Plessey Exchange Works
|
|
The factory was built
in the 1950s in order to hide classified design and manufacturing
of the UK Air Defence System away from prying eyes. It was an
extension of the ATE (Automatic Telephone & Electric Company)
based nearby in Edge Lane, Liverpool. Alas, with the end of the
Cold War the British defence industry collapsed and the building
was turned into luxury flats. The site was alongside Cheapside
at the front and Cunliffe Street at the rear. Adjacent to the
site across Cheapside Alley was a small factory that made candles.
Looking at the picture below, the far end of the factory housing
laboratories, some manufacturing, and the factory canteen, before
being bought by ATE, made margerine and (so I'm told) during
WW2 bombing a direct hit resulted in a stream of molten margerine
running down Cheapside. |
|
|
|
What went on in the building
where I worked for 20 years? I've managed to find some rare pictures
of the equipment (thanks to Terry Pate who was my boss for quite
a few years before moving to the radar division on the Isle of
Wight where he was the Managing Director).
The XL2 and XL4 computers used
in Project Linesman were developed from the XL1 shown below.
This picture shows a late version in 1961 known colloquially
as the "XL one and a half" handled the wages for Exchange
Works staff and had about twice the number of cabinets of the
XL4. Below, centre is Tony Patterson who kindly supplied the
picture. Note that engineers back in the 1960s wore suits and
ties. In fact, I recall, our boss at Exchange Works in later
years was one of the "best dressed men in Liverpool".
Below right is Jack Brown.. not sure about the chap on the left..
anyone know who this is?
Tony tells me this XL machine
is contained in the leftmost 5 racks, left to right: Arithmetic,
Transfer, Control and two 1K Stores. Compare those latter two
with say a tiny 64Gbyte memory stick not to mention the power
required which used heavy copper bus-bars whose connections are
visible at the top of each rack.
Further down this page you can
see the electronic plug-in units used in the XL computers. |
 |
|
The next
picture is the XL4 computer of which a couple of dozen were manufactured.
This was a 64 bit machine that worked asynchronously rather than
being governed by a master clock. As you can see it has five
cabinets and each was stove enamelled with chrome-plated fittings. |
 |
|
I understand the XL4 was
designed by Ron Threadgold and Dave Hartley (who together also
designed the XL2); and Terry Pate (who supplied these pictures)
worked on the XL4 processor. Linesman was known as PLCS to muddy
its purpose.
Each cabinet housed a number
of plug-in units carrying about 25 printed circuit boards whose
connections were wired to a frame by gun-wrapping. The circuit
boards carried standard logic functions constructed from germanium
transistors and diodes with their interconnections made in the
wiring of the unit frame. This method made the design quite complicated
but allowed for standardised logic functions equivalent to todays
integrated circuits ie. each circuit board, say for example six
inverters was equivalent to a modern logic chip such as a 7404.
Like a modern PC the XL4 needed memory, and this was 32K x 64
bit x 2uSec, and occupied a couple of 7 foot cabinets.(nowadays,
double this amount of RAM would occupy just a few square millimetres
within a microprocessor chip) |
 |
|
Below is a rare picture of a "touchwire"
display being used by an air traffic controller. Whether this
wasever produced in quantity I can' say but I have vivid memories
of it because it was the subject of my first job at Exchange
Works. In 1965 my then boss, the late Bill Boyd asked me to build
an electronic box of tricks from surplus Linesman circuit boards
to handle the touchwires. These were gold plated wires embedded
in a plastic overlay fitted on a cathode ray tube. With some
help this was accomplished and the new touchwire display demonstrated
to MoD at Malvern, where the prototype Linesman computer was
located. Was this an XL1? Whatever it was used lots of neon tubes
(a device bridging the gap between valves and germanium transistors). |
 |
|
Next a rare view inside RAF
West Drayton computer room... Note the "wiring" carrying
power to the computers.. enormous copper busbars in the ceiling
driven by a number of hundred amp power units backed up by lead
acid batteries. and the trunking carrying cooling air to keep
the fragile germanium transistors safe. There were several power
supply voltages and it was vital to turn these on/off in the
correct order to avoid thermal runaway in the germanium transistor
junctions. The computers operated in real time and were responsible
for tracking and predicting the course of enemy aircraft. Radar
sites around the UK coast picked up echoes from all aircraft
in the vicinity of the UK and transmitted data to West Drayton
where it was processed by the Linesman computers with results
passed to display consoles.
If
I recall correctly, in the room above this were something like
70 radar display consoles. |
 |
|
If you'd like to see an
outstation, RAF Neatishead, just as it was when Linesman was
scrapped visit the Radar Museum close to Horning in Norfolk.
https://www.radarmuseum.co.uk/
The next major development was
the XL6 computer (designed by Terry Pate), a 24 bit machine.
