After typing a lengthy note on this topic, I clicked the preview button and rather than previewing the note the system asked me for my pw again and my note had completely vanished!

Anyway, to summarize my main points - new high-end graphics chipsets such as the recently launched ATI Radeon 9700 will require separate power feeds and higher rated PSUs. The ATI rep I talked to told me that the 9700 should only be used with a 300 watt or higher PSU. Add in a high end system with a 2 big drives in a Raid array, 2.8 gig P4, Audigy sound card, 768 megs of PC1066 ram, like I am doing, and you might want to go to 400 watts - especially since with a Seasonic or (probably) Truepower the 400 PSU due a higher heat floor for fan speedup is likely to be quieter.

Comparing the 430 watt Truepower and the 400 watt Seasonic - Mikailtech has tested the voltage regulation on the 300 watt Seasonic and found that it had much greater variability on the voltage rails that the equivalent Truepower, which only varied by .6% on the +5 voltage and not at all on the +3 on high load. This can be critical for overclockers and even for regular users of problematic boards like the Asus P4T533, as voltage drops on rails can cause crashes and other flakey behaviour on these systems. Antec uses several tricks with the Truepower line, such as fully independent rails for the different voltages and compensation for voltage drop at the connectors, that result in tighter regulation. Silent PC needs to add voltage variation assessment to its reviews to be truely comprehensive in mapping out a PSU's strengths and weaknesses. It's true that the Seasonic unit has one less fan - the recommendations indicate that this makes it less noisy, but I wish this site would give a subjective percentage - 10%? 70%?

Anyway, thanks to all concerned for the site - it's very useful for those of us trying to build a quieter system (my current Dell box should be levitating on its fan exhaust given the racket it makes).

Herb, I have to disagree with you on this issue. MBM shows the voltages monitored by the motherboard. The points at which this monitoring is done differs from mobo to mobo, and more importantly, comes AFTER the highest potential votage drop points, which are 1) the ATX connector to the mobo and 2) the wires carrying the voltages from the PSU to the ATX connector. This thread explains -- http://forums.silentpcreview.com/forums ... 1&start=15 -- and quotes John Hoot of overclockers.com extensively. PLEASE read through the post & http://www.overclockers.com/tips845/.

Then you will understand that while MBM5 tells you the voltages on the mobo, it does not tell you the cause of low voltages. The drop could easily be happening in a couple of poor contact points in the ATX connector, caused by looseness, contact surfacr deterioration from overuse and age, or oxidation. It does not really tell us anything about the quality of the PSU voltage regulation. This is particularly true for test systems (such as Mikahil's or mine) which are subject to much higher wear and tear than normal system because of all the plugging and unplugging required to test components.

MBM is the worst way you can test voltage regulation from a PS, as mike said. it needs to be done at the atx connector.

however, spec-wise, antec TP is rated at +/-3% (mostly due to additional remote sensing lines). i think seasonic is at the standard +/-5%.

almost always missing from PS review is transient load testing, even when proper static load testing is done.

basically, every PS review i've seen so far has been inadequate in terms of load regulation testing performace.

consider that when you read any PS review based on load regulation if you're using it as a determining choice when buying your PS.

Thanks for pointing this out...Of course if the tests at Mikail's had been more carefully conducted - such that the same motherboard + system load had been used for both PSU tests, preferably in an AB-BA type ordering to control for contact wear - they still could provide useful information. If PSU A were reported by MBM to provide +4.85 v on the +5v line under load, vs. PSU B's +4.97, we'd have a pretty good estimate of which one was superior... but of course no absolute values could be calculated without having reference figure(s) obtained from MBM when the motherboard being used is known (through measurement at the connector) to being "fed" the exact standard voltage. I wonder why so many sites test voltage with MBM instead of conducting the simple test Hoot outlines in that OverClocker article....

Looking at the three reviews for PSUs on the site, I see no voltage load testing of any type...

Well, even that will not help because the ATX connectors are not consistent enough. If you examine carefully the way the pins and jacks mate, you can see that there is a fair amount of play in each mated pair. The ATX plug from PSU#1 may fit fine in mobo#1; but a couple of pins in the plug from PSU#2 are slightly off. MBM5 results would favor PSU#1.

