Memories of Burroughs

Subject: Series L/TC - From: Henry Griggs of Virginia, USA

Subject: B700 - From: Bill Roberts of New York State, USA

Subject: Burroughs and SAM Systems - From: Mike Piper, Menai Bridge, Anglesey, United Kingdom

Talking of Altham's Travel, I took over the majority of the programming work for them from Brian Fairies (a colleague at SAM's) in 1982, who'd written the original batch processing travel agents system, which SAM's also sold to Star Travel. I converted the system to real-time and added various new features, such as BACS output for ticket payments and an airport transfer system to provide transport scheduling to and from Manchester Airport. All this work was specified by Altham's financial accountant, David Ball who might still be working there - I spoke to him about 5 or 6 years ago. Apparently they moved to IBM AS/400 equipment in the early 1990s.

Stollers is another company I did programming work for. By then, they had a B80 running a modified KeyBMS accounts receivable system. I went there a couple of times with Martin Harbour (of SAM's, also an ex-Burroughs guy - Stockport office I think) to install extra sales analysis tailored for their furniture business.

Subject: Memories from Tom LaPorte

Subject: Memories from Fred Little, Preston Branch

I think that's enough for now...I could go on.

Subject: "Too far ahead of its time": Britain, Burroughs and realtime banking in the 1960s

Subject: Memories from Judith Weekes, Machine Operator in Burroughs Australia from the 1950s

I joined Burroughs Field Service in 1958 on demob from RAF/National Service, working out of Northdown Street, Kings Cross.   London West-End Branch transferred premises to Baker Street, LondonW1.    Branch Service Manager was George Breed, then Andy McCulloch, who promoted me to Zone Service Manager. I was with Burroughs for 10 years, then moved into a full sales position with H.J.Heinz 57.  I don�t expect any of you guys to know me, but I�ve been trying to make contact with anybody who remembers Sensimatics, etc, for ages.  I hope I�m not barking up the wrong tree, and look forward to hearing from any of you.

Subject: Memories from Steve Garry of Burroughs in Bristol area

Been reading through some of the Burroughs site memories - it's probably time for me to add a bit to the mix.

I joined Burroughs pre Decimalisation, in 1969, working for Bristol Branch in the UK,  The service manager was Bernard Parsons, two of the Zone managers were Wilf Parkhouse and Roger Rumble, and I spent a number of months living in a bed-sit in Bristol while doing initial training, on things like the P6000 cheque encoder (horrible machine), the J series adding machine, and then the TC500 which was being installed in massive numbers across the Midland Bank, National Westminster Bank and Barclays Bank branches in the run-up to decimalisation.  Due to the timing, I never got involved with the older mechanical and electro-mechanical accounting systems, although there were quite a few around the place

I had my own car, which was unusual in those times, so quite often I was the driver for an experienced engineer who didn't have his own transport, but it meant that I got to see and experience a lot more than some of the other trainees at the time.  There were massive numbers recruited around then due to the huge bank order for TC500s, which in most cases were on 4 hour response-time maintenance contracts, which was a problem in the rural areas due to the distances involved.  It  would be nothing for me to claim 600 or 700 miles a week of travelling, and that was without a 180 mile round trip to the branch office from home, and this was in the days before the motorways, so getting home on a Friday evening in the summer could take 3 to 4 hours due to the volume of traffic on the roads.  I think the highest I ever claimed was 1,200 miles in one week, but that was dealing with a major system crash on a B700 system that was a long way away from me, and in those days, the company didn't do things like pay overnight hotel accommodation, which would have been more cost- and time-effective, given how slow the B700 COBOL compiler was (8 cards a minute if I remember correctly, so a small program took a number of hours to compile, and if there was an error, you didn't get to know about it until the last pass, which could be incredibly frustrating).

When I started, home was not in the Bristol Branch area, but it was agreed that I would operate from home, which was Exeter, as the Bristol branch covered Somerset and Dorset, and I could get to both from Exeter relatively easily.  In those days quite a few of the engineers didn't have their own cars, but I did, so being able to cover a large rural area was a huge advantage.  The he down-side was that I could go for days without seeing another engineer, and I had to be very much self sufficient in spares in order to meet the response times that had been agreed with the banks.  So, the car ended up carrying a considerable load in spares weight, especially when the L8000 & L9000 series came on stream, due to the different circuit boards on those systems.

