Originally posted by: jonnyGURU
See, at first I thought you were just being sarcastic. But now I realize that you're just ignorant. Not ignorant as in stupid. You're clearly not stupid. You just don't know what you're talking about here. What made you the end all authority on what power supplies publish true continuous wattage on their label and which ones publish peak?
The power supply industry is older that you are. Standards and specs have been permanent fixtures in the industry and no matter how much one individual, or even 30, want to deny them, they don't change. Get over it, your personal interpretation is only that, a subjective opinion against an entire industry. Ever wonder why the major players in PSU manufacturing become such? They don't play games with numbers, the customer gets what they spec'd instead of some kind of game to guess and derate and pull a number out of air to determine the value.
The whole "Continuous Power" thing on the Ultra PSU's is that the PSU can sustain that wattage. It's a continuous rating. Not peak.
It's supposed to be continuous. You have no evidence that it IS continuous. Runnning a test for an hour, or even a day, isn't even close to evidence that it meets that as a continuous rating. You lack even a basic understanding of what specs are. A spec is not "if you agree", consider it set in stone, a guarantee against which lawsuits occur. Consider the word "continuous" to mean "permanent", "forever", or "until it dies". All PSU will die eventually, but some far sooner than others. Just about any will run for a half hour. Motherboards with known bad caps ran for over a year quite often. Do you feel your tests would've revealed a faulty motherboard after 1/2 hour? This is not meant as a criticism, rather there has been a trend by many that simply does not serve the intended purpose of qualifying PSU based on their wattage rating. I would like that to change, or rather, mirror what can already be expected from the largest PSUs manufacturers, can be expected in OEM systems, as well as SMPS in almost every other product.
For example, open up a Xerox copy machine, you might find a (ballpark but it varies per machine) 150W PSU inside, one of them at least... Note the construction quality and parts specs relative to these supposed 500W PSU. Measure their output. Even a TIVO sub-100W PSU looks an order of magnitude better-engineered
per capacity than these Ultras.
You're right, a lot of power supplies DO in fact rate peak power output numbers on their labels. Some put both peak and continuous. So it is true that you can have a 500W that's really only worth it's weight as a 300W power supply, sure. You're correct there. But that's not the case with the Ultra. See my link where I hooked it up to the load tester? That's a 522.8W load for a full half hour. Not a quick shot at 522.8W
You are confusing "instantaneous" and "peak", or at least to discriminate between the two. 522W for half an hour is not evidence that this PSU can sustain 500W, that it's a valid continuous rating. Sustain means sustain, not for awhile, but "forever" until the PSU dies as all eventually will- but some far sooner than others, within the useful life of the system. Your test of a mere half hour is no evidence of anything but a peak output and that the fans at least moved enough air that it didn't POP within a half hour of doing it.
Pick any mechanical or electrical part, do testing on it at beyond it's long term capabilities. It is quite typical and expected that the further beyond the spec the test is, the shorter the life. Your test provides no evidence whatsoever that at 500W output, the PSU would last even a week, let alone a month, year, or the typical MTBF, which is closer to 10 years.
BY IGNORING THIS YOUR TESTS ARE SIMPLY INVALID TO SUPPORT THE CONCLUSION.
They are not evidence of anything except that it can output the labeled wattage for 1/2 hour.
Just as i'd written previously, I can get a 300W PSU to do that at 400W, but that doesn't make it a 400W PSU. Same thing but 100W higher label.
Any power supply that doesn't have overly conservative temp or voltage threshold points for shutdown (IE- any typical power supply), can manage to output a wattage higher than it's (if accurate) labeled spec would suggest. A wattage rating is not only a "can it output that" rating, it's a "can it output that and still meet MTBF rating".
There's a big difference. Any manufacturer could ignore the MTBF and simply add ~ 100W or more watts onto the labeled output figure and still call it continuous, but it would not meet the same MTBF spec if that remained accurate/honest. It is expected, in any typical PC PSU, that running one at higher output will result in shorter lifespan. Likewise, running (an accurately rated) 500W PSU at only 300W output, has the potential to increase the realized MTBF rating, so long as the extension doesn't conflict with other parts like the fans' lifespan. It is crucial that output wattage always be seen in the context of how long it can produce that output, otherwise the figure is rather arbitrary in being +- a few dozen %.
Here's the link again:
http://www.ultraproducts.com/forums/showpost.php?p=1742&postcount=3
The other thing you have to worry about with these "max wattages" is how CLEAN the voltage is on the way up. That's my big pet peeve agaisnt the OCZ. I used to think the 520ADJ was pretty sweet for high 12V load systems, but then I started to analyze how much the drop in voltage from 0 to max and although it's within spec.. it's still more than MOST of the good PSU's out there. The other thing that sometimes happens is a lot of ripple in the voltage. I don't have an O-scope, but my load tester reads voltages true RMS and it can show voltage fluctuations faster than the human eye can see them and if I put a load on a power supply and those voltages start bouncing, even only by as little as .1V, every 1/8 or 1/4 of a second then it's failed in my book.
Hmm. You are of course aware that one cannot use an RMS reading to determine ripple? That will necessarily make the reading invalid for this purpose.
Your conclusion about a .1V variation is also subject to more strutiny because the resolution of the voltage measurement will typically result in this kind of deviation even when the output is far more stable that the fluctuation appears to suggest. A few ideas here would be to use a voltage measurement device with more counts, and have it calibrated or checked with another (calibrated) meter, or calibrate it yourself with a precision voltage source. Something like a
MAX6350 should get close enough for this purpose. Even then, a voltmeter is not sufficient for ripple or noise determination, but at least this is one of the few areas where lack of capacitive load means one potential for deviation is gone.
It's still a static load though, the difficulty a unit has in regulation is not seem well at all with the static resistive load, that is as easy as it gets, contrasted with a variable load from a PC. PCs have a dozen amp current swings in a fraction of a second and the PSU must necessarily respond fast enough, and certain qualities of the PSU become paramount with today's ACPI capable PCs, such a high density (Or very large) transformer core. Otherwise, this slow recovery time is present and cannot be measured with a static load- and for the purposes of a PC, simply switching your load tester to a different wattage is still only a static load except for those few ms the moment the tester is switched to the different load.
