BACK TO THE INDEX OF MY OTHER PAGES

Riding the office equipment boom.

This is a brief history of my career as an office equipment technician during the years where every new device was less mechanical and more electronic.

Prelude

Electronics was my hobby and many happy hours were spent pulling old radios apart and building 1 valve sets. Somebody gave me the books from a Marconi course and I worked through the 2 foot high stack of manuals that started with spark transmitters. I read hundreds of magazines, haunted Deich Bros and Waltham Dan searching out disposals equipment etc. etc. In 1962 at the age of 16 I started a degree course in electrical engineering at the university of NSW .

I failed one subject, mathematics and decided to repeat it at night (eventually I gave up) while I took a job (in those days you could pick and choose) at G. W. Engineering as a dogs body who ran around making sure that the mainly women workers had what they needed to assemble the various switches they made. When I found myself actually pulling on the handle of a press I scoured the paper for something more technical.

My introduction to the spring hook

Sydney Pincome was an old established firm that held the agency for Facit equipment, about one third of the firm was devoted to the repair of the renowned Facit rotary calculator another third repaired typewriters. The section I started in repaired adding machines and cash registers. I was in my element, the equipment was state of the art and I progressed under supervision to pulling down things to hundreds of parts using my first spring hook, fixing what was necessary and getting them all back together. Occasionally I was sent out on service calls which on public transport could take most of the day.

Logabax 

The ultimate machine that this section had to service was the Logabax.  See model 200 at  http://logabax.free.fr  There were about 10 located in various companies and about 5 in a service beaux on a lower floor. There was one person capable of fixing these and I was detailed to hold the light for him. A logabax was a 1 one meter square mass of metal that could store 198 12 digit numbers in two "boxes" of 99 registers. A keyboard allowed you to add and subtract from any register and transfer between boxes. The Logabax was perhaps unique in that the registers that stored the numbers were not the usual group of cogs but a flat layer of "reglets" that looked like hacksaw blades the position of which represented the number that was stored.  The layers were stacked 99 high and one layer in each box at a time was selected to be acted on. Whenever something went amiss there was a safety mechanism that stopped the beast from destroying itself and a service call was required to put things right.

When the expert was not available I was sent out alone. One frequent job was at a gravel pit at Greysteins on the outskirts of Sydney. This was a a real trek by train then local bus or taxi then a huge walk to the site office from the gate lugging a tool kit (you always checked your bag before departure for lumps of extra steel that jokers added). The actual work was usually only a few minutes and on return you were often told (as a joke-sometimes it wasn't) that it had broken again down just after you left.

To see what was going on in the machine you could screw in a handle and wind the beast over by hand. There was a machine located at Knock and Kirby a large hardware store in the city. One day I had screwed the handle in and started to wind it over when the power cut in and after a few seconds of flailing metal inches from my nose the handle was flung over the heads of the multitude of office staff and landed neatly between desks. Nobody seemed to notice as I nonchalantly retrieved it. Later Mr. Nock or Kirby mentioned to my boss that he had seen a young fellow almost wound up in the machine.

In the middle of the workshop was a sort of stage about 1.5 meters square and half a meter high that I had not seen used but soon would. The Logabax could print and the latest model in the service beaux down stairs had an electrical ribbon mechanism that I eagerly (something electrical at last) volunteered to fix . All I had to make was one measurement with the meter to locate the problem. The moment the probe touched the contact it started to arc and burn. Turning the key-switch off on the front of the machine made no difference to the arc and the wall switch was behind and difficult to reach without moving 1/2 a ton of live metal. Smoke filled the air and the girls realised that a bit of coughing would give them an excuse for an extended break. Most of the under spec. French wiring was burnt out and the machine had to be removed to the workshop and placed on that stage , a process that involved two large wooden poles bolted to the sides and 4 men in the mode of a sedan chair. After a lot of ribbing and comments I set about rewiring it using some of the then new heat-shrink tubing for extra insulation.