In 1971 we moved to Wellington in Somerset when I got married, which meant I was in the branch area, and I spent the next 5 years working in Somerset and Dorset with occasional trips to other areas depending on the urgency of the calls and availability of an engineer to respond.  Bristol covered a huge area, from Minehead in West Somerset to Blandford Forum in Dorset, and up to Moreton In March in Gloucestershire, then across to Swindon, down to Devizes, and then down to the South Coast.  Pre-motorways, it could take 2 or 3 hours to get to one call, and I could easily do over 200 miles in a day if someone was out sick or on holiday.

In the early days of the TC500 there were no parts vans to deliver things like print-head decoders to us, so if a decoder failed, which they did regularly due to some design issues in the print head that were not found for a while, it was a case of doing an on-site rebuild of the decoder.  That was a big job to do, and sometimes messy if the oil seals were bad, so there was oil everywhere inside the machine.  Sometime later a central rework facility for decoders was set up, and (Securicor I think) a van service to deliver parts was introduced to make it easier to meet response times.  So the engineering skills needed to repair the machines was significantly diminished, most repairs were a major module swap - good from the bank perspective, but not good for the engineers who joined when that system came in, as they never got the in-depth training that the early people did.

In the early 1970s things like the L5000 magnetic stripe machines, then the L8000 and L9000 series came on-stream, and started selling to commercial customers with all manner of peripherals on them, including line printers, paper tape, and 80-Column punch card.  We usually had to put a constant voltage transformer on to the larger L8000 & L9000 series to help smooth out power fluctuations.  The abiding memory of the transformers was that you didn't go near them, they were hot enough to fry an egg on when in service, and needed 2 people to move them safely, they were so heavy.

I was lucky as I'd done the in-depth Series L training, so I was ideally placed to move on to also covering the commercial Series L sites and the later L8000 and L9000 machines.  On them, there were some aspects that still needed good diagnostic skills, especially on the L5000 and later series that used magnetic-stripe ledger cards.  The paper tape punches were tricky little things, and as for the 80-column card punches, (A149 if I remember correctly) the bulk of the logic there was relays, with the reliability issues that went with them.  I still remember spending over a week analysing a problem at Yeovil District Council, where they were having problems with an incorrect value being punched on an 80-column card, and after a LOT of head scratching, I was able to find out that it was a problem with a firmware overlay that had a wrong value in it, which was causing it to punch a 0 (zero) value incorrectly.  I was able to Memory Modify it, and sent the resulting change to head office, which did get recognised with an appropriate letter of thanks from someone a lot further up the pecking order than I was used to dealing with.  It had apparently been causing problems on a number of sites, but had up to that point eluded the software support teams at the factory.

Then there were the Cash Dispensers, based on the TC500, and they were a nightmare to service due to the security and the problem that they were often in the public area of the branches, so working on them was a lot more difficult.  They were often vandalised, which made for a long job to repair things like a pin keyboard filled with Superglue, which happened quite regularly.  There were also huge problems with the "umbilical cord" of cables that connected the "computer" trolley to the rest of the hardware that was physically built into the wall of the bank.  It was too easy for that to get damaged, and if it did, finding the one wire among 100 or more that was broken took a long time.  I didn't miss them when I moved on to other things!

Then came the B700, with all sorts of strange options, and then the B80 and later the B90 range, but by then I'd made what would be regarded as an unusual move.  At the request of the Sales Manager, Dick Booth, I transferred from engineering to Sales, and I spent a number of years selling things like the L8000 and L9000, then the B80, before becoming more involved with a more technical role on software support, and later with an emphasis on Manufacturing Production Control System on B700 & B800 systems.

