2003 Power Supply Roundup Part II: Better Faster Cheaper
by Kristopher Kubicki on July 31, 2003 1:58 PM EST- Posted in
- Cases/Cooling/PSUs
This is kind of a unique test we have worked on for a while. One of the more important qualities of a power supply is to make sure it does not produce harmful interference to the components. Perhaps the most sensitive of these components is the memory (video card, RAM, and CPU cache), which is just millions of little transistors locked in an on or off position, held in place only by a delicate balance of electricity.
The driving theory behind ECC memory is that it corrects errors that occur in memory. Perhaps a transistor is faulty and flipped for the wrong reason, or a faint electrical signal pulled a transistor into the incorrect position. While researching this review and others, we have noticed the largest factor for incorrect memory blocks is faulty power supplies.
How it works:
We started off by modifying MemTest86. We bumped the delay between tests up to several hours, rather than seconds. MemTests86 works by writing various patterns into the memory. After our programmed delay of six hours, the program goes back over the memory and analyzes how many of those bits were flipped out of the original pattern. Both the memory and motherboard used passed the unmodified MemTest86 without any problems.
Not only is this a good measure of shielding on the PSU, but also a good measure of shielding on the motherboard and memory. However, we used the same motherboard and memory for the entire test. This way we assure that the small amount of interference coming off the motherboard is consistent with each power supply. Unfortunately, it took 6 hours to run this test and we ran it three times for each supply to assure accuracy. On the next page is a table showing the trial data.
Facebook
Twitter
RSS
70 Comments
View All Comments
Anonymous User - Tuesday, August 5, 2003 - link
I personally thought the Antec True Power 330 was shining a little bit in this review. With 31.0 - 34.5 dBa noise, 26.5 - 33.8 C temp, on the lower end of memory errors, and a little tighter voltages than the average, it seems like it gives you the best combination of all categories. The only thing we're missing is some of the wattage testing. What do you guys think? And what about some of the other True Power product line - perhaps the 430 since 330 is at the lower end of PSU size that I would like to get for the power usage of any new computer (the True Control 550 is in a different product line than the True Power).Anonymous User - Tuesday, August 5, 2003 - link
Seasonic Super Silencer 400 would be interesting to test later - see if their claims of efficient, cool and quiet are accurate.abr27440 - Tuesday, August 5, 2003 - link
does it strike anyone else as odd that their memory if flopping bits left and right on their test? Dont know about you but i would prefer NO bit flips. :)Anonymous User - Tuesday, August 5, 2003 - link
A few clicks of my calculator indicate that only the Zalman ZM400A-APS and Fortron FSP400-60PFN did significantly better than average on the Interference Test, and only the TTGI/Superflower 420SS and TTGI/Superflower 520SS 4Fan did significantly worse.What were the results when different memory modules were tried, preferrably other makes of modules with other makes of chips? I'd like to know because the error rates indicated are roughly 10,000 to 1,000,000 times what memory manufacturers claim in actual systems.
Anonymous User - Monday, August 4, 2003 - link
I'm a bit confused over what it takes to be a good power supply? You slammed the Antec TruePower 330 for its "modest" +12 volt rail (204 watts). Yet you give the Frotron FSP400 your editors award and it only pumps 180 watts on the +12 volt rail. I would think this is a very important part of the formula for your pick since so many systems by your readers have high demands on the +12 volt rail due to the popularity of high end video cards. Comments?Anonymous User - Monday, August 4, 2003 - link
Another agreement with #15 and #18. If you want quiet, go for the silencer series. They are the same PS with different cooling.I discovered this when I found out what an outstanding company they are. My first PCP&C supply was a silencer 235, which is still running perfectly after 8 years. However, it wasn't enough juice for a P4 system. I wanted a new silencer with more power, but I balked at the price of the 400. I bought the turbocool 350 and boy was it loud! I called PCP&C and told them of my dilemna: what I really wanted was a silencer 350, which they didn't offer. NO PROBLEM! They put together a custom, full-warranty 350 silencer for me for $10 extra and shipped it right out.
Anonymous User - Monday, August 4, 2003 - link
The page on the TruePower 330 says the power on the +12V rail is 'extremely modest'. Did you mean to say the "+5V rail", rather than the "+12V rail"?KF - Monday, August 4, 2003 - link
The startling part of this review is that every power supply produced memory errors within six hours. It would seem every PC is unrealible and therefore worthless, regardless of the power supply. The cause could just as easily have been junk (noise, spikes, drop-outs, oscillations) on the power supply leads as electromagnetic interference. Or maybe it was a problem with the motherboard.Another interesting fact is that all the power supplies preformed insignificantly different under the loads used. Good news for people using cheap power supplies.
It would be interesting to know what the loads on the different voltage supplies are for real PCs by actual measurement. Not guess, not calculation from the labels. Monitor the currents while you do some of those stressful mobo benchmarks. How much on 12V, 5V, 3.3V? Without numbers, it is impossible to gauge what realistic power supply specs should be.
Anonymous User - Monday, August 4, 2003 - link
Erm, don't know of any mobo using the +3.3 as source rail for an AMD cpu; perhaps you might look at the +5.0 as a candidate.What 'theory' is behind that "theoretical combined" value? Meaningless, except to show a lack of understanding of how a 'standard' PSU is constructed (the only PSU where this would be comprehensible are the non-standard design Truepowers, where, of course, the value was ommitted).
A multimeter (or mobo sensor) gives only vague 'averages' of the madly fluctuating voltages that occur as millions of transistors switch on and off on each clock cycle; these broad values are only useful as an indicator of capacity/quality by the amount of sag between idle and load. You need a trapping 'scope' to nail the maximum variations to see the quality of the regulation (also to actually measure ripple).
Generally speaking, about 20W are consumed by the control (-) voltages, so that the total actual capacity of the three power rails is less than the claimed power. I have never seen a manufacturer claim a wattage not rounded to 5W, and just as the individual +3.3 and +5 wattages can 'sum' to a value greater than the TCO, so can the TCO and +12 wattages 'sum' to a number greater than the combined +3.3 +5 and +12 total wattage capacity. All of this leads to the impression that some table values (+12 in particular) were 'calculated' under some 'theory', and not the manufacturers' specs.
Anonymous User - Monday, August 4, 2003 - link
/RANT ON HOW THE MODS MODERATE...IF you mods want to play the PC card, they play it fairly or don't play it at all.
Why is a thread locked, when a person mentions a "blackie is aggressive, or these "africa americans are always aggressive".
Yet when another individual makes a similar thread, but this time ranting on "hispanics are all crazy etc etc" it doesn't get locked on.
Or this one is my favorite, i just saw a thread where someone mentioned his "retarded neighbour" lit fire to a paper roll or something. Now i thought calling someone retarded is not Political correct.
Like i said if you mods want to play the Political correctness game, then play it fairly or stop choosing at your will which threads offend you personally.
I assume this message won't get thru, because its sent from an anonymous. oh well.