Left the computer and right its memory. This was supplied as
a message switch for ICI. |
 |
|
Then the XL7. |
 |
|
Next was the development
of the XL9 computer
This was a commercial venture
designed by Terry Pate and Dave Paculabo, and was used in a number
of projects including an air defence system for Burma, a telephone
exchange in London, a message switching system based at GCHQ
Oakley, and "wide area traffic control systems" in
Liverpool and West London (awaiting pictures). |
|
Below is an XL9 computer which was our first to use
all silicon transistors instead of the germanium types. The latter
were insisted upon by MoD because they had reliability data on
these.. not silicon types (unfortunate because the latter were
very much better than the former), although I believe some silicon
transistors managed to find their way into the XL6.
Basic design, as you can see, was based on plug-in
units, each carrying a set of printed circuit boards. These boards
(carrying gold-flashed pins) were gun-wrapped to bakelite strips
set with gold-flashed pins fitted in the unit frame. Reliability
was fair and any problems tended to be intermittency in the wiring.
Input was via paper tape (note the paper reel on the left of
the picture which is likely to have been the latest set of program
fixes colloquially known as "the bugs tape". Output
was generally by teletype, typically an ASR33. As hard drives,
as we know them, hadn't been invented we used either magnetic
drums which were pretty big and couldn't store very much or magnetic
tape (seen on the right).
What sort of computer architecture was used in those
days? Well, all the XL computers used a "hard-wired"
set of logic boards for their program instructions. Programs
were held in "core store", a locally designed precursor
to modern "RAM", which used tiny ferrite rings that
could be energised to store a field used to define either a "1"
or a "0". The same core store held operating data with
long term data (generally not time-critical) stored on magnetic
tape. Although general purpose programming languages were becoming
established XL computer programs were written either in machine
code or a locally developed language designed around the particular
instruction set wired into the machine. For example the early
XL computers used a locally designed high level language named
"Mini-Coral" or, in the case of the XL9, "Exec-9",
which with their compilers were able to produce the machine code
loaded into core store.
Below the posed picture in our computer room is an
advertisement for the XL9. Not too many were sold but one important
sale was to Liverpool Corporation who automated their city traffic
lights.
I looked up the reference on the page, "The Computer
Journal Volume 9 Number 1", and discoved this was published
on May 1st 1966. This was a year after I joined Plessey. |
 |
 |
|
Following up from
the XL9, a venture into a more advanced machine was the XL12
which unfortunately fizzled out... no picture yet... but Exchange
Works did design and develop a new computer for the management
of telephone exchanges. This was known as "Sprat" when
three computers were manufactured and installed at the Plessey
site of Taplow near Heathrow. These computers were again based
on ideas from XL9 and XL12 and were designed to be fault tolerant.
A loss of one or two machines from three would result in the
remaining system of two or even a single machine picking up the
workload without any hesitation. To demonstrate this to the customer
it was arranged that test software was written and an audio amplifier
connected to part of the computer to produce tones. Using the
test software a musical tune was produced, that was "English
Country Garden". During playback one computer was turned
off and then a second with no hesitation in the tune. Then the
computers were re-introduced and again no hesitation in the tune.
Another venture in the 1970s was the CP400 shown below. This
was developed after "Marge", the XL9-based message
processing system sold to GCHQ in the 1960s. |
 |
|
The last (proper) computer designed at Exchange
works was the PP250. This was a joint venture with other Plessey
sites and was aimed at making a general purpose machine suitable
for British MoD future requirements and was the first to depart
from a hard-wired logic instruction set, instead using EPROMs.
It made the grade for Project Ptarmigan but, although proposed
for Project Wavell, politics dictated that GECs home grown computer
was selected instead. I understand that this proved to be a really
bad decision. This was around 1972.
Not long afterwards (in the mid 1970s) Exchange Works
designed and built elements of satellite ground stations for
ESA/ESOC (the monitoring and control parts) plus the multiplexers
for European air traffic control systems. Both areas of development
used processors employing standard 74 series logic circuits,
rather than microprocessor chips which appeared later (Intel
4004 etc), with programs held in ROM. These first ROM chips proved
very unreliable giving rise to countless intermittent errors
until the manufacturers changed the basic design of their chips
and produced pin-for-pin compatible versions. |
|
|
 |
Meanwhile, sharing the top floor
of Exchange Works with the Drawing Office, and even more secret
than Linesman, took place the design and development of crypto
equipments, a few of which are pictured here. Left is the BID610.