It was an interesting time.  I managed to sell the first B80 in Bristol branch area, much to the chagrin of the product sales manager for B80s, Bernard Ward.  It went into a housing association, and then I sold a number of other B80 systems running a package known as KeyBMS to small wholesalers for use as invoice and accounting systems.  We had one of the highest volume B80 sites in the early days.  They were doing nearly 200 invoices a day, with up to 150 lines on each invoice.  They were a sweet and tobacco cash-and-carry, and the B80 (initially on 2x 8" floppy discs!) worked incredibly well and was subsequently upgraded to add things like a line printer and 30 MB Disk Drive that was the size of a filing cabinet.  Having been involved in the engineering side was a huge advantage, as it meant I had a much better understanding of what was technically possible on the hardware, as well as a programming background.  The early TC500 courses spent quite a bit of time on the SL3 programming language, so I was able to get very close to the software side of the Series L, which was a big advantage when discussing what could or could not be done, and it wasn't a big change to get involved with COBOL on the B80.

At the other end of the scale, there were things like programmable calculators, (C6000 I think)  which used a magnetic stripe card with 16 memory accumulators, and I sold a number of these doing a payroll application.  I was able to put together a full build-up to gross pay, with overtime rates, then gross to net calculation, and a full coin analysis, all within 16 available memory locations.  And if they got a payslip wrong, they could reload the memories from the previous card, and process the employee again without losing the grand totals for the run.  For a small business, it was a superb and relatively cheap system.  The payslip was specially designed by the sister company of Burroughs Business Forms - a 2-part carbonless-transfer pre-printed form, so the whole thing became a lot less complex for small companies that needed to pay a sometimes complex payroll.  The first was a small bakery with 50 staff, who all worked varying hours at different rates, so to be able to automate this was a huge saving for the owner, with a list of what notes and coin to get from the bank being the icing on the cake for him.

As someone else mentioned, the money for an engineer wasn't particularly good, and it was one of the reasons for moving over to Sales in 1975, which worked out quite well until in 1978 Head Office caused me huge grief with a site in Weymouth.  We'd taken an order for a B800 running a complex production control system, but for reasons that I never got to the bottom of, while the software was in our price lists and "available" to sell, they never placed the order for a number of critical modules, so when the customer started pushing for delivery, head office informed us that it wasn't available and would not be.

The customer, a large electronics coil winding company in Weymouth, was obviously very upset and took legal action.  I left Burroughs and joined my former Zone Sales manager in a company (Deane Computer Services) he'd taken over some time earlier, as I wasn't prepared to lose the commission that I'd earned on the B800 sale, as it was through no fault on my part that the order was being cancelled.  It took a while for it all to get sorted out, and in the end the company I'd moved to was responsible for the installation of a machine to replace the Weymouth B800, which worked very well and we subsequently also replaced the B80 in the sweet and tobacco cash-and-carry with a larger multi-screen system that we programmed in-house.

We still did a lot of work on Burroughs kit, buying and selling used Series L machines, and in some cases providing maintenance support for them.  That led to my next role in 1983 -  Data Processing Manager for a large poultry processor in Tiverton, Devon.  We had been supporting a number of Series L machines which were damaged beyond repair by a flash flood, and I spent the 3 months installing a multi-screen minicomputer there to replace the four machines they'd been using for invoicing and sales ledger.  That led to an offer to join them and manage the data processing function for them, with a new system to be designed to cover all aspects of their operation.  That was a 3-year project and a very good position, and led to a later move to a position as European Technical Support Manger for a major financial services software house that was using the hardware platform that I'd been responsible for in Tiverton.

The training and experience at Burroughs laid a good foundation that still stands me in good stead some 45 years later.  Modern hardware is smaller, faster and technically more capable these days, but the underlying principles are still the same,  and sticking to them has prevented me from being caught out on many occasions.  We had a good team in Bristol, and all in all Burroughs wasn't a bad company to work for.  The last of the L9000 machines were incredibly capable and reliable, and provided a good platform for many companies.  They provided the perfect platform for the B80 and B90 systems that came along to replace them.  Looking back, the only difference between the B80 and things like the Commodore 8000s and then the first IBM PCs was that the B80 had a much better operating system than any of those early PC systems.  It's a pity that Burroughs was not able to take advantage of the advantages that the B80 and B90 platform had.  In hindsight, they were a long way ahead of the competition, but didn't know how to capitalise on that advantage.