Then there is the question of when the tests are conducted. If they are done at the same time, then it's a bit better; but if you have a gap of more than a couple of weeks, I guaranteed you a test review system will have been assembled & disassembled enough times to have at least a bit more contact deterioration.

Another thing: a single screen shot of the MBM5 tells you almost nothing at all. You really need to see what the voltages are doing over a period of time (30-60 mins) under a variety of loads, with sampling being done as often as possible.


Time, tediousness (to watch the voltages on the various lines over time), lack of adequate understanding. Most PC review sites are staffed by PCers, not electronics guys. (I have a wee advantage; I did study EE for a few years eons ago. )

If the tester's motherboard pins have deteriorated from wear then I agree they certainly shouldn't be testing voltage via the motherboard under any circumstances, since any faulty connection can't be attributed to a potentially poor PSU connector - there is no way of knowing where the source of the error is. However as an end user it's useful to me to know that what comes through the plug when it's plugged into a "proper" motherboard connector is at the right voltage level; testing of only the output at the tips of the plug can't tell you that - you would need to run you run the voltage test through a quality mb connector part first or the quality of the PSU connector would remain unverified. A great power supply with lousy connectors isn't much of a bargain....


I thought that likely, the ABBA reference was meant to informally convey a counterbalanced research design (measure A, then B; then B, then A; average both) which controls for time-dependent effects...



Mikhail's site, among others, isn't doing a snapshot approach - from their Seasonic tests: "I put each power supply through two different testing sessions. The first session was a one hour of the Prime95 torture test followed by an extended period of idling in order to measure the voltages from the power supplies". The screen shots of MBM shows the high, low, and average voltages derived from hundreds of samplings taken over the test periods....probably the major reason (aside from ignorance) that MBM is used, since as you say doing this by hand would drive you nuts!
[[/list]

Fair enough, and you're right about Mikhail, I have seen that comment on his reviews before. I guess to really do a thorough job, if I see differences between the PSU connector voltage and MBM, I should run a couple of different mobos with the PSU to account for any differences between connector interfaces... Hey did I just double up on my work??

I have actually taken a couple of ATX connectors apart and haven't been able to spot much in the way of differences; doesn't seem like there are miserable quality connectors and top notch ones. (Someone will correct me if I am wrong, I'm sure) I do notice gold plating on some expensive PSUs, but this is not really an advantage unless the mating connector on the mobo is also gold plated, because contacts between different metals actually cause faster deterioration. Can't recall exact details of why right now; something I learned in high end audio.

Re - AB-BA -- your comments are right, of course, if it was done that way, but in your reference to the Antec & Seasonic PSU reviews by Mikhail, aren't they separated by like 6 months?

I was just wondering if the Seasonic reviewed on this is really better than the antec truepower when it comes to silence. I know that this wasn't the point of comparison between the two, but I'm just a bit disappointed at my truepower purchase. I dunno if this calls for modding my psu with a spare panaflo l1a or not. But I think Mike mentioned that it had little to nothing to do with the stock psu fan because he tried putting a piece of thread to stop it momentarily to check if the fan was the source of noise. I'm just hoping to resolve this issue especially since I'm working with recording audio with this computer. Thanks.

Yes, mySeasonic samples are quieter: only one fan (instead of 2)and no coil whine. What is the nature of the noise you hear with the Antec?

it's a high pitched whine that's soft yet audible. Not to mention there's a droning in there. Possibly the latter is coming from my three l1as that are running at 12v. I did make a 5v/12v switch but I have yet to fully put it in since I'm doubtful that I did a good soldering job on the rubber shielding around the atx wires and the switch itself. I'm just hoping that there's a way to salvage my current boxes psu and fans since I unfortunately have no more money left to buy another psu. Thanks.

sonarcade -- try stopping the fans in the antec manually, with like a plastic straw. Stick it into the blades while the fans are running. If you still hear noise when the fans are stopped, well, trouble. Describe the noise. Re other noises, just systematically go through and isolate each component and identify where each noise is coming from. After all this, you should have a real good idea what to do.

If the PSU noise is high pitched whining, it's most likely a coil or transformer, as the Antec fans are fairly quiet, they only get ~5V up to ~150W output, which basically means they will never speed up under normal use.