Slashdot Mirror
Tom's Hardware Compares Power Supplies
Some guy wrote in to say "Tom's Hardware Guide takes a hard look at power supplies to find out if we are getting what we paid for. The results of the testing were very surprising." Very useful to anyone who has built their own machine from scratch or burned out a cheap power supply.
Users who stand to fare the worst are those who have purchased their computer from a computer super store and/or discount retailer. These systems are generally fitted with cheaply made, low-cost power supplies, which often can sustain damage even under minimal loads.
I realize that me and the half of my friends and family that are not hardware-savvy only make up a small sampling. But none of the people that I know that have bought pre-built machines have had a problem with power supplies. Whereas at work, where my office is filled with machines constructed from the parts of other machines, we've had two instances of power supply failure.
Of course the machines at the office are older and are used more often than the home-based ones, but I figured I'd put in my 2 cents.
Something I've been ranting about for years: It's not just that power supplies are rated in "music watts." It's also that basic engineering apparently went out the window when micros came in--and has never come back.
Why isn't every board and component clearly marked with its power consumption?
Why isn't every system clearly marked with the amount of power available to devices on the bus (power supply minus consumption of preinstalled components?)
Why isn't there some kind of built-in INDICATOR that WARNS you when the drain is approaching the power supply capability?
None of this is rocket science. It requires fourth grade arithmetic, a multimeter, and a little honesty.
On minicomputers, the power supply was sized for the worst-case set of boards that could be installed in it. That's probably too much to expect from PC vendors, but at the very least there should be an easy way to TELL.
"This is a real good power supply and it should be OK unless you put in an awful lot of boards that take a lot of power" just isn't the way to do things.
We expect this stuff to be clearly marked on our light bulbs, our vacuum cleaners, and our fuse boxes. Why shouldn't we expect it in our computers?
"How to Do Nothing," kids activities, back in print!
Very true. The problem's one of brand differentiation.
At the low end (vast majority), there's Joe Sixpack, who doesn't know anything about what's inside his box, and who doesn't even think there are modular components in it. "Dude! I got a Dell!"
At the midrange (20%), there's most of us. "It's an ASUS or ABit mobo, an AMD/Intel CPU, and an nVIDIA or ATI card. Umm, and a power supply that came with the box."
At the high end (5%), there's folks who know that no-name power supplies suck (IMHO) and Enermax rox (IMHO). But only because we've read from...
Unlike ATI-vs-nVIDIA, where the midrange and high-end folks can plunk in an "upgraded" video card and immediately see the difference between "good" brands and "shitty" card manufacturers, brand names in power supplies have trouble gaining traction, because while they're working, they're indistinguishable from each other.
And of course, when they stop working, it's too late. For that reason, always keep a spare supply handy. Even a "free" no-name supply ($5 bucks from a surplus store) will last you a week until your "real" power supply arrives from the web merchant.
Rule of thumb: Put your hand over the PS's exhaust fan. If the air's significantly warmer than the air in the PC's case, get a bigger power supply.
(I learned it the easy way - a d00d at work got a dual Athlon mobo and just tossed it into his old case, and wondered why it was so warm under his desk while his CPU temperatures were normal. Answer: 2-year old 300W supply + dual athlon + two 7200 RPM disks = one massively overloaded P/S. We threw in a 365W "spare" for the weekend and ordered a 450W that got there by Monday. Air flow out the back was much cooler. Frankly, I'm amazed the 300W supply was even able to boot before blowing itself to hell.)
The relevant safety standard is UL 60950 (or the identical EU 60590 in Europe), "Safety of Information Technology Equipment". One of the required tests is a full-load worst-case temperature test. No way should those units have received UL or CSA certification.
UL's certification search engine is broken today, so I can't check the power supplies listed to see if they really passed. But those certifications are public information; you can check.
Current CPSC product recalls in the computer area include PowDec power supplies for NextLevel DSL modems and several batteries for laptops. Sounds like that list needs some additions.
UL's testing agrees with Tom's Hardware. All devices that really passed UL certification were found OK by Tom's Hardware. No UL-certified device blew up, caught fire, burned out, or failed to perform at rated load. All the units that failed lacked valid UL certification. The Leadman LP-6100 E did fail under full load at Tom's Hardware, but it shut itself down properly without damage. UL hasn't rated it, although they've rated previous Leadman models.
The Tom's Hardware article shows the data plate from the Chieftek power supply, which bears a UL marking. It's not in the database. It looks like many of those power supplies have fake UL certification, and for good reason - they don't meet specs or they're outright hazardous.
So if it's not in the UL database, don't buy it. There are plenty of good power supplies that have real UL certification. Corporate shops probably should check for those phony brands and take appropriate action. And tell UL; they will take action for phony markings.