I can still remember some of the presentations I put together on the early B80s.  We had a line printer in the branch, and a demonstration of real-time invoicing while also printing statements on the line printer was, for its time, light years ahead of anything that could be done by the competition, but we couldn't get access to the software source code, so were restricted in what we could offer without spending massive sums on customising.

The funniest was probably having to demonstrate a B80 in the skittle alley of a car distributorship in Yeovil, ( big L9 customer, and very friendly).  It was a smart place to use, carpeted, etc, with a small mini bar etc, but I had to demonstrate for 3 days without shoes on.

If I didn't take them off, the static from the carpet would cause the machine to reboot unpredictably, which wasn't helpful!  It wasn't much fun for my finger tips either - they were pretty big sparks, jumped about an inch, but unpredictable. 

It caused some merriment at the time, and made it easier to get customers to set up their sites correctly when we came to install, they remembered the issue, which made for stable sites going forward.

Ahhh, memories! - but they might be of interest to some of the other people who were involved during what was a turbulent time in computing.

SUBJECT: Memories from Rodger K Beaton, Arkansas, USA

I started in the Plymouth factory as a DA tech in 1968.  When I started there they were manufacturing the Sensamatic.  They had 6,000 employees.  Then they started the TC500 - I helped build that.  The factory went down to about 1,200 employees.

I became an ALGOL DA programmer.  I designed programs to build the computers and test the resulting comps.

I went to WHQ for a while, then transferred to Carlsbad, Calif.  Then they moved us to Rancho Barnardo 2, and then they closed that and we went to RB 1. Then Cadence bought the small engineering group in RB 1, and we moved to another building.

They fired anyone over 50 after 2 years, and a few other tokens.  Then they closed the facility a year later.  We made chips, then chip manufacture was all moved to Taiwan, with no environmental control, but our chips could not compete.  Jobs gone.

I moped for about a year.

Then my daughter mentioned Y2K and that they were looking for old Burroughs programmers.  My lucky day, and an interview in Toronto!  They were dazzled, and could not believe I could program an A machine (I could make it stand on its heels!).

So they sent me to work for CTA in Lansing.  It was late 1998, and they had to be done by end of 1998.  We made it - I worked many hours and made lots of money.  I was a tester, looking at documents and comparing till I was seeing double.

Then in 1999 they fired everyone except me - they were still dazzled.  Should they keep me, just in case???   So I had nothing to do for about a year, and to stay amused I sat down and tried to figure out the state's software for its retirement systems.  They were dumb and poorly written.  I wrote ALGOL programs and in a few pages duplicated their software and fixed their problems!!!  Extracted data, and FTP to their Excel spreadsheets.  ORS was ecstatic.

DMB was not impressed - they hated my guts.  They were an XGEN shop - worst idea I ever heard of.  4th Generation indeed! - makes a COBOL program the size of the Encyclopaedia Britannia!

So I reached age 58, my last-ditch retirement target.  I said I�m retiring!!

They freaked out.  ORS said they would create a new department just for me at ORS.  I deeply thanked them and said "nope", bought a home in Arkansas, paid cash and retired. Y2K cash!!!

Treasury did entice me to go back about 6 months later, for $100 an hour.  I thought that was pretty fair.  They ordered me to write in XGEN.  I lasted one week!!

Back here in Arkansas, life is good.

SUBJECT: Programming Burroughs L series from Jeffrey Haran

My first programming job was programming L series for a company that sold document processing systems to the mortgage industry in Castro Valley, CA circa 1975.

You'd mentioned 3 different ways of programming an L.  We did it sort of like this description you provided:

"in COBOL.  In this case the program was written on programming sheets, which were then sent to a Burroughs data centre to be punched onto 80-column cards.  These were then processed with a COBOL/GP300 compiler application, that produced the final program on punched paper tape - if the program compiled!  If it didn't (which was usually the case first-time round), the programming sheets had to be changed, and the whole process had to be gone through again.  This could take several times, and several weeks.  As a result, this approach was rarely used."

