Yngve Rapp - The history of Ericsson

ERICSSON
REVIEW
1
SODERTALJE-THE FIRST AXE EXCHANGE
YNGVE RAPP-A BIOGRAPHICAL SKETCH
DESIGN OF ERICOFON 700
COMPUTERIZED OPERATION AND MAINTENANCE SYSTEM
SPC GROUP SELECTOR IN MEXICO CITY
PAPER-INSULATED TWIN CABLES AT HIGH FREQUENCIES
1977 AXB 20 - SYSTEM FOR TELEX AND DATA TRAFFIC

ERICSSON REVIEW
Vol.54, 1977
Contents
Cable Technique Page
Cross Stranding of Telephone Cable 105
Magneto Switchboards
ABJ 101 -the Modern Public Magneto Switchboard 136
Telephone Sets
Design of ERICOFON 700 10
New Telephone Set 112
Telephone Exchange Systems
Sodertalje-the First AXE Exchange 2
Computerized Operation and Maintenance System for Telephone
Networks 14
The Stored Program Controlled Group Selector ANC 11 and its
Introduction in Mexico City 22
ARE Systems in Modern Networks 54
ARE 11 —System Description 67
The Software and its Handling in ARE Systems 77
Crossbar Exchanges in Arhus Become SPC Exchanges 86
ARE 11 in Australia 90
Selection and Testing of Electronic Components for LM Ericsson's
Telephone Exchanges 94
Once Again Australia Chooses LM Ericsson-AXE Becomes
the System of the 1980s 142
Reed Switch with Integrated Micro Circuits 162
Operation and Maintenance Characteristics of AKE 13 125
Operation and Maintenance Characteristics of AKE 13 168
Telex
AXB 20 —All Electronic, Stored-Program-Controlled System for
Telex and Asynchronous Data Traffic 32
AXB 20-Description of System 41
COPYRIGHT TELEFONAKTIEBOLAGET LM ERICSSON STOCKHOLM • SWEDEN 1977

Transmission Technique
Transmission Properties of Paper-insulated Twin Cables at High
Frequencies 28
Digital Line Equipments for 8 Mbit/s and 2 Mbit/s 114
Generation of the Basic Frequencies for FDM Systems 174
Branching Equipment for FDM Systems 180
Miscellaneous
Yngve Rapp —a Biographical Sketch 7
The Role of Telecommunication in the Formation of the Society 144
Telecommunication Traffic Problems in Developing countries 149

ERICSSON REVIEW
NUMBER 1 1977 VOLUME 54
Copyright Telefonaktiebolaget LM Ericsson
Printed in Sweden, Stockholm 1977
RESPONSIBLE PUBLISHER DR. TECHN. CHRISTIAN JACOB/EUS
EDITOR GUSTAF O. DOUGLAS
EDITORIAL STAFF FOLKE BERG
EDITOR'S OFFICE S-12625 STOCKHOLM
SUBSCRIPTION ONE YEAR S6.00ONE COPY $1.70
Contents
2 • Sodertalje —the First AXE Exchange
7 • Yngve Rapp-a Biographical Sketch
10 • Design of ERICOFON 700
14 • Computerized Operation and Maintenance System for Telephone
Networks
22 • The Stored Program Controlled Group Selector ANC 11 and its
Introduction in Mexico City
28 • Transmission Properties of Paper-insulated Twin Cables at High
Frequencies
32 • AXB 20 —All Electronic, Stored-Program-Controlled System for
Telex and Asynchronous Data Traffic
41 • AXB 20-Description of System
COVER
Stored-Program-Controlled exchange type AXE
in Sodertalje, 40 kilometres south-west of
Stockholm, Sweden.
(The front covers for the two nearest raws are
removed.)

Södertälje—
the First AXE Exchange
John Meurling
The AXE system has been described in a number of articles in Ericsson Review
No. 2, 1976^ 5 . The system exhibits a number of interesting technical innovations
and properties. In particular the modular structure of both software and hardware
has been highly acclaimed. This structure has made the system easier to handle as
regards planning, production, installation and operation.
The first AXE exchange has been installed in Södertälje. Experience from the
installation and operation of this exchange has verified many of the system's
properties. This experience is briefly recounted in this article.
UDC621 395 34(485-3)
Fig. 1
The AGF 500-switch exchange at Södertälje
Fig. 2
Södertälje trunk code area
— — - High usage route
The Södertälje project
Södertälje is a medium-sized industrial
town some 40 kilometres south-west of
Stockholm. Since 1940 the town has
had an AGF 500-switch exchange,
which today has slightly more than
31,000 installed subscriber lines.
Södertälje is centre of a trunk code
area (trunk code 0755). There are thus a
0755 SÖDERTÄLJE
number of group centres and terminal
exchanges (of the Swedish Telecommunications
Administration's crossbar
type) connected to the AGF exchange.
There are also routes to Stockholm and
other adjacent towns, but also to such
distant places as Gothenburg and
Malmö (see network configuration fig.
2).
The AXE exchange in Södertälje consists
of a 3000-line unit in the same
building as the AGF exchange and constitutes
the first step in its total replacement.
Owing to the small size of the AXE
unit in relation to the AGF exchange, all
traffic is being preliminarily directed to
and from the AXE unit via the AGF group
selectors (see trunking diagram, fig. 4).
The AXE exchange at present interworks
with other exchanges on an
AGF basis but will later be independent

JOHN MEURLING
Telephone Exchange Division
Telefonaktiebolagel LM Ericsson
Fig. 3
The exchange building. The AXE exchange is on
the ground floor. The remainder of the building
accommodates offices and part of the AGF
equipment
Fig. 4
AXE, Södertälje
Simplified trunking diagram
AXE 10
LIC Line circuit
SSN Subscriber stage switching network
AJC Junctor circuit, A-subscriber
BJC Junctor circuit, B-subscriber
KRD Receiver for pushbutton signals
GSNI Group selector stage switching network,
incoming side
GSNO Group selector stage switching network,
outgoing side
ITC Incoming trunk circuit
OTC Outgoing trunk circuit
CSD Code sender
ASD Answering service device
AGF
S Finder
LV Final selector
SNR Cord circuit
IGV First group selector
IIGV Second group selector
GIV Incoming group selector
GUV Outgoing group selector
FUR Outgoing trunk circuit
of the AGF conditions. The new exchange
does, however, afford an interesting
example of the adaptibility of
the AXE system. No additional interworking
equipment has been introduced
on the AGF side, but the
characteristic revertive pulse signalling
of the AGF system is entirely handled by
AXE.
The installation work took slightly
longer one year. The reason for this extended
period was that as Södertälje is
the first AXE exchange in commercial
service, the desire was to use it for verifying
a number of new installation
methods and aids. The handling of the
software has, of course, been a matter
of special interest.
Since April 1976 a group of 20 trial subscribers
has been connected. These
were requested to generate as much
traffic as possible. The test traffic has
proceeded satisfactorily and has actively
contributed to the testing of the exchange.
During the first week of December 1976
the Administration made an operational
test of the exchange. On this occasion
an additional 20 subscriber lines were
connected by special agreement, in this
case high-traffic PBX lines. This test
was the first phase in the Administration's
acceptance of the exchange.
3
Installation and testing
It has been very important to verify the
principles for testing of functional units,
i.e. the procedure by which the functional
unit-the magazine-is equipped
with PC boards, put through its final
production test, and delivered to the installation
site fully equipped and tested.
This is of special value as regards the
processors CP and RP, which are thus
delivered equipped and tested. In a few
days the hardware is fully tested (with a
specially developed off-line test program
AVS), including the I/O system
with associated RP, after which the
operative and maintenance system is
loaded. This is run through in another
few days. Thereafter APZ —the control
system— constitutes the main instrument
for the subsequent installation
testing. APZ is accordingly kept in normal
operation during the greater part of
the installation testing period. As APZ
hardware and operative systems have
been tested in conjunction with their
manufacture, APZ specialists are not
normally needed on site during the installation
stage.
Three other important installation test
aids have been used:
— STG (Subscriber Traffic Generator)
consisting chiefly of hardware.
— SOLIT (Switch Operation and Link
Tester), consisting mainly of off-line

Fig. 5
The switchroom. Space for 12,000 lines
Fig. 6
Control room. In the rear are racks with five cartridge
decks and I/O matching equipment
software, i.e. the program is loaded in
APZ during testing only.
- CIRTI (Circuit Tester for Installation),
primarily for testing the junctor
circuits AJC and BJC. Software and
hardware.
The use of prefabricated cables should
be mentioned in this connection. Most
of the cables were delivered cut to
lenght and plug-ended. In future it is
intended to try out a further development
of this principle by delivering
from the factory entire cable forms
("octopuses") corresponding to the
wiring within the respective groups of
magazines.
The small number of parts has, of
course contributed to the ease with
which the AXE system has been installed.
This applies also to the cabling. Compared
with earlier systems an AXE exchange
requires less cabling, as a large
part of the connections already exist in
the wiring units of the magazines. The
separation of the cabling from the
mechanics is, of cource, another factor
that greatly assists installation work and
its planning. With these excellent handling
properties and testing aids, and the
low fault rate in equipment leaving the
factory, the installation time for AXE is
substantially shorter than for conventional
equipment.
Production of software
Within the software field the main point
of interest has been to verify the handling
improvements resulting from the
new AXE system structure, and to check
the efficiency of the new test methods
and equipments.
The software, i.e. programs and data,
for an AXE exchange consists of five
parts, each of which is produced independently:
— programs (the functions)
— exchange data, i.e. categories of
trunks and the routes to which they
belong, connection data for signalling
devices of different kinds, etc.
— analysis data, i.e. tables for route determination,
tariff determination, etc.
— subscriber data, i.e. the relation between
subscriber number and multiple
position (number group) and indication
of subscriber category
— supervisory data, i.e. threshold values
and permitted ratios on supervision
counters.
These five software packages, delivered
in the form of magnetic tape cartridges
are combined for the first time in the exchange.
No simultaneous testing is thus
needed during the production of programs
and data, which greatly facilitates
their handling.

Fig. 7
Part of signalling subsystem with two magazines
containing outgoing trunk circuits and two ringing
generators with distribution unit. At the top is
seen part of one of the two regional processors
of the magazine group
Fig. 8
Layout of the AXE exchange in Södertälje.
The final capacity of the switchroom will be 12,000
subscriber lines. The positions already equipped
for 3 000 lines are shown in colour
Magazines belonging to one subsystem are combined
into a magazine group; hence the notation G
CPG Central Processor Group
SSG Subscriber Switch Group
TSG Trunk and Signalling Group
GSG Group Switch Group
IOG I/O Group
TW Teletypewriter
PR Printer
Experience from development of software
Most of the programs are written in the
high-level PLEX language, some in assembler
code, ASA. Program quality is
usually measured in number of errors
per 1000 statements. The AXE software
has been tested mainly in two phases. In
the first, the interpreter test, about 30
errors per 1000 statements were detected.
The interpreter simulates a
machine working in source code, i.e. in
PLEX or ASA. From the interpreter test
one proceeds in the second phase direct
to functional testing in a system test
plant. Here about five errors per 1000
statements were detected. The delivered
programs have a quality corresponding
to a maximum of two errors
per 1000 statements.
The total source code volume is subdivided
as follows:
CP/PLEX statements 82,000 (71 %)
CP/ASA210C
instructions 16,000(14%)
RP/ASA210R
instructions 18,000(15%)
It may be of interest to note the expansion
to machine code:
Binary 16-bit words/PLEX
statements 3-3.5/1
Binary 16-bit words/ASA 210C
instruction 1.5/1
Binary 8-bit words/ASA 210R
instruction 2/1
5
Relocatability of the software
It has been extremely interesting to verify
the handling properties of the AXE
system in conjunction with equipment
extension and with the introduction of
new functions. The excellent handling
properties in these respects have been
achieved through the fact that the software
(programs and data) is entirely relocatable
and that packing of the stores
isdoneautomatically, on command. Relocatability
signifies that software modules
can be freely transferred between
the respective program and data stores.
It is the reference store that keeps a
check on the addresses.
The principles have been verified by,
among other means, installing the
memory units module by module, building
them up stepwise during the installation
period.
Another factor of interest is the number
of patches, i.e. provisional program corrections.
The characteristic feature of
the AXE system in this connection is the
module-by-module correction of errors.
Change of blocks is done individually,
without necessity for complete reloading
of an exchange, and is a simple procedure.
In other words, quick correction
of software errors has been attained.
The maximal number of patches at any
one time during the testing of the
Södertälje software was about 200. This
figure will, of course, be considerably
reduced in future exchanges as the

6
Fig. 9
One of the two central processors CPU
The control panel at the top right is provisional
Fig. 10
Control room in the Turku AXE exchange.
Test are at present in progress
system comes into wider use. The mean
life of a patch has been about four
weeks.
New AXE projects
The next AXE project isa4000-line local
exchange at Turku, Finland, which will
be cut over later this year. An AXB 20
telex exchange is also to be put in
service in Malmö this year. AXB 20 is
made up of digital switching equipment
controlled by APZ 210, the same data
processing system as in AXE.
In 1978 the first AXE exchange is to be
opened in France, at Orleans, and the
first digital AXE tandem exchange in
Finland.
The cut-over of the first AXE exchange
marks the termination of
the first stage in a very large project,
involving the development of
an entirely new generation of
telephone exchanges. The entire
project has comprised:
— system design
— high-level programming language
— new packaging structure
— several new special components
— new documentation system for
highly automated handling.
The birth of the AXE system has
further accelerated the development
of new methods of production
and testing. This has involved
the training of personnel also. An
indispensable requirement for
implementation of a project of
this magnitude is, of course, extensive
training affecting all parts
of the organization.
References
1. Meurling, J.: Presentation of AXE
10 Switching System. Ericsson Rev.
2, 1976, pp. 54-59.
2. Braugenhardt, S. and others: Introduction
of the AXE 10 Switching
System in the Telephone Network.
Ericsson Rev. 2, 1976, pp. 60-69.
3. Eklund, M. and others: AXE 10-
System Description. Ericsson Rev.
2, 1976, pp. 70-89.
4. Hemdal, G.: AXE 10-Software
Structure and Features. Ericsson
Rev. 2, 1976, pp. 90-99.
5. Alexandersson, R. and others: New
Packaging Structure for Electronic
Switching Equipment. Ericsson
Rev. 2, 1976, pp. 100-107.

