1883), pp.
444-445.
18. Elektrotech. Zeitschrift (April 1884). pp.
183-184.
19. Kingsbury, J. E.: The Telephone and Telephone Exchanges - Their Invention and Development. Longmans, Green and Co.,
London, New York, 1915, p. 400.
20. Uppfinningamas Bok. (The Book of Inventions.) Vol. Ill, Stockholm 1896, p. 581.
21. Fagen, M.D.: A History of Engineering and
Science in the Bell System. Vol. I. BTL, Murry Hill, N. J., 1975, p. 546.
22. Hill, R. B.: Early work on dial telephone systems. Bell Laboratories Record 31 (1953),
pp. 22-29.
23. Ericsson, E.A.: LM Ericsson 100 years.Vol.
III. LM Ericsson, Stockholm, 1976, p. 77.
24. Swedish patent no. 10662. Application
dated January 16, 1899.
ERICSSON
ISSN 0014-0171
Telefonaktiebolaget LM Ericsson
10284 Liungforetagen. Orebro 1