Testing PSU's real output.

[Tolya]

Is there a way to do it using software?

Also, I have an ohmmeter but since PSU's output can be load specific (or at least thats what I always thought) then what do I need to do to see if my PSU delivers what the label says?

[Fingolfin_Noldor]

You need a Amp meter, not an ohmmeter. Ohm meters measure resistance and not current.

You will also need a voltmeter. In short, get a multimeter, with a sufficiently good data recording utility. They don't come cheap though, even for those with no data recording utility.

[Starglider]

Is there a way to do it using software?

There are numerous diagnostics programs available that will show you whatever the motherboard is recording - this will normally be accessible via a page in the BIOS menu as well. However this is usually the post-VRM voltage for the CPU, memory etc. I've seen enthusiast class systems with integrated monitoring of the actual PSU rails, but most motherboards don't seem to have it.

Also, I have an ohmmeter but since PSU's output can be load specific (or at least thats what I always thought) then what do I need to do to see if my PSU delivers what the label says?

Ideally a variable load (either a beefy rheostat in an oil bath or a fancy digital programmable one) for each rail you want to test simultaneously, some plugs and some cable splices. Those aren't cheap, but usually the rails you want to load are the 12V ones, and if you have any low-voltage halogen lights handy you can use those as an improvised load. Worked for me at least, but this isn't something you should be messing about with if you don't know what you're doing, it's easy to blow the PSU.

[Sea Skimmer]

You really need an automotive battery load tester hooked up to each power rail if you want to test the true current output, anything smaller probably won’t be safe to use. Some 100 dollar plus multimeters could handle the 20-30 amps each 12 volt rail may be able to supply, but those are very expensive and typically only rated to handle such a large current for a matter of seconds (30 or so) with the continuous load rating being much lower since the thing has to convert all that energy into heat. Furthermore, you must test all rails at once for the test to be valid. So only one multimeter wouldn’t work. Most multimeters cannot handle more then about 1 amp or even only fractions of an amp in continuous load in any case.

Also keep in mind if you fuck this up you can destroy the power supply, and potentially kill yourself by electrocution.

Basically if the load tester does not have a fan on it, it will not be able to handle this job. It’s not for nothing that computers typically have a minimal of one fan on the CPU and one fan on the power supply.

I cannot see any real reason to run such a test though. If you suspect a PSU problem then hook up a different PSU and see if that solves it or not. For normal operations the PSU should just simply work if the computer was designed probably. You should not need every last watt the thing can output, if you do then its going to run hotter then ever and decrease its own lifespan and that of other nearby hardware. If this is just because you’d like to know, then just find reviews online of the product. Odds are someone with the proper equipment did run a load test and put the results online if it’s a reputable power supply.

[Tolya]

You need a Amp meter, not an ohmmeter. Ohm meters measure resistance and not current.

You will also need a voltmeter. In short, get a multimeter, with a sufficiently good data recording utility. They don't come cheap though, even for those with no data recording utility.

Right. I meant I have a multimeter

[phongn]

Most of the reputable PSU review sites use expensive automated test equipment like the SunMoon SM-268. Sea Skimmer's suggestion is also spot-on.

[Tolya]

Thanks for the tips. It all boils down to one thing: it is cheaper just to get a new PSU than invest in all this equipment.