On 20 Jun 2001, David Dyer-Bennet wrote:

> Phil Mendelsohn <mend0070 at umn.edu> writes:
> 
> > Look, it's just a switching mode power supply.  There's no magic about
> > attaching a motherboard.  The reason Ben's trick below works is because
> > sometimes they cross couple the current limiting / protection circuitry. 
> > Some chips that require both +5 and +/-12 will die if one rail dies, so the
> > conservative P.S. designer will kill the whole supply rather than let one
> > rail stay up and kill chips.
> 
> I thought there was a "power good" signal fed back from the
> motherboard, which the power supply looked for very shortly after
> startup, and shut down if it wasn't present?  Mine certainly behave
> that way!

We're saying the same thing.  You're talking about remote sensing.  If one
of the rails dies, the remote sense falls below threshhold and the supply
shuts down.  But it's a logical OR -- any one of the rails will cause the
same behavior.  That's a kind of protection -- don't let things run on one
rail if the others fail, 'cause some IC's will smoke if you do.
 
Probably the same thing happens if one of the rails shorts to ground too.  
Rather than just go into current limiting on one rail, it shuts the whole
supply down to keep from unevenly stressing components.  If you want to
test that one, use an alligator clip to make a dead short between ground
and one of the rails.  If the supply does fault protection that way,
you'll remove the jumper and everything will come back into line.  

If it doesn't survive, then it must have *been* a witch! <g>

-- 
"To misattribute a quote is unforgivable." --Anonymous