Except we wrote the source code on the coding sheets in the GP300 assembler rather than COBOL.  The advantage to using the assembler directly was we avoided the problem of getting it to compile the first time around that you mentioned.  The assembler was much more forgiving than the COBOL compiler in getting us a listing and object code on paper tape.  Even in the presence of errors like mistyped instructions and references to non-existent labels the assembler would generate a listing and object tape, so we usually got something we could work with on the first build.

We'd take our coding sheets to a keypunch outfit a couple blocks away in Castro Valley.  The next day or two they'd have the punch cards ready, so we would then drive to San Francisco and do one build on the big Burroughs mainframe at the Burroughs office on Battery Street, take our error-filled listing and paper tape back to Castro Valley and then patch in fixes for the build errors using the Memory Modify utility there.  These patches would be recorded in pencil on the paper listing.  Bug fixes, minor and sometimes not so minor enhancements would be done the same way, so over time the source code a given customer was running would end up being represented by a wrinkled, original paper listing covered in endless penciled-in patches and coffee stains.

That was our "git".

I was still in college when I started that job (summer between my sophomore and junior years).  By the time I graduated I had turned the job into a contracting gig where I wrote L applications for customers that Burroughs sold systems to (since they didn't employ anybody who knew how to program L-series, apparently).  That lasted until about 1980 and was pretty lucrative for an early-20s snot-nosed kid who didn't really know much about computers yet.

One of the companies I wrote a billing application for about 1977 (a cement company in Berkeley) was still using that application on their L9000 system in the mid-90s.  They told me they'd had several other companies come in and try to replace it with something more modern like PCs on an Ethernet, but those attempts all failed.  The new guys couldn't quite get the billing system right.  I should have charged monthly maintenance fees like Burroughs did!

I still have an old Burroughs 80-column card reader in my garage that I picked up as surplus back in the late 70s from what was left of the mortgage documents company.  It's about the size of 2 toasters.  Nothing to connect it to though.  Which is just as well I suppose. Burroughs must be charging a fortune for maintenance on those old Ls today!

SUBJECT: Field Engineering in Burroughs Manchester from Vince Chadwick

I joined Burroughs Manchester branch in January 1970 as a trainee field engineer.  The branch manager was a mad Irishman called Paddy Patton, and the FE manager was Brian Cunliff.  The initial interviews had been at the Midland Hotel in Manchester where a couple of hundred applicants were whittled down to 10 or so.  My final interview was with Brian Cunliff in his office, when he handed me a basic adding machine without its case and asked me various questions on it, such as "how does it add?", "how does it total?" etc.

I joined the Banking Division and after initial training worked mostly in the city centre on adding machines and the dreadful P6000 cheque encoder.  I had no car but could walk or get the bus between most calls.  Then came TC500 training and years of changing decoders, tachometers, re-loading software, etc as well as the occasional electronic fault.  John Doyle was my Zone Manager and I remember characters like Pete Dawes, Kevin McGlaughlin, Pete Davies and John Noone.  I was promoted to FE Group Leader and having married and moved to south Cheshire I spent a lot of time at Barclays Bank Rabroke Hall in Cheshire, and travelling my territory all around south Cheshire with my other engineers in the group.  I remember the Nat West TC310, the RT1100 cash dispenser (I can tell you some tales about those!) and the B700 (at Radbroke).

An earlier correspondent to the site remembers head office tech guru Keith Grant-Scrutton who I came across several times.  He was held in awe due his technical ability, and generally known as Keith Grand Scrotum!.

As some have said, Burroughs did not pay engineers very much, so in 1978 I left to join DEC.  That was a fun company - much better paid, an excellent social side, and rocket-like growth.  Then came microprocessor control and the end of fault-finding to component level with a scope - it became a board-swapping exercise.  Time to move on.