Yngve Rapp
— a Biographical Sketch
Christian Jacobaeus
Yngve Rapp, who passed away in March 1976, had taken part in most of the
international teletraffic congresses (ITC). His contributions to these congresses
aroused great interest and came to have a great influence on the general
development within the sphere of network planning and network optimization. As a
markofhonourto Yngve Rapp's memory an invited paper was presented at ITC VIII
in Melbourne in November 1976 with tribute to Yngve Rapp.
UDC92
Yngve Rapp, D.Sc, born in 1903, died in 1976
As we all know, Yngve Rapp departed
this life on March 12, 1976. He was then
72 years of age.
Yngve Rapp was associated with LM
Ericsson during the whole of his active
life. He joined the telephone operating
company in Naples in 1927. At the beginning
of 1928 he moved to Smyrna
(Izmir), where he became Technical
Manager somewhat later. During the
years 1931 —32 he was in charge of the
line construction work in Athens. He returned
to Izmir on November 1, 1932,
now as president of its telephone
operating company. From 1939 onward
Rapp was stationed in Stockholm in the
Telephone Operations Department. On
October 1, 1950, he was transferred to
the Management Staff Department,
Sales, and became Vice President in
1954. From 1958 onward he reported to
the Chief Technical Officer, working on
matters within the technological and
scientific fields.
Yngve Rapp's work was concerned
almost exclusively with the planning
and construction of telephone networks.
He had a very strong economic
bent, and almost all of his publications
deal with questions of how administrations
can minimize their long-term investments
without neglecting customer
service. It was his work during the thirties
and forties which provided the
foundation for his later scientific production.
It became clear to him in his
practical work on line construction that
economy was of paramount importance,
particularly as at least 50 per
cent of the investments in a telephone
system are for line plant. Rapp became
conscious also of the significance of
long-term planningforeconomy. Sound
investments are essential, as capital
costs are the main expenditure item for
ministrations.
The first of Rapp's publications, from
1939, "Economic stages of extension of
a telephone network", dealt with a
number of problems to which he later
returned in greater detail on several
occasions. One of them, as the title indicates,
was the economic intervals in extension
of telephone plant, e.g. how
much and how frequently should the
cables in a network be extended.
Another was how to compare alternative
investments with different lengths
of life. He treated these problems by
means of the present-value method. We
now consider this method self-evident,
but it was not so well known when Rapp
started to use it. He also wrote several
papers on the present-value method,
without special reference to telephony
applications, in various journals during
the forties.
Rapp's doctoral thesis
The discoveries Rapp made through
practical work on line construction in
Naples, Athens and Izmir were the
sources on which he drew in his doctoral
thesis, "The economic optimum in
urban telephone network problems". In
this he arrives at a happy combination of
theory and practice. The thesis is in
three parts. The first deals with local line
plant provisioning in an exchange area,
based on LM Ericsson's network layout
system. In this system the network consists
of primary cables running to division
cabinets, from which secondary
cables run to distribution points, and
thence distribution lines to the subscribers.
After going through the procedure
of network planning, he considers
the question of how large the distribution
areas should be, and also their
shape. He discusses the question of the
location of division cabinets and the
size of the areas to be served by them.
He then gives some examples of the use
of equalization lines between division
cabinets. The section concludes with a
discussion of economic stages of extension
for different parts of the urban
network.
The second part of the thesis deals with
multi-exchange areas. Rapp states that
the problem is similar to that considered
in the first part. But it is more complicated,
as the number of variables is
much larger. Apart from the number and

8
Yngve Rapp at his doctoral dissertation in 1950
locations of the exchanges, it is necessary
to determine the conductor dimensions
in the local and junction cables
and how the junction routes are to be
arranged. Rapp refrains from treating
the problem in one context in its entire
breadth, but instead proceeds from certain
simplified assumptions and then
takes up the subproblems one by one.
He first determines the cable diameters
having regard to transmission conditions
and economy. He also deals with
the question of different diameters in
cable runs. He investigates the conditions
applying to loaded cables. He then
goes on to determine the number of exchanges
and their location.
The third part is entitled "The HT
system —a new system of local line plant
provision". Here he describes methods
for further improvement of the economy
of local line plant by means of a partially
new arrangement of the network, using
equalization lines between the distribution
points. He also achieves better utilization
of the equipment by placing feed
points and feed cables in the primary
network.
Rapp's thesis is a work of very high
class. Practical engineering skill and
mathematics are combined to produce
an eminent result. It many be said that in
this work Rapp solved the main plant
provisioning problems in urban networks.
The concepts of the HT system
have had a great influence on the development
of line construction.
Rapp's publications
1961-1973
After an interval of 10 years when he was
engaged on other assignments, Rapp
published in 1961 "Extension of telephone
plant with regard to the value of
subscribers' time". Here he deals with
the inconvenience experienced by subscribers
as a result of insufficient
number of circuits and poor quality of
transmission. He estimates the time lost
by subscribers owing to congestion and
unsatisfactory transmission and assigns
a value to the time so lost. From
this starting point the number of circuits
on a route and the diameter of conductors
in subscribers' cables can be
determined so that the sum of the plant
cost and of the economic value of the
time lost by subscribers is as small as
possible.
The starting point for this study must be
considered interesting. In practice, however,
the method proves to be somewhat
inexact and the study may be regarded
as a contribution to the discussion,
hardly more.
In a number of papers during the period
1962-1973 Rapp dealt more thoroughly
with problems which he had worked
on earlier. He had now refined his mathematical
models. He found in the
computer, furthermore, a tool which
opened up new possibilities. The computer
soon became his standard aid.
Rapp undertook the task of systematically
going through all problems relating
to the planning of urban networks.
His "Calculation of traffic distributions
in multi-exchange networks" proceeds
from different assumptions concerning
the community of interest between exchanges
and then uses matrix algebra
for dealing with the changes in traffic
resulting from a change in the network
plan.
In "Planning of exchange locations and
boundaries in multi-exchange networks"
Rapp uses to some extent the
results of his preceding study. He first
investigates simple cases such as the
location of an exchange when there is
only one exchange, and the determination
of the boundary between two exchanges
with fixed locations; also the
locations of two exchanges when the
boundary between their areas is known.
Finally he considers the genetal case
indicated in the title. Here he works with
an iterative process with successive
approximations starting from an empirical
first network plan. The solution
of this problem was made possible by
use of a computer.
"Planning of junction network in a multi-exchange
area" presents methods for
planning of large urban networks (Metropolitan
Areas). He starts with the case
of alternative routing and calculates the
minimum cost for such an arrangement.
The result is then used to deal with the
general case. He takes into consideration
all factors which may affect the

Yngve Rapp at the rostrum
result, such as the existing network,
where transit centres should be placed,
which exchanges should be served by
each transit centre, the transmission
plan, etc. The methods of this study have
been used in several metropolitan areas.
ITU also uses computer programs based
on Rapp's work in its consultancy service
to developing countries.
In "Planning of junction network with
non-coincident busy hours" Rapp
studies plant provisioning in a network
offered traffic with different busy hours.
He presents rules for the construction of
networks of this type. A finding of
particular value is that the traffic conditions
make a network of this type more
sensitive to overload.
Yngve Rapp as human being
Nearly all of Rapp's work has a technoeconomic
foundation. It is always
founded on practical experience. With
his realistic outlook he is always on firm
ground. One can be sure that his results
can be translated into practice. The
mathematical reasoning is always clear.
He proceeds very logically, one may
even say pedagogically. It appears as if
he drives every study to the limit permitted
by the mathematical and computational
apparatus available. He also goes
as far as is reasonably possible having
regard to the quality of the input data
that can be expected. The results he achieved
have often been fundamental
for subsequent developments. Rapp indicated
solutions of several major problems
in network planning and line construction.
He acquired a place of rank
among researchers in this field and his
name is likely to live on among network
planners for a long time to come.
Many of those present today have met
Yngve Rapp at earlier congresses. Several
have undoubtedly exchanged
thoughts with him inside or outside his
speciality. We all had the impression,
I am sure, that Yngve had interests
extending far beyond technology and
science. He had a very all-round knowledge,
including impressive linguistic
abilities. He spoke Swedish, German,
English, French and Italian and could
make himself understood in Spanish,
Greek and Turkish.
He also had a pedagogical ability. He
thus became a successful consultant
and was a popular lecturer. He had a
large measure of humour and sympathy.
Relations with him, therefore, always
had a human stamp. We, his colleagues,
not only feel respect for his achievement,
we miss him as a friend
and comrade.
9

Design of ERICOFON 700
Carl-Arne Breger
LM Ericsson's new version of ERICOFON-ERICOFON 700-was described in
Ericsson Review No. 3, 1976\ The present article contains a personal account of
how the designer viewed his task, what problems he considered important, and how
he found solutions to them. The article illustrates in an interesting way how the
demands of the user-man-set their stamp on the product.
UDC 621.395.6 An industrial product often consists of
different components, each of which
has an important function. This applies
particularly to so advanced a technical
product as the telephone. Through its
close contact with man, the telephone
must also have a design which makes it
functional and attractive to use. The
classical man/machine interface is really
well illustrated in the case of the telephone.
An extremely complicated technical
product must be adapted to several
human organs —the hands, ears,
mouth, eyes, even the face and the head
as a whole. The difficulty is particularly
great if the entire telephone is a single
unit, as in the ERICOFON. Then there is
the question also of adaptation to varying
tastes and notions of beauty.
Fig. 1
(Right). The new model, ERICOFON 700, and (left)
the old
It is for this reason that it is so important
correctly to compose the individual
form of the various parts, e.g. to get the
austere rectangularity of the earcap to
conform to the underlying circular
diaphragm. In contrast thereto, and to
fit into the whole, the sound apertures
are consciously given a rectangular
form. In ERICOFON 700 the receiver
must pass into a handle, the cross-section
of which, on technical grounds,
must be rectangular. The handle has a
very vital function in the entire telephone.
It must unite the receiver, with its
characteristic form, to the base of the
telephone. Looking at the handle from
the side, it emerges from the base and
continues in a natural manner into the
receiver unit. If ERICOFON 700 is turned
a quarter turn, so that the handle is
viewed from straight in front or straight
behind, it must form a part with soft,
aesthetic lines, fitting properly into the
hand, starting where the receiver ends
and gliding almost unnoticeably out
into the base of the telephone.
A few words also about the conditions
for design of the base of ERICOFON
700. It was desired that the base should
be small, almost on the bounds of the
impossible, and so designed that it
emphasizes to the user in a natural way
that he must speak into the (incredibly)
small hole on the fore edge of the body
of the base. This was done by placing
the hole on a surface of its own.
Work in sketch form
When I started on my design commission
by composing some one hundred
sketches, based on the above ideas and
decisions, I soon found that I had in LM

CARL-ARNE BREGER
Breger-Design AB
Fig. 2
The placing of the microphone was given, but
there was side scope for different formation of the
holes in the earcap and in the bottom portion. The
thumb-grip was still included at this stage
Fig. 3
A study made to obtain a strict and effective subdivision
of the telephone. This foundered, how-
11
Ericsson an almost ideal client. I felt the
support of all concerned in the project.
All had a great understanding for and
experience of industrial design.
As the work proceeded. I understood
that the so-called thumb-grip was unnecessary
in the new ERICOFON 700,
which was both smaller and weighed
much less than its predecessor. The
thumb-grip was entirely removed when
a number of models with differently
formed thumb-grips showed a thumbgrip
even to be somewhat disadvantageous
from the ergonomic
aspect.
In this context it is interesting to recognize
the difference between working
in two and in three dimensions. As is
seen from the sketches, one can be inspired
to very interesting and beautiful
forms in two dimensions. But when the
sketches are to be realized in three dimensions,
adaptation to man immediately
enters into the picture. The
telephone is today one of the most used
of man's implements both for work and
leisure. In the styling and design of
ERICOFON 700, this has been the point
of departure. No sacrifice has been
made on the altar of beauty which was
opposed to ergonomics or technology.
Model-work
Work now started on plaster and
wooden models. The wooden models
were of great importance when discussing
the contours of the telephone and
how one surface affected an adjacent
surface: which surface or line was most
important for the overall impression,
but also for handling the instrument,
and so on. All wooden models were
painted white, as an unsparing revelation
of light and shade was thereby
obtained. At the same time as we designers
were working on the external
form, LM Ericsson's engineers were
working from within to adapt the technical
elements-the components-to
this form within the scope of the functional
requirements. Through this interplay
between different specialists,
who had a deep respect and understanding
for one another's knowledge,
the form of ERICOFON 700 grew up in
an organized and correct way. The
models were continuously subjected to
examination by a reference group-and

