ARA Project 1957-58

ARA opening

In preparation for the interview for a position on the Zebra team I swotted up on computers and found that up to then virtually all of
them had been developed in universities or other technical institutes mostly in USA and UK; and that few if any were available commercially.

For example, in the UK at the National Physical Laboratory they had developed ACE Automatic Computing Engine. The key staff had been Alan Turing, probably the most famous British computer pioneer ever, and a namesake of mine Donald Davies. At Cambridge they had developed EDSAC Electronic Delay Storage Automatic Calculator, the name of Dr Maurice Wilkes was prominent. At Manchester they had developed the Manchester 1 Machine and then the Baby or SSEM Single Scale Experimental Machine, and the name of Prof Tom Kilburn was prominent.

One of the very first machines, Colossus, had been developed in 1943 at Bletchley Park near London. Colossus was the machine used for helping cryptographers to crack German military codes, but information was restricted for security reasons. The name of Alan Turing appears on this one too. Surprisingly, Lyons the cake maker and restauranteur, had decided to proceed on their own and developed LEO Lyons Electronic Office..

In the USA, pioneers at the Moore School of Engineering of the University of Pennsylvania, had developed a computing machine during WW2, and another called Univac 1 in 1947. The names Eckert, Mauchly and John von Neumann were prominent. Remington had bought Univac in 1950.

In Holland, Dr v.d.Poel and others had outlined the concept of a small practical stored program computer called Zebra – Zeer Eenvondige Binaire Rekon Automat – and had offered it to several companies including Bell Telephones in Antwerp. Bell was a sister company of STC and did not consider itself capable of developing a computer and so referred the proposal to London. There it was decided to proceed with the idea starting with a logical design made at STL Standard Telephones Laboratory in Harlow by Dr Joe Rice, but many others claimed the honour of doing this highly original and important work including my boss. When the logical design was done a decision was then taken to continue to engineering design, manufacturing and marketing based in Newport under the name of the STC Information Processing Division.

For the technically inclined, the Zebra computer was a stored program machine having 33 bit words with rotating drum that provided 15 registers and random access storage of 4000+ words. Programming was made using a 13 bit drum address, 5 bit register address and 15 bit operation field .This made programming an interesting skill.

The Zebra Job

Some months later I joined STC Information Processing Division in a building on two floors that had been a flour mill next to the Newport Rugby Club ground with a power station on the other side.

The ground floor of the building was used for constructing and testing the computer hardware. This comprised a 6 x 2 foot cabinet of logic with a cathode ray tube at the front with an array of switches used for inputting key data like bootstrap instructions; plus a drum storage cabinet of similar size and, some feet away, a desk unit with a paper tape reader and two teleprinters on top.

Upstairs there was a working area with twenty wiring girls plus the toilets, these were separated by a half wall from an open plan area and a handful of offices for bosses and senior designers. In the main the bosses had come from STL the laboratory where the logic design was done, and the senior designers from RRE the Royal Radar Establishment at Malvern where radar was invented. It was these RRE boffins that had played intellectual ping pong with their counterparts in Germany during the war and had been clever enough to get the better of them. Their chief had been Dr R.V.Jones who had attracted a number of Welsh electronic engineers to Malvern. Some of these same men had been tempted to become involved with another first machine and had done the engineering. The programmers were also up there, near to Jimmy Ord-smith the chief programmer.

I, and the other five new boys, sat in the open plan area in a three by two arrangement of desks. We had started thinking about ideas as to what other applications a scientific computer could be put to. My own hunches were that computers would considerably improve the company administration processes and could also be used as the control units of the message handling systems that were under development.

My interest in messaging had began when I was in the Army. I had been a signalman for two years at the 3GHQ Signal Centre at Fayid in the Canal Zone of Egypt. There, working on shifts around the clock, each of 20 soldiers was responsible for managing an incoming or outgoing link in a chain of communication stations leading from and to other British and Empire Army units around the world at Aden, Nairobi, Singapore, India, Burma, Japan, Canada, Australia and South Africa etc. So I had first hand experience of how a manually operated message exchange centre was handled. (This all came in useful when STC started to develop automated messaging systems and I became the project manager of one such in Canada some years later).