Decimal conversion

About this time office equipment companies had a windfall when they were paid to convert or replace all of the old non decimal office equipment. The equipment was first placed in a solvent bath for cleaning and the boss was looking for a label method that would survive the bath. I found the solution was the "write-on" metal tags that you can get for garden use. This show of enthusiasm I think led to my being selected for a course on a new piece of equipment.

Relay Multiplication

The course on the Siemag accounting machine was run by an instructor with no English and there was no one on the course who spoke German. The Siemag consisted of a solenoid operated typewriter that entered and printed out results, the position of the carriage determining what function had to be performed. The storage, addition and subtraction computations were performed by a bar fridge sized collection of mechanical racks and gears but multiplication was performed by relays. At the time it was amazing to see, as you entered a number solenoids pushed out pins the position of which represented the number , push the tab, and a set of racks moved until they hit the pins thus transferring this number to some contacts, after a second number was loaded to a second set of contacts the multiplication could start. The relays started chattering and digit by digit the answer was punched out on those pins, It was then transferred by toothed racks that were moved until they contacted the pins which in turn wound the number onto some gears for addition or sent to the contacts and printed on the typewriter. The instructor would cause some fault and our class of 5 would spend the next few hours poring over the circuit diagrams. Often the fault was diagnosed by looking for the sticky tape that he used to place between the contacts, but we made progress and eventually we could find real faults. The technology was moving quickly and this machine was soon obsolete and was never sold. So it was back to the spring hook.

IBM Customer Engineer

I moved to IBM to to use the skill I thought I had in electronics and applied for a job as a customer engineer. The entry test was mainly about things mechanical which seemed a little odd. We spent a few weeks at head office working our way through self teaching manuals while the numbers built up enough for a course to begin. IBM had wonderful employee benefits eg. 6 months full pay sick leave followed by 2 years half pay. One person on the course had been with IBM for 4 months most of which was on sick leave after a skiing accident. When we numbered about about 10 we were told to report to the training centre to begin the typewriter course. The proportional spacing typewriter had some nifty internals but the thing of wonder was the new bouncing ball model and its spring clutches , clevises and self the adjuster for the rotate band. At the end of the course we had to strip down the pawl block of the proportional model into it's 40 or so bits and bring it back the next day assembled. I actually got it together but with one bit left over, still I was one of the 2 that passed.

I was given a couple of city blocks with 500 or so installed machines. On the first day I was set up when I was asked to go out of my area to "Help" . The job was an old model with the dreaded broken pawl block. The usual work consisted of phoning in to get service jobs and keeping the 3 annual maintenance visits up to date. One job was to replace the rubber platens that when old and hard caused the paper to be cut. They had to be sent back for recovering and carrying them around was a pain. I discovered that if I placed a bundle of them next to busy bus stop they would still be there at the end of the day, I just had to get away from them without someone telling me I had forgotten them. Once a week our city group of 5 met before work with our supervisor in a plush coffee lounge for a meeting. The biggest drag was filling out the time-sheet books, there was a code for everything, one day we had to attend a meeting at head office after work and I coded it in my time-sheet as a meeting which must have pushed the total into overtime and I was called in and it was explained that as the meeting was for my benefit it did not need to be documented. I found IBM to be a really productive company but the nearest I came to electronics was fixing the magnetic belt dictation machines that some customers had. I could not understand why IBM did not come out with electronic calculators.

Electronics at last

Pye Industries was a manufacturer of TV sets who had started to diversify into office equipment and had employed some ex Sydney Pincome staff. When electronic equipment was scheduled to arrive I got a call and I was soon working with most of the old gang. It is hard to describe the feeling that a teco gets when he first opens the box of what is the latest most advanced lump of technology but it is teco heaven. The first Sanyo "DK" (Discrete component) calculators were the size of a typewriter but twice as high with a row of "Nixi" tubes for the display. The 10 or so vertical circuit boards contained the 1000 or so transistors. The logic that ran the calculators of this period was crude and if you subtracted to below zero you did not get a minus sign but you were faced with a 9's complement just like you would if wound a car odometer too far backwards. Pushing the minus key again took you back to a real number and you had to remember that it was a negative. Servicing consisted of swapping boards and some in circuit transistor testing.