12
Fig. 4
The base with keyset and cradle-switch
mechanism was very thoroughly studied both
from aesthetic and functional aspects
Fig. 5
The dotted line shows the earlier appearance of
ERICOFON and how. within these boundaries, it
could be contracted and yet retain its character
at decisive stages I_M Ericsson's management
was consulted.
A demand I place on myself and my colleagues
is that the product we are working
on shall have an inbuilt dramatic
quality, a refined play between light and
shade. In a global product such as
ERICOFON 700 its art form must be understandable
by all of whatever nationality.
It is also important, in view of its
long life expectancy, that its form
should be austere and self-evident. It
must not be such that, in 1985 forexample,
people say: "This was designed in
1975" or "typical 1975". The entire form
has therefore been tightened up and,
one may say, adapted to the requirements
of the day and the morrow. The
profile of the earlier ERICOFON may
stand for everything inherent in the
word profile. It is well known throughout
the world. It has appeared in films,
newspaper reports —wherever there
has been the desire to add a detail of
unusual interest.
In the world of telephony ERICOFON is
an established, self-evident, "different"
telephone with a good technical reputation,
already a classic. When LM Ericsson
asked me to redesign this telephone
to adapt it to modern requirements,
among which the change from dial to
keyset, less bulky components, etc., it
was obvious that I would have to carefully
study the structure of the now classical
form of ERICOFON. I and my colleagues
had, within extremely small
margins, to produce something entirely
new. We had, with little free scope, to
design a new product, so new that the
world market would see the fact, but,
paradoxically, the main impression
would remain the same as in the case
of the earlier instrument, as this was
well known and accepted.
The new features of
ERICOFON 700
What did I find in the classical ERICO­
FON which I did not consider to be consonant
with modern design language?
As people change and adapt to new social
forms, new technical advances,
etc., there was quite a lot I wished to
change in the ERICOFON design. And
as this new pushbutton telephone —
ERICOFON 700 - is to remain "modern"
lintil tahirMit 1 QQO lOnnrw ~.r.r.nrA. t_ L n

Fig. 6
The base surface has acquired a quiet, harmonious
appearance through combination of the subscriber
number frame and service button
Fig. 7
Another example of the styling at the stage when
the thumb-grip was still retained
Ericsson's technical and marketing
evaluation, it is important to see with
"eyes of the future". What I especialy
wanted to change was the base of the
instrument, which I judged to be optically
too heavy and which also partly
blocked the speaker's view. Nor was
there any difficulty in reducing the size
of this part, as the new electronic components
were much smaller. The handle
was a point of particular interest. In the
design of the earlier instrument there
had been full freedom, as the handle
had to accomodate only the thin conductors
to the receiver. In ERICOFON
700, on the other hand, it would have to
provide space for a whole double-sided
printed circuit board. Here a compromise
had to be found between several
extremely relevant demands, namely.
internally (large)
— space for the "innards"
— possible to manufacture
— impactproof design
externally (small)
— ergonometrically correct
— harmonious union of base and top
— beauty of line
What we finally achieved appeared
from the start to be guite impossible.
The great change for the subscriber is,
of course, the substitution of keyset for
dial. Here there was an opportunity for
some ingenuity, something that could
give a stimulating feeling. To work here
with form and colour to obtain an
aesthetic whole, I thought self-evident;
but it was necessary, too, that the form
of the instrument should lead to no misunderstanding
in handling. The three
parallel-operating cradle-switch buttons
for clearing a call must strike the
eye. For this reason they were coloured
red. LM Ericsson explained how important
it was that keying of numbers
should be easy and reliable. In retrospect
it is therefore evident that the entire
base (the keyset) was made black
with only the digits white. This gives
maximal contrast and the fingers are
drawn automatically to the correct digit
(key). It may be freely acknowledged
that the desire for as small as possible a
base made it difficult to accomodate on
the base all functions that are technically
necessary. When, therefore, an LME
engineer hit upon the idea of combining
••e subscriber number frame and the
13
service button, the base could be given
a symmetrical and guiet appearance.
The feeling of beauty that the outer form
of ERICOFON 700 should impart at a
distance should be reinforced by the
close-up impression when the instrument
is raised to key a number. The
strictly functional "rectangular" form of
the keyset is stressed in a graceful yet
emphatic fashion by the characteristic
form of the base plate.
Finally a couple of words on the
"pedestal"-the visible intermediate
link between the desk and the telephone.
It is thin and discreet, its function
being to "lift up" ERICOFON 700 so as
visually to emphasize its airiness.
Summary
In conclusion I should like to summarize
my impressions of working on ERICO­
FON 700, which was a stimulating task
precisely because it was very complicated.
- It was a matter of designing a working
implement within a given
framework and with very strict
specifications.
— The difficulty was enhanced by the
fact that the earlier ERICOFON was
so well known.
— Change of a line might be a tricky
matter; an improvement or impairment
might arise in another surface
or line.
- LM Ericsson's reguirement was a
"large inside", mine a "small outside".
This was difficult but stimulating.
— LM Ericsson's technicians made
ERICOFON 700 good and strong. I
wanted to create an implement
perfect in form. It was a matter of
pushing this combination as fas as
possible.
- It was stimulating to work with many
real experts.
References
1. Boeryd, A. and others: Electronic Pushbutton
Telephone Set. Ericsson Rev 3
1976, pp. 118-133.

Computerized Operation ana
Maintenance System for
Telephone Networks
Torbjörn Johnson and Lennart Söderberg
Controlled corrective maintenance (CCM) has for a long time been employed for
LM Ericsson's telephone exchanges with good results^- 2 . Further centralization
and automatization of operation and maintenance open the way to additional
cost reductions. The Computerized Operation and Maintenance System COMS,
designed for parts of the network in which SPC exchanges have not yet been
introduced, builds further on the CCM principles. COMS also offers the staff the
means for more effective supervision from an Operation and Maintenance Centre
(OMC) owing to its new and valuable supervisory functions, its considerable
analysis capacity, the means it off ers for remote-controlled fault isolation in central
units, and methods for effective collection and presentation of traffic measurement
data.
UDC 621.395.74.-
681 3
Fig. 1
Centralization of operation and maintenance
data link
Centralization of controlled
corrective maintenance
by means of COMS
COMS is a data processing system for
automatic collection and analysis of
operation and maintenance data from a
telephone network. With COMS an
administration can centralize operation
and maintenance also for the parts of a
network which do not contain SPC exchanges.
The costs of operation and
maintenance of telephone networks are
constantly growing owing to the rapid
rate of development and to the rise of
wages for operation and maintenance
staff. The long-term solution is the introduction
of SPC technique. But before
this has been brought about, there
will be a long period with growing sections
of network without SPC exchanges.
During this period-when,
furthermore, centralized operation and
maintenance from an Operation and
Maintenance Centre (OMC) is coming
into use for SPC exchanges —it will be
Operation & maintenance centre
economically advantageous to modernize
the operation and maintenance
properties of exchanges which it is not
planned to replace by SPC systems
within the immediate future.
For LM Ericsson's AR family a modernization
to SPC level is obtained through
ANA 30. For administrations which plan
to modernize only operation and
maintenance, the computerized operation
and maintenance system COMS
(fig. 1) is an appropriate solution.
The existing supervisory systems in the
AR family are based on CCM and have
given very good results from the outset.
COMS is a further development of the
supervisory systems of the AR family.
What has been improved is, in particular,
certain supervisory methods and the
manner of presenting and evaluating
collected test data by means of modern
computer technique. Certain new
supervisory functions, moreover, such
as circuit supervision, have become
economically possible.
The basic principle underlying CCM is
to take maintenance action only when
the service quality has fallen below a
predefined level. On these occasions
the staff are alerted and a fault tracing
procedure, followed by repair and reporting,
is started. In COMS the same
principles apply both to the traditional
and to the new supervisory functions.
The large quantity of crude data concerning
the functioning of exchanges is
collected by COMS, analysed and compared
with fixed and floating threshold
values. Alarm is initiated in response to

TORBJÖRN JOHNSON
Ml-division
LENNART SÖDERBERG
Telephone Exchange Division
Telefonaktiebolaget LM Ericsson
Fig. 2
Example of COMS network
K Concentrator
OMT Operation and maintenance terminal
abnormal conditions. No extensive,
regular printouts are presented which
require manual analysis. Traffic measurements
can also be made with collected
data as base.
As COMS is a further development of
the operation and maintenance system
of the AR family, there are simple interfaces
in the exchanges and no extensive
alterations are therefore required.
Installation work is facilitated by the fact
that test points are normally available in
the intermediate distribution frames.
Most maintenance, repair and testing in
electromechanical exchanges must be
done on site. Therefore COMS incorporates
equipment and functions which
permit attended exchanges to receive
all maintenance information they need.
The existing organization thus receives
a tool which, through distributed data
processing in the COMS network, provides
a reliable and economical solution
to the operation and maintenance problem.
Centralization of operation and
maintenance to, for example, OMC becomes
possible also to an extent which
is economically and practically warranted
for electromechanical exchanges.
As the staff are freed from such routines
as reading and analysis of centralograph
and counter data, and collection
of traffic measurement data, their number
can be reduced; and in some cases
even larger exchanges can be unattended,
which is in conformity with the
CCM principles.
Modular System Structure
15
COMS is a data collection system with
an extensive capacity for processing
and evaluation of collected data. It was
developed especially for operation and
maintenance of network sections without
SPC exchanges. The system has a
modular structure both for hardware
and software, and the modules can be
combined to meet individual requirements.
Minicomputers are used
throughout, in order to provide wide
flexibility at low cost.
The chief units of COMS are autonomous
Operation and Maintenance Terminals
(OMT) in the exchanges and data
concentrators connected to a data processing
centre. A typical COMS system
is shown in fig. 2. The operation and
maintenance terminals collect data
from test points in individual devices.
These data are preprocessed and reduced
in volume before being transmitted
on communication lines directly or
via concentrators to the central computer.
To make the total system availability
high without expensive redundance in
the hardware, some processing of
crude data has been located as far out in
the COMS network as possible.
The results of computations for
supervision, alarm detection and traffic
measurement are printed on centrally or
locally placed printers. Data displays —
colour or black-white —can be used if
desired. An important characteristic of

16
Fig. 3
Principle of disturbance supervision by sequence
analysis
Stepping ot Counter: seizure
disturbance
disturbance rate
COMS is that the operators are warned
of an alarm condition by an audible or
visual signal and that all extensive
printouts must be requested by the
operators.
Another important characteristic is that
all grouping of individual devices into
routes is done with the aid of tables in
the computer memories. This means
that no alterations of the cabling to the
operation and maintenance terminals
are required when the exchanges are
extended or altered.
Functions
More effective supervision of
operational disturbances
With the traditional method of disturbance
supervision on common control
devices such as registers, code senders,
code receivers, markers and route markers,
the number of seizures and disturbances
(time outs) is totalled on electromechanical
counters. A supervisory
equipment then generates alarm in response
to an abnormal ratio between
disturbances and seizures for groups
of devices. By instead collecting and
processing these data in COMS the following
advantages are obtained:
— supervision per individual device
— improved methods of statistical
supervision
— simply changeable threshold values
per type of device
— convenient and effective handling of
the results of the data processing
The disturbance supervision is carried
out primarily by sequence analysis,
which has proved to be very effective for
detecting abnormal behaviour of devices.
The principle of sequence
analysis is shown in fig. 3. Apart from
the purely supervisory function, the
number of seizures and disturbances
can be stored for later statistical
analysis.
Supervision of time congestion
Route congestion may be an indication
of fault. COMS therefore collects information
on congestion time from exchange
systems in which it is easily accessible.
When an adjustable limit for
permissible congestion time is passed,
an alarm is issued. The resulting congestion
information is a valuable supplement
to the other supervisory functions.
Supervision for detection
of faulty circuits
Although the fault rate is low in the AR
systems, it has been found that the
overall service quality may be unsatisfactory.
The cause lies in all probability
in the trunk and junction network.
The traffic route tester (TRT) 3 is used for
detecting impairments of the service
quality in different directions by checking
the entire chain of links between Aand
B-subscriber. By supervision of individual
circuits COMS offers, in sup-

Fig. 4
plementation of TRT, a means for quick
fault location in the trunk and junction
network. The following functions exist:
- identification of killer trunks, i.e.
trunks with abnormally short seizure
time, which, by attracting calls, may
greatly lower the service quality on
the route
- identification of trunks which are
never seized
- identification of trunks which are
seldom seized
- identification of trunks which are
continuously seized.
The system collects information from
each individual circuit concerning the
number of seizures and seizure time
and checks whether the mean seizure
time is normal or not. To prevent unnecessary
alarms due to temporary
overload, the individual mean seizure
time is compared with the mean seizure
time for the route before alerting the
maintenance staff.
Supervision of coin-box telephones
Defective coin-box telephones are a
constant problem requiring expensive
routine manual checks. By applying the
same supervisory functions for individual
coin-box telephone terminations
in the exchanges as for individual
circuits, faults can be automatically detected
at an early stage. In this case no
comparison is made with the group
mean value, but the threshold values are
fixed.
17
Collection of data from other
alarm sources
Apart from their already mentioned
supervisory functions, the AR systems
have other alarm sources, such as power
failures and blown fuses. Alarms
from, for example, monitors of temperature
in buildings and from cable pressure
sentinels are also of vital importance.
COMS collects all such information
via its operation and maintenance
terminals. The supervisory functions
thus become integrated in theirentirety.
The computerization makes it easy to
change threshold values and alarm
classes when required.
Issue of alarm
Common to all supervision in COMS is
the principle for issue of alarm, which is
done with audible and visual signals.
LM Ericsson's ordinary classification
A1, A2, A3, 01 and 02 is used. The
operator then requests supplementary
information via printer and data display
unit.
Alarms can be received both from
central and local points, and the alarm
indications can be directed to different
places at different times of the day. At
nighttime, for example the supervision
of the network can be centralized to a
large attended exchange and be
switched overto local control in the daytime.
The alarm printouts contain information
on source, type of fault, and category of
staff concerned. A typical printout is
shown in fig. 4.

