Soon At Tom's Hardware: Full-Scale Power Supply Testing

[ctmk]

[citation][nom]Gamer-girl[/nom]Is it possible to include sound levels in decibels?[/citation]

i am hoping for this also. and also tests designed for 24/7 idle load. (like file server with low CPU usage but busy network.)

i ahve couple of silverstone PSU (400 to 750W) they makes noise often and i found out it is not the fan noise. it is coming from the electronics.

 

[ctmk]

Also try answer a question, how long it takes to change for a new PSU? 1yr? 2yr? 3Yr? and so on

 

[erendofe]

I think this is an excellent idea (as on who trougle shoots many mysterious power supply deaths) One thing; is it also possible to test AC line noise rejection/ripple (input to output)as well as noise levels ????

 

[Guest]

Most important stat is voltage regulation. It takes better construction and design to hold voltages within a 1% range under all loads. And dock points for units that artifically inflate the zero load voltages up to the max allowed under ATX specs, such as starting at 12.5V for the +12V rail.

Second is ripple/noise suppression. Lower ripple/noise suppression figures are always preferrable to looser figures, but if it's within specs, the unit with tighter voltage regulation will get my vote.

Not much difference in performance between units with 10mV ripple differences (such as 35mv vs. 45mv, both well less than half ATX spec and won't show any real world performance differences), but a lot of difference when one unit has 3% regulation.

Third is efficiency. There are a slew of power supplies out there garnering Gold certification for efficiency but have horrible electrical performance.

Fourth is fan noise. Honestly, if fan noise is such a concern, buy a fanless unit like the Silverstone Zeus line or a Seasonic X-series (fan doesn't operate until half power draw placed on it.) Outside of PC Power and Cooling's leaf blower Turbo Cools and Silencers, fan noise isn't much of an issue for performance computers. Granted, for a completely silent computer, it may be, but the vast majority it's a metric that's almost irrelevant with the large, slow turning fans used on power supplies these days.

 

[Proximon]

Where exactly is this testing equipment located and who exactly is operating the equipment?

Why does the test list not include voltage regulation and ripple?

What will the ambient temperature be, and will it be kept the same through all tests?

 

[cutterjohn]

Nice to finally see another PSU review, as when I need them they're usually about older products no longer available and I cannot trust companies to be consistent in their quality.

This means that I usually pick out generic supplies from one of the common OEM suppliers to the pricey models, and have had good luck with those date excepting super cheap POS PSUs which sometimes come with genero cases. Speaking of which I've only ever owned one "big" brand PSU, an Antec PSU that came with the case and I only got that one as the cases available locally(needed it right away) were all about the same price and the Antec case had nice accessibility features for the time. (I'm still using the case, but the PSU is long gone, was only 350W, so I retired it to a server running older hw whose el cheapo, i.e. trash, PSU was in the process of croaking.)

I'm especially interested in the catastrophic failure behavior of various PSUs, and would like to see the larger OEM/ODMs included as well as the re-branded "brands". (Most of them just buy PSUs from OEM/ODMs and at best may spec slightly better components on existing designs.)

 

[jfem]

This is good news! I'm looking forward to this PSU testing.

 

[Mousemonkey]

Great news.

We should of course give the reviewer the nickname "Scotty" ... surely?

The PSU is technically "engineering" and ratings should include references to "warp factor" and dilithium crystals" ...lol.

Nice test gear there ... good looming job too.

Looking forward to some definitive reviews now you have the gear.

I am wondering how some of the vanilla PSU's will go against the more expensive units?

I think mine is better. [:mousemonkey]

 

[bildo123]

I think mine is better. [:mousemonkey]

I would say photoshopped, but thats too perfect for even photoshop...somehow. What is that? Some audio setup?

 

[bill1514]

Hi folks, The test I would like to see is the
PSU fan failure scenario. All you need is a paper-clip
and a stopwatch, but I hesitate to perform this on
my own stuff. Smoke and fire has been reported, and
mobo damage too. A PSU that monitors PSU fan rpm
would probably have the best chance to respond quickly.

 

[zeterian]

Great idea, a fan failure test would be easy and provide important info. If the power supply fan stops does the power supply overheat, melt, catch fore, or otherwise fail catastrophically? Or does it shut down gracefully and wait for repairs?

 

[wa1]

It would be nice if we know about all brands PSU quality & performance...

 

[Mousemonkey]

I think mine is better. [:mousemonkey]

I would say photoshopped, but thats too perfect for even photoshop...somehow. What is that? Some audio setup?

They're Audio/Video racks and no, there was no photo trickery involved just about eight days on each one but there were only about forty in total.

 

[ram1009]

Your PSU testing is fatally flawed since it does not include the effects of ambient temperature. Any power supply can only supply it's rated maximum up to a certain ambient temperature. At than point the maximum available power begins to decline as ambient temperature increases. Commercial power supply specs provide a plotted curve to demonstrate the effects of ambient temperature on available output power. This plot is called the "derating curve". All power supplies derate as temps increase. The makers of consumer grade power supplies (such as those found in computers) get away with ignoring the derating process mostly because consumers are unaware of the physics involved. I would have thought Tom's would have been more astute. The only way around the effects of derating is to use a supply that provides ample power at the ambient temperatures it will actually see. I think we all know that the temps inside a computer case are well above the outside room temp. I see nothing in Tom's testing procedure aimed at providing this info. Back to the drawing board folks.

 

[ksampanna]

Great attempt at familiarising us with the testing procedure

 

[perishedinflames]

how soon will the article be online? i wanna buy a new PSU so hurry 😛
now seriously, great idea about that article and the test methods taken into consideration but as written on previous posts a noise level value would be mostly welcome 😉 (who doesn't love a silent p.c.?)