My neighbour in the open plan was Geoff Hunt who had been an RAF pilot in jets screaming about the sky and loving it but had had a motor bike accident which took off his left leg and put paid to his flying career. A very personable man much admired by the wiring ladies who, when he limped off slowly to the toilet, was invariably serenaded by the girls singing ” Why was he born so beautiful? Why was he born so tall? Why was he born so wonderful? Why was he born at all?”.

Sitting in front of us was Harry Becker who, unlike the rest of us, had decided to become expert at programming. He was the first take himself away from the rest of us new boys who tended to keep together, by either dining by himself or with members of the programming team.

Later a member of NAAFI Navy, Army and Airforce Institute, by name Alex D’Agapayeff, was sent to Newport to expedite the arrival of their computer because the delivery date was slipping badly. He and Harry Becker became pals and later formed the company CAP Computer Analysts and Programmers that became as CAPGemini one of the largest software houses in Europe.

Logical designer quits

We six, the new system designers, were still getting our feet under the table when the man responsible for the logical design maintenance quit his job. Apparently he and a few other fun-seeking boys and girls had been disovered playing doctor too realistically at a house in the mud flats of the Severn Estuary. and got into serious trouble for it. But why exactly he had to leave work and Newport because of this faux pas I never really understood. There must have been something else as well.

But leave work he did and the very last thing he organised, before going, was to a one-day emergency training course on Zebra logic. Unfortunately I missed the course and when I returned was told that only one person, Geoff, had understood a word of it. When he came in next day Geoff told me the logic design was easy to follow and began to explain it. I, too, found it easily understandable even though I had never seen anything like it before, with its ‘and gates’, ‘or gates’, ‘flip-flops’, ‘registers’, ‘timing pulses’ etc.

So Geoff and I became, de facto, the official logic designers for any additions and changes to Zebra.

New Work

The Hollerith punched card reader and punch machine

In this new role Geoff immediately started the design for a ticketing system for SAS Scandinavian Airlines System. This included a new hardware unit called a Route Card Reader that he designed into which ticket sellers could push a mechanical Route Card and then easily select the flight legs and number of seats required by pressing co-ordinate keys on the device.

The output was interestingly simple, it consisted of two lights one red one green; if the green light was lit the ticket could be sold, if the red light was lit the it couldn’t. But, interestingly, if both red and green were on together the seller could sell but should realise that although the plane was fully booked cancellations were expected.

For my part I was given the task of designing an interface between Zebra and a combined card reader/punch for ARA Aeronautical Research Association in Bedford where the study of aeroplanes breaking the Sound Barrier was the work in progress and the output data from their wind tunnel was in a special home-grown card format.

When the card reader/punch machine arrived it was accompanied by an engineer called Tom Cheshire who had just finished an exercise at NPL National Physical Laboratory working with another Donald.W. Davies.

Tom was one of the old school of tabulator engineers whose expertise was getting overtaken by computer technology but was working hard to keep up before he retired and making a good job of it. The other Donald’s W stood for Watts, my W was for Wynn. He was a Director of NPL in charge of computer science including ACE computer, and becoming famous, while I was just starting my first computer job. For example, the other Donald Davies is nowadays high on the listof famous Welshmen not only for his pioneering role in computers but also for his work on security systems and package switching which is the basis of the Internet.

Tom also told me it was fortunate for me that he and the other Donald Davies had spent many months deciding and defining how to attach the punched card equipment to the ACE computer, as now I could have all that information as a starting point.

As a result of working with Tom and his recently gathered data, I completed the required logic design within weeks and checked it with Geoff Hunt. Wondering what the next steps would be, I asked the Head of Systems ” Who will built this?” He said ” You will”. I asked “Who will test it?”He said ” You will”. And in desperation I asked him” Who will accompany the hardware to ARA and commission the system?” And once again he said ” You will”.

Project manager?

In that moment I had become what became known later as a Project Manager without knowing it – it was a case of being thrown in the deep end of the swimming pool, or as someone later said, being expected to give a concert while learning to play the violin – and I may well have become the first commercial computer project manager in Europe. But I often wondered later what the boss had been thinking when he pushed me into unknown waters like that. Did he do it for my good or his own?. For the moment let’s give him the benefit of the doubt.

Anyway, I had to get on with things, work with the build engineers to layout the hardware units in a new third cabinet, order the plug in units and define the wiring between units on the back plane then wait for it to be built and assembled.

