power supply benchmarking

[stinger608]

Just wondering if anybody knows of a good power supply benchmarking software???

[Tyreal]

The best way to test a power supply is to monitor it during idle and load usage. There are several ways to monitor it, the most accurate with a digital multimeter. You can also use speedfan or other monitoring software but depending on your hardware these readings can be innaccurate.

To use a multimeter to test your PSU, read my guide:
http://hardwarelogic.com/news/136/AR...006-03-25.html

If you look at Rich's review on the Enermax Galaxy, he states:

When testing a power supply, we focus our attention on the +3.3V, +5V, and +12V rails. The ATX specification allows a 5% variance in either direction, but the closer these rails stay to their voltage rating, the better. Equally important, if not more so, is whether or not these rails remain stable or have a tendency to jump around. For example, we'd much rather have a power supply whose +12V rail remains at a constant 11.90V (well within the 5% variance allowed) over one that measures +12V on the nose but repeatedly fluctuates.

The testing and methodology page will give you a good idea of what you can do to test a power supply: http://hardwarelogic.com/news/137/AR...006-08-14.html

As for any actual software to benchmark a power supply... there isn't as far as I know.

[One4yu2c]

Power supplies are one of the harder peripherals to review in terms of accurate 'benchmarks.' Few sites have the necessary equipment (which is pricey) to truly test a PSU, and even then, the results may or may not be indicative of real world performance. I personally like SilentPCReview's methodology, which can be read here (LINK).

Here on HardwareLogic, Rich and I have been working on revamping future power supply reviews, both in terms of methodology and the equipment being used. While I can't give a timeframe, I can say that you folks can eventually look forward to more extensive reporting on a PSU's performance sometime down the line.

[stinger608]

That would be great one4, but what I am wondering is, is there a program that will give the idel and load details as you are testing a power supply?

[Lead Head]

Originally Posted by stinger608

That would be great one4, but what I am wondering is, is there a program that will give the idel and load details as you are testing a power supply?

There are some DVM that can connect to the PC and record results in real timel.

@ 1 4, that is how JonnyGuru does it to, except with slightly different hardware...very cool stuff

[polobunny]

I like the current testing method and results.
You know, good psu, bad psu, dual x1900's will not make the 12v move... :P

Very interesting review at silentpc, but it is honestly boring. I'm geeky, but not enough to like reading tables and compare at which temperature the power supply operates best. =/

[One4yu2c]

Originally Posted by polobunny

Very interesting review at silentpc, but it is honestly boring. I'm geeky, but not enough to like reading tables and compare at which temperature the power supply operates best. =/

No worries there - review content and methodology are two very different things. We'll always keep our content interesting to read, but we'll definitely be improving on our PSU testing methods to paint a more accurate picture of a PSUs capabilities. It's just simply not enough to say that the rails were in spec and didn't move much and is therefore a good unit.

[polobunny]

Lol I know, it was funny to read though. :P

[gvblake22]

Yeah, as mentioned, you will get the best results by measuring with a digital multimeter. If you want to use software, you could use something like Motherboard Monitor or SpeedFan. However, software readings are notorious for inaccurate readings, whether it be temperatures or voltages, so take it with a grain of salt.

[MantaBase]

Originally Posted by One4yu2c

No worries there - review content and methodology are two very different things. We'll always keep our content interesting to read, but we'll definitely be improving on our PSU testing methods to paint a more accurate picture of a PSUs capabilities. It's just simply not enough to say that the rails were in spec and didn't move much and is therefore a good unit.

I know this thread is a tad old - and frankly, being new here I have my hands full already. But recently I've been polishing up on multimeter construction and use (nothing serious and I am a danger with a soldering gun). It seems to me that the DBS-2100 PSU load tester - used by certain sites and touted as costing up to 10K - isn't all it's cracked up to be. I don't doubt the use of it - but after looking close at it, I think one could be made pretty cheaply - say $500.

I mean, accuracy isn't the issue - you want precision and reproducibility (cetainly spelled wrong - sorry). All you need are some things to produce load and some inline meters right? Or am I missing something? So what if you are off by 5% - as long as its always 5% you can compare. And really, the way I read it, if you stick to digital meters, error should be <1.5% consistantly.

I'll read some more - and keep in mind that I ofetn wear safety gear at work for the sake of others rather than myself (or so I am told - "Manta - the red hat is so we can see you! (from a distance)"

Manta