Is PSU fan speed controlled by temperature only?

[netmask254]

In most reviews, I saw a picture showing the relationship between power load (in Watt) and fan speed or SPL (http://xbitlabs.com/articles/coolers/di ... dup_5.html). It looks like the fan is adjusted according to power load. The higher load, it runs faster.

From some SPCR's review articles (though they also prefer to provide a table between SPL and DC load), I heard that fan speed is controlled by temperature (absolute value, or delta value between Intake and output?). Therefore, it runs faster in summer than in winter, lower noise can be achieved by lowering intake air temperature (e.g. using Antec P18X series). Is this true?

If so, if I find my PSU is not quiet as expected, maybe I shouldn't replace it with a quieter PSU immediately, but try to improve my case's air flow first. I saw the low SPL is generally achieved in very low temperature (21 ~ 30 Celsius, http://www.silentpcreview.com/article684-page4.html), but I don't have such confidence for my closed case in summer

BTW, there are two different types of PSU fans, 12cm and 8cm. The former is located inside the case, and the noise of latter is directly emmitted to the outside of case. Given they have the same noise level (I guess SPCR's review is done in open air), will the former (fan is inside) have smaller perceived noise than the latter (fan is outside).

[Klusu]

By the temperature of the heatsink with output rectifiers.

[SebRad]

Hi, yes I believe all PSUs that vary the fan speed do so based on a temperature sensor. Generally attached to one of the heatsinks in the PSU. As the PSU load goes up the waste heat increases so the temperature rises and the fan spins faster to compensate.

This means that if the PSU has cooler air going through it the heatsink will be cooler and thus the fan will run slower and quieter. This is the purpose of the "PSU duct" sometimes mentioned. It stops warm/hot air from the inside of the PC going through the PSU. The P18x and other cases that seperate the PSU from the rest of the system should have the PSU run a little quieter.
The opposite example would be a system with no case fan where all the hot air in the system goes out through the PSU. In this case the PSU fan will run faster as the PSU inside will be hotter than it would otherwise be due to the hot air being fed in.

SPCRs PSU test rig feeds all the PSUs output power as heat back in to the testing box. There is a case fan also but as power load goes up so does the intake air temp the PSU is exposed to. This makes SPCR's PSU test very demanding, in high load situations perhase unrealistically so. In a real system drawing very high load there would be more exhaust fans (or a P18x type case) so the PSU would have cooler air.

There is full description and discussion of the SPCR PSU test rig for more info.
Regards, Seb

[netmask254]

Thanks for your explanation! Later I read the SPCR test rig carefully and it does make sense to me.

But I still have several puzzles regarding the fan control:

1 Why do almost all PSU use temperature rather than power load or delta temperature to control the fan speed?

- Fan is used to cool the heat generated by PSU itself. The larger power load , the more heat generated, the higher fan speed is needed. A delta temperature between intake and output can also be used to control the fan speed. Assuming it is technically implementable.

- But using temperature (absolutely value) has annoying issue, i.e. different environments, different seasons & different cases can have very different temperature (the difference can be 10~20 Celsius), and the fan may run at very high speed even in light load, which is totally unnecessary. For example, my case temperature may rise to more than 30 Celsius in summer even in idle mode, I don't think it is dangerous for any components, but most PSU fan will run at a high speed than its real need.

2 How does heatsink make sense to PSU?

- For CPU & GPU heatsink, the answer is straightforward. Larger & better heatsink has higher efficiency, and we can use much lower speed fan to get the same cooling effect.

- For PSU, I saw many quiet PSU such like VX450 doesn't have a huge heatsink, meanwhile some PSU with impressive heatsink (such like Coolermaster M700) doesn't be significantly quiet than others. So, can larger heatsink result a quieter PSU?

[nomoon]

1 Why do almost all PSU use temperature rather than power load or delta temperature to control the fan speed?

Damage to the electronics depends on the absolute temperature of the electronics. The difference between the internal and external temperatures are irrelevant.

- But using temperature (absolutely value) has annoying issue, i.e. different environments, different seasons & different cases can have very different temperature (the difference can be 10~20 Celsius), and the fan may run at very high speed even in light load, which is totally unnecessary.

If the ambient temperature is hotter, then the PSU fan SHOULD run faster in order to keep the electronics at a safe temperature. Likewise, if the case has restrictive air intake, or if there is lots of power dissipation inside the case (high power video card), then the PSU SHOULD run faster. If the PSU fan is properly controlled by the PSU temperature, then it the fan won't run any faster than is necessary.

- For CPU & GPU heatsink, the answer is straightforward. Larger & better heatsink has higher efficiency, and we can use much lower speed fan to get the same cooling effect.

- For PSU, I saw many quiet PSU such like VX450 doesn't have a huge heatsink, meanwhile some PSU with impressive heatsink (such like Coolermaster M700) doesn't be significantly quiet than others. So, can larger heatsink result a quieter PSU?

In theory a better PSU heat sink will allow slower fan speeds. However, as the good people at SPCR has shown, some PSUs have better temperature control mechansms than others. Some probably run cooler than necessary and thus, run louder. Also, a bigger heat sink inside the PSU isn't necessarily a better one, so it's hard to judge the effectiveness of a heat sink by looking at its size. Some PSUs use cheap fans than run much louder than a a good fan at the same RPM. Some have loud bearings that are loud at any speed.