Usac accounting machine

One day I was called up to head office and offered a contract, a pay rise and a one month course in Japan on a new electronic accounting machine from the Ushida Yuko. My wife of one month was not really impressed. It was a fantastic trip especially when I visited Sanyo, they were not told that I was a humble teco and I had the red carpet treatment. The Usac was similar to the Siemag mentioned previously but had no mechanicals and used TTL logic chips and was programmed by a plug board that was prompted by the number of tabs on the typewriter.

The factory was in outback Japan (Kanazawa) and my hotel was an english free zone. Breakfast was a poached egg on toast and although I was the only person eating this food I never managed to get anything else. On the nights I had to eat alone I wandered about the small town and selected something from the replica meals that are in the windows of all of the restaurants. After 3 weeks I was homesick and one night as I walked down the hall I heard a english woman speaking and could not resist knocking on the door. Japan has one TV channel devoted to education which I think is a fantastic thing and the english lessons are very popular as I found out.

On my return I was mainly involved in plugging up (programming) various demonstrations in the low level, left shift, store, add sort of language. One request was to convert numbers from base 10 to base 4 and I worked out a programme where you simply entered one number and the machine printed out the conversion, went to the next line and waited for the next number. The customer came in with a list of test numbers and was impressed with the instant results and placed an order. About 4 machines were sold and they never broke down.

About this time Phillips bought out Pye and as Sanyo was a competitor of Phillips the Sanyo calculators had to be spun off and I followed them.

The calculator boom

I joined the newly formed Sanyo Office Machines. Few of the "DK" calculators were sold but the new "ICC" (Intergrated Circuit Calculator) started selling, initially at 2 a month. I can remember doing a service call at the opera house site, it was a huge project but there was only one electronic calculator. The popular ICC141 had about 200 DTL chips and contained a state of the art component a dual 64 BIT shift register manufactured by Philco Ford. Any dead shift registers had their cans cut open and were in demand by the salesmen so they could show customers how advanced the technology was. The shift registers were used to store the 14 digit numbers in the calculator. Just how advanced this was is highlighted by the fact that a competing calculator used a rotating magnetic drum instead.

Nixi v 7 Segment

All our competitors used NIXI tubes for the display, these are glass tubes with the numbers 0 through 9 formed by wires one behind the other. When power is applied to the required number it emits a neon glow. We had the new 7 segment system and it was an uphill battle to convince prospects that they could read numbers made from straight lines. Our 7 segment display used a small light globe for each segment and changing these globes was a common job. Each display had 112 globes and if you accidentally knocked the board when you were changing one you blew 3 more. Below is a digit showing its 18 parts. It also required a socket. No wonder modern displays are cheaper.

Digit using light globes

Fault finding

Our use of the advanced 64 bit shift register meant that we could not single step the logic like some competitors, this coupled with the repetitive use of the various logic components when for instance a division was calculated made for a sticky problem. Most faults were caused by open circuit input diodes in the DTL logic chips or faulty through hole joints and were tracked down laboriously with a dual beam oscilloscope. Later better methods were developed.

The first tools

The first attempt at easier logic solving was a simple logic probe, this developed into what was known to us as an "IDIOT BOX". With this matchbox sized device you could set a few plugs and answer questions like "is this high when this is low and this is high". Field repair of faulty logic could now be attempted but it was still hard to determine just what to check when faced with some problems and intermittent problems still posed problems.

The ultimate logic tool

I realised that we could synchronise a good and a faulty machine, feed some problem to them both and compare the data stream at any point. A probe was made that indicated when there was any difference in the data. This was helpful but as soon the calculation process came to some faulty logic the calculation went off the rails and the data stream at almost all parts of the logic would from then on be different on both machines. All we had to do was find the first instance of a difference and we had the fault. To this end a counter was rigged up to count the clock pulses from the start to when a difference was detected. A plug board was made to set up the test calculation and a foot switch rigged to start the system (you had a probe in each hand). To use this device you put your probes on some point (on the good and bad machine) that was faulty (you could always use the display as it was after all faulty) pressed the foot switch and searched to find the lowest number. You did not need any knowledge of the logic, if you were on an output and you could find a lower number on an input to a given chip that chip was not faulty. We had diagrams showing the tracks so we just worked backwards to the next chip, ran the probes around it and followed the lowest number. A chip with a faulty output before a faulty input was faulty.