18
Fig. 5
TV monitor showing alarm situation
Alarms are indicated on lamp panels or
on modern media such as colour television
monitors (fig. 5).
In addition to indication and printout,
alarm events are recorded so that the
processes can later be analysed and
constitute basis for statistics.
Supervision and recording
of disturbances in markers
Centralograph equipment today record
faults in markers and route markers in
ARM. The printouts require manual
analysis for identification of the faults.
COMS automatically collects centralograph
data and makes
- a computation of seizures and disturbances
per type of fault and
marker
- a detailed recording of disturbances
for combinations of devices in which
a fault is suspected
Printout is received on a printer or display
unit with time of day and easily understandable
codes for type of device
and fault.
Isolation of faults
When faults have been identified by
COMS in central devices such as registers,
code senders, code receivers and
markers in unattended exchanges, they
can be isolated via COMS by remote
blocking from, for example, OMC. Disturbance
of traffic by certain types of
faults is thereby limited, which reduces
the need for a high state of readiness
and quick repair.
Operational statistics
Some of the administrations' operational
statistics are at present obtained
by manual reading and analysis of the
data presented on seizure counters and
disturbance counters. COMS does this
automatically on request.
Traffic measurement
From COMS, traffic measurements can
be ordered in advance for different
periods of 15, 30 or 60 minutes.
The following results can be obtained
for:
Routes (trunks):
O traffic flow in erlangs
• number of seizures
• mean seizure time, in seconds
D time congestion, in per cent (in certain
types of exchange)
• call congestion (require extra cabling
in exchanges)

Fig. 6
Group of registers and code senders
• traffic flow in erlangs
• number of seizures
• mean seizure time, in seconds
• call congestion (as option)
For other common control devices the
number of seizures is indicated. The
seizure time need not be measured, as it
is roughly constant. An example of
printout is shown in fig. 6.
All totals per group can be recorded on
magnetic tape for the ordered measurement
period for later statistical
analysis.
All definition of groups and routes, and
allocation of devices and circuits to
them, is done through tables stored in
COMS. The same tables are used both
for the supervisory and traffic measurement
functions. This means that the
cabling to the operation and maintenance
terminals need not be altered in
conjunction with alterations in the exchanges.
Measurement of call dispersion
As an effective instrument for planning
especially of networks with alternative
routing, traditional traffic measurement
on routes is not sufficient. If it is sup­
19
plemented by measurement of the call
dispersion for originating traffic, i.e. the
breakdown of calls by destinations (exchanges),
the required traffic matrix can
be calculated, i.e. the traffic in Erlang
between originating and terminating
exchanges in the network. It is therefore
planned that COMS shall incorporate
functions for random busy-hour sampling
of the destinations of calls in
originating exchanges. For this, an
equipment is required for connection to
the exchange registers or markers. The
call-dispersion and erlang values presented
by COMS per route will be used
for calculation of the traffic matrix.
Traffic route testing
Traffic route testing (TRT) by generation
of test calls is the primary method in
present networks for assessing the
service quality and obtaining indications
of faults. LM Ericsson's TRT
equipment is already centralized, but in
large networks it is an advantage to
have a common control centre for all
TRT equipments, which is possible with
COMS
Component units
The units of COMS are designed on the
principles of the modern, compact

20
Fig. 7
Operation and maintenance terminal
packaging structure BYB 4 . Operation
and maintenance terminal, concentratorand
central computerare built
up of modules. Computer technique
and modern integrated electronics are
used.
COMS is powered from the exchange
battery, with the exception of certain input/output
units, e.g. printer.
The communication between operation
and maintenance terminals and concentrators
or the central computer
takes place on transmission lines. The
transmission procedure allows connection
of several units to the same transmission
line. The normal speed is 1200
bits/s, but other speeds are also possible.
Higher speeds are often used between
the concentrator and the central
computer.
Operation and maintenance terminal
(OMT)
The main functions of the operation and
maintenance terminal are, for each inlet,
to accumulate the number and the
total time of events. It also contains a
certain capacity for local processing of,
for instance, alarms. The control unit is
a minicomputer with the possibility for
introduction of new functions in future
or for change of existing functions.
The programs can be stored in a nondestructible
read-only memory, while
parameters and the like are stored in a
random access memory which is loaded
locally or, usually, from the central
computer at start-up. Devices such as
counters and status indicators are also
placed in the random access memory. A
simplified block diagram of the operation
and maintenance terminal is shown
in fig. 7.
Up to 16 inlet units with up to 1024 inlets
in each can be connected to the control
unit. It is also possible to connect up to
512 digital outlets to a control unit. If
local printout of data is desired, a printer
can be connected
The operation and maintenance terminal
is controlled by the central computer
which, apart from counter values, can
also request the momentary state of
each connected inlet. The central computer
can also send data to the digital
outlets and order different tests in the
operation and maintenance terminal.
Concentrator
The concentrator is based on the same
minicomputer as the operation and
maintenance terminal. It is used for
concentration of several transmission
lines to a single line or for local printout
at places where there is no operation
and maintenance terminal.
Central computer
The COMS network is controlled and
supervised from the central computer.
This is a minicomputer specially designed
for the requirements of telecommunication
plant. External units for
data storage and man-machine communication
can be connected to the
central computer as shown in fig. 8.
Most supervisory and traffic measurement
functions need access to a corn-

Mass memory
Fig 8
Central computer-block diagram
Fig. 9
(Central computer-
Cartridge
tape units
Magnetic
tape units
Line printers
Tele type writers
Video display units
Colour TV monitors
Communication
links
Special
adapters
21
mon data base. The bulk of the data
processing is therefore done centrally,
but some preparatory processing and
reduction of data is done locally.
The software system is modular (fig 9)
and most programs are written in the
high-level PASCAL language. The base
system consists of communication and
data-collection programs which update
the data base with data from the operation
and maintenance terminals. The
data base also contains grouping information-parameters
and alphanumerical
strings. These data are then used
by different programs for supervision,
traffic measurement, printout, etc. The
software system is so designed that the
user can add new functions in a simple
manner.
References
I.Eriksson, V.: CCM-A Well-tried
and Economic Maintenance System.
Ericsson Rev. 3, 1976, pp 134
-137.
2. Hamers, J. A. and Scheurwater, C:
Ten Years Experience of CCM in
the Dutch Telephone Network.
Ericsson Rev. 3, 1976, pp. 138-
141.
3. Broby, S.-B.: Electronic Traffic
Route Tester m 70. Ericsson Rev 3
1974, pp. 80-87.
4. Alexandersson, R. and Rörström,
H. O.: New Packaging Structure for
Electronic Switching Equipment.
Ericsson Rev. 2, 1976, pp 100-
107.

The Stored Program Controlled
Group Selector ANC 11 and its
Introduction in Mexico City
Björn Lasson and Sune Lindblad
LM Ericsson has designed a stored-program-controlled group selector ANC 11
made up of reed matrices. The switching stage, which is intended for very large
multi-exchange networks, is used for outgoing, incoming or combined outgoing
and incoming traffic in local exchanges. It can also be used as an independent local
transit stage for two-wire speech transmission.
ANC 11 has already been described in another publication* and is therefore dealt
with only summarily in this article, the object of which is to illustrate how the
switching stage is used in practice, particularly in the Mexico City network.
UDC 621 395 74 ANC 11 has properties
adapted to large
multi-exchange networks
Fig. 1a
ANC 11 as outgoing group selector with a common
route from all 1C.000-line groups to each
destination
Fig. 1b
ANC 11 as incoming group selector distributing
the incoming traffic to several 10,000-line groups
ANC 11 was designed to serve metropolitan
areas. Its basic properties which
fit it for this purpose are the following.
The switching stage has a large maximal
capacity but can nevertheless be built
up in small modules. The maximal
capacity of an ANC 11 stage is 6,144 inlets
and 6,144 outlets and can be extended
in steps of 64 inlets and 64 outlets.
If an even larger capacity is needed,
two or more ANC 11 stages can be
combined.
It has a good circuit economy, as all
circuits are accessible from each inlet
(full availability). The internal congestion
in the link system is very low since
there is conditional selection throughout
the switching system (six partial
stages) and up to four reselections are
possible. The permitted congestion can
therefore be almost entirely assigned to
the circuits.
As the group selector is not graded, it is
easy to install and extend. Regrading is
never necessary. The routes can comprise
up to 256 circuits each. The maximum
number of routes that can be inserted
is 256.
In older exchanges with more than
10,000 lines, each 10,000-group usually
has its own first group selector stage I-
GV, with one route running from each
10,000-group to each of the destinations
concerned. If l-GV is built up on
the ANC 11 principle, however, several
10,000-groups can be connected to its
inlet side, in which case only one common
route is required from all of these
10,000-groups to each destination (fig.
1a). As incoming group selector, ANC 11
can also serve several incoming
10,000-groups, with corresponding reduction
of the number of routes (fig. 1b).
The reduced need for routes is illustrated
more clearly in fig. 2a-c. This
results in a further reduction of the need
for circuits.
SPC technique is used. Duplicated
microprocessors are used foreach control
and signalling unit, in addition to a
separate operation and maintenance
processor. Changes in the operational
conditions are made by an operator
sending a command from an electric
typewriter (fig. 3). The system also has
good traffic-rout ing properties owing to

£«*•,
BJÖRN LASSON
SUNE LINDBLAD
Telephone Exchange Division
Telefonakliebolaget LM Ericsson
Fig. 2a
Without ANC 11 one route is normally required
from each 1C,C00-group in exchange A to each
1C,CC0-group in exchange B. and vice versa, i.e.
in all 2 (mxn)
Fig 2b
If both exchanges are equipped with ANC 11 as
l-GV. one route is required from exchange A to
each 10,COO-group in exchange B. and vice versa.
i.e. in all n + m
Fig. 2c
If both exchanges are equipped with ANC 11 both
for l-GV and ll-GV/GIV. a single route is required
in each direction, i.e. in all 1+1
Fig. 3 (right)
Change of the operational conditions in ANC 11,
c n insertion of alternative route B, can be done
its well developed analysis capacity; it
has, besides, many operation and
maintenance facilities.
Schematic structure
The structure of ANC 11 is shown
schematically in figs. 5 and 6, which
contain the following blocks and units:
Incoming selector GV-I consisting of up
to 12 identical GV-I groups, each with
512 inlets. Each group hasan individual
control unit of SPC type.
Outgoing selector GV-O, consisting of
up to 12 identical GV-0 groups with the
same grouping and type of control unit
as GV-I
The duplicated central processing unit
CE, common to twelve GV-I and twelve
GV-0 groups, controls the traffic handling
functions of ANC 11, such as digit
analysis, link selection, circuit selection
within a specific route, and switch operation.
Operation and maintenance processor
OMP performs supervision, traffic
measurement, fault-tracing and issue of
alarm, and constitutes a communication
link with the input and output devices.
23
Input and output devices (I/O) of up to
four types-electric typewriter, visual
display, tape reader and tape punch.
Grouping plan
The basic module of the reed switch
contains 24 cross-points which form a
4x6 matrix. Fig. 4 shows a grouping
plan for GV-I with 512 inlets and 864 outlets
The latter constitute links to GV-O,
which has the same grouping as GV-I
but reversed.
ANC 11 in Mexico City's
local network
Mexico City is the capital and administrative
centre of the country. With its
roughly 10 million inhabitants it is one
of the biggest cities in the world and is a
centre for commerce and communications.
The existing network
The first automatic telephone exchange
in Mexico City was installed in 1929 and
was of AGF type. The local network has
since been continuously extended and
in June 1976 there were 62 automatic
local exchanges with a total capacity of
880,000 subscriber lines. Of these ex-

24
Fig 4
Grouping plan for an incoming group (GV-I) with
512 inlels and 864 outlets
Fig, 5
Schematic structure of ANC 11
GV-I Incoming switching stage
GV-0 Outgoing switching stage
BUS Bus system, duplicated
CE Central processing unit duplicated
OMP Operation and maintenance processor
TW Electric typewriter
OLY Visual display
TR Tape reader
TP Tape punch
changes 17 are of AGF type with 500switches
and the other 45 of ARF with
crossbar switches. The locations of the
exchanges are shown in fig. 8. As will be
seen, they are spread over a wide area,
so that the junction network represents
a large proportion of the total costs for
exchanges and line plant.
Twelve new exchanges are planned up
to the end of 1978 and the total number
of lines in the city will then be 1,125,000.
The annual subscriber growth over a
number of years has been 10 % or more.
This places great reguirements on the
flexibility of the equipment and on
foresight in planning.
Structure of the local network
The local network is divided into four
tandem districts, each containing a local
tandem exchange equipped with a
number of ARF group selectors type I
with 80 inlets and 400 outlets.
Direct routes may exist between the
local exchanges both in the same and in
different tandem districts, depending
on such factors as:
• the traffic dispersion
• the network structure
D type of exchange eguipment (certain
AGF exchanges of older type have no
possibility of alternative routing).
All tandem exchanges are interconnected
and are used for traffic between
local exchanges without direct routes
and for rejected traffic from direct
high-usage routes. Alternative routing
is used in most cases when the originating
exchanges are of more recent AGF
or ARF type.
Reasons for introduction of ANC 11
The group selectors of the ARF exchanges
have been successively improved
through the introduction of
group selector type II with 160 inlets and
1,600 outlets. At the same time the
tandem exchanges have been divided
into incoming and outgoing sections.
This has meant that a larger number of
routes with better availability could be
connected.
At the beginning of the seventies it was
clear that, owing to the site of the telephone
network and its rapid rate of
growth, it would be advantageous to in-

Fig 7a
Direct route and alternative route over two
tandem exchanges
Fig 7b
Direct route and alternative route over tandem
exchange in the tandem district of the originating
exchange
Fig 7c
Direct route and alternative route over tandem
point in the tandem district of the terminating
exchange
Alternative route
Direct route
Tandem exchange
Local exchange
Fig. 6
—. -• __i »C n nf AMP 11
troduce an even larger group selector
with full availability on all routes. The
restricted capacity of the existing group
selector stages made it impossible to
connect direct routes to new local exchanges,
all of this traffic had to be passed
over tandem routes. Owing to the
resulting increase of tandem traffic and
the limited capacity of the tandem
stages, an increasing proportion of the
tandem traffic had to be connected
through two tandem exchanges (fig. 7a)
instead of through one (fig. 7b and 7c).
Introduction of ANC 11
The first exchange to be eguipped with
ANC 11 was cut over in early 1976 and an
additional local exchange has been
eguipped with ANC 11 during the year. It
is planned that altogether 22 local exchanges
and 2 tandem exchanges shall
be equipped with ANC 11 group
selectors by the end of 1979 (fig. 8).
Some advantages of ANC 11 in Mexico
The junction line network can be extremely
well utilized, since ANC 11 has
— full availability, which of course allows
fewer outgoing circuits than
with limited availability (gradings)
- a very large capacity, so that several
10,000-groups can be connected to
the same outgoing or incoming
group selector stage. The routes are
thus fewer and larger, with increased
circuit utilization in consequence.
This results in large savings in a city
like Mexico City.
The good circuit utilization reduces the
requirement of inlets and outlets in the
exchanges concerned, so saving the
corresponding switching equipment as
well.
25
ANC 11 is based on SPC technique,
which simplifies administration and
planning of alterations and extensions.
The network structure can also be made
simpler owing to the advanced traffic
properties of the system.
In Mexico City's large and quickly growing
telephone network alterations in the
traffic between the exchanges occur at
frequent intervals, e.g. in conjunction
with the building of new suburbs and
industrial zones
When a new 10,000-group is put into
operation in such a suburb or zone, new
routes must be established with a large
number of existing local exchanges and
with one or more tandem exchanges.
This calls for time-consuming work in
the exchanges concerned. The intermediate
distribution frames of the
group selectors must be regraded.
Restrapping must be done in the common
control devices. In the larger exchanges
such work is in fact constantly
in progress. This constitutes also a
source of faults, sometimes disturbs the
traffic, and reguires much planning
work.
As already pointed out, however, there
is no grading work in ANC 11 and new
route data and number series are very
easily inserted by command from the
operator. The aforementioned tasks are
therefore eliminated in exchanges
where ANC 11 has been introduced.
ANC 11 also offers extended operation
and maintenance facilities. For Mexico
City the possibility of different types of
traffic statistics are of great interest in
this respect.

