A power supply with no load OR with all outputs shorted
together causes no power supply failure, assuming the power
supply was properly designed. Furthermore, if someone says a
bench test with no load will ruin the supply, then that person
did not even learn power supply fundamentals.

You want to discover if a power supply is defective. That
means either tested in or outside the computer, you must still
obtain numbers. The multimeter is still required.
Furthermore, a power supply that appears OK on the bench can
still be defective. Power supply must be fully under load to
be tested.

No load is also why those power supply testers are myopic.
They cannot test when a power supply is most likely to
demonstrate its failure.

Furthermore, not all power supplies require a load to be
tested. This 'load' requirement is unique to each power
supply design. How much load is required is also a function
unique to each supply.

Keep it simple. Even on the bench, a test is insufficient
without a meter. With the meter, a better test is conducted
with supply still inside the computer - under load.

You saw the power supply do something particular. But you
did not report back what you saw. Appreciate that you are
starving me of details that could solve your problem quickly.
Appreciate that without removing anything, we can pretty much
isolate the problem down to one of three components in a power
supply 'system'. So which voltages rise on power up? Which
voltages do not? And what are numbers for other voltages
that, for example, must always be there?

The secret to solving simple power supply problems so quickly
- first get the facts - the numbers. Others make it vastly
more complex by shotgunning (putting a power supply on a
bench). Procedure to identify the failed component of that
power supply system takes only minutes. But it means working
smarter - not harder.

Reason to pull a power supply: after a few minutes of meter
readings, then it is obvious the power supply has failed.

Al wrote:
Tim, I'd read warnings that running the PSU with no load could ruin it.
Do you mean I should attach some load, or that I should turn the PSU on
for only a short duration? Or were the warnings overly cautious?