Arfa Daily wrote:
As far as your contention that most of the caps that fail are running
at under 60 deg and are not in hotspots, that may be true if you are
talking just mobos, which are a bit of a special case in that apart
from all the problems that have been caused in the past with boards
built using caps with fake electrolyte, the caps that fail are all
decouplers on constant DC rails, and are rated voltage wise pretty
close to the continuous voltages that are applied to them. Even
accepting that, many of the decoupling caps that do fail on mobos, are
sited very close (by nature of the job that they have to do) to LSIs
which *do* run very hot.

The caps that I am talking more about in general, are on switch mode
power supplies, where they are subjected to huge stresses from the
high frequency pulse currents that they have to endure, and the self
heating caused by this in all but the most expensive types, very
specifically specced for use in these positions. Further, the ones
that fail most regularly are, without doubt, the ones positioned close
to heatsinks. I replace hundreds every year in the course of my daily
work. Although the caps positioned on switchers are by far the most
common ones to fail, they are by no means the only ones. I also
replace many in other circuits, for instance audio output stages,
where failed ones are almost invariably close to heatsinks.

Arfa

I'm in complete agreement with Arfa. Internal heating caused by the
ever increasing switch mode power supply frequencies is the most common
cause of failure. High external temperatures don't help one bit since
it reduces the components ability to get rid of internally generated
heat.

A place where I'm finding more and more capacitor failures is in the
memory supply voltage regulator circuits, sometimes causing CPU failure
in addition to memory damage.

--
Best Regards:
Baron.