26
Fig 8
Local and tandem exchanges in Mexico City
already in operation and planned up to 1979
^exchanges equipped with ANC 11
Mother exchanges
These functions relate not only to the
home exchange, but valuable operational
information and traffic statistics
are obtainable also for the surrounding
network of exchanges. These statistics
and information are used in the home
exchange, but at a later stage will be
sent over a data link to the various
centres planned for future centralized
maintenance in the city.

Technical Data
Maximum values per ANC 11 stage
Number of inlets 6,144
Number of outlets (incl. KM) 6,144
Traffic per inlet C.7 erl.
Traffic per outlet C.7 erl.
Availability per route full
Number of routes 256
Number of outlets per route 256
Number of alternative routes 4
Switching time Rapid throughconnection
time, so that
the switching
time is determinedprincipally
by the
signalling
speed
Space requirement for ANC 11 (fig. 9)
COMPONENTS
— reed matrices
— special circuits for testing and
operation of reed switches
— integrated circuits type TTL
— core stores used as tabulated state
memories
— semiconductors for program stores
and temporary stores
— miniature relays
References
1. SPC-Telephone Switching System,
Metropolitan Group Selector
ANC 11. LM Ericsson Brochure No.
204526.
Fig. 9
Space requirement for ANC 11
Fig. 10
Testing of ANC 11 equipment in Mexico City
27

Transmission Properties of
Paper-insulated Twin Cables
at High Frequencies
Staffan Fredricsson
An account is given of the effect of the insulation of paper-insulated twin cables
on their attenuation and phase distortion. These characteristics are affected by the
dielectric properties of the paper which, at high frequencies, differ markedly from
those of polythene. The author also discusses the consequences for equalization in
digital transmission.
UDC621 391 8
621 315 2
Fig. 1
Attenuation for 0 9 mm twin cable with, respectively,
paper and polythene insulation
Paper
Polythene
Twin cables are employed to a large extent
as transmission medium in telecommunication
networks. Copper is
used exclusively as conductor material.
The insulation may be plastic —normally
polythene —or paper. In the great majority
of networks the cables are paperinsulated.
Twin cables are used chiefly for physical
connections, utilizing the frequency
range up to 4 kHz. But multiplexed signals
(PCM, FDM) are not uncommonly
transmitted. An increasingly high degree
of multiplexing is striven for to increase
the capacity of existing cables. In
this connection the transmission properties
up to 5-20 MHz are of interest
(digital h.f. lines for 8 and 34 Mb/s and
analogue h.f. lines for 120 and 480
channels).
For transmission on twin cable, attention
must be paid to the transmission
properties of the individual pair and to
the influence of signals on the other
pairs in the form of crosstalk. In this
article the interest is confined to the
properties of the individual pair, as it is
principally in this respect that the insulation
material exerts its effect.
By way of example there is shown in fig.
1 the attenuation for two geometrically
identical cables with, respectively,
paper and polythene insulation. Note
the marked increase of attenuation in
the MHz region for paper insulation and
the rather lower attenuation in the
speech band for this type of cable. The
latter property is one reason why
paper-insulated cable is still used in a
large number of cable installations today.
Dielectric properties
of the insulation
The explanation of the curves in fig. 1
lies in the dielectric properties of the insulation
materials. It is appropriate first
to relate these properties to the so-called
primary parameters of the line:
R Resistance per km double-circuit line
L Inductance
C Capacitance "
G Leakage
From the primary parameters the
electrical properties of the line in the
form of, for example, attenuation, phase
distortion and characteristic impedance
can be determined. All of these
parameters are dependent on the frequency.
The first two, R and L, however,
are completely independent of the insulation
properties.
It is principally two properties of the insulation
that are of interest from the
transmission point of view, the
dielectric constant i and the dissipation
factor tan 6. The capacitance is given by
C = A-F
where k is determined by the geometry
of the cable. The leakage can be written
G = i.)C tan o
where (o de notes t he angular frequency.
In polythene-insulated cable e is around
1.9 and practically independent of the
frequency. With paper insulation the
dielectric constant is affected to some
extent by the hardness of packing of the
material. Normal values at 1 kHz are
1.3-1.6. At frequencies above a few kHz
i diminishes by 1 —2 % per frequency
decade. It is the dielectric constant/
capacitance that gives rise to the lower
attenuation in the speech band for
paper-insulated cable in fig. 1. (The
dielectric constant and attenuation for
polythene insulation can be reduced by
foaming.)
The dissipation factor in polythene-insulated
cable is less than 5.10" within
the frequency range of interest. Owing
to its low value it has only a marginal
effect on the transmission properties.
For paper insulation the situation is very
much less well defined. The dissipation
factor is affected both by temperature
and by moisture content in the material.
The temperature coefficient is of the order
of 10 2 per degree C in the MHz
range. The dissipation factor is also
greatly frequency-dependent. Typical
values at different frequencies are indicated
in fig. 2. As will appear later, it is
the dissipation factor/leakage that
gives rise to the high attenuation for
paper-insulated cable in the MHz region.
This attenuation is troublesome

STAFFAN FRECRICSSON
Transmission Division
Telefonakliebolagel LM Ericsson
Fig. 2
Dissipation factor for paper-insulated cable
Fig. 3
Measured temperature coefficient Tk for attenuation
in paper-insulated twin cable (0 9 mm,
29 nF/km)
not only because of its magnitude but
also because of its great variation with
temperature. The temperature coefficient
for attenuation in the MHz region
is shown in fig. 3.
The four primary parameters referred to
above are often treated as entirely independent
quantities. When one regards
their frequency dependence, however,
it is more relevant to speak of two complex
quantities, the shunt admittance of
the cable
and its series impedance
in
[2]
Admittances and impedances cannot
have an arbitrary shape. Under very
general conditions it can be shown that
the real and imaginary parts of an admittance/impedance
are coupled together
via the so-called Hilbert transform.
Another name is Bode's integral relations
1 2 . This coupling applies also to
the shunt admittance of the cable.
The coupling implies that the frequency
dependence of the leakage G (to within
one constant term) is entirely determined
by the frequency dependence
of the capacitance C and vice versa.
Corresponding relations apply to R and
L
As a result of these various relations the
dissipation factor and dielectric constant,
for instance, are intimately associated.
A frequency-independent
dielectric constant is possible, for
example, only if the dissipation factor is
inversely proportional to the frequency
(or zero).
Attenuation and phase
distortion
Attenuation and phase distortion for a
transmission line can be determined
from the primary parameters. We shall
start by considering the attenuation and
use the approximation
[3]
This expression gives a very good
approximation of the attenuation for
frequencies above about 100 kHz.
29
The first term in [3] can be related to skin
effect losses and is essentially proportional
to VT. The leakage attenuation,
which is associated with the second
term in [3], gives rise to the heavy increase
of attenuation in the MHz region
for paper-insulated cable. This attenuation
is comparatively independent of the
geometrical dimensions of the cable,
since the factor \'LC is not appreciably
affected by the dimensions. As a rough
rule of thumb the leakage attenuation
(at 20 C) for paper-insulated cable can
be estimated at 2-3 dB/km at 1 MHz
and 20-30 dB/km at 10 MHz.
With digital transmission attention must
be paid not only to the attenuation but
also to the phase distortion introduced
by the cable. The phase distortion can
be divided into a term which grows
linearly with the frequency and a smaller
but significant non-linear term. The
linear term corresponds to pure delay of
the signal and is of little interest from
the transmission point of view.
It can be shown that, apart from the
linear phase distortion, the transfer
function of the cable (determined by
attenuation and remaining phase distortion)
is of minimum-phase character.
This property is fundamentally dependent
on the earlier mentioned coupling
between C and G and between R and L,
and implies that attenuation and phase
distortion are also inter-related via the
Hilbert transform.
In other words, the minimum-phase
property implies that, if one seeks to
approximate the cable attenuation by a
rational function with poles and zeroes
in the left half-plane (i.e. minimumphase
type), a close agreement in attenuation
gives at the same time a close
agreement in (non-linear) phase distortion.
The minimum-phase property also
implies that, if one seeks tocompensafe
the attenuation with a rational function
of this type, good equalization of attenuation
results of the same time in a
linear phase distortion, i.e. in all essential
respects in equalization also of the
phase distortion.
The minimum-phase property is of interest
not only because it makes it possible
to determine the phase distortion
from the attenuation. It also gives a hint
as to how an equalizer can be designed.

30
Fig. 4
Attenuation for paper-insulated twin cable
(0.5 mm, 37 nF/km)
\ t term
I term
Total attenuation
Fig 5
Attenuation (or paper-insulated star quad cable
(1.2 mm. 25 nF/km)
\1 term
I term
Total attenuation
Simple filter links of L, 7 and JT type, and
cascade connections of them, are in
fact of minimum-phase character.
Figs. 4-7 show measurements of attenuation
and phase distortion in two
different paper-insulated cables. Within
the frequency range the results can be
described with good approximation
with only four parameters:
[4]
[5]
In the smaller-diameter cable in figs. 4
and 6 very exact measurements have
been made for f

Fig 6
Non-linear phase distortion for the cable in fig, 4
\ i term
Mn -term
Total non-linear phase distortion
Fig. 7
Non-linear phase distortion for the cable in fig 5
\I term
f In - term
Total non-linear phase distortion
suffices to equalize with one parameter,
which must then correspond to an average
ratio between A0 and A,.
This solution does not provide
satisfactory performance in a digital h.f.
line for 8 Mb/s with max, 65 dB attenuation
at 4 MHz. This may be attributed
both to the higher frequency, which
makes the linear attenuation term and
its variations more troublesome and to
the greater absolute attenuation, which
of course also increases the variations
of attenuation.
In LM Ericsson's regenerative repeaters
for 8 Mb/s, ZAD 8-2, a manual
strapping procedure is used to adapt
the equalizers to different types of cable
(i.e. different ratios between A0 and Au
including polythene-insulated cable).
An automatic equalizer, controlled by
the pulse amplitude in the detection
31
point, compensates for minor deviations
such as temperature variations.
The characteristic of the automatic
equalizer corresponds to a fixed ratio
between A0 and A-,. Since minimumphase
networks are used in the variable
parts of the equalizer, there is
simultaneous compensation of attenuation
and phase.
Concluding remarks
Paper-insulated twin cable appears to
be unsuited to digital transmission
above 8 Mb/s. At 34 Mb/s the equalization
problems are further accentuated.
Automatic and independent equalization
of variations both in A0 and A:
seems to be necessary, which greatly
complicates the equipment. At the same
time the distances between regenerators
become very small owing to
the heavy leakage attenuation.
References
1. Balabanian, N. and Bickart, T.A.:
Electrical Network Theory. John
Wiley, 1969.
2. Bode, H.W.: Network Analysis and
Feedback Amplifier Design. Van
Nostrand, 1945.
3. von Hippel, A. R.: Dielectrics and
Waves. John Wiley, 1954.
4. Schutze, W.: Transmission on lines
technical report T 1170, LM Ericsson,
1970.
5. Mattsson, 0.: 8 Mb/s P.C.M. Line,
Equipment, ZAD 8-2. Brochure '
T/S 8436-207 e, LM Ericsson.

AXB 20-All Electronic, Stored-
Program-Controlled System for Telex
and Asynchronous Data Traffic
Eric Strindlund, Lars-Ake Andersson, Lennart Brennick
LM Ericsson has developed an all electronic, stored-program-controlled system
AXB 20 for switching of telex and asynchronous data traffic. For synchronous data
networks LM Ericsson has designed a corresponding system called AXB 30. The
latter has been selected by the four Nordic countries for the joint Nordic data
network. This article briefly presents system AXB 20 with some emphasis on its, in
terms of volume, greater use for telex traffic. A more detailed description of the
system is given in the following articled System AXB 30 will be desribedin a coming
number of this journal.
UDC 621 394 3
681 327 8
Fig. 1
An advanced control system APZ 210 has been
developed for control and supervision of LM
Ericsson's new generation of switching systems
for telephony, telex and data traffic
The rapidly increasing telex and data
traffic places new and more advanced
requirements on telecommunication
networks. To meet these demands LM
Ericsson has developed a new type of
telex and data switching exchanges
with the system designation AXB 20 and
incorporating a number of new functions.
Among these new functions may
be mentioned new types of subscriber
facilities, the possibility of higher
transmission speeds, integrated operation
and maintenance functions, the
possibility of introducing new signalling
systems, etc.
System AXB 20 is based on LM Ericsson's
long and extensive experience of
both telex exchange systems and
stored-program-controlled systems for
telephony.
The switching system in AXB 20 is controlled
by the control system APZ 210 2
which was developed specially for control
and supervision of switching equipment
in telecommunication systems
(fig. 1).
Application
Telex networks
System AXB 20 is intended for use
primarily as a combined local and trunk
exchange both for national and international
traffic (fig. 3). The first exchange
of this type will be opened in Malmo in
1977.
System AXB 20 can be extended
stepwise from about 1,000 up to 30,480
circuits (subscriber and trunk lines). It
has a very high traffic capacity, being
designed for 60 calls per second, or
around 6,000 erlangs. The system is intended
for telex and for asynchronous
data traffic up to 300 bauds.
Telex, the dominating service today,
uses a standardized alphabet (International
Telegraph Alphabet No 2) and a
uniform speed, 50 bauds. This has been
the reason for using a separate switching
matrix for telex. The switching
system, APT 601, is based on a bitoriented
time division multiplex technique
and is regenerating, i.e. received
characters are retransmitted in ideal
form.

ERIC STRINDLUND
LARS-AKE ANDERSSON
LENNART BRENNICK
Data Communication Department
Telefonaktiebolaget LM Ericsson
Fig. 2
LM Ericsson is today one of the world's largest
suppliers of equipment for telex and data
switching exchanges. Of LM Ericsson's crossbar
system ARB/ARM for telex, which was introduced
in the early sixties, there are in 1977 more than
250,000 multiple positions for subscriber and
trunk lines in operation oron order in more than 20
countries
Fig. 3
Modernization of telex network with system AXB
20 for both national and international circuits
Existing national exchange
National/international exchange
National telex exchange
Terminal exchange
Multiplexor/Concentrator
Asynchronous data networks
For asynchronous data networks a
separate switching system, APT 603,
has been designed, based on the same
technique as switching system APT 601
and, like the latter, regenerating. The
switching system can handle the codes
and speeds laid down by CCITT in Recommendation
X 1, User Classes 1 and
2.
Switching systems APT 601 and APT
603 can be combined in one exchange
and controlled by one pair of processors
APZ 210 (fig. 4).
System properties
As mentioned in the introduction,
system AXB 20 has a number of new
properties, which can be broadly listed
under three headings: flexibility, ease of
handling, and reliability.
The properties presented in this context
relate primarily to telex applications of
the system, since telex is the entirely
predominant service in terms of traffic
volume.
Flexibility
— In transmission
owing to optional line adapters for
high or low level signalling for adaptation
to different methods of trans­
mission, and owing to integrated
connection of division multiplex
transmission equipment. The regeneration,
furthermore, permits teleprinters
and lines to work higher
distortion than was previously possible.
In signalling
owing to the structuring of the software
system into separate function
blocks, which can be altered, extended
or exchanged, for example for
compliance with new international
recommendations.
In new subscriber facilities
which are programmed in independent
function blocks. These new
facilities will be described below.
In traffic routing
owing to unlimited number of possibilities
offered for the forming of
routes, alternative routes, one- and
two-way circuits, the establishment
of closed user groups, etc.
In charging
owing the various possibilities that
exist, e.g. multimetering, toll ticketing,
and a combination of the two.
The calculation of charge for a set-up
connection can be based on several
different criteria, e.g. destination,

34
Fig. 4
In system AXB 20 separate switching systems are
used for telex and asynchronous data
duration, services used such as store
and forward, etc.
— In numbering
since AXB 20 is designed for free selection
of subscriber number to every
line adapter in the exchange area.
— In packaging structure
which is adapted to the modular
structure of the function blocks with
electrical and functional interfaces
coinciding with the mechanical interfaces
of the building modules
(figs. 5 and 6).
The packaging structure is described
in Ericsson Review No. 2, 1976 3 .
— In adaptation to future requirements
which can be done without extensive
alterations of software or hardware
owing to the functional modularity of
the system.
Handling properties
Each functional module has standardized
interfaces with all other functional
modules. A functional module is thus
fully defined by its interfaces. Work can
therefore be done on the system without
knowing in detail the structure and mode
of functioning of all interacting functional
modules, which is a great advantage
from the administrative aspect.
It is especially in the handling of the
software that this structure in the form
of functional modules has great advantages,
since each corresponding program
block can be compiled and tested,
and also maintained in operation, as a
separate unit independently of the other
blocks.
A consistent principle in system AXB 20
is that functions which are normally
subject to change are implemented in
software, while those which will normally
not be changed are implemented in
hardware.
This principle makes system AXB 20
easy to handle from the operation and
maintenance point of view.
Modifications of exchange data or functions
are made mostly by changes of
programs and/or data without disturbance
of the traffic.

Fig. 5
A common, entirely new packaging structure,
called BYB, is used (or LM Ericsson's new
generation of communication systems for
telephone, telex and data traffic
The modular packaging structure makes it
possible to adapt the mechanical arrangement
Reliability
Compared with commercial computers
in automatic data processing systems
the control system in telex/data switching
exchanges must comply with very
much higher requirements of reliability.
In system AXB 20 the necessary hardware
reliability has been attained by
duplication of all traffic-handling subsystems
and by the use of thoroughly
tested, dependable components.
The system is also designed to eliminate
the effects of a major fault. For example
a fault in the control system or in one of
the parallel/synchronously operating
switching systems does not affect the
traffic handling capacity.
Reliable and easy-to-handle software
In stored-program-controlled systems
software properties such as reliability
and ease of handling are of the very
greatest significance for the operational
result.
In AXB 20 the programming has been
greatly simplified and made more efficient
through a software structure with
well defined interfaces between the
program blocks and through the use of
a high level language, PLEX, for the encoding
of programs.
35
The software structure permits testing
of individual program blocks by simulation
in the programming system, using
IBM or UNIVAC computers, regardless
of whether interworking blocks are
available or not.
The interworking between different
function blocks is effected by microprograms.
Address calculations, which
were a fairly common source of fault in
earlier SPC systems, and also a lengthy
procedure, have thus been simplified in
system AXB 20.
Programs for a function block have access
only to the specific data belonging
to that function block, i.e. the effect of a
program error is limited to a single function
block. Program errors can also be
rapidly detected through the means that
exist for checking the signals between
the blocks.
In other words, system AXB 20 is so designed
that software reliability has been
easy to retain. It is also simple to correct
program errors, or to change a program
sequence at the individual exchanges.
Subscriber facilities
Apart from the facilities offered by earlier
Ericsson telex systems, i.e. keyboard
selection, printed service signals, PBX

36
Fig. 6
The hardware in a function block corresponds in
mechanical construction to a building module
called a magazine, composed of a PC board frame
and a wiring unit
groups, closed user groups, etc., a number
of new facilities have been introduced
in system AXB 20 which give subscribers
even better means for exchange
of messages.
There follows a brief account of some of
the most common of these new facilities.
Abbreviated address
The abbreviated address facility enables
a subscriber to store in the AXB 20 exchange
a number of telex numbers
which he often calls, e.g. of branch offices,
agents, suppliers, etc. An abbreviated
telex number is called by the subscriber
keying a code consisting of 1 —3
letter or numerals at his own choice.
Hot line
When a subscriber presses his call button,
he is automatically connected to a
predetermined number. This is usually a
more advantageous facility than the
present method of leased circuits. The
use of this facility can also be assigned
to specific periodically recurring intervals
(scheduled calling).
Store and forward facility
The store and forward facility enables a
subscriber, when he encounters busy
condition, congestion or the like, to
store his message(s) in the AXB 20 exchange,
which forwards them automatically
at a later time.
A subscriber knows that, towards the
end of office hours for example, there
will be little possibility of reaching certains
destinations and can then instead
send his messages to the AXB exchange
for later forwarding. This has a trafficequalizing
effect in telex networks.
Broadcast and conference calls
The broadcast and conference call
facility permits the transmission of messages
to several subscribers simultaneously.
This facility can be combined
with the abbreviated address facility, so
enabling a call to be made to several
subscribers on one abbreviated number.
Operation and maintenance
The costs for operation and maintenance
of a telex/data switching exchange
today consist predominantly of
salaries and wages. The constant rise of
these costs is another reason for automating
the operation and maintenance
functions as far as possible. This applies,
of course, not only to the switching

Fig. 7
Great attention has been paid in system AXB 20 to
the selection of components in order to ensure the
use of reliable units. Here is a PC board for the
equipment but also to other parts of the
telex network such as subscriber equipment,
trunk and local lines, etc.
As already noted, the operation and
maintenance functions in AXB 20 are to
a high degree automatic and integrated
in the normal mode of functioning of
the system. Preventive exchange maintenance
in system AXB 20 is required
only in the form of a minor inspection of
certain input and output devices I/O.
Anotherdifference compared with older
systems is that alterations, extensions,
traffic rerouting, etc., are effected by
commands from an I/O device, which
involves very much less work and allows
simpler and better documentation.
Operational functions
The administration of such functions as
subscriber facilities, tariffs, charging
routines, etc., is greatly simplified in
system AXB 20.
These operational functions are initiated
and supervised from input and
output devices I/O placed either in the
AXB 20 exchange building or, fora large
area, in a common operation and maintenance
centre linked to the AXB 20 exchange
by a data channel.
37
The operational functions consist of the
administration of
- subscriber terminations
- routes
- charging, international accounting,
tariffs
- traffic routing
- traffic statistics
Great pains have been taken to simplify
the control and supervision of these
functions. By way of example for alteration
of a route analysis table the modified
table is prepared and tested separately
down to the last detail before being
put into service.
Charging is based on the multimetering
ortoll ticketing principle. The necessary
data are collected in the exchange and
stored in an external store, usually magnetic
tape. These data later go on to the
administrative routines for processing
of billing data. In this form the material
is also suited for production of various
statistical data.
Maintenance functions
System AXB 20 contains a large number
of functions which simplify the maintenance
of the switching equipment, and
to a large extent too the maintenance of
the subscriber and transmission equipments.

38
Fig. 8
Computerized functional testing of PC board
The bit-oriented regeneration of incoming
characters is an example of such a
function. This regeneration takes place
in conjunction with transfer, the incoming
characters being regenerated on a
bit basis and retransmitted in ideal form.
System AXB 20 thus offers the important
advantage that a higher distortion
can be allowed than was earlier acceptable
both on subscriber and trunk
lines.
A considerable item in the maintenance
costs for telex networks is precisely the
maintenance of the subscribers' teleprinters,
which requires regular adjustment
of the send and receive margins.
As the AXB 20 exchange sends undistorted
signals to the subscribers' teleprinters,
the receive margins need not
be adjusted as often as is necessary today.
The send margins of the teleprinters
may also vary within wide limits,
as the AXB 20 exchange has a receive
margin better than 46 %.
The regeneration function therefore
considerably reduces the maintenance
especially of teleprinters, but also of the
transmission equipment.
System AXB 20 contains a number of
other automatic functions which effectively
contribute to simplified maintenance
of the line plant, e.g.
— automatic adjustment of the line current
on all lines
— automatic supervision of loss of line
current
— continuous automatic supervision of
distortion on all lines
— automatic test facilities for all lines.
For automatic supervision of distortion,
measured values are compared with
predetermined threshold values and, if
the latter are exceeded, an automatic
alarm is received in the form of a printout
on an appropriate output device.
Maintenance of hardware
In system AXB 20 built-in fault-detecting
functions are used for automatic
tracing of faults in the hardware. Examples
of such functions are:
— comparison between two parallel/
synchronously working units
— continuous generation and transmission
of test sequences on characteror
bit basis on special supervisory
circuits. A transmitted sequence is
compared with the received sequence
and any deviation gives rise
to alarm.
— parity checks.

Fig. 9
In many telex exchanges the available space is in
great demand owing to the constantly growing
traffic. Above is shown on the right a
national/international telex exchange based on
crossbar switches. On the left is an AXB 20
exchange equipped for the same traffic capacity
Maintenance of software
An important aspect of exchange
maintenance is maintenance of the
software. Modifications of the software
normally occur in conjunction with operationally
conditioned changes of exchange
data, e.g. connection of new
subscriber groups, rerouting, changes
of subscriber facilities, new tariffs, etc.
System AXB 20 has a software structure
with consistently standardized interfaces
between the program blocks. Internal
addressing of programs and data
is used within each block. Recalculation
to a desired address in a store takes
place on execution of the program and
is based on start addresses and sizes of
data areas specific to the program block.
An error in a program can thus not affect
other data than those associated with
its own block. There are also admirable
tracing facilities and means for separate
compiling, testing and loading of program
blocks. In combination with the
relocatability of the program blocks this
has essentially contributed to simplification
of the exchange software.
AXB 20-costs
The total cost of a telex/data installation
may be broken down into three parts:
1. Capital costs
for investments in buildings, switching
and transmission equipment, local
and trunk line plant, etc.
2. Operational costs
salaries of operators, operational
staff, charging routines, power consumption,
etc.
3. Maintenance costs
for fault repair in subscriber equipment,
lines, switching equipment,
etc.
The introduction of an AXB 20 exchange
in a telex network has an immediate
effect on all of these costs, as
illustrated below.
Lower capital costs
Reduced space requirement
System AXB 20 is space-saving. The
-aatample in fig. 9 illustrates the saving of
39
space that is possible In the figure a
crossbar exchange for 4,200 trunk lines
and 3,600 subscriber lines is compared
with an AXB 20 exchange equipped for
the same traffic capacity
Better utilization
AXB 20 permits better utilization of existing
line and switching equipment,
among other means through such
facilities as hot line, closed user groups,
alternative routing, low level signalling,
etc.
One result of low level signalling is that,
from the interference aspect, less restrictivity
need be adopted in the choice
of cable pairs for telex in the subscriber
line network.
Lower operations costs
Labour-saving operational routines
Alterations of functions, rerouting, extensions,
etc., are done in software by
commands from an I/O device
Rational programming aids are furnished
for operation of the exchange.
The function-oriented command language
provides rapid and reliable communication
between staff and the processor
system.
Data for charging and accounting
collected automatically
The charging and accounting information
is delivered from AXB 20 on magnetic
tape, which allows rational processing
in, for example, a data processing
installation.
Simplified traffic measurement and
operational statistics
Traffic measurement is initiated by
command. The desired output data are
recorded on magnetic tape, from which
the relevant information can be quickly
obtained.
Lower maintenance costs
Cheaper maintenance of teleprinters
Larger distortion margins can be allowed
in subscribers' teleprinters both
in the receiving and sending direction,
as the AXB 20 exchange is regenerative
and has a receive margin better than
46 %.

