Big Iron (was VLSI of classic architectures)
roger.holmes at microspot.co.uk
Thu Jan 17 13:51:27 CST 2008
>> Current custodian of the University of London machine, having
>> bought the
>> 200,000 pound machine for 150 pounds in 1976.
> Still a 200,000 pound machine, regardless of the price.
> That was big iron back then. :) How hard is to keep and get spares
> for that system?
I bought a machine which I stripped for spares and also the previous
owners had got most of the parts from the factory when they had
cleared their parts store.
Some SES light bulbs are still available from classic car parts
stores, and wedge lamps were used in old telephone exchanges which
sometimes come up on eBay, I even got some 17volt ones a while back,
and some are still available new from RS components and they even
make LED equivalents at a price, though they're the wrong shade of
The rarer transistors sometimes come up on eBay too. The commonest
(Mullard GET871/GET872/A/B/C) are difficult to get, though I passed
on a half dozen on eBay where the blue plastic had turned brown -
looked like they had blown up or got hot when de-soldered.
Fortunately I have enough for the foreseeable future.
Diodes (mainly OA5 and OA10) did have sky high prices and now are
unavailable but I've got thousands.
Of course resistors, capacitors and inductors are still available,
though they do not look the same, and I still have to find out why
the resistors have double green bands where there should be a gold
band for the tolerance. I have been lucky so far with the
electrolytics, not one has blown up, despite being 46 years old. They
bought the best components available back then, and hand built them
and tested every soldered joint and marked it with lacquer to show it
had been tested. The biggest electrolytics are bigger than a hand
grenade, so I hope they never fail!
There two types of valves in the tape drives. I got about a hundred
of one type with the machine, and not used one, the other type I've
picked up in the States on eBay, they are used in old CNC spark
I am rapidly getting though my stock of 'rubber' pinch rollers for
the mag tape drives, which go hard and crack up. The same thing
happens with the rubber-like roller in the punch card reader, but
that is about 4 inch diameter, so it is currently wrapped in four
layers of motorcycle inner tube, though its not very effective and I
may have to mould a new one in synthetic rubber.
Most of the mechanical parts are simply unavailable, but I have a
lathe and a milling machine.
I bought some flat belts for driving the tape capstan rollers, they
are commonly used as knitting machine drive belts apparently. V belts
for the drums and line printer are fortunately standard car parts.
Line printer ribbons will be a problem (when I get the printer
working again). I have a box of a dozen or more but if they have now
dried up by now I will be amazed. There used to be re-inking firms
around, or maybe I could just buy the ink and get my hands dirty.
Should not be unsurmountable, even if the fabric has failed, its just
(nylon?) cloth, should be able to get something to do the job.
I've a good stock of blank punched cards and paper tape (5 and 8
track) and enough mag tapes to keep me going for a long time. Fan
fold paper is probably still available new and when I see any cheap
locally on eBay I buy it up.
The logic indicator tubes (DM170 or something like that) never seem
to fail. The microswitches (particularly the start switch which takes
a hammering when single stepping a program) are in short supply,
though I don't see too much of a problem in fitting modern sub-micro
versions and extending the actuating arms. Multi-segment rotary
switches just need a squirt of switch cleaner/lubricant from time to
One problem, and it sounds stupid, is the rotary switch knobs. They
are made of plastic and they fit on a 1/4 inch shaft with a flat
section, nothing unusual there. There is a boss on the back which
fits into a recess in the formica covered blockboard (yes the
computer has wood in it) front panel. These bosses break off. The
shafts are not long enough for most modern knobs, and anyway I want
to keep it looking original. Because of the recess, it is not
possible to use knobs with a screw. I am thinking about making some
aluminium knobs which I can glue or screw into the original front
parts of the knobs. The problem is making the flat part which fits
the flat of the shaft. I suppose I could make them of brass and then
put sufficient soft solder in the hole so that I can shove it onto
the shaft so that it won't come off easily. I guess there must be a
tool for making D-shaped holes somewhere, or maybe there should be
some spring steel in there somehow.
The relay contacts need cleaning sometimes, though they are platinum
so are pretty reliable.
There are no edge connectors in the computer itself, every one of the
4000 PCBs is wire wrapped in, which means there are about 50,000 test
points. The racks in the computer are also wire wrapped together, no
plugs and sockets. This all improves reliability and I think accounts
for the machine's survival (well that and my eccentricity). The tape
drives have some plugs and sockets and edge connectors, but have so
far been reasonably reliable as well.
Castors were a nightmare. The tyres on the ones on the (800 lb) tape
drives crack up. The ones on the computer have a flat top with a stud
poking out. The stud has a washer and a nut and fit into an open
ended slot with another nut on top. Any jolt and the stud bends. If I
have to move anything heavy now, I remove the washer and first nut
and make up a three inch piece of wood the equivalent thickness to
take their place, which spreads the load over a larger area instead
of concentrating it on the stud.
Fuses: small ones still available from Maplin for peanuts, large ones
are NATO standard and cost silly money, so I drill the ends, empty
out the sand and solder 10,15 or 20 amp domestic fuse wire into the
Meters, so far very reliable apart from one where the needle would
not go above a certain point, so I had to dismantle it and glue the
glass back in which was what the needle was jamming on.
I think that just about covers all the parts which go wrong.
More information about the cctalk