Tests for a bunch of PSU's

[burebista]

Hmm, here.Only for voltage regulation and efficiency.

[Hutch]

Very, Very interesting results. The GameXstream failed another PSU test. The 5v rail failed the test, very suprising. Both Antec PSU's failed, the Be Quiet BQT failed both the 3.3v and the 5v test.

[burebista]

I don't know what to say. Epsilon/GameXtream have some ripple issues (shown by jonnyGURU) but rails were OK.
Again, on jonnyGURU review Antec Trio just went pop on overcurrent but rails were OK.

[jonnyGURU]

Did I miss something, but.. where were the loads?

I saw the PDF's, but they didn't say what load they put on each rail.

Notice the Enermaxes all came out on top. Not saying the testing is rigged, but based on how the load is spread out across the rails, some power supplies are more efficient than others at different load.

They say "full load," but is that Full Load like the PCP&C tests where none of the rails are 100% (as I feel the 12V should be) and all of them are cranked up?

All the info may be there. I'm sort of in a hurry right now. But for now, I haven't seen any of that.

[Oklahoma Wolf]

The results of the Epsilon and GXS lead me to think they had it crossloaded or overloaded somehow... without more info on how they loaded them it's hard to tell.

I am somewhat amazed though the two Tagans lived through the experience

[jonnyGURU]

Okay. I gave it a good read just now.

It seems to start off with some good noob information about PSU's. The only real "testing" is described on the last page and consists of:

To test the supplies we stress tested them at the manufacturer’s quoted output with a 100 per cent load for an hour (in some cases, due to some very unpleasant smells coming from the PSUs, the tests didn’t last as long). This was to isolate any weakness in the units including rail stability, heat generation, noise, or internal components starting to fail, any one of which might cause failures when installed in a system.

"Manufacturer's quoted output?"

As anyone knows, rails aren't additive. If you have a 600W PSU that can do 500W on the combined +12V rails and 200W on the combined +3.3V and +5V rails, that doesn't mean you load it up to 700W. Did they max out the 3.3V and 5V or did they max out the +12V? If they maxed out either, they could product a crossload situation, so I doubt they did that. So did they just put a certain percentage of a load across all of the rails? That's fine, but not fair. Fair would be putting the same load on all power supplies within the same power output range (like taking all of the 600W PSU's and putting a 40A load on the 12V rails and the remaining 120W on the 3.3V and 5V.)

During testing a number of supplies had failures with the 3.3V and 5V rails, although it’s unlikely that in normal use these rails would get so stressed. The same cannot be said about 12V rail failures. Failures from 12V rails producing voltages under the 12V ATX specifications could cause systems to freeze or crash, while 12V rails producing voltages over the specifications are likely to cause more serious damage to system components.

Ok.. So if they had "failures with the +3.3V and +5V rails" then they probably had the 3.3V and 5V rails maxed out. As they state, this is "unlikely" under "normal use" as a modern PC wouldn't max out the 3.3V and 5V rails and then "top off" the rest of the load with +12V. Quite the opposite, in fact.

To test the PSUs, we were kindly granted access by Enermax UK to a Chroma 6314 power supply tester. This testing unit enables each of the PSU’s power rails (3.3V, 5V, 12V, -12V and 5V Standby) to be individually stress tested. Supplying the AC input was a Chroma 6430 Programmable AC source. To measure the amount of power input to the PSUs a Zentech 2100 Digital Power Meter was used.

So they're using a good testing methodology, but unless the author is the one actually programming the loads and unless the author reports what each load is on each rail for each PSU, then this "review" is worthless.

Another thing I want to point out that burebista stated in his OP is the misconception a lot of reviewers have of "voltage regulation."

Not just in this review, but in other reviews, I keep seeing the voltages from one power supply plotted against the output voltages of another. This is horse crap and it doesn't explain or exhibit voltage regulation at all.

Reviewers... guys.... I'm trying to help you here.

If you have an Enermax Liberty 500W that puts out 12.4V when running Prime95, that doesn't make it better than the Coolmax CTG-850 that putting out 11.9V. That has nothing to do with load regulation.

Load regulation is the PSU's ability to maintain a voltage from idle to load. So you want to compare the delta between idle and load between different power supplies... not just voltage output.

Here's a good example of how stupid that is: The OCZ EvoStream doesn't have a 3.3V rail. Well.. it does, but they call it a "3.4V rail." It's "default" idle voltage is 3.4V. DOES THAT MAKE IT BETTER THAN A PSU THAT'S PUTTING OUT 3.3V ON THE 3.3V RAIL?!?!?!

And none of this means ANYTHING if you have 150mV of ripple tearing the beejesus out of your motherboard and video card either.

I'm no PSU god. I'm not even an electrical engineer. Hell.... OKW and Madmat have more experience and background in electronics than I do, but if common sense can't be applied to a simple power supply review, DON'T DO THE REVIEW!

That might sound a bit arrogant, etc. but when you have the tools at your disposal, i.e. a Chroma ATE, and you half ass the review, you lose all points with me. If Enermax or FSP or whoever is "letting you use their equipment" is going to watch over your shoulder and/or do all of the ATE programming so you can't perform an unbiased and thorough review, shake the man's hand, say "have a nice day" and walk out of the office.

If the only tools at your disposal is taking a PSU and putting in a PC and measuring idle and load voltages, THAT'S FINE! Especially if the rig is a kick ass rig and you have multiple power supplies to throw in the rig. I AM NOT wanting to discourage ANYONE from doing a PSU review. But don't take a load tester and do a half ass review. That's just bad.

I shouldn't post when I'm sick...

Anyhoo... like I said the first few pages do have some good noob information about efficiency, etc.