40
Simplified maintenance of switching
equipment
AXB 20 has automatic fault-detecting
and supervisory functions which relieve
the exchange staff from routine maintenance
work. Continuous supervision of
the traffic quality in traffic-handling subsystems
is integrated in the normal functions
of the system, so reducing the
number of man-hours on fault tracing
and correction.
The function-oriented command language,
rational training, etc., are examples
of technical aids and services
through which the staff can quickly
learn to handle an AXB 20 exchange effectively.
Automatic supervision of distortion
The switching system continuously
supervises the distortion of incoming
lines and an alarm printout is received if
the preset threshold is exceeded, i.e.
action to correct distortion is called for
only when required.
Automatic distortion measurement
The AXB 20 exchange is equipped with
automatic distortion measurement
equipment complying with CCITT Recommendation
R 79.
Distortion measurement can be programmed
to take place automatically
during, for example, low traffic periods.
The result of the measurement is presented
by a printer.
Automatic line supervision
AXB 20 automatically indicates a loss of
line current and automatically adjusts to
the correct line current.
All lines can be tested and specially investigated
from a test desk, from which
other functions as well can be manually
tested. A line to be measured is connected
metallically to the test desk by
command from an I/O device. This test
desk is referred to as Engineering Control
Board, ECB and is shown on page
45, fig 6.
References
1. Strindlund and others: AXB 20-
Description of System. Ericsson
Rev. 1. 1977, pp. 41-51.
2. Eklund and others: AXE 10-System
Description. Ericsson Rev. 2,
1976, pp. 70-89.
3. Alexandersson, R. and Rörström,
H. O.: New Packaging Structure for
Electronic Switching Equipment.
Ericsson Rev. 2, 1976, pp. 100-
107.

AXB 20—Description of System
Eric Strindlund, Lars-Åke Andersson, Lennart Brennick
A summary account is given of the structure of system AXB 20 with the emphasis
on its most characteristic property- the functional modularity both in hardware and
software.
The applications and general properties of the system are summarized in a preceding
article in this number.
UDC 621 394 3
681 327 8
Fig. 1
System AXB 20 has a four-level functional
structure:
System structure
Telex and data traffic are in rapid
growth. At the same time a technical development
is taking place in the field of
telex and data switching exchanges,
with the emphasis on electronics, time
division multiplex being used in the
switching network and stored program
control in the central processing system.
An important property of the system is
its ability for flexible adaptation to
changed conditions, e.g. to changes of
traffic volume, new signalling systems,
new subscriber facilities, new types of
teleprinter, etc.
System AXB 20 possesses this flexibility
to a high degree owing to the functional
modularity that has been introduced
throughout the hardware and software.
This means that an AXB 20 exchange is
easy to handle primarily from the operation
and maintenance point of view, but
also during the various stages preceding
cut-over, i.e. specification, production,
documentation, and installation.
The modularity also facilitates extensions
of the system.
The modular structure of AXB 20 is of
fundamental significance and will
therefore be illustrated from three
aspects.
Operative modularity
The operative modularity of the system
means that it is flexible in providing for
needs of increased traffic capacity, increased
line capacity, rerouting of traffic,
etc.; in other words, it is easily
adaptable to changes in the traffic envi-
••rment in which the exchange works.
Administrative modularity
The functional modularity of AXB 20,
which results in simplified administration
of the exchange both before and
after cut-over, is here called administrative
modularity.
After cut-over of the exchange the administrative
modularity permits simplified
routines for maintenance, service
supervision, programming, fault tracing,
etc., which together signify higher
reliability.
During the stages before an exchange is
put into operation administrative modularity
implies more rational methods
for technical adaptation, production,
documentation, installing, training,
etc., which in total result in a better and
more economical system.
Technical modularity
Technical modularity is no new concept
in telecommunications. It isthe property
which has made it possible to create
global telecommunication systems with
their interworking and interchangeable
local exchanges, trunk exchanges and
transmission systems. This technical
flexibility has been achieved by using
well defined interfaces between the
component parts of the system. This
principle has been adopted in system
AXB 20, which is correspondingly made
up of function blocks with strictly defined
interfaces.
System AXB 20 can therefore be easily
adapted to new functional requirements
and, owing to its exchangeable functional
modules, there is no practical difficulty
in the introduction later on of
new solutions or system components if
warranted by the technical development.
The basic structural principle which
runs through the whole of AXB 20, as
shown schematically in fig. 1, is thus a
consistently implemented functional
modularity. At the highest level, the
system level, AXB 20 consists of the
switching systems for telex and asynchronous
data and of the control
system APZ 210. Some of the basic features
of these systems are presented below.

42
Fig. 2
Schematic presentation ot through-connection in
a switching matrix of TDM type. A connection
seizes two time slots, one in the forward and one
in the backward direction
The figure shows how a unit pulse is connected through
from incoming circuit no. 3 to outgoing circuit no. 1 by
change of time slot in the next addressing cycle.
The switching system
APT 601
LM Ericsson's extensive experience of
the special requirements placed by
telex traffic has been drawn upon in the
development of system AXB 20 and its
switching system APT 601.
In telex applications the following properties
of the switching system are of
particular interest:
— high traffic capacity
— stepwise extension of capacity to
30,480 lines
— regenerative switching function
— full availability without internal congestion
— rapid character transmission
— autonomous switching functions
— high reliability owing to parallel
synchronous operation.
The switching system is dominated by
the duplicated switching matrix which,
via busses, interworks with the traffichandling
programs in the central processor
(fig. 8).
The switching matrix works on the time
division multiplex principle. Each
channel has its own time slot, so that no
internal congestion occurs in the
switching matrix (fig. 2).
Incoming telegraphy characters are
handled in the switching system in such
a way that each unit element is separately
decoded and transmitted via the
switching system to the outgoing circuit
(CCITT advocates this decoding
method in Recommendation R 101).
The delay through the switching system
is therefore small, amounting to at most
two unit elements.
Apart from setting up of connections,
the switching system performs other
very essential functions such as regeneration
of characters, supervision of
distortion, signal reception and transmission,
etc.
Functions such as detection of calls,
character recognition, reception and
transmission of service signals, and
supervision of established connections,
are implemented in hardware. These
functions are to a large extent
performed autonomously by the switching
system and therefore do not load the
central processor. The capacity of the
control system can thus be used more
rationally for complex tasks requiring
intelligence in conjunction with the
analysis of selection signals, signalling,
charging, service supervision, etc.
The division of the tasks between
switching and control systems for the
establishment and clearing of a connection
is shown schematically in fig. 3.
Supplementary functions can be connected
to the switching system, e.g.
operators' positions for manual service
and assistance, store and forward
equipment, etc.
Structure of the switching system
The switching system performs the traffic-handling
functions but also under-

Fig. 4
The modular structure of system AXB 20 makes it
possible to extend an exchange in economical
stages
Fig. 3
Division of tasks between switching and control
systems during the establishment and clearing of
a connection through an AXB 20 exchange. The
presentation is summary and is simplified by the
elimination of certain functions
Call
Free
H Switching system APT 601
Control system APZ 210
A-subscriber
B-subscriber
A-subscriber
B-subscriber
Free
Free
takes certain associated operation and
maintenance functions.
A summary account follows of the modular
structure of the switching system
and its functions, and of the principle
for establishment of a connection by
TDM technique.
Functional modularity
The switching system is made up of
functional modules and consists essentially
of multiplexor/demultiplexor stage
TCM, decoding modules TDM, and central
logic with switching and buffer
stores (fig. 5).
The first multiplexor/demultiplexor
stage TCM isgrouped in modulesof 256
inlets, 254 of which are used for connection
of line adapters and 2 are reserved
for service observation.
Call arrival
The switching system continuously scans all line
adapter circuits and sends a call confirmation if a call
arrives.
Scanning and call confirmation do not load the control
system.
Setting-up of connection
At the initiative of the control system the switching
system sends a proceed-to-select signal to the calling
subscriber. The selection information is evaluated by
the control system, which thereafter establishes the
connection through the switching system.
Established connection
The switching system autonomously transmits characters
from incoming to outgoing line and identifies
the clearing signal.
The control system is not loaded by character handling
for an established connection.
Clearing
The control system sends via the switching system a
clearing signal, records charging data, and marks the
circuits free.
43
The line adapter circuits are grouped in
building modules, so-called magazines,
each accommodating 12 PC
boards. Each board is equipped for two
telex lines. The magazine thus accommodates
24 telex lines.
The magazine is equipped with the
number of PC boards required, so that
the smallest extension unit in AXB 20 is
a line adapter board, i.e. two telex lines.
Decoding modules, switching and buffer
memories, etc., are built in units
common to 2,048 inlets, so-called
2000-groups.
Other central switching equipment is
common to three 2000-groups, e.g. the
interface module with the control
system, service observation equipment,
etc., and forms with the other modules a
6000-group.
A 6000-group is the largest unit in the
switching system APT 601. Five 6000groups
together form a full-capacity
AXB 20 exchange for 30,480 lines.
The modular structure of the switching
system makes it possible to increase the
line capacity in economical stages (fig.
4).
Through-connection with TDM
technique
The principle for the flow of information
from incoming to outgoing circuit is
shown in fig. 5.
The information on the incoming circuit,
i.e. the signal state, is read via TCM and
TDM into the switching memory SM, in
which each incoming circuit has its own
word for buffering of this information.
This word includes other information as
well, e.g. as to whether t he circuit is free,
busy, blocked, through-connected for
telex transmission, etc. This information,
which is written in via the control
system, indicates, how the incoming information
from the line is to be handled.
In this case it must be forwarded to the
desired outgoing circuit.
The information in the switching memory
is transferred during the next
addressing cycle to the buffer memory
BM, where it is stored in the address
corresponding to the outgoing circuit.

44
Fig. 5
Hardware structure of the APT 601 switching
system
BM Buffer memory
LA Line adapter
SM Switching memory
TCM Multiplexor/demultiplexor
TDM Decoder
TPM Interface with the control system
The through-connection in fact consists
precisely in the addressing of the
incoming information on a circuit to an
optional address, i.e. to the desired outgoing
circuit.
The address of the outgoing circuit is
read by the control system into the
circuit word in the switching memory on
the basis of the selection signals recieved
at the time of setting up of the
connection. The address remains in the
circuit word until the connection is
cleared.
From the buffer memory the information
is read out to the demultiplexer
stage TCM, where the called line is accessible.
In this way the information has
been transferred from incoming to outgoing
line adapter.
Main functions of the switching
system
The main functions of the switching
system may be divided into the following
three groups 2 :
— traffic handling functions
— operation and maintenance functions
— supplementary functions
Traffic handling functions
In system AXB 20 the interface with the
line side consists of the line adapter
circuits LA, where the incoming and
outgoing trunks and subscriber lines
are connected to the first multiplexor
stage in TXS (Telex Switching Subsystem)
(fig. 8).
The TXS subsystem carries out the setting
up of connections, clearings,

Fig. 6
Connected by command, any line or circuit to an
AXB 20 exchange can be measured from this Engineering
Control Board ECB
transmission and reception of signals,
etc. It also performs certain service
observation and supervisory functions,
e.g. regeneration, continuous test pattern
generation for functional checks,
check of distortion, time measurement,
etc.
The more complex switching functions
are performed by TTS (Telex Traffic
Handling Program Subsystem), which is
made up of software.
The TTS subsystem consists of a
number of function blocks, which may
be divided into five groups with the following
functions:
1. Handling of data relating to trunks,
subscriber lines, and to the TXS subsystem
2. Handling of central analysis functions,
e.g. analysis of selection signals,
route numbers, etc.
3. Handling of charging, accounting,
statistics
4. Handling of subscribers facilities, e.g.
hot line, broadcast/conference call,
call diversion
5. Handling of signal transmission during
the setting-up and clearing
phases of a connection.
Operation and maintenance functions
of TMS and TLS subsystems
The TMS subsystem administers chiefly
the operation and maintenance functions
which are implemented in software.
By commands to TMS, changes
are made in subscriber and circuit data,
and in routing, and other normally occuring
operational measures are carried
out.
Through TMS the mode of functioning
of the switching system is supervised on
the basis of the periodically recurring
character pattern. On encountering deviations
from the normal mode of functioning
TMS takes action in the form of
alarm and possibly change-over to the
parallel switching system.
TMS is engaged also in start-up of the
system, updating, program modifications,
etc.
The TLS subsystem comprises hardware
for measurements on trunks and
subscriber lines. The measurements are
made from an engineering control
board, to which an optional line is metallically
connected via a connection relay
in the line adapter circuit for that line.
Measurements, e.g. distortion measurements
of routes, are programmed to
be carried out automatically with the aid
of the corresponding maintenance
program.
From the engineering control board direct
measurements are made of line
current, voltage, and leakage resistance
for each line. The distortion on
the line can also be measured and read
directly on an instrument in the control
board (fig. 6).
Supplementary functions
Manual service
Operators' positions equipped with visual
displays for manual assistance in the
setting up of connections can be connected
to the switching system.
These positions can be connected to
the AXB 20 exchange from a remote
point via modems, which permits this
service to be located at any point in the
network. Examples of functions performed
at the operators' positions are
- international telex traffic to countries
requiring manual assistance
— information services
45

