Coal Burning Computing.

I went into the computer industry in 1986 as a “Technical Support Engineer” and worked on the bench doing repairs to a variety of different machines. I never got to work as one of those highly paid support guys who walked in, declared the unit bust and took it back for idiots like me to repair. The decline in the industry in South Africa finally ousted me and I no longer work with computers. Maybe its a good thing because when I look back……

The company I joined (Visiondata (Pty) Ltd) in 1986 were dealers and distributors for the Ontel, Hartley and Point 4 brand of computers in South Africa. The PC as we know it today did not exist as yet and small micro’s like the Commodore and Atari were the only ones available on the home markets. On my first day at my new job I was dumped right into the deep end and given an Anadex Rapid Scribe printer to repair. It had an intermittent fault on it which we were trying to localise using a variety of techniques. I sat and stared at that machine like an idiot, I had no clue about things like “on line”, “self test” “form feed” “VFU” and carriage and feed motors. After all, I had joined the company with the intention of learning about this stuff. 

I had been allocated to the workshop of the Ontel department and that implied fixing Ontels and all their peripherals as well as any other odds and sods which may have come my way. 

The Ontel came in a few models, the OP1-R, OP1-SA, OP1-70, and I seem to recall an OP1-15 as well. It usually had a large hard drive attached to it, and that would be either a Diablo 44B or CDC Hawk 10M drive, or as in the one model, an 11 (Could be 8″) Inch dual floppy disk drive (Tandon?). Some of our customers had the Phoenix 96M HDD and they were usually running the OP1-SA, all of which were connected to a separate file controller by means of a flat multi-pin Winchester cable. The one version of the Ontel could handle its own semi dumb Ontel terminal which was attached by means of its own cable. The CPU was a huge machine with a massive VDU for the time, possibly 12″ and had a cardcage which housed the processor board, a display board, 2 memory boards and 4 addition controllers, one being a drive controller and the other being a printer controller.
The South African machines had paper white screens and many tubes were very burned in after all the years. Repairing these machines was a mission, the VDU had its own video board which had components which were not available in SA so we had to resort to a generic equivalent in the horizontal drive circuitry, we couldn’t get line output transformers for them so we cannibalized old machines. The repairs were straight forward horizontal field collapse or vertical stability. 
A main power supply was housed underneath the VDU and it was only accessible with major surgery, lucky it wasn’t too temperamental but it had some huge caps in it. I do recall that each time we removed the VDU section we hauled out our biggest screwdriver and discharged the tube. The controller boards were a different story, and we were fortunate in having spare boards to use for comparison and there weren’t too many component level repairs to carry out. Most of the problems we seemed to have were related to communication between the CPU and the printer, drive or terminals. Luckily we did have diagrams. mnemonic lists, and very basic circuit descriptions. The unit would come ready when the 4 status lights came on and if I remember correctly was obtained by the 6th key on the keyboard, about where F6 is on a modern keyboard. Then we were running HDOS and all that it entailed. Our software seemed to be written in a language called SACBOL which was supposedly a South African version of Cobol.

The one version of the Ontel (OP1-15?)  was a glimpse into the future with its green screen and ergonomic beige cabinet, it was much better looking than the old OP1-SA. Its boards were all clipped onto each other and slid into the case on slides which ran underneath the screen. We only had 2 of these machines, the one suffering from a mysterious intermittent contact that swift slap would cure. Nobody ever found the fault on that one.

Fault finding the memory was another chore, there was a special Eprom that loaded a memory test, and by looking at the combination of status lights and the graphical display we could get a fair idea of what chip was faulty. I suspect the memory boards were 16K per board. The other way was the finger test… any ram chip that fried your spit wet finger was sure to be faulty. We made the printer controllers ourselves and it was easier to replace all 16 chips on it than look for an individual faulty chip.

The biggest problem we seemed to have were with the HDD’s and file controllers as well as the flat ribbon cable between the 2. Many of these were very old and cranky and the cables had a tendency to loose connections. Our customers were also very fond of ignoring backups and when the drives crashed it was always a major mission to restore from the last known good backup with its attendant data loss. I also encountered the huge gap between programmer and technician and we spent our lives running after them.

