The only proper way to test the power supply is to try it in another working
computer, or verify it by trying another one. The only other way is to bench
run it. I cannot remember for your supply how to start it without being
connected to the computer. When bench running it, it has to be properly
loaded. An unloaded power supply test is not viable, because they may fail
under load, and read normal with no load.

Most of the time, the failed components are not visible under inspection.
Proper troubleshooting must be done in order to determine the defective
parts.

Considering the low cost of a power supply, it is not a viable thing to
service.

--

Jerry G.
=====

"Al" wrote in message
. ..
Hi. This is on a Gateway Essential, 950 Mhz Celeron, FlexATX, mfg date in
fall of '91. The Power Supply is SFX, Gateway's part# is 6500545, 90W.
(AKA Newton NPS-145PB-117 A Rev 03).

The computer would start but then go only into standby mode (with the
power indicator being always yellow, never green). After a few days of
that behavior, it would then start only for a moment, and only with a few
spins of the ps fan (there is no CPU fan). Pushing the start button again
would have no effect - unless I removed the power cord and let it sit for
a minute or two. Then I'd replug and the pattern would repeat, with only
a momentary start for the first button press.

With an analog voltmeter, I would see 1V for that brief moment, on
POWER_GOOD (pin 8). (I got the exact same result measuring between pins 3
and 9, which should have given 5V.) When I removed all drives and tried
again, I got the same behavior.

So, I took out and opened up the PSU. There was some dust, which I blew
out. (Nothing looked burnt, that I could see.) But thereafter, the
computer behaved differently: when I merely plugged the power cord in,
I'd get the momentary light and ps fan spinning. It didn't need me to
actually push the power button anymore. Next, to see if the power on
switch might be shorted, I removed its leads from the MB - but that
didn't change anything.

It seems to me that maybe the CPU is not getting a good signal on
POWER_GOOD, and so shuts down the PSU. Or could it be that the CPU, the
MB or some other thing is bad, not the power supply? The more I research
on this, the more it goes round and round.

I'd rather not waste $40-50 on trying a new PSU, not to mention the week
or more it would take to be shipped to me, if it's not likely to be a PSU
failure. How can I tell if the PSU is bad or not? Thanks.

Oh, and one more thing: weird behavior showed up in the last few weeks,
like the mouse and/or keyboard or even drives not showing up on POST.
This msg would also sometimes be given by Windows: "An unknown device has
exceeded the current limits of its hub port". And a USB attachment for an
MP3 player had been added after Christmas. But the "current exceeded" msg
still sometimes appeared, even after the MP3 attachment had been removed.