Buying a used Powr supply

  • Thread starter Thread starter Semi Head
  • Start date Start date
Semi said:
Is there a safe way to test an (ATX) PSU without hooking it up to a
motherboard?

S_H

You can connect PS_ON# to an adjacent COM pin with a paper clip.
If the switch is ON on the back of the supply, when that connection
is made, the PSU fan should spin. All outputs will be operating at
that time.

Is that safe ? I cannot guarantee that it is. I was corrected by one
poster a while back about that fact (cannot guarantee it), and as a
result, I do not try to promote that technique actively. So while
I told you _how_ to do it, I don't know if it is guaranteed a safe
thing to do with all possible supplies on the market.

In my home testing, I have a load circuit with a bunch or resistors
I got at our local "good" electronics store (not Radio Shack). The
load circuit draws about 100W total from all the outputs, and each
output has some load on it. Thus, with that connected to a power supply,
I can connect PS_ON# to COM with perfect safety. I can then also connect
my voltmeter and verify all voltages are present and correct. After
two hours on the test load, and all voltages verified, then the PSU
gets thrown into the new computer.

You can buy a "load dongle" listed on some web sites, but as far
as I know, it only loads one rail. While that does absorb some
energy and presents some load to the PSU, I don't know if it is
guaranteed to be any safer than operating the thing with no load.
That is why I didn't take any chances, and built my own load circuit
with power resistors and a fan to blow over them.

http://www.amazon.com/Antec-Atx-Power-Supply-Tester/dp/B00006B6U0

Another way to create a load, is take a couple of old disk
drives you don't care about, and connect them to the Molex drive
connectors. That will draw a small amount of current from +5V
and +12V, and should be roughly equivalent to the Antec tester
shown above. Then you can connect PS_ON# to COM and do your
voltmeter checks. With no load on 3.3V, it might not meet the
3.3V +/- 5% spec, as that is the limitation of not loading
the output rails. Some unloaded supplies don't put out the
correct voltage when they are not loaded, and don't be
surprised if the voltage is not exactly 3.3V when the 3.3V is
not being loaded.

A few companies list both minimum and maximum current for power
supplies. Some rails on those supplies have a non-zero minimum
current spec. That implies incorrect (but not necessarily
destructive) operation of the supply if no load is present.
But not all companies place that info on the label of the supply,
so that cannot be relied on as a means of vetting the safety of
an unloaded test. If the label on your supply shows 0 amps minimum
current on all output rails, then yes, you could connect PS_ON#
to COM and do the test unloaded. Otherwise, I cannot give you
an iron-clad guarantee, unless you have your test load connected
and are drawing a bit of current.

Paul
 
Back
Top