Hardware Secrets Power Supply Test Methodology

[Hardware Secrets Team]

There has been a new article posted.

Title: Hardware Secrets Power Supply Test Methodology
URL: http://www.hardwaresecrets.com/article/522

Here is a snippet:
"We were one of the first hardware review websites to alert users that the vast majority of power supply reviews posted around the web and even on so-called "specialized" magazines were completely wron..."

Comments on this article are welcome.

Best regards,
Hardware Secrets Team
http://www.hardwaresecrets.com

[ProphetCHRIS]

(they haven’t published power supply reviews in the past 2 years)

... which is a very good thing because THG uses the worst approach in history of testing methodologies. They were so wrong that they f&%ked up half of the power supplies in their last stresstest because it was just wrong how they tested.

[Kanchenjunga]

If you know about any other website that uses a load tester to test power supplies please let us know so we can add it on the list above.

pcper.com and extremeoverclocking.com are two more that use a static load tester...also technic3d.com (German)

Thanks for the quality reviews...I really like your style overall but I would like to see the voltage and ripple & noise measurements rather than simply pass or fail. Am looking forward to seeing some of the units reviewed here in the past get put to the test.

[Gabriel Torres]

Hello,

Thank you both for the comments. I added Extreme Overclocking's methodology to the list. Unfortunately PC Perspective doesn't give enough details about their methodology besides saying that they use a 2,000 W load tester.

As for ripple and noise measurements I will try to add more info on that. If you read the OCZ ProXstream 1,000 W review you will see that I added more information about noise than on the other two reviews posted before (PC Power & Cooling Silencer 610 and Antec NeoPower 550). The problem is the amount of generated data. We are talking about six outputs with five tests each, i.e. 30 screenshots... Anyway, for the next review I will try to add at least the most important screenshots, which are for the +5 V and +12 V outputs under full load, because here is where we should something "wrong", if you will.

Also, a guy at www.jonnyguru.com pointed out the fire hazard I have here at my lab for using a cardboard box as my "heat box". I am thinking about other materials and I will probably build something on wood, maybe even adding some kind of fire extinguisher to it, like race cars do (i.e. some sort of system that would activate the fire extinguisher located on the top of the power supply by pulling a wire or pushing a button). Comments on this issue are more than welcome.

Once again, thanks about the comments.

Cheers,
Gabriel.

[Kanchenjunga]

I think ripple & noise at full load would tell us what we want to know...and the high and low voltage measurements. As it is now a theoretical unit with a high of 12.6V and a low of 11.4V still gets a "pass" even though that would be really shoddy load regulation.

[Gabriel Torres]

I just want to avoid posting tons of tables...

But I could change the methodology to narrow regulation specs from 5% to something lower like 2% or 3%. So I'd say "pass" only when the power supply was really close to the labeled voltages. This is in fact a good idea and I will look into it.

Gabriel.

[Kanchenjunga]

yeah I hear you about not wanting to post a lot of tables...and I like that about your reviews. Maybe something like "Excellent, Good, Fair, Poor" would be useful to the reader without going overboard with data...or a 1-5 Stars rating system might correlate well with the ±5% tolerance.

[Spectre]

Originally Posted by Gabriel Torres

Hello,

Thank you both for the comments. I added Extreme Overclocking's methodology to the list. Unfortunately PC Perspective doesn't give enough details about their methodology besides saying that they use a 2,000 W load tester.

Lee at PcPer built his load tester from progammable DC loads (CSI3710A and CSI3711A) for each rail (you can find them here (http://www.circuitspecialists.com/prod.itml/icOid/7169 and http://www.web-tronics.com/prello030c.html) to load in the increments he shows on each review.

Also, a guy at www.jonnyguru.com pointed out the fire hazard I have here at my lab for using a cardboard box as my "heat box". I am thinking about other materials and I will probably build something on wood, maybe even adding some kind of fire extinguisher to it, like race cars do (i.e. some sort of system that would activate the fire extinguisher located on the top of the power supply by pulling a wire or pushing a button). Comments on this issue are more than welcome.

Once again, thanks about the comments.

Cheers,
Gabriel.

That would have been me since I have had a couple of small fires and a number of units arch and spark when dieing. Like I said in the email I sent I would use something metal like a PC case and insulate that if you want or use the pump and dump method that Jon/Oklahoma/MadMat use.

[Gabriel Torres]

Thanks for the input, Paul, it was really appreciated.

Also, congratulations for your terrific job at HardOCP. You guys have, in my opinion, the best rig for testing power supplies among all websites.

And also thanks for having the time to visit us and post your comments personally.

Cheers,
Gabriel.

[Gabriel Torres]

Just to let you guys know that I bought an ABC fire extinguisher and also re-programmed the load tester to use a 3% margin instead of a 5% for the voltage stability test. I've just made minor changes to the text to reflect these two additions.

Cheers
Gabriel.