In the meantime the next item in my work queue was to work with an independent mechanical engineer on the design, build and interface of a fast serial printer to replace the very slow, four characters a second, teleprinters. We got his first model working but there was a glitch – and I handed the job over to somebody else while I went back to the ARA project.

First I had to learn how to test digital circuitry. My teacher in this subject was Laslo Kvasz who was the chief commissioning engineer and the first choice to accompany systems delivered to key customers. Laszlo was a brilliant man filling in time in England with STC having just come through the barbed wire from Hungary during the revolution there. He later joined the Max Planck Institute as a researcher in atomic science. My keenest recollections of Laszlo were his happiness at being free from the shackles of communism and his newfound freedom to go wherever he wanted in Europe.

Laszlo could only spare me one day in between commissioning jobs so I had to be a quick learner. The trick of testing digital circuitry was to use a Tektronix oscilloscope to freeze the digital signals as they skitted around the computer. As if by magic a word of 32 digits could be stopped and inspected as it passed a particular point. By this method it was possible to follow what was happening in and out of “and gates”, “or gates”, “flip-flops” and “registers”

Remember in binary there are only two characters 0 and 1. Flip-flops were single digit storeage devices, they were either in the 0 state or the 1 state. When flipped by an input pulse, say a 1, they stayed in that state, the 1 state, until flopped by an input pulse, an 0, to the other state, the 0 state, if you see what I mean.

Swan hotel bedford

Bedford

Then I found myself booked into the Swan Hotel in Bedford fora couple of months ready to go to work at ARA. This was a big moment in my life, like getting out of gaol must feel after being incarcerated for a few years. Five years working in roughly the same place with travel only to London and that by train because use of ones own car was not allowed even though we often had to carry heavy equipment to be calibrated, was constricting to say the least.

I don’t think Bedford is in the top ten most beautiful cities of England but it must be in the top part of the list. And being allowed, or even encouraged, to use ones own car to drive there from Newport through some of England’s most admired countryside was a luxury. The river in Bedford runs through the middle of the town with handsome banks on either side and the Swan Hotel is on the riverbank.

ARA

The Aircraft Research Establishment was outside town in a new industrial estate, a ten minute drive by car from the hotel. Their new computer room had been prepared with air-conditioning, under floor wiring etc and our Zebra computer complete with the extra cabinet of logic that I had created, plus the installation team, were on their way by road from Newport. I felt I was approaching a critical moment in my career: could I hack it or would it crack me?

On the site next door to ARA, builders were putting the finishing touches to a new factory building and were erecting a large hoarding with name plate proudly announcing TI – Texas Instruments. We asked

” Who on earth are TI and what in heaven’s name do they make” but nobody knew the answer, so ignorant were we all of the word and the coming all-conquering product – The Transistor.

The installation team arrived, worked like crazy for four or five days, and finished the job. On the first evening they were there, we left work after eight at night and couldn’t find anywhere to eat except pubs with potato crisps and cold pork pies, that was before pubs served hot food, remember, this was 1950s. The Swan Hotel’s restaurant closed at eight with last orders at seven thirty so we were too late for that and it was with great relief when we eventually found an Indian eatery just out from Town Centre, decorated as if it was in the harbour in Bombay, that welcomed us with open arms. That was because we were the only ones there and may have been the only ones all evening.

They served ice-cold Carlsberg lager beer as soon as we sat down, which we appreciated because in those days most beer in England was warm. Then we all ordered chicken curry, and heard squawking in the yard and bin lids clanking, it was as if there was a fight going on out there and they wanted to prove that they truly did cook to order as advertised. The Indian food was so good and the curry so hot that we congratulated the chef when we demolished it, and I must admit that it was so good that I found it difficult to eat dinner anywhere else during the rest of my stay in Bedford.

One of the difficulties that I had not foreseen was that the installers who were all without cars had been booked into different boarding houses around Bedford, while I and my car were at the Swan Hotel, and ARA was out of town a few miles away. My solution was to get us all into the same boarding house as the senior installer was booked into, so that we could lodge together and travel in my car to ARA and back.

Finally the installation was completed and the installation team left, leaving me alone on site and accompanied only by the Zebra system, the Tektronix scope and the sheaf of logic design documents.

I discussed with the ARA Director the subject of how he intended to achieve an acceptance test of the system and he said he hadn’t thought about that and did I have any ideas. I told him that the standard Zebra pre-delivery test was to run a mathematical calculation program that printed out answers on the teleprinters every hour or so for twenty-four hours without errors. He thought that would prove a good acceptance so we shook hands on it and I started the implementation work.