I took a change of tack and joined software house SPL as a project engineer in 1982.  That was a super place to work and really widened my experience to include project management, dealing with suppliers (we built our own kit), and running progress meetings etc.  SPL got taken over by Systems Designers, then we took over Scicon and became SD Scicon.  By then I was working in sales support mostly on bid work.  Then, in 1992, US computer services giant EDS took us over.  Many of my colleagues in hardware, drawing office, lab staff, wiremen, etc were “let go” as EDS didn't make stuff. I survived through my sales-related role.  However, I didn't like the EDS culture, but by then age, salary, and pension pot made leaving difficult, so I stayed.

I eventually became a Solution Architect, the technical lead on multi-million £ IT bids into UK government and top end industry.  Long hours, a lot of working away from home, but quite lucrative if we won the deal.  My job was to pull together “Subject Matter Experts” within EDS such as storage, networking, security, data centre, etc experts.  I had to ensure the total solution would work without any 'gaps' and also no 'overlaps', and present the solution to the client in terms of business enhancement, not technology.

Nearly 11 years ago EDS was going through a bad patch and the offer of an “early bath” came my way (voluntary redundancy with a 'deal').  I was burning out by then and grabbed the VR with both hands!

The last 11 years have flashed by.  Retirement was the best thing I ever did!

SUBJECT: Burroughs TCs at NatWest from Roger Barton

You've omitted the Burroughs TC310.  This was a stripped-down version of the TC500 used by NatWest in their branches.  The mechanics and basic electronics were the same, the difference being in the comms and firmware.  The TC310 had a built-in modem which connected directly to the GPO wires: this would have been very unusual for the time as the GPO was very touchy about what was connected to its network.

The 2-wire GPO connection went to a local IBM concentrator, unlike Midland, Barclays and Yorkshire banks who used TC500s connected directly to mainframes via a 4-wire connection (with 2-wire dialled standby) and GPO modems.  The concentrators were connected to the mainframe.

These concentrators were installed in the larger branches and served a local area: for example the concentrator in NatWest Halifax served the TC310s in smaller towns like Todmorden and Hebden bridge (I installed the TC310 in Todmorden).

NatWest had a computer centre in Bradford with a room full of TC500s (not TC310s).  I think these were used to process data physically transported from branches without a TC310.  The computer centre had an IBM mainframe, but I'm not sure where this fitted into the network.

Also based on the TC500 was the Burroughs cash dispenser (ATM).  I was never trained on these, they weren't a great success and only used by Midland Bank in the UK.

The inbuilt tape reader you mention was weird: it actually operated the keyboard (although the keys themselves didn't move), hence the low speed.  It was the only way of loading firmware and took ages.  Before loading firmware we had to install jumpers to write-enable the firmware area of the hard disk.  There was a facility called "Keyboard to Q", "Q" being the write register for the disk, which allowed you to write from the keyboard directly to disk, really only used for de-bugging.

Final ramble: for fault finding we had a system called "MTR" meaning "Maintenance Test Routine".  This consisted of a massive manual which was effectively a flowchart written as a series of tables, and a meter.  This meter measured the mark-space ratio (0 to 100%) of the logic signal it was connected to.

To use the system you'd look up the basic symptoms in the manual and connect your meter as instructed.  The book would ask e.g. "is the signal greater or less than 50%" you were then directed to another part of the book depending on your answer, and finally you found the fault and fixed it. Or not.

The flaw in the system was, what if the meter showed 49%? These meters had an analogue scale: were they calibrated?  Had they been knocked about a bit?  So you'd assume it read less than 50%, but you made a note just in case it wasn't and you had to return to that point and revise your decision.

SUBJECT: Burroughs Cumbernauld from David Weaver

I joined Burroughs in April 1974 as an Electronic Technician at Burroughs, Wardpark, Cumbernauld.

At that time the B80 was still in its infancy, a lot of it still theoretical.  My first sevaral months were spent qualifying components - resistors as it happens.  You get 100 resistors from a vendor, of the same value.  You label them all and then measure them, size and resistance.  Then you cook them at designed wattage for 100 hours and measure them again, checking for failures.  Whilst they are cooking you get 100 resistors and start over again ...

