Another very common one is that the startup resistor of a switch-mode PSU
goes open-circuit while the machine is left on The PSU keeps working as
long as the mains is there, but if it gets turned off, or the mains
fails, of course it won't restart.
Some terminal servers and the like were very prone to this problem.
It's a very easy fix once you know which resistor to check.
I've had a slightly related fault in the switch-mode power supply (wall
wart) of a Decserver 90 terminal server. A small electrolytic smoothing
the supply from the startup resistor had lost capacitance. Lisening closely
to the power supply, I could hear a high pitched whine which made me think
there was a good chance this was normal vibration of the chopper transformer
and threw me right off finding the real problem. When the capacitor was
replaced, the whine went away and the power supply worked correctly, for a
few weeks before it failed again. I haven't had the heart to investigate
further.
Regarding Dec equipment, recent experiences are making me speculate that
leaving some items turned off for a few months might be bad for them!
- An Alpha 3000/300 in continuous use for several years was retired. When
I next turned it on several months later, it gave a loud bang and emitted
smoke from the PSU. This turned out to be from a VDR which had exploded due
to a triac shorting. (This wasn't easy to track down as the VDR had been
sleeved with rubber which completely hid the problem.) The function of
the triac was to automatically switch between 115V and 230V input. Its
shorting left the PSU permanently in 115V mode and the power around here
is 230V. Replacing the VDR (after a good bit of fun tracking down a suitable
replacement) and snipping out the triac (which I couldn't find a replacement
for) solved the problem at the expense of being able to operate on a 115V
supply.
- A HSZ80 disk controller, like the 3000/300, also failed with a loud bang,
without smoke though, making it difficult to figure out which of the many
items plugged into the extension lead had blown the house fuse this time.
An unmarked item across the mains which looks like a VDR or disc capacitor
had shorted. I can't replace it as I have no idea what its spec is.
- A Vaxstation 3100, also in continuous use for several years was similarly
retired. In service, it had suffered very occasional memory failures but
nothing persistent. When I switched it on a few months later, several
problems were apparant. Firstly, one of the fans in the PSU ran for a short
while and seized. Secondly, the memory power-on test failed to complete. It
didn't report an error but the machine reset in the middle of the test and
went back to the beginning of the power-on tests in an infinte loop. It's
not supposed to do that! Removing both 4MB memory expansion boards got past
that problem but resulted in a horrible error from the SCSI driver on
attempting to boot, presumably because there was not enough memory on the
motherboard. Replacing one memory board (either one!) got by this problem
but it then became apparant that the disk was failing with lots of clunking
and clicking and various errors reported. Finally, the "compose" light on
the keyboard remains lit all the time which as far as I recall may be
intended to indicate that all is not well in some part of the system in
the event that nothing appears on the graphics console.
- A Microvax 3100 model 80 retired for some years was quite happy to
perform correctly for a few hours when switched back on. However, after
later shutting down and restarting, it died in the middle of the memory test
and produced no console output since. It's been a while since I looked
at it and I can't remember what indication the diagnostic leds gave. I'm
sure if they gave any useful lead I would have followed it up though.
Sometime I feel like my kit is failing quicker than I can get around to
fixing it.
Regards,
Peter.