NCR Tower PSU again

From: Tony Duell <ard_at_p850ug1.demon.co.uk>
Date: Wed Dec 1 19:08:06 2004

>
>
> Hmm, getting somewhere; found a dead triac near the input which was
> supposed to bypass that 10 ohm resistor that was burning out once the
> PSU had started up. Presumably the intention was to allow the PSU to
> start a little more gracefully.

YEs, quite a common soft-start circuit.

>
> Nice bodge story: I just found out why the system wasn't starting up at
> all. It had been semi-dismantled since it had died years ago; I'd
> reassembled it recently when trying to get it running again.
>
> After much wiring tracing I've just discovered although the cable
> between the control panel and the backplane is keyed at both ends, at
> one end the connector was on back to front. That hadn't stopped whoever
> assembled it at the factory from hammering the keyed connector backwards
> into its socket!

Argh!. You want to find the idiot who built that system and connect his
private parts to selected points on the mains side of that PSU :-)

> Now, I'm getting a good 5.1V output, but there's a nasty wheezing noise
> (not a whine!) from somewhere on the main PSU board. The +12V and -12V
> rails are seriously up the spout. With a light load (just a couple of
> fans), the fans spin up for a couple of seconds before everything shuts
> down for a couple more seconds, then the cycle repeats (the wheezing
> sounds like breathing at this point!). With no load the wheezing's
> constant and the PSU doesn't shut down, but -12V rail is at -4V and +12V
> rail is at +5.3V.

It's not at all uncommon for the regulation to be taken from the 5V line,
and the 12V lines either to 'tag along' or to have linear regulator
circuits. I nthe latter case, particularly, the 12V lines will only work
if there's a significant load on the 5V line, otherwise the PSU runs at
such a low duty cycle (to keep the 5V line right with no load) that the
12V lines are very low.

I would therefor try a respecable load on the 5V line and see what happens!

-tony
Received on Wed Dec 01 2004 - 19:08:06 GMT