[LAD] [Somewhat OT] Strange failure mode of a PC

[Jörn Nettingsmeier]

On 07/17/2010 10:46 PM, fons at kokkinizita.net wrote:

> What I don't understand is how the contacts got so dirty.
> If a resistance of a few kOhm is enough to make it look
> as a closed contact then it can't be handling large currents,
> so there should not be any arcing. And the construction of
> the thing is such that it is virtually closed, no dust or
> whatever could ever creep in.

it can't be totally sealed, as that would require an expensive gasket. 
my guess is that the way servers are designed, air is drawn in through 
the front (the "cold aisle" in datacenter lore), and expelled through 
the rear (the "hot aisle"). after some years of use, it's not so 
unlikely for gunk to accumulate pretty much anywhere in the path of the 
airflow.
and the conductivity of pure industry-strength gunk should never be 
underestimated.

btw, if the bios does not allow the boards to be configured as "always 
on after power loss" (common on older hardware), a capacitor in place of 
the momentary power switch works nicely, provided its charging current 
has dropped to zero before the bios thinks it's a 
more-than-four-seconds-turn-me-off event.