Hello, all!
I came up with some measurements on the PSUs when they are switched off. Usually, we measure the exhaust temperature while the PC is on in idle or while CPUburning. I think a relevant measure of efficiency would also be the temperature of the PSU when the PC is off. Of course, it probably depends a lot on the motherboard consumption but still, the figures are quite surprising (for me).
I measured by letting the PC stay off for the night and then I put the temp probe into the top part of the PSU exhaust vent. Here is what I have for my PSUs (ambient temp 23 C):
- Compaq 250W PSU: 27 C
- Superflower 350W: 32 C
- Enermax 350W: 28 C
Now, of the three, Compaq is by far the quietest, Enermax is sufficiently quiet and SuperFlower is quiet only the first 2-3 minutes (takes off after that to the full speed of the fan). There seems to be a clear correlation between the OFF temperature and the temperature of the exhaust under load (and that means noise of the fans too).
Any thoughts or more measurements, anyone?
PSU OFF temperature
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar
-
dukla2000
- *Lifetime Patron*
- Posts: 1465
- Joined: Sun Mar 09, 2003 12:27 pm
- Location: Reading.England.EU
So you are saying those are the temps after a night of +5V SB and convection cooling through the case & psu only?
Like you I worry about the mobo loads on 5VSB etc, but if the correlation to noise is there ...? Especially the Enermax coming 2nd: is that one of the 2 fan jobs? I would have thought the experiment is invalid because of the unknown static pressure of the psu box itself, but that a 2 fan version should have a low static pressure and thus perform well under convection. Except it appears it didn't and so maybe this is a reasonable indicator of efficiency as you say.
Like you I worry about the mobo loads on 5VSB etc, but if the correlation to noise is there ...? Especially the Enermax coming 2nd: is that one of the 2 fan jobs? I would have thought the experiment is invalid because of the unknown static pressure of the psu box itself, but that a 2 fan version should have a low static pressure and thus perform well under convection. Except it appears it didn't and so maybe this is a reasonable indicator of efficiency as you say.
-
larrymoencurly
- Posts: 170
- Joined: Tue Dec 03, 2002 11:29 pm
Did all the PSUs drive the same load in standby? How did they do when each had a 5 ohm load on the +5Vsb?
I think that the locations of the vent holes can make quite a difference because a PSU with lots of holes on the top or sides would allow a lot more natural convection air flow. But such a PSU may run hotter than one with vent holes only in front.
I have a pretty bad 250W PSU that stayfed very cool in standby, even with a 1A load. Its +5Vsb is handled by a completely separate PSU mounted on a tiny circut board -- it doesn't even share the main PSU's, filter capacitors or diodes. I don't know how efficient it is, but it runs at 200 KHz, or about three times the frequency of the main supply (not to imply that higher freq. means higher efficiency).
I think that the locations of the vent holes can make quite a difference because a PSU with lots of holes on the top or sides would allow a lot more natural convection air flow. But such a PSU may run hotter than one with vent holes only in front.
I have a pretty bad 250W PSU that stayfed very cool in standby, even with a 1A load. Its +5Vsb is handled by a completely separate PSU mounted on a tiny circut board -- it doesn't even share the main PSU's, filter capacitors or diodes. I don't know how efficient it is, but it runs at 200 KHz, or about three times the frequency of the main supply (not to imply that higher freq. means higher efficiency).
This is, indeed, the temperature of the PSU after a night of +5v standby.
Of course, it would be difficult to generalise after a simple measurement like I did. I did the measurements on 3 different computers in the same room but the loads are probably different and the different cases probably also contribute to the difference by their airflow patterns. I agree this completely unscientific but my wife won't let me disassemble her computer to take the PSU out and make a more precise measurement anyway.
Even so, the results were striking for me. Let's discount the PSU of Compaq - that's a very old computer and it is difficult to compare it to the others. The other two are both in cheap cases with moderately bad airflow. They both support AMD motherboards. Both have more or less the same configuration - mobo, AMD processor, same memory, one HD, one CD. The biggest difference is the PSU itself. Oh, yeah, and the one that keeps cooler - Enermax - probably has more load because the Ethernet card is set to WakeOnLan (all time powered).
Basically, what I am saying is that the difference is too big to discount it as a result of external influences. I think the major factor is the quality of the PSU itself.
Of course, it would be difficult to generalise after a simple measurement like I did. I did the measurements on 3 different computers in the same room but the loads are probably different and the different cases probably also contribute to the difference by their airflow patterns. I agree this completely unscientific but my wife won't let me disassemble her computer to take the PSU out and make a more precise measurement anyway.
Even so, the results were striking for me. Let's discount the PSU of Compaq - that's a very old computer and it is difficult to compare it to the others. The other two are both in cheap cases with moderately bad airflow. They both support AMD motherboards. Both have more or less the same configuration - mobo, AMD processor, same memory, one HD, one CD. The biggest difference is the PSU itself. Oh, yeah, and the one that keeps cooler - Enermax - probably has more load because the Ethernet card is set to WakeOnLan (all time powered).
Basically, what I am saying is that the difference is too big to discount it as a result of external influences. I think the major factor is the quality of the PSU itself.
Ok, folks, here is one more PSU. This time it is Zalman 300A-APF. The idle (PC OFF) temperature compared to the Superflower in the same box with exactly the same setup:
Superflower - 32 C
Zalman - 28 C
Zalman is by no means a silent PSU but it is a huge improvement compared to Superflower. Now I noticed that the construction of the two is really different, I expect a much better airflow thorugh Zalman. And it shows.
Interestingly, the temperature inside the case while running CPUBurn increased from 28 C with Superflower to 29 C with Zalman but this can be discounted. Imprecision of measurement, variations of ambient temp and so on.
It seems now to me that the idle temperature of the PSU (when the PC is off) can be a good indicator of how well the PSU is designed.
Superflower - 32 C
Zalman - 28 C
Zalman is by no means a silent PSU but it is a huge improvement compared to Superflower. Now I noticed that the construction of the two is really different, I expect a much better airflow thorugh Zalman. And it shows.
Interestingly, the temperature inside the case while running CPUBurn increased from 28 C with Superflower to 29 C with Zalman but this can be discounted. Imprecision of measurement, variations of ambient temp and so on.
It seems now to me that the idle temperature of the PSU (when the PC is off) can be a good indicator of how well the PSU is designed.