If so, if I find my PSU is not quiet as expected, maybe I shouldn't replace it with a quieter PSU immediately, but try to improve my case's air flow first.

Improving airflow into the case might be cheaper than buying a new power supply. You might try that first. Has your PSU been tested at SPCR? If so, then you may get an idea of potential of your PSU to run quietly.

Jason

[netmask254]

Generally I can't agree more with you in most points, thanks a lot for clarification!

If so, if I find my PSU is not quiet as expected, maybe I shouldn't replace it with a quieter PSU immediately, but try to improve my case's air flow first.

Improving airflow into the case might be cheaper than buying a new power supply. You might try that first. Has your PSU been tested at SPCR? If so, then you may get an idea of potential of your PSU to run quietly.
Jason

It's Antec EA-380, whose measured SPL is 22dB in light load (under 30°). I found it is louder than my expectation, when system is in idle state. Originally I was going to replace it with a better one such like HE-430 or VX450, until I heard that the fan is controlled by output temperature rather than power load (be confused by the test result table).

My case is a Coolermaster RC-534 which doesn't have a good airflow, comparing to Solo or P18X. Instead, it is almost closed except some tiny intake holes in the front. But after evaluating some alternatives, I decided to keep it since it works great in Beijing's air full of dust, I don't want to cleanup my case every month

Then I met this tough issue: It's relatively hot (30~40°) inside my case. I think it's safe for almost all components, and set other fans at low speed (SFLEX-E 12cm@600rpm for E8300, Case fan is Notuca S12@800rpm, and passive AC S1 for HD3650). But as you know, it's difficult to control the PSU fan, and seems like replacing PSU doesn't make much sense for my case.

[NeilBlanchard]

Hi,

The temperature control is the only/best way to go -- the difference is in the fan and in the fan control circuit. Depending on the design of the heatsinks, the air flow through the PSU in your system, and other things out of the control of the PSU designer; going with a load based fan control would be problematic...

I suggest that you listen to the recordings of the various units in the SPCR reviews?

[pcy]

Hi,

Temperature is used to control PSU fan speeds because it is cheap and reliable.

The temperature used is one that is typically high compared to the coolant air temperature, with the effect that the fan speed is more related to PSU load (which is reponsible for the heat) than to ambient temperature (i.e air temperatuer inside the case).

Depite what other here have said, this is far from ideal, because the fan does not cool the PSU directly - what it does is cool the air that cools the PSU by replacing it with cooler air.

If the temperatuer delta (PSU inlet to outlet difference) is low, then it does not matter how much faster you run the PSU fan, it won't actully make any significant difference. There is a stroing arguement for saying that if the temperature delta is (say) about 5C or less you should not increase the PSU fan speed however hot the PSU gets, simply because it will make more noise but not reduce the PSU temperature (much).

The only arguement for not using temperature delta - i.e running the PSU fan to keep PSU temperature delta constant at about 5C - is that it's expensive.

If the PSU is hot because the air temperature inside the case is high, the right thing is to increase the speed of the case ventilation fans if the PC is working in a room which is at a reasonable temperature, and switch on the air conditioning if the room itself is too hot.

Ultimately, the air in the room is the coolant fluid for the entire PC. Turning up fans inside the PC in reponse to high room temperature is stunningly in-effective.



Peter

[netmask254]

As a software programmer, I prefer using some intellegent logic (which can be implemented in a simple micro-controller) comparing to temperature only solution, though I know it's totally different between software development and PSU implementation.

If output temp is smaller than predefined safe threshold (e.g. 50 C)
Use delta temperature or PSU load to control the fan speed
Else
Use absolute temperature to control the fan speed.
Endif

// The desired capability is to run the fan at only the necessary speed to cool the PSU itself, in order to achieve the best silence, no matter whether the user has an excellent case or ordinary case.

[danimal]

It's Antec EA-380, whose measured SPL is 22dB in light load (under 30°). I found it is louder than my expectation, when system is in idle state. Originally I was going to replace it with a better one such like HE-430 or VX450, until I heard that the fan is controlled by output temperature rather than power load (be confused by the test result table).

My case is a Coolermaster RC-534 which doesn't have a good airflow, comparing to Solo or P18X. Instead, it is almost closed except some tiny intake holes in the front. But after evaluating some alternatives, I decided to keep it since it works great in Beijing's air full of dust, I don't want to cleanup my case every month.

the 120mm fan underneath the power supply on that case should be pulling out a lot of the hot air before it rises to the power supply... if you have the side fan mounted on the case, tho, it's going to be noisy, and the cpu cooler hole is also going to be noisy... since that case doesn't appear to come with an air filter, it's going to be dusty inside no matter what you do, unless you don't have any fans inside? the ea-380 that i'm running has a pretty quiet fan, but if it's the only fan in your system, it will have to work a lot harder.

one big advantage of a p18x case is that the 120mm fan in the lower compartment helps cool the power supply, if you have it set up correctly... so the internal fan on the power supply doesn't have to work so hard, because the cooling is being done by the case fan.

i think that you need to get rid of the case, get something with filtering capability, and preferably a lower compartment power supply, like the p18x.

[netmask254]

Thanks for your advice! I will consider buying a P18X again, it's very difficult for me to make a choice between P182 & P183

Regarding RC-534, I don't have a fan mounted on the side panel, actually I use some sound proof material to close the side panel since last summer.