On May 27, 21:37, John Honniball wrote:
> vassilip_at_dsl.cis.upenn.edu wrote:
> > So when the BBC + Torch arrived, the genious who was responsible
for
> > assembling the kit, promptly added a mains plug to the power cable
> > and plugged it in, feeding 240V AC to the +5 and +12V rails.
>
> Same thing happened at my Uni when one of the Prime operators took
> home a BBC for the holidays. She had an external disk drive for it,
> that plugged into the BBC's power underneath the machine (not at
> all obvious). So, she went ahead and fitted a mains plug to the
> three-core cable that fed the drive. Usual release of magic smoke...
Ah, yes, that would be one of those drives suplied by a company who
figured that you don't really need two 0V lines, and 3-core mains flex
is cheap.
I can add another to the "240-into-5 doesn't go" list. I ran the
repair and engineering department for a large Acorn dealer/distributor.
One Saturday, a customer with his son bought a BBC B, Torch Z80 with
disk pack, and a high-res Microvitec monitor; well over UKP1000 in
those days. He was back on the Monday, looking very sad. His son had
connected everything together before finding that the "mains plug"
didn't fit. When he changed it, the fuse blew, so he replaced it and
tried again :-(
He was very nice about it, quite philosophical, and so we tried quite
hard to rescue what we could, and find second-hand parts. We salvaged
one of the two drives. We repaired the monitor. We found a spare BBC
B board -- the original was deemed unrepairable. The disk pack PSU was
a write-off, but we got the original Beeb PSU back from the shop. The
Z80 board didn't look healthy, and he let us keep that too.
Eventually, we stripped every IC and some other damaged parts off the
board, fitted sockets everywhere, repaired several tracks with quite a
lot of wire-wrap wire and epoxy, and it became a test bed. The Z80 got
similar treatment, except it didn't need track repairs, and we turned
it over to a teenager who was with us for work experience. We thought
it would be good (de)soldering practice. He was delighted when he
ended up with a clean board, and even more so when the fully-socketed
version worked at the the end of it, so I suppose some good came of it
all -- he'd apparently been something of a problem to the college
tutors before he discovered there was something he could actually
achieve.
Those Torch Z80's were a royal pain. After that incident, we told the
shops not to take out or replace the BBC supply under any
circumstances, and eventually Torch stopped the practice of replacing
the original PSU. The Beeb SMPSU was perfectly capable of running a
fully-expanded machine with a Torch processor, so using Torch's rather
inferior (but more powerful) unit was not a good idea.
The Torch Z80 board is about Eurocard size, fitted with a short 40-way
ribbon cable, and was supposed to be mounted in the lid of the BBC B,
using four PCB mounts of the type that have self-adhesive pads. They
used to fall off after a while. Now you might think that's just
irritating (it's certainly awkward to fit correctly) but it's more than
that. When it falls, it tends to short things out. Unfortunately, the
regulation on Torch PSUs is very poor when they crowbar, and when the
5V rail comes back after it's shut down, it tends to overshoot -- by
about 4 volts, according to our scope. TTL doesn't like 9V. I've seen
at least two Beebs destroyed that way.
One of them looked fine, but didn't work -- something of a puzzle, as
the owner had replaced the original Acorn PSU and we didn't know he had
been using the Torch one. It just went "tick-tick-tick", as they do
when they're shorted, or as in this case, detect over-current. So we
put it on a bench supply, thinking either to blow off a faulty
capacitor or maybe find a hot chip. No, but it drew 9 amps (a Beeb
usually takes about two) and after a little while it was obvious
several LS TTL chips were cooking. So we removed them. Back on the
PSU, still several amps over normal. A few more hot chips after a
short while. And so it went on, until all that was left was the NMOS
(all of which still worked). Another expensive repair.
--
Pete Peter Turnbull
Network Manager
University of York
Received on Tue May 27 2003 - 19:18:00 BST