After about two weeks I felt I was getting close to finishing, because the card reader was moving data into the computer and the computer was beginning to punch holes in cards, when there was an unholy bang accompanied by smoke pouring out of the main logic cabinet, thermionic valves were popping like old light bulbs and the CRT display console was showing rubbish. I switched off the power as fast as I could and sat there gutted.

What could I have done during the testing of logic, a relatively thoughtful and risk-free exercise one would think, that could have caused a disaster of this magnitude to happen? It was as if an earthquake or at least the early tremors of an earthquake, had happened. The computer was clearly ruined, and I thought I must or might be responsible for ruining it. With trepidation I phoned the bad news through to Newport. ” Oh hell” they said “Another one’s gone, that’s the third, there’s a problem with the high voltage from the CRT line getting into the low voltage logic circuits.”

So, it wasn’t my fault. I didn’t think I had done anything stupid but you’re never quite certain especially if it’s your first computer job, first digital testing, first logic design. So I reported the news to the Dirctor of ARA and drove back home to contemplate the next steps.

In fact I helped in the construction and testing of three new machines and in the preparationof a Zebra operating manual which had clearly been missing up to now. And was back in Bedford a few weeks later.

Zebra show ing my added cabinet

Acceptance

I got the system to the stage of nearly working well. Harry Becker had arrived with updated software for the application, and claimed that he had done it literally on the back of an envelope on the train journey down and to this day I don’t know if I believe him or not. He finished his work quicker than I thought he would and didn’t stop overnight and so I missed a chance of getting to know him better. But he did have time to update me as to what was happening back at the ranch.

The good news was that the proposed new hardware unit – an automatic multiplier – that had proved hard for Geoff to design, had just been completed by the new logic designer, a man called Francis, whom I had recently met while in Newport, and was being built. The bad news was that Francis had already announced he had received an offer from Imperial College London to join them as a lecturer on Computer Design.

When Harry left, the system seemed to follow his programmed instructions and my logic but only up to a point. The point being that the results that were printed were nearly always wrong and not all results were printed. I was puzzled, but instead of struggling with the problem I decided to ask for help in the person of Laslo Kvasz the chief commissioning man who had taught me how to test digital circuitry. His reputation had increased to star level during the time I was at Bedford so I was lucky to have him allocated to me for a week.

During my time with him learning how to test computer circuitry he had also gave me some lifemanship as well as technical advice. One of his habits, he told me, was to finish a job as quickly and professionally as possible but to take his time before reporting back to base. And then to use that stolen time to do things he hadnt been able to do in Hungary, like visit Paris, go down the Rhine, or whatever. Otherwise he explained, he would be sent out on another job immediately with no time for himself. As stupid as it seems, he said, many of the other installers do exactly the opposite, instead of getting on with the job they take their time getting there and then dawdle or go skiing or sightseeing before they get down to the job in hand. In this way they are late starting and even later finishing and often cause customer disatisfaction when the reverse could so easily be the case.

On arrival Laszso immediately got my logic diagrams spread out on the desk and studied them, then after asking a few questions about why I had included this or that or done something this way rather than that way and concentrating quietly for an hour or so more, he said ” I think I have it. The timing pulse you used to start the transfer of data to the punch is incorrect (there was a choice of 1 out of 32 pulses I could have used), it should have been the next one, that’s why the arithmetic is wrong. Also, the new cabinet is more than six feet from the main cabinet and you should have added a buffer unit to allow for that distance, that’s why the punching is unreliable. That was was impressive, but more cleverness was to come.

He arranged for an installer to bring the necessary extra equipment from Newport and within days they had installed the changes and we started testing again. The early tests indicated that the logic was now correct and so we proceeded to a full end-to-end test, from the input reading the cards to the outcome of punching cards. Disturbingly the punched card output still did not correspond to that expected, it was as if the process of punching caused the system to go haywire. Laszlo thought that this new problem was probably due to the high power energy punching action causing interference to the computer logic.

In order to prove this he wrote a small program to punch a single digit and entered it into the computer. And then, because it was the weekend and we were the only people there except for the security guards, and didn’t have a capacitor handy, he constructed a capacitor from the foil and cardboard of a cigarette packet he found in the dustbin, and then soldered wires to it so that it could be connected across the punch mechanism of the appropriate single digit. We tested that and sure enough it proved his theory to be correct, and it showed clearly that each punch mechanisms had to be shielded by having a capacitor wired across it.