Next I was moved into the prototyping section.  Building boards to go in the first test machines.  I learnt wire wrapping to construct the circuitry - there were no printed circuits at that stage.  I recall spending a couple of months wire-wrapping memory cards for the first systems.  These were then put in a backplane and debugged by the engineers.  The one I had most contact with was Jim Stephenson, although I sometimes liased with Ted Skwara (probably spelt wrong).  I believe Colin Crookes was Lead Engineer.

The CPU(processor) also started out as a wire-wrapped system.  It was huge.  From memory it was a rack with several layers and some six or eight feet long.  I was told it was so big because the chips used contained only four or five transistors, which were linked into gates and registers to make the logic parts of the cpu.  At the same time the mechanical parts were constructed to hold the backplane, printer, keyboard and power supply.  The supply was a huge beast shoving out 200A for the +5v(logic) and +/-12v for serial circuits.

Prior to starting manufacture we built six 'first offs' for the engineers and programmers to sort out bugs before taking the project to full manufacture.  When these got through the proving stage manufacture was begun in the UK and in the USA coupled to a big sales drive.  A conference was held in California to show the system to the sales force.

The 180cps printers were made in-house for a while, before being moved to Seneffe(Belgium).  The main backplanes were also done in-house on Gardner-Denver machines with all slots and connectors being wire wrapped.  Keyboards were also made in house with a 100-ton press stamping out the base plates for the keys.  Eventually manufacture was begun, with actual PCBs and lines of ladies building boards (double-sided PTH).

At the start I was assigned duties on the line debugging boards, mainly dry joints and misplaced components.  It was a steep learning curve all round and meant an increase in staff.  As production ramped up auto -nsert machines were brought in, as were semi-automatic crimping machines - the system used a number of interconnecting harnesses.  A GenRad 2270 ATE (Automatic Test Equipment) tester was installed (we were the first to get one) and I went on a course to learn how to program it to test the finished PCBs.

A word about the system boards.  When the breadboard CPU was deemed correct and bug-free the circuits were sent to Burroughs at Ranch Bernardo in California for committing to silicon.  That is, the circuits were split up and came back as a set of 'chips'.  I cannot remember what type - NMOS/CMOS or whatever - they did need two supplies, +5v and -12v which probably seems strange.  They were fifty-one pin devices, four lots of twelve pins and three power pins. vI was able to argue with one instructor who said ALL integrated circuits had an even number of pins ... and showed him one.

Originally the processor set came in as nine chips, all with different names, such as MAR, MicroMAR etc.  Two frontplane connectors went from the cpu board (double height) to adjacent single-height boards.  One was the memory controller and the other was the arithmetic unit.  From memory the CPU ran at 3Mhz and there was 64K of memory, which could be expanded.

Other double-height boards generally had one large fifty-one pin chip.  These were typically peripheral boards such as printer drivers or multi-I/O boards for serial terminals (keyboard + screens).  It was designed as a multi-user system (terminals being time-shared/multiplexed).

File storage was in two cassette drives, either MFM or PE types.  Later systems had one or two floppy disk drives (8").  I cannot recall capacity of the drives, maybe 240KB.  We did see drives which had been modified by another Burroughs plant at Glenrothes, which had a 6502 CPU with hardware to allow storage of 6Mb on a floppy disk.  Problem with these was time taken to initialise - it had to calibrate itself for every disk used by finding the first and last tracks so it could calculate track spacing to read all tracks in between, as for thermal expansion ...

Just as an aside, Burroughs at Glenrothes had a management buy-out and reappeared as Rodime, who made hard drives.

The CPU board was updated with the CPU redone on three chips and included the arithmetic and memory address circuitry.  This led to the B90 startup.

In the late 1970s to the start of the 80s we were making systems like there was no tomorrow.  But sales dropped off and we simply built and sent to store.  Ended up with so much unsold stock it was embarassing.  This led to the first and second lot of redundancies - bad times.  I went out in the second lot in 1981.

For all that I not only enjoyed my time at Burroughs - I felt we were leading the company with a product designed and built in the UK.

