Prophylactic replacement of electrolytic capacitors?

From: Jerome Fine <jhfine_at_idirect.com>
Date: Mon Jan 14 08:45:30 2002

>Pete Turnbull wrote:

> >On Jan 13, 21:27, Louis Schulman wrote:
> > Old computer power supplies generally have big old electrolytic
> > capacitors. When these go bad, they can
> > cause real problems, and damage other components.
> > The literature indicates that many of these only have a working life of
> > 2000 hrs., or a shelf life of ten years.
> > Obviously, this will be exceeded in old computers.
> I think you may have lost a digit off the working life, Louis, at least if
> you're referring to the sort of electrolytics found in PSUs :-)

Jerome Fine replies:

I was hoping you would answer this question. Thank you! Of all the
hobby users around here, it seems that you have the most experience
with repairs.

Now, that lost digit for the working life, I presume you mean that 20,000
hours is more reasonable? Would that vary if the usage was 24/7 (on all
the time for the whole 20,000 hours) as opposed to being on for 10 hours
continuous once a week on average, i.e. about 50 times a year or 500
hours a year?

The other MUCH BIGGER question for the stuff most of us tend to work
with is shelf life. That was estimated at 10 years. What is a reasonable
estimate? Also, more specifically, I have a number of BA23/BA123 boxes
which are about the only PSU I am concerned about for the long term.
Most are probably more than ten years old already - as far as shelf life
is concerned. Is a PSU (in general an very specifically a BA23/BA123
box) more likely to have a longer shelf life of used every so often - any
if so, how often - once a week, month, year, decade?

I am asking since my goal is to keep running RT-11 for as long as possible.
Three walls loom:
(a) I am 63. If I figure another 30 years before I get senile or my sons will
not help me run my hardware, that will be reasonable.
(a) The life of a BA23/BA123 box and the Qbus boards and disk drives.
(b) The life of the emulators - probably since "simh" is written in C, they will last
although Ersatz-11 written for DOS/W95/W98 on a current PC may have future
problems.

> > So, should these be replaced if they exceed a certain age?
> If it ain't broke, don't fix it. As always, there's an exception to prove
> the rule. If one of a pair goes, it may be worth replacing both to get the
> values to match. Or in some PSUs, if one or more fails, they may all have
> got too hot or dried out, and at least for small ones (relatively
> inexpensive), I tend to replace the lot. In any case, check the others,
> they may be on their way to join it.

This seems like good advice for working life - what about shelf life?

> The working life is an estimate based on mean time between failures (MTBF).
> All this says is that for a large sample, it is statistically likely that
> a certain proprtion will fail in a certain time. Variation can be
> enormous.

Could shelf life of some PSU also be as long as 30 years?

Sincerely yours,

Jerome Fine
Received on Mon Jan 14 2002 - 08:45:30 GMT