The CDC and Diablo 44B 10M we had were huge machines, easily weighing in at about 150lbs each. They had a fixed platter and a removable diskpack of 5M capacity each. The 4 heads were mounted on a voice coil and a very big heavy magnet along with a huge transformer and steel chassis made up the weight of these machines. Between the Diablo and CDC, I found that the CDC was the more reliable of the 2. We very rarely did component level repairs on the drives, but head and platter changes were the norm on the Diablo’s. These drives had very high quality filtration systems and the read/write heads floated on a cushion of air when they were loaded and a head crash was when the head physically contacted the platter.  Head crashes were a common occurrence for these units, because they were supposed to be housed in a climate controlled environment and would ordinarily be very reliable, however our customers very rarely seemed to adhere to this practise. A speck of dust could literally crash a drive.

Diablo 44B font panel

When a crash occurred the drives were brought into the workshop (manually carried up 3 flights of stairs I might add) and part of my job was to get it back to working condition once again. First I would strip out the old platter, heads and filters, then service and clean what had to be done, then install a new platter, filter and heads and then start her up.  A bad crash would leave the drive full of iron filings and I found Prestik very useful.  We would leave the machine to purge and get to operating temp before attempting to do the alignment. The variations in temperature were catered for in the drive but our workshops were neither dustfree or air conditioned so we had to make sure that we monitored the ambient temperature. The removable platter in the Diablo was located under the circular aluminium cover in the bowl.

Disk bowl

The head alignment was critical, it involved getting the heads to be not only 90 degrees to the axis of the machine, but at the correct distance from the sector mark.

Top (removable) platter R/W heads

The top heads were very critical as they had to read the boot sector of the removable platter. A removable disk was theoretically readable on any correctly aligned disk drive, and a removable disk that did not read on a correctly aligned drive was either faulty or the alignment of the drive it was written on was out of spec. The drive was hard sectored and the sector marks were on the disk hubs. The sensor that read the sector marks on the hub had to be the correct distance from the sector marks and aligned at the correct angle to the marks.  You can see the sensor (white object in the disk bowl) in the first image above.  We used a special removable disk called a CE pack to do the alignment with as well as an oscilloscope and lots of patience.  Once you put the diskpack into the bowl and closed the side latches the drive would be started, as it started the latches would be locked with solenoids and  as the disks spooled up a set of motor driven brushes would sweep out of a recess next to the heads and brush across the platters as the machine got up to speed. The next event would be the heads loading, with a resounding “thunk” noise.  A grinding noise was bad news.
The voice coil could also be manually operated by a switch and our hearts would give a little bump when we triggered those heads to load. 
Then we would run diagnostics after formatting the bottom surface. There were a few things we could do, but mostly we would copy one good working disk to the bottom platter and then run a compare between it for a few hours. We were also able to read individual sectors using a utility called “hzap” and could manipulate the data within. Often the programmers were working down to this level when the customers encountered a problem.

I also used to work on the Phoenix 96 MB HDD’s  although I was only limited to filter and head changes as well as cleaning and purging.

Phoenix 96M HDD front view

The head alignment on these drives was even more critical, and the wrong move could crash the drive, and, when they crashed they did so spectacularly. I seem to recall that they had 2 fixed disks and 1 removable disk (6 heads in total), and they had to operate within even stricter parameters of temperature and dust. One of our customers had at least 18 of these beasties and their computer room had its own underfloor air con running 24/7 to keep them within the operating temperatures.

Read/Write heads

The drives were connected to a file controller via flat ribbon cables with Winchester connectors on and these were very prone to going faulty, more so than the drives were. The Phoenix was very similar in operation to the 10M drives, although much more critical because it was a quicker drive and had a much greater capacity (96M as opposed to 10M). I seem to recall that the one fixed patter was written with the sector information which made the fixed surfaces soft sectored. The Phoenix also had a lo air/no air sensor in its filter and a special latch on the front door that would be locked by a solenoid. The diskpack was also slotted into a front loading tray which meant that you could stack drives on top of each other, the 10M’s you had to move the drive forward on rails to drop the diskpack in.