46
Fig. 7
Line adapter equipment for two telex subscribers
with double-current signalling
— assistance in setting up of broadcast
and conference connections, e.g. for
subscribers who have not subscribed
to the corresponding automatic
services.
Other functions can naturally be added
to AXB 20 when required.
Store and forward
The store and forward facility provides
the means of storing messages received
by the AXB 20 exchange for later transmission,
e.g. during a low-traffic period.
This function is performed by the TSS
subsystem, which includes disc stores
for storage of messages and miniprocessors
for handling of the disc stores.
TSS also includes the necessary software
for this function.
Line terminations
Trunks and subscriber lines can be
connected to an AXB 20 exchange in
arbitrary proportions.
The interface between line and exchange
sides consists of the line adapter
circuits. Fig. 7 shows a PC board for
termination of two telex subscriber lines
with double-current signalling.
As AXB 20 works with low voltage levels
on the exchange side, the line adapter
circuits have been provided with opto-
couplers which metallically separate
the line side from the exchange side.
Several new facilities have been introduced
in the line adapter circuits,
among which may be mentioned:
— automatic adjustment of the line current
— alarm on loss of line current
— overvoltage protection against transients
up to 5 kV
— physical through-connection to engineering
control board for measurements
on the line
— automatic clearing of line-current
alarm.
The control system APZ 210
The control system APZ 210 was specially
developed by LM Ericsson for
control of switching equipment. It is
characterized by high traffic-handling
capacity, high reliability and good handling
properties.
The hardware structure of the system is
illustrated in fig. 8, showing the following
subsystems:
— the central processor subsystem
CPS
— the regional processor subsystem
RPS
— the input output subsystem IOS
— the maintenance subsystem MAS

Fig. 8
System AXB 20 consists of the APT 601 switching
system and APZ 210 control system. The illustration
shows the hardware structure of these two
systems
APT 601 Telex switching system
APZ 210 Control system
CPS Central processor subsystem
CPU Central processing unit
DS Data store
IOS Input output subsystem
LA Line adapter
MAS Maintenance subsystem
MAU Maintenance unit
PS Program store
RP Regional processor
RPS Regional processor subsystem
RS Reference store
TLS Engineering Control Board
TOS Operator for manual assistance calls
Store and forward equipment
The central processor subsystem CPS
is in charge of t he complex, intelligencedemanding
functions, while t he regional
processors perform simpler, repetitive
and capacity-demanding functions.
All exchange of information between
exchange staff and AXB 20 system passes
through the input output subsystem
IOS. The greatest possible simplicity in
the man-machine communication and
high flexibility in respect of future technical
developments have been fundamental
requirements in design of IOS.
47
The maintenance subsystem MAS comprises
built-in functional checks for dependable
operation of the APZ 210 control
system, chiefly supervision of the
synchronous parallel operation of the
central processor subsystem CPS.
Central processor CP
The central processor contains the
central processor unit CPU with associated
stores and operative programs
for job scheduling and loading of programs.

48
Fig. 9
The central processor is made up of a number of
self-contained units
DS Data store
PS Program store
RS Reference store
CPB Central processor bus
RPB Regional processor bus
MAU Maintenance unit
CPU Central processor unit
TCU Table and counter unit
RPH Regional processor handler
ALU Arithmetic and logic unit
BAM Maintenance buffer unit
MIG Microinstruction generator
PCU Priority control unit
TRU Trace unit
DSH Data store handler
LIU Link and instruction address unit
PSH Program store handler
UPM Updating and match unit
RSH Reference store handler
SBU Shift and bit handling unit
The central processor unit CPU, shown
in fig. 9, is made up of self-contained
function units connected to a common
bus.
The work of the central processor is
controlled by microprograms stored in
programmable read only memories
(PROM) in the microinstruction generator
MIG.
Microprogramming has made it possible
to use powerful machine instructions
adapted to the high level language
PLEX.
The central processor interworks with
the APT switching system and with the
regional processors RP. The central
processor has three stores,
— the data store DS
— the program store PS
— the reference store RS
The central software in a function block
consists of programs and associated
data. Data belonging to afunction block
can be accessed only by the programs
of that function block. Combined with
the register structure and an effective
procedure for interwork between program
blocks (which is always effected
with special software signals), this has
resulted in a very high degree of software
reliability.
Only one function block at a time is active
in the central processor, and this
block has its number temporarily stored
in the block number register in CPU.
Storage in the block number register
gives access to the area in the reference
store RS which is reserved for the block.
From this area are taken the start
address of the function block program
in the program store PS and the start
address of the base address table for the
block in the reference store RS.
The base address table contains the absolute
addresses of the variables in the
data store, the length of the variables,
etc.
All software and all data areas can be
relocated by changing the information
in the reference store. This results in
optimal use of the available storage
areas.
This repeatability is valuable for all
handling of programs and data, e.g. for
change of functions, program modifications,
extensions, etc.
An extension, for example connection
of a new subscriber group, is extremely
simple to introduce: it is initiated by a
command from an I/O device.
If a data area of sufficient size has been
reserved for this extension, the central
processor can execute the command
without hindrance. But if data area has
not been reserved, a free area of sufficient
size must first be assigned.
When this new area has been assigned,
the information is transferred (or rather
copied) from the old area to the new.
By a special indication in the base
address all changes of data in the old
area caused by normal traffic handling
will also be transferred to the new area.
An extension in progress will therefore
in no way disturb the normal traffic
handling by the program block.
After all information has been transferred
to the new area, the base address is
switched over to it, after which the old
area can be released. Data area of sufficient
size is now available for the desired
extension and the extension
command can be executed without
hindrance.
When required, the data store is repacked
in order, for example, to obtain a
continuous free area.
In order that a functional change maybe
effected without disturbing the traffic
handling, the signalling interfaces between
the program blocks as well as the
internal data structure of the blocks
must be retained unchanged.
The control system APZ 210, however,
permits more radical changes of function,
and for such purpose the duplication
of the central processor is utilized
for so-called separation. One central
processor is taken out of service and
loaded with the new software, after
which it is switched back into service

Fig. 10
Priority of central processor tasks
PCU Priority control unit
MFL Malfunction level
TRL Trace level
THL Traffic handling level with sublevels 1-3
BAL Base level with sublevels 1-2
MAU Maintenance unit
TRU Trace unit
JBA Job buffer A
JBB
JBC
Job buffer B
J° b buffer C
inh buffer D
.nn
Job scheduling
and priorities
Job scheduling in the central processor
CP is carried out mainly by microprograms,
with four program priority levels
as follows:
MFL Malfunction level
TRL Trace level
THL Traffic handling level
BAL Base level for supervisory functions
The malfunction level has the highest
priority. When a fault detecting circuit
has found a hardware fault which must
be quickly corrected, an interrupt signal
is generated to MFL, which then takes
the necessary action to limit the effects
of the fault.
49
The trace level TRL is used solely for
program testing and is an admirable aid
for quickly correcting any defects in the
software.
The normal tasks of the central processor
are carried out on the traffic handling
level THL and base level BAL
The job scheduling in the central processor
CP is controlled by a clock interrupt
system using the four job buffers
JBA, JBB.JBCand JBDshown infig. 10.
In these buffers are stored incoming
signal messages which call foraction by
the central processor. When such a
message is placed in a buffer, an interrupt
signal is automatically generated
and interrupts the work in progress on

50
Fig. 11
Block diagram of the input output subsystem IOS
the BAL level. Work in progress on the
THL level, on the other hand, is not interrupted
but is completed before the
next work is started.
All signals from the switching networks
and regional processors, and most internal
signals between the central processor's
program blocks, are stored in
the job buffers.
Certain program blocks in APZ 210
perform time measurements and these
functions are controlled by clock interrupt
signals to THL level at 10 ms intervals.
Regional processors RP
The main job of the regional processors
is to relieve the central processor from
simple, frequently recurring routine
functions such as supervision of input
and output devices I/O, alarm scanning,
etc.
A regional processor consists of a
central processing unit CPU, a data
store DS, a program store PS, and an
operative program for administration
and supervision of the program handling.
CPU is made up of self-contained function
units connected to a common bus
RPB. The work of the regional processors
is controlled by microprograms
stored in PROM memories.
On detection of a fault in a regional processor,
work in progress in the processor
is immediately interrupted and a
fault signal is sent to the maintenance
programs in the central processor. Depending
on the extent of the fault the
central processor can then transfer the
work to a fault-free regional processor.
The input output subsystem IOS
As already noted, the input output subsystem
IOS is used for all exchange of
information between the exchange staff
and the AXB exchange. Thus all input
and output of programs and data for
loading of programs, commands, alarm

printouts, readings, change of store
contents, etc., pass through the IOS
subsystem.
The greatest possible simplicity has
been striven for in the design of the
routines for man-machine communication.
The functional structure of the IOS subsystem
is shown in fig. 11, from which it
will be seen that various types of input
and output devices can be connected,
eg.
- typewriters
- visual displays
- magnetic tape cartridges
— disc stores
— data channels
— alarm indicators, etc.
Particular attention has been paid to the
problem of producing an easily understandable
and therefore easy-to-handle
command language. The command
language used in AXB 20 has the same
structure as the command language in
LM Ericsson's new generation of telephone
switching systems, AXE.
Various functional codes exist for controlling
printouts of a certain type of information
to predetermined output devices,
which simplifies the handling of
output data from the processor system.
Maintenance subsystem MAS
The maintenance subsystem MAS ensures
that a non-functioning central
processor unit is taken out of service.
Fault diagnosis and elimination of the
effect of faults in AXB 20 is thus incorporated
in the normal automatic
functions of the system.
The maintenance programs work in intimate
contact with the hardware and
make use of the redundancy of the
central processor and the regional processors.
Each central processor has the following
types of built-in fault detecting
circuits:
1. Supervisory circuits for indication of
faulty processor side through parity
check of stores and buses, voltage
and time supervision, etc.
2. Supervisory circuits for comparison
51
between the processor sides detect
faults by comparing data on the internal
buses of the two processors.
3. Program supervising circuits which,
by time measurement, detect when
the processor program-handling is
not functioning normally.
4. Routine checks of the central processor
hardware through continuously
recurring maintenance programs.
When a hardware fault has been detected
in the central processor, a fault
signal is sent to the maintenance unit
MAU. If the fault is a comparison mismatch,
MAU interrupts the work in progress
and initiates a side-indicating
program in both sides.
When the fault has been located to one
side of the processor, MAU automatically
stops this side and the other side takes
over and continues the traffic handling.
A hardware fault in the central processor
is thus remedied without traffic disturbance.
Since the function units of the central
processor can be blocked and deblocked
under program control, a simple
and reliable method has been
created for fault diagnosis. After detection
of a hardware fault the faulty processor
side is blocked and its function
units are automatically deblocked, one
by one, until a new fault signal is received.
This ensures rapid and direct
pinpointing of faulty units.
The software structure of the data processing
system has made it possible to
introduce a number of checks of program
handling direct in the microprograms.
This permits early detection of
faults and minimizes fault dispersion,
thus securing a higher software quality.
References
1. Strindlund, E. and others: AXB
20-AII Electronic Stored-Program-Controlled
System for Telex
and Asynchronous Data Traffic.
Ericsson Rev. 1, 1977, pp. 32-40.
2. AXB 20-System Survey. Brochure
LMEno. 8104/08.
3. Klintwall, R.: Modernization of the
Swedish Telex Network. TELE 1
1977.

The Ericsson Group
With associated companies and representatives
EUROPE
SWEDEN
Stockholm
1. Telefonaktiebolaget LM Ericsson
2. LM Ericsson Telemalenel AB
1. AB Rifa
1. Sieverts Kabelverk AB
5. ELLEMTELUtvecklings AB
1. AB Transvedex
4. Svenska Elgrossist AB SELGA
1. Kabmatik AB
4. Holm & Ericsons Elektnska AB
4. Mellansvenska Elektnska AB
4. SELGA Mellansvenge AB
Allngsas
3. Kabeldon AB
Gavle
2. Vanadis Entreprenad AB
Gothenburg
4. SELGA Vastsverige AB
Kungsbacka
3. Bofa Kabel AB
Malmd
3. Bjurhagens Fabnkers AB
4. SELGA Sydsvenge AB
Norrkoping
3. AB Norrkdpings Kabelfabnk
4. SELGA Ostsvenge AB
Nykbping
1. Thorsman & Co AB
Spanga
1. Svenska Radio AB
Sundsvall
4. SELGA Norrland AB
Vaxjo
1. Widells Metallprodukter AB
EUROPE (excluding
Sweden)
DENMARK
Copenhagen
2. LM Ericsson A/S
1. Dansk Signal Industri A/S
3. GNT AUTOMATIC A/S
1. I. Bager& Co A/S
Tastrup
2. Thorsman & Co Aps
2. LM Ericsson Radio Aps
FINLAND
Helsinki
2. Oy Thorsman & Co Ab
Jorvas
1. Oy LM Ericsson Ab
FRANCE
Colombes
3. Societe Francaise des
Telephones Ericsson
Paris
2. Thorsmans S.A.R.L.
Bologne sur Mer
1. RIFA S A.
Marseille
4. Etablissements Ferrer-Auran S.A.
IRELAND
Athlone
1. LM Ericsson Ltd
Drogheda
2. Thorsman Ireland Ltd
ITALY
Rome
1. FATME Soc. perAz
5. SETEMER Soc per Az
2. SIELTE Soc. perAz.
The NETHERLANDS
Rijen
i c-;«c

ISSN 0014-0171
TELEFONAKTIEBOLAGET LM ERICSSON