At this point I considered Laszlo to be little short of a genius and told him so. He then reversed the compliment and told me I was equally clever. I said “At what for God’s sake?” And he said something I will never forget. He said ” You are a marvel at making things happen”. Amazed by this I said ” Really, why do you say that?” He said “You are one of the key guys in the Company for getting things done” to which I replied “You’ve got to be kidding”. he said “No, I’m not, you ignore politics and other stupidities that most people get stuck on, and you get on well with everyone, so can work to get important things done. And you do it all in a very organized way”. I didn’t really believe him to be honest but it stuck in my mind as you can imagine. It was heady stuff.

On the Monday, we advised the factory of the need for a wiremen with a box full of capacitors. An installer arrived and fitted the capacitors and then we tested again and for the first time obtained the required end to end result in all positions. As everything seemed to work we started a dry run of the acceptance test fairly confident that it would run for the agreed 24 hours and printout the correct answers.

Much to our horror the acceptance test stopped running after a few hours. So we started it again to run overnight, but in the morning when we arrived back at work, it had already stopped. We decided to start it again but not leave it alone for even one minute, specifically to get pillows and sleep on site with our heads on the two teleprinters and to check the results every hour right through the night, and the next day, to try to find out why it stopped.

We did this and the system ran perfectly through half a day and through the night, that meant it had been going for over twenty hours. But when full daylight returned we, quite naturally, turned off the lights. The system stopped immediately. Could it be that the system stopped because we turned the lights on or off? We thought about it and concluded that it could be. The system had stopped at about the same time each morning and evenings ever since we started the 24 hour test. It must be an earthing problem.

Zebra is designed to have an effective earthing system with a three foot metal spear buried into the ground outside the machine room and it appeared to be in place. Inspection outside in the garden showed the earthing spear lying uselessly on top of the grass, the installers had neglected to bury it, so we buried it instead.

After that the acceptance test ran faultlessly so we invited the Director to witness the 24 hours test which he did, and pronounced the system to have been accepted. The technique of Acceptance had proved itself to be as effective in computer projects as it had in communications implementations.

Lazlo and I left and dawdled on the drive back to Newport, in his recommended style, stopping at the Manor House Hotel in Moreton in the Marsh for lunch which Laszlo pronounced as being virtually perfect. In those days the hotel was run by a group of ladies of impeccable taste, so that the house, gardens and even the gentleman’s lavatories were exquisitely furnished and maintained, and the meals were cooked and then served in a charming dining room by young women being coached to be wives of rich and famous men with their own stately homes to run. They were learning fast.

Manor house hotel moreton in the marsh

Finale

I would like to report that on return to Newport we were hailed as heroes but that was not what happened. After being congratulated for finishing six months ahead of expectations – not planned dates you understand but only unwritten, unpublished expectations – I was reprimanded for moving the engineers from their cheap digs around town into an expensive guest house where we were together and could use my car to shunt us to and from work. In short, I was reprimanded for forming an effective team.

Their actual words were “Who do you think you are? Don’t do it again, engineers are not treated like that in this company”. I was appalled, what a load of snobby, rubbishy, mistaken thinking, I thought. Maybe I should find a job in another company if that’s an example of how they think they should treat people.

And that is precisely what I did, but not before being sent to Stuttgart to design and implement the interface between Zebra and the Standard Electric Lorenz computer, this because my boss, who had been out there for some months under false pretences to do the job, but had failed.

I got a job in the London area with DeHavilland. A little later, as if to prove Laszlo’s prediction, I was invited back to Newport, and offered a job to participate in a senior role in the development of a new version of Zebra. I agreed to their proposition up until they stated that if I returned it would have to be on the same salary I was paid when I was last there.

Quite calmly I reported that I had doubled my earnings in that time and would certainly not consider returning for a penny less than my current salary.

They declined to make a bigger offer and I declined to consider their small one. The STC Information Processing Division was disbanded soon afterwards.

Lessons Learned

Acceptance management, even in its infancy, proved to be a wonderfully effective technique.

Involving the cleverest, most effective specialists, was a very wise move.

Forming an effective, hard-working team was essential even if it didn’t match company culture.