Besides manufacturing complete
equipments, Exchange Works developed the small unit forming the
most important part of most UK cryptographic equipments, the
noise generator; based on a semiconductor junction or zener diode
which was able to provide completely random binary code used
to produce the key for encrypting messages..
Below, newer machines.. first
BID1000.. |
|
 |
|
Then, below the more compact
BID950. |
 |
|
When Margaret Thatcher
decided to sort out troubles in Northern Ireland, Exchange Works
won contracts for the design and development of specialised military
equipment. These included P6 (a mine detector), P7 (a device
for detecting command wires for things like A mines hidden in
trees), IA2 (a device for recording sounds from terrorst activities),
IA5 (a sniper detector), 3PS (an intruder detection system) and
others (awaiting pictures). |
|
At roughly the same time,
Exchange Works entered into the satellite business. Monitoring
and control equipments were supplied to support the ground station
handling the GEOS and METEOSAT satellites and others, managed
by ESOC in Bretigny and Darmstadt. This area of expertise ended
once the design, development and manufacture of equipment for
RAF Oakhanger (the anchor station for Skynet) had been completed
and installed (awaiting pictures). |
|
Because of expertise from
previous air defence systems, Exchange Works provided the communications
cabins for a number of 3D-radar based air defence systems including
those for South Africa (1981 Project Rodent), Egypt (1981 Project
Lion), Ecuador and Qatar (Project Falcon), plus helping out when
the Falkland Islands were invaded by diverting to there one of
the systems for Qatar (1983 Project Penguin). Below.. a comms
cabin for Project Rodent. Because of their large physical size
these cabins were manufactured on the second floor of the building
because lower down the building support pillars were too close
together. A doorway needed to be cut into the side of the building
through which a crane could deliver empty containers and collect
equipped cabins for transit, usually to Bournemouth Airport. |
A Project Rodent communications cabin
for DoD in South Africa
|
 |
|
|
Is that everything? Not
quite.. Exchange Works also supported other Plessey businesses
by designing and manufacturing all sorts of interesting things.
These included parts of radar jamming equipments (for Syria would
you believe...), message heading generators for NATO (because
we had experience with ACP127 for GCHQ), the Store and Forward
Message Switch for Ptarmigan, studies for Ptarmigan and Wavell
for the British Army and specialised equipment for Eurocontrol.
This last equipment (mentioned previously) comprised a pair of
128 channel multiplexers using lots of microprocessors developed
before those from Intel. The design of these was based on the
type of work that had gone into the XL12 computer which was being
built for commercial systems but had been cancelled when Plessey
and others formed ICL. The pair of multiplexers were installed
at Karldap and used for combining peripheral equipments into
Telefunken computers for the control of middle airspace over
Europe. |
Air Defence Systems
|
|
 |
 |
Not easy finding a record
of what used to be highly confidential military projects but
here are some snaps from old pals Norman Fung and Allan Marshall.
Click on these to see more and
click the link below to see the radar history of Plessey (thanks
to Allan Marshall who commissioned the systems seen by clicking
the pictures here).
Below, a scene from a quiet
Sunday morning with a communications cabin being lifted from
Exchange Works en-route to Bournemouth Airport for onward delivery
to a customer. Further down a view inside
the cabin (pictures from Chris New). |
|
|
 |
 |
A view south down Cheapside
towards the Liverpool Municipal Offices
This is a Project Rodent communications
cabin en-route to South Africa. On the right is the old margarine
factory housing the canteen, and upstairs the Accounts Department,
Purchasing and laboratories.
Below, views to the north with
the Liverpool College of Commerce on the right in Tithebarn Street. |
|
 |
 |
|
|
To see how these equipments fitted together into
systems look in the brochure
Then finally with the cold
war coming to an end and the defence industry collapsing the
Exchange Works factory was no more... click the picture to see
the buyers plans.
|
 |
|
Under the building was
a garage which had reserved spaces. The general principle was
you had to wait for someone to die or retire before you could
get a space and then only if your status was high enough to qualify.
Working shifts could sometimes earn you a temporary space and
I recall one day in the late 60s trying to get to street level
after working all night on a project called "Sprat".
The exit was fairly steep and my car an old 1953 Rover 75. It
was when attempting to drive out I discovered all was not well
with the car. I could almost reach street level before the engine
cut out and I had to roll backwards. After several unsuccessful
attempts I managed to race up the ramp before the engine spluttered
out. Many months later at home on an extremely cold morning the
car refused to start and it was then I found the fuel filter
glass bowl was full of ice. Ordinarily driving flat and level
petrol would pass over the water filled bowl, but meeting even
a slight hill caused water to enter the carburettor and stop
the engine. Once unclipped and emptied of water no more problems... |
Anyone with more information
or corrections please contact me.
|
|
|
