On Sat, Mar 22, 2014 at 2:45 PM, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
If you have a fualt, like nist spikes on a power
line, that could be due
to a faulty capcitor, then of course you _check_ the capacitors. And
replace (or reform) them if necessary. Capacitors can fail. But changing
parts for no logical reason will get you nowhere fast.
If old electrolytics fail for want of reformation, you'll know about it.
Absolutely.
The failure mode is punch-through; a portion of the
capacitor becomes a
dead short, though usually only briefly, and it can be spectacular. (At
Yes. Contrary to popular belief, the electrolyte is not the dielectric
(insulator) in the capacitor. It can't be, an electrolyte is a conductor.
In fact the electrolyte is the -ve 'plate', the dielectric is the oxide
film on the positive electrode. So when that oxide re-disolves, the
capacitor goes dead short.
If you are lucky, it'll blow a fuse. If you are less luccky, it will blow
the rctifier feeding it and then blow the fuse. If you are even more
unlucky, the transformer will protect the fuse by failiung furst. And if
you are very unlucky, nothign will fail. But the capcitor will pass an
excessive current, the electrolyte will boil amd it will explode. Many
large capcitors ahve a safety vent which is supposed to prevent this, but...
least that's what I've observed for large ones
with 16V and 25V ratings;
smaller, lower-voltage ones might not be obvious.)
With the PDP-1, we had the expectation that some capacitors would need
reformation. If we hadn't checked them, and reformed the ones that needed
it, some of them would have blown up and needed modern replacement. The
purpose of our checking and reformation was to avoid replacing any more
than absolutely necessary, due to the interest in preserving as much of the
historical artifact as possible.
Sure. Nobody, least of all me, is suggestign that chcecking things before
applying power is a bad idea. In fact I wil lgo the otehr way and say it
is very foolish to apply power to a classic computer before doign a
number of 'cold tests', which incldue checking for shorts on every supply
line (regualted and unregualted), which should pick up shorted smoothing
capcitors, 'meggering' the transformer, and so on. You may also want to
test the capcitors on a current-limited bench supply of a suitable
voltage.
No, what I am commenting on is the idea that when a machien fails, it is
a necessary to re-cap it. No matter what the symptoms are, and without
doing any tests to find if the fault could possibly be a failed capacitor.
-tony