The Jim Austin Computer Collection

Marconi TAC

Transistorised Automatic Computer

This is one of only 5 TAC machines originally developed by Marconi in 1959 for Radar (History of the Marconi Company 1874-1965, By W. J. Baker). It ran in Wyfla Nuclear power station on Anglesey, UK, from 1966 for 38 years (to 2004) (from 1968 non-stop), which could make it the longest running computer in the world. It was used to monitor the nuclear reactor.

It was originally a pair of machines. I have one of the pair, Bletchely park (national museum of computing) has the other half.

The following is text from the switch off report:

Wyfla Automatic Computer (TAC) leaves the station after 38 years.

1963 saw the construction begin of Wylfa Power Station. In 1965, English Electric commissioned Marconi to build a computer that could handle the monitoring and alarm system at Wylfa Power Station.

Work began initially in 1965 to assemble the system in Marconi's Head Office and then was moved in May 1966 to English Electric's Head Office in Kidsgrove. In October 1966, the computer was moved to Wylfa, and was given the name of Transistorised Automated Computer, which became known as TAC.

5 TAC's were built by Marconi, 2 of which were at Wylfa.

The TAC was kept in a room 100ft long, and had orange display screens, very different from today's PC screens.

Phil Runciman  recounts:

Short Wylfa story:
One of the on-site electrical engineers said to me, “Phil, your presence is the only common factor we have identified where many alarms to go off in the control room. Do you mind if I just hang around while you work?” To which I replied, “Be my guest.”

After some time I had to reload my program using the high speed paper tape reader. The program tape was held in a Perspex holder. During reading the paper tape flashed through the reader and spewed out onto the ground. I had discovered that the Perspex holder held a substantial static charge by the end of this operation. It was a matter of personal curiosity to see how large the static charge got.  I had learnt to hold my thumb on the end of my pencil while the pointed end was moved closer to the holder. Simultaneously the spark and the engineer leapt into the air. The latter uttering expletives and the former the usual , “Crack.”

Sure enough in the control room a host of alarm messages were being generated.

Reprise:
I was somewhat amused, when I arrived in the computer room the following week, to find that the suspended ceiling had collapse under the weight of accumulated debris. (Cement dust, dropped fittings and the like.) This caused havoc with the drum and the relays. My amusement was not due to this, in fact it was quite frustrating. I had travelled from Whetstone in vain. No, my amusement was in spotting a worker, suspended some way up in the huge space above the computer room, doing arc welding. I thought that might have caused a few alarms too!

Did the CEGB, or whatever they are called now, supply you the Wylfa source programs?

I used to work on the Whetstone KDF9 and had kept a copy of my 1D Transient Heat Flow Program. (Sadly, I did not have the 2D version.) The former is one of two surviving application programs for KDF9. I got it going using the emulation of KDF9 written in Ada2005 by Bill Findlay. Mercifully I had some test data and results to check it with.

KDF9 was a brilliant piece of work from many perspectives. It was all downhill from an assembler programmers’ perspective. For example: Doing spherical trig, backwards, and, on the Hawker Siddeley Dynamics DCC2, was hellish. (Backwards? It was for a navigational training simulator. We knew where we were, but had to tell the instruments what they should indicate.) Unlike KDF9, DCC2 had no floating point.

Andy Corrigan, Station Manager said:

'The TAC has been an important part of equipment on site since the day the station was built. I am glad that both TAC's are going to [the jim Austin Collection] and [Blechely Park] where future generations can see how far technology has advanced in 40 years.'

Capell Aris notes:

I was the computer software specialist at Wylfa 1976-8. Your TAC timeline contains a trivial error: the Flexowriters were not replaced with line printers in 1977. All we did in that year (and really, into 1978) was to replace the drum stores with Dicoll RAM - I didn't have too much involvement with that project other than a minor rewrite to the operating system. I think the line printers replacements were fitted in 1979-80, which was after I'd moved on to Dinorwig.

I was the guy they honoured with turning the on-line machine off in 2004 - merely because I was the most decrepit person they could find!

[I have corrected the errors in the table below]

TAC timeline

This is the machine arriving:

and the console:

up right now on wheels

TAC showing all units open. Left image is the core memory and drivers. Then there is the register store and then the microcode store, finally the power system. Below shows

the same from the cards side, rather tan the interconnections.

The control desk for computer 2 and the switch over desk.

This is how we got it upright! (with the help of Richard and trevor).