The flaw in the system was, what if the meter showed 49%? These meters had an analogue scale: were they calibrated?  Had they been knocked about a bit?  So you'd assume it read less than 50%, but you made a note just in case it wasn't and you had to return to that point and revise your decision.

SUBJECT: Burroughs Calgary from Philip Black

I worked for Burroughs in the mid 1970s in Calgary.  All the training was done in Toronto.  They made me pass an accounting exam prior to the first course.  Although I worked in sales as a Territory Manager, we employed a somewhat unique approach to the whole design process.  Once you sold the machine, you programmed the system and trained the operators.  I had a rural territory so most of the customization and bug fixes were done in machine language.  You always brought a tape punch with you so you could pump out a new tape.  I still have the debugging/trace tapes and the “little orange book” although you got to learn the hex instructions in short order.  Tax time was always interesting with the payroll programs.  Many of the payrolls max’ed out the memory so if they introduced a new tranche in rates you had to look around for some features you could delete.  My last memories were a Payables/General Ledger program for a municipality south of here.  Written in COBOL and utilized a 3-cassette sort.  Still gives me nightmares!

SUBJECT: B80/B90 Experiences from Tanj Bennett

I worked on the B80/90 at Cumbernauld during 1979, the period when B90 was being developed.  I did drivers for the 3MB floppy (which was designed and built at Glenrothes site), as well as swapping the segments of extended memory which B90 introduced.

Glenrothes built the 3MB floppy using a very flexible membrane which flew over the head rather than rubbing, held at correct distance by Bernoulli effect.  They tried but never succeeded (AFAIK) to make a cartridge version with multiple layers just close enough for an oval head to push in between and have the platter flex to accommodate it – aiming for 30MB?  It was a clever group up there.

I never worked on CMS, it was a mystery to me, done somewhere else.  There was also a native system which we worked with and was used for a lot of the data entry sites, which were a common use, since it was essentially instantaneous for that purpose.  This system was primitive – the disk was used for sequential files essentially emulating tape.  So, Burroughs did not make much noise about it but the banks knew what it was and commonly bought it.

I worked in Assembler, there was no compiler for that OS.  It was the most bizarre instruction set ever.  The design of the silicon started from the architecture of the Burroughs calculator (Sensimatic?) which was an amazing mechanical beast made on the shop floor of the Cumbernauld site up until 1980.  The accumulator was decimal (16 or 20 digits?) and all of the registers were special purpose, like index to memory, or binary arithmetic, or binary to decimal or decimal to binary.  You really needed mind warp to program it.  I can quite believe they used a P-code approach for CMS and Cobol.  I have no idea how they even got that off the ground, although for the time it was very highly integrated into just a couple of LSI chips so it may have impressed the accountants, who decided everything in Burroughs.

For all I know the B90 never shipped with the bare OS like the B80.  I worked there exactly the year of 1979 and was not on the sales side.  My skip-manager ran a peculiar setup where we did B90 work to justify our budget, but then moonlighted our excess time working on wafer-scale integration where a whole 4” wafer was going to be a single content-addressable memory of a couple of Mbyte size.  Crazy, but this was the location who had built the LSI chips so they had access to do more chips while the company actually just raised the clock rate and added the memory extender lines going to the B90, so they wanted something interesting to do.  On the moonlighting side I worked on tiered fault-tolerant serial bus that could route around failures.  They were still shipping to banks so they might have fulfilled existing customer upgrades with the bare OS, I don’t know how that played out.  I suspect Detroit never found out about the WSI moonlighting and it simply vanished.

Burroughs Cumbernauld was such a messed-up place.  I was young and just went with it, but looking back now it was so dysfunctional.  The office tower 3 stories above the shop floor (which spread out over acres) held the electronics teams.  When we went to the cafeteria, we walked through machine shops still making gears and stuff and the management/union face-offs were just what you would expect from Glasgow of the time (Cumbernauld was built to relocate people out of the Gorbals). Everything ran on a budget, and every manager knew how to game their budget.

By summer I felt the weirdness and decided to give it a bit more time, but started studying Japanese to keep my brain in gear.  Around the new year my former CEO found my phone number and invited me to his startup in Winchester, and I was gone.