The ultimate logic tool for intermittent logic

The system above soon worked it's way through our stack of previously unsolved boards but intermittent boards still posed a significant problem. As an example consider numbers stored in a circular shift register that is continuously clocked, if one of the components fails for a fraction of a second the whole register may go high or low but you do not know which part failed. Instead of counting clock pulses all we needed to know was did this go faulty before that. The final system replaced the clock pulse counter with a before or after circuit. The problem is you needed 4 hands. To use this system you found a difference then used another set of probes to find an earlier difference, you then used the first set of probes to find an even earlier one. You could leave the probes connected and continually run the test. Each time a fault occurred you made progress because you had answered a question with a definitive answer. Not requiring any counter meant that the whole device could fit in a shirt pocket. Solving intermittent problems was now just a matter of slowly tracing the fault back.Each time the fault occurred you made progress.

Smaller and smaller with bigger and bigger parts


A mid sized (MSI) simple calculator box and a modern scientific.

Gradually the number of integrated circuits was reduced as their complexity and cost rose. The first real breakthrough for us was the 5 chip 4 function calculator (82D) that could be held in the hand (a big hand) it was powered by 5 C size NiCad batteries and it came in a red felt lined box, it sold in the thousands. The main faults were the micro switch in the pop up hood, the battery pack and the DC power lead. We never discovered a way to determine which of the G.I.A.N.T. (General Instruments Advanced Nitrate Technology) integrated circuits were faulty when they failed as it seemed that the logic was drawn on 5 pages and each page went into a chip. The need for electronic servicing was sharply declining as the component count went down and the cost was soon to follow the downwards spiral.

The computer

When I heard that Sanyo was bringing out a computer I thought about fridge sized units but what arrived was desk sized box with display and 3 1/4 " floppy disk and a separate keyboard. The processor was a 16 bit TMS9900 (Texas Instruments) and Sanyo had written an operating system. Later this model was teamed with a solenoid driven typewriter and sold as a word processor. This was a 16 bit machine way before the competition.

The first real computer we sold used the 8085 processor, had 8" floppy disks and ran the new CPM operating system. A model with 360K 5 1/4 floppy's (MBC1000) was more popular and thanks to a multiuser accounting package (written in assembler) was sold in fair numbers.

CPM gave way to MSDOS and we had a boom time with a low cost machine with similar specifications to the new IBM PC. Sanyo did the right thing and made the computer MSDOS compatible and it came with a full set of Wordstar software. This MBC555 was sold in big numbers as a word processor combined with daisy wheel printers. It is hard to fathom but Sanyo wrote the Basic interpreter that came with it, unlike another Basic this gave correct double precision trig results and I remember surveyors buying it for this reason. Various models followed, a "lugable" with colour display, various 286 "AT" machines and notebook computers with 20M hard disks. Sales of these models were retarded because they were MSDOS compatible but not IBM clones. The lack of processing power in the computers of the day meant that a lot of software was written to talk directly to IBM hardware and this required the machine to at least be a clone to run it.

Servicing computers

Very little went wrong with circuit boards and Sanyo used only top quality floppy disks that never gave any trouble. Most of the work consisted of installing, up grading hardware and sorting out customers software. There was a great demand for hard disks and we sourced many brands. The first units were 5 megabyte full height devices that gave endless trouble especially when pressed into running the 5 user accounting package that we were selling. Finally Sanyo brought out an 8 megabyte unit that was fast, silent and never failed but by then the customers wanted the latest 20 megabyte non Sanyo units that died like flies.

When the SANYO ICON Unix computers arrived I concentrated full time on supporting them but following the previous pattern little if any actual electronic servicing was needed and any failed boards were shipped back to the factory.

BACK TO THE INDEX OF MY OTHER PAGES