Phoenix from above with top cover removed

I would say at least 1 day of my week was spent working on these drives, but as more customers began moving way from the Ontel so we ended up with more spare machines. I also used to repair the printers for the company, and as I became more proficient I ended up working on a variety of machines, from the Okidata, Seikosha, Citoh, Centronics, and Anadex right through to the big line matrix printers like the Printronix P300 and P600. I also started to help with the modifying and repair of dumb terminals which were used with the Point 4 equipment which we also used to distribute. By the time I joined the company in 1986 many customers were moving away from the Ontel and replacing them with other machines. We tried to sell them the Point 4, but alas many felt that they had had enough of the company. Oddly enough though, many customers were determined to not get rid of the Ontel as most of the software was custom written for them and migrating it and the data wasn’t always successful.

The Point 4 was more of a “mini” computer and was the technical directors personal empire. He was loath to pass on any information on the machine and handled the repairs to them himself. I was saddled with the QUIC interface tape streamers and the terminals and printers. The Point 4 came in a few formats, usually housed in 19″ racks or in their own self contained case. If my memory serves me correctly there was the MK2 which wasn’t widely used and which had a old quarterback tape streamer and a 90M Kennedy drive, the MK5 which used the Phoenix 96M drive, the MK9 which had small Maxtor drives in them (possibly 45M) and the MK12 which had a variety of combinations of drives. These units were mostly installed in a chain of private hospitals and we also supplied the terminals and printers for them. We used mostly Televideo TVI925/910 and Qume terminals but later introduced Tatung and Wyse terminals. We ran them in VT100 emulation mostly, although later on we were using Wyse emulations. The Televideo were the most reliable of the lot and were easy to repair, most faults were video and communications related. It had a large electrolytic capacitor in the horizontal circuitry and periodically this would burst with spectacular results. The biggest problem with the Televideo was that it took up a whole desk when opened up. The Tatung was in our opinion not an ideal machine, especially when it got older. Its keyboards were very prone to problems and the video side of them was a bit of a nightmare. They were difficult to strip and I eventually had to manufacture a jig so that I could work on the boards outside of the machine. The Qumes were a pleasure to work on, their only failing was that the lettering would wear off the keyboards. 

Lightning is a problem in the summer months and I must have changed hundreds or 1488, 1489 and 75154 chips in my time at the company. We also used to maintain a variety of multiplexers and dreaded summer for the many lightning strike victims. When we moved across to the Altos machines we had what were known as TCU’s (terminal cluster units) with 8 terminals per unit, and at some of our customers lightning would take out 5 TCU’s at a time. Oddly enough we never saw this kind of damage on an Ontel.  The standalone terminal was very prone to spikes and often we had to repair its boards down to component level, unfortunately, when these were blown then they were really blown badly and often we had to scrap the boards. Overall the Ontel was a very reliable machine, it seemed as if the peripherals were the most finicky of them all. I do recall that they used a parallel keyboard and it was a mission to repair them due to the sheer size of the unit. There were 4 allen bolts to undo to access the keyboard and as we use the metric system we never had the correct allen keys, and if we did have them they were usually too long and we would have to balance the CPU on the edge of a desk and try release the keyboard from the unit. Fixing printers was a challenge, some of the early dot matrix printers were large and heavy, and like the Anadex, very temperamental. I did find that the most reliable seemed to be the Citoh range, and many of the older Okidata’s ran for many years. As time passed though the tendency was for the printers to become more cheaply made and with more features but none of the reliability that we saw in the earlier models. The early laser printers were expensive and somewhat of a mystery to us and only many years later would I get to fix these units regularly. The printers which by necessity I became a fundi on in later years were the Printronix range.

Printronix P600

Printronix P600 Line Printer

I left Visiondata in 1995 and joined a company that were the agents for Printronix in South Africa. There I took over the component level repair of these complex but reliable machines. There was a lot to learn and in between 1995 and 1997; by then I was doing the board repair and workshop support for the whole of South Africa on Printronix. I never saw or heard about another Ontel after that.

This was written a number of years ago, and is as I remember it.

Images of the Phoenix and Diablo are courtesy of Andre Schaefer of  Nova 3/12

Trade names and brands are copyright to their respective copyright holders.

© DRW 2012-2021. Images recreated 24/03/2016
This entry was posted in Hobbies and Interests, Personal, Retrospective, South Africa and tagged , , , , , , , . Bookmark the permalink.