Bob Blick wrote:
> On 25 Feb 2002 at 19:14, Robert Rolf wrote:
>
> > The 'pro' supplies used in high end boxes have serious 'crowbar' SCRs on
> > their outputs to prevent this sort of thing from happening. When you have $10k
> > of high performance server/hard drives, you don't worry about $10 of fail-safe
> > parts.
>
> I've never seen a supply with multiple crowbars. Usually(like this

The Compaq's I once attempted to service did have multiple crowbars,
as did the DEC PDP 11/xx supplies. Had lots of PDP supplies try to
go AWOL, but the crowbars did their job.

> one) there is a crowbar on the 5v supply, plus a 339 quad
> comparator sensing all supplies. The 5 volt line never got too high, it

If designed correctly, the 339 would have triggered the 5V crowbar and
killed the supply with high loading. Of course a few milliseconds of 50V
isn't going to sit well with 12V parts.

> was the other supplies that did all the damage. That's why the video
> card, ethernet card, and floppy drive survived. Everything else was
> toast.
>
> Computer was plugged into a UPS, it wasn't a surge, the power
> supply just picked a good time to go full duty cycle. Or perhaps the
> crowbar somehow triggered, and it forced everything else up as it
> tried to restore +5 to the proper voltage?

Yes, a lot of supplies use the 5V as a reference for everything else.
Pull it low, but not dead, and it's quite possible the other voltages
would go high enough to pop something. The original IBM PC AT supply had
a low ohm, high watt resistor to load the 12V supply if you didn't
have a hard drive. With the ballast resistor disconnected you got 12V,
and 3V on the 5V line. Brilliant design, NOT!

What does a high power zener cost these days?
Why can't the manufactures charge an extra $5.00 to have a bit
of 'robustness' in a design?

> Redarding expensive power supplies, I'm not convinced they are
> any more reliable. When I bought a system with an Athlon, I

Brand name? With a long warrantee that they'd have to 'make good'
on if their supply failed? Tell us what we should avoid in future.

> followed the recommendation and bought the best power supply I
> could, and went so far as to buy ones on the list at AMD. Three
> power supplies from two different brands all failed for various
> reasons within a few months. I now have a piece of crap power
> supply in it and it's been working three years. The expensive power

Check that the fans are still spinning. Every cheap supply we've had
has lost it's fans after 3-5 years. Of course our machines are on
7x24 since reboots take too darned long.

> supplies were heavy, had lots of nice parts in them, but they were
> less reliable.

That is unfortunate. All it takes is ONE weak component if the design
has intrinsic flaws.

Robert

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu