26 power supplies reviewed by matbe.com - Best seasonic ?

Uriox · Post by **Uriox** » Thu Dec 01, 2005 5:47 am

This review (in french) is located here:
http://www.matbe.com/articles/lire/260/ ... entations/

EDIT
At idle :
http://www.matbe.com/divers/zoom/compar ... 23951.png/
As people have said all is OK.

Now the real problem. During charge :
http://www.matbe.com/divers/zoom/compar ... 23949.png/

END OF EDIT

Look at all Seasonic. I thought the seasonic S12-430 was meant to be quieter than the S12-500 and S12-600 ??? This is not the case here... Have other fan changes occured recently that could explained these results ? Is this due to the very high load from the configuration used ?

NapalmDeath · Post by **NapalmDeath** » Thu Dec 01, 2005 7:38 am

Likely test errata for the 1 db discrepancy.
Keep in mind it's an idle load test.

Aris · Post by **Aris** » Thu Dec 01, 2005 7:49 am

40.6, 40.7, and 40.9 dbA, would, in my mind, be the exact same sound level. the slight differences are within the margin of error. all 3 probably use the exact same fan.

Bitter Jitter · Post by **Bitter Jitter** » Thu Dec 01, 2005 8:20 am

Have to say reading reviews like this just make you glad that SPCR exists because those sound reading don’t tell you alot.

One thing which seems strange is that the NEO HE in this test seems to have similar or better efficiency compared to the Seasonics. Maybe Mikes speculative comment about 220v in the review, really does makes a big difference to the Neo.

I just brought a S12-380 to replace my Nexus NX-3000 which i believe is making a rather annoying screeching sound. A friend wants me to build him a computer in about a months so I think I will recommend he gets a P150 then I can test them both and see which one is better.

Uriox · Post by **Uriox** » Thu Dec 01, 2005 8:23 am

Sorry, wrong cut&paste, I meant this one :

http://www.matbe.com/divers/zoom/compar ... 23949.png/

That's a 6 dB difference in charge, and cannot be attributed to measurement
error...

Here are their comparison between 430, 500 and 600 W models :
http://www.matbe.com/images/biblio/art_ ... 023775.png

Post by **MikeC** » Thu Dec 01, 2005 9:14 am

It's too bad they show nothing about the loading for that test except that Vcore is 1.5V. So what? It tells us nothing about the power draw on each unit -- whcih they do indicate on the other charts.

In any case, where such a discrepancy in noise might occur is at very high load, where the better efficiency of the 500/600 would tend to keep the fan speed down while the fan in the 430 would have to run faster due to higher temps. Given how well the 430 behaved in this regard on our test bench, I don't know how Matbe managed to make this happen. In my experience, the 430 always has a slight advantage over the 500/600 because its fan is quieter, and it's very hard to make the fan in any of these PSUs to ramp up in real systems.

Uriox · Post by **Uriox** » Thu Dec 01, 2005 9:20 am

You're right Mike...as usual !!!!
At IDLE they were at about 200 W, and 420-450 W during charge. So the S12-430 had ramped up, while the 500 and 600 had not.

OUTLAW · Post by **OUTLAW** » Thu Dec 01, 2005 3:39 pm

Just bought the seasonic s12-380w
This is quiet

Althgou i expected better finishing and bundl im very pleasd with psu itself.
Its runing dfi nf4-lanparty whitout any problems so the problems guys over at df-strret forum were talking about is plain rubbish(480w psu at least

) and boot problems

Ironic · Post by **Ironic** » Thu Dec 01, 2005 10:41 pm

Yeah, I really liked matbe psu tests but lately I have begun to lose a little trust in their accuracy and protocol.

First, they don't test psus inside a case.
Then, their configuration is just a power hungry beast, not reflecting common setups and even less, silence oriented ones.

It remains interesting for overclockers still.

SnooP · Post by **SnooP** » Fri Dec 02, 2005 12:47 am

on a lighter note, the google translation (French to english) is interesting reading

A few years ago, the tendency of the food was with the color. The persons in charge marketing left us the blocks increasingly more beautiful, with pretty cables sheathed and sometimes of the reactive connectors to the ultraviolet rays.

Then, their configuration is just a power hungry beast, not reflecting common setups and even less,

I don't see a big issue with that for testing how stable the PSU is, if it handles the worst case scenario, it can handle common setups. Be nice if there was more detail in results though, and a lighter system to test acoustic levels (more representative of silent pc / htpc).

EricTerminator · Post by **EricTerminator** » Sun Dec 04, 2005 11:30 pm

Hi all !

I carefully read the Matbe.com's article, and it's simply a conclusion for ultra-hungry PCs.

The best PSUs are simply the Seasonic S12 series, although the 430W seems a little bit noisier than the others. But the S12 series cost a bit too much.

I have the Fortron Green Power FSP400-60GLN, and I agree with all known tests on the web ; its efficiency is extremely good, and its noise is very low. I made a measure of the efficiency here : http://www.pcsilencieux.com/forum/viewt ... 9&start=15

I have an efficiency of 83.25% at full load, with an error margin of 1.92%. Note that I use the European AC 220V.

I don't know what is the power draw of the components, but it's a more realistic load than the P4 560 3.6GHz with an ultra-high VCore with 2 GeForce 6800GT...

The FSP400-60GLN is simply the best compromise between cost and performance, while being one of the cheapest PSUs...

See you

ckolivas · Post by **ckolivas** » Mon Dec 05, 2005 3:15 am

Normal Sound Pressure Level gauges only read down to 40dB. Only SPCR have openly said they purchased a professional audio one which reads below this. Does this review say anything along those lines? (I can't read French). If the gauge only goes to 40dB and their noise floor is around that, how can we realistically compare PSUs that are around 20dB at idle? Hint - the Phantom should be almost silent barring coil whine yet it gets 40dB which supports my theory.

BillyBuerger · Post by **BillyBuerger** » Mon Dec 05, 2005 7:35 am

I was looking at their setup. They've got a high speed Delta fan running on the HS. That and they have their SPL meter 20cm from the PSU (but on the side to avoid wind noise). The system itself must be producing about 40dbA. Although in theory, if they were consistant, the amount above 40dbA should be attributed to the PSU.

ckolivas · Post by **ckolivas** » Mon Dec 05, 2005 7:43 am

BillyBuerger wrote:Although in theory, if they were consistant, the amount above 40dbA should be attributed to the PSU.

Being a log scale, adding 20dB to 40dB would make no measurable difference; 1000+10 = 1010.

BillyBuerger · Post by **BillyBuerger** » Mon Dec 05, 2005 12:46 pm

ckolivas wrote:Being a log scale, adding 20dB to 40dB would make no measurable difference; 1000+10 = 1010.

I wasn't suggesting that it was a linear amount. Just that 40dB was their base sound level. Which seems terrible to be making any other kind of meaningful measurments.

Also, I forgot to mention that I wonder if system noise is consistant? It looks like the CPU fan is probably wired directly to the PSU. Running a 36 watt fan through a fan controller probably wouldn't make for a very long test. But I wonder if the GPU HS fans were consistant. I have a hard time believing that they could really be heard over the HSF. But then again, the same goes for the PSU.

Either way, I can't say I would trust the SPL measurments at all.

EricTerminator · Post by **EricTerminator** » Mon Dec 05, 2005 1:18 pm

They measured the sound level at 20cm of the PSU, but on the side, in order to have a noise level that is not perturbated by the airflow.

At such a distance, I suspect them to turn off all the fans of the PC, except the PSU's one.

Of course, I ask my friend Stéphane Charpentier what was exactly his test protocol.

See you

EricTerminator · Post by **EricTerminator** » Mon Dec 05, 2005 1:43 pm

OK guys, I've got the answer of my friend Stéphane : http://forum.matbe.com/materielbe/Matbe ... htm#t15452

All the fans, except the PSU's ones, were turned off for 20 secs in order to measure a stabilized noise level.

Stephane · Post by **Stephane** » Mon Dec 05, 2005 1:54 pm

Hi,

I am Stephane, the Editor in Chief of Matbe.com. First, I have to say it is nice that my reviews are discussed here on a american website. Then here is the answer for the noise measurement.

All measurements concerns only the PSU since I stop the CPU Fan with a Sunbeam Fan controller and that I stop with my fingers the fans of the VGA card. As the Asus P5ND2-SLI has no fans to cool the chipset, the only noise remaining comes from the PSu's. Even the hard drive is place in a Scythe HDD Box to avoid any noise from it.

dB/A are measured while the Pc is under very high load : CPUBurn is running and a homemade 3D test is burning the VGA cards (fillrate test for those who know).

The fans of the CPU and VGA cards are stopped for 20 seconds, the time to stabilize the noise level on my soundmeter...

Sorry for my poor english...

EricTerminator · Post by **EricTerminator** » Mon Dec 05, 2005 2:04 pm

Not bad, although I'm not a true English-speaking person, I only detected 4 mistakes.

And Stéphane gets and A- for his 1st, his maiden performance in English on the World's reference for deadly silent PCs !

With some practice, you should get an A+.

Don't worry Stéphane, it's quite good, since there are sometimes non-English-fluent people who post on this website, or some people who use an impossible-to-understand slang.

At least, this website isn't like Hardware.fr's one. (A true "bordel" (mess) in terms of orthography and grammar.

).

See you

EricTerminator · Post by **EricTerminator** » Mon Dec 05, 2005 4:35 pm

Just a question off topic for Aris ; was Offutt AFB, NE, the former SAC command center ?

If not, it was a major base of the SAC in the 60s, right ?

See you