28 May
2010
28 May
'10
12:49 p.m.
Have you read
the report in Computer Resurextion? (This is the _only_
indformation I have seen on this project). There were a couple of
things
that really upset me...
=20
I had previously heard this story directly from the gentlemen involved, pre=
tty much verbatim. What do you find upsetting? =20
from what I recall it said something like (this is a
paraphrase, but I
hope I've kept the meaning)
'After basic electircal safety tests we applied power and the machine
appeared to start up correctly'
'However one part proved unreliable, we replaced all the PCBs, but it was
still unreliavly. In the end we discovered the power supply was faulty,
so we replaced that'
Now, as to my problems with that:
The first thing is to consider why htis machine is 'special' After all, a
PDP11/73 while a nice compouter, is hardly rare. This installation may
well ahve had special data input or display hardware, of course, but I
would cuggest the main reason that it's significant is what it was used
for -- its provenance, if you like. With that in mind :
They don't say what the 'electrical safety tests' were, but I would guess
it means using something like a PAT tester (an instruemtn, which IMHO
should be banned, but that's another rant). That is nowhere near enough
before powering up a significant machine. If it were mine, I would _at
least_ completely dismantle everything that could be dismantled without
damage (i.e. not open up HDAs...), then do a complete visual inspection,
looking for any poissible futrue problems. Then, if there is custom
hardware, consider making backups of any programmed parts (EPROMs, etc)
-- if they are socketed. Note down the configuration. Reassemble the PSU
and check it on a dummy load (This would have found the intermittant
fault, I suspect). Check and clean (and maybe lubricate) all cooling fans.
Now we come to the 'repair' : My dislike of board-swpaping in legendary
here, but I find it particularly bad in this case for a couple of
reasons. With an internmittant fault, the only way you can know you have
cured it is to know what cuased it (and then to correct that). Swapping
boards won't help. An intermittant fault could be caused by a bad
connection, whcih is distured and then works well enough so that you
think you've fixed it, only for it to recur some time later. And as I
said earlier, this machine is significant for its provenance. In which
case (in fact for any museum machine), the smallest possible part should
be replaced -- the machine should be kept as original as possible.
Which brings me on to the faulty PSU. Firstly, why wasn't this checked
before the machine was even assembled and brought into service?
Secondly, even if you take tht risk, the first thing that IMHO you do
with a malfunctioning machine is check the power supply rails with a
voltmeter _and 'scope_ at the backplane. And then why replace the PSU.
Why not repair it? Again, keep the machine as origianl as possible.
-tony