Test: PSU fan up or fan down?

[lehtv]

I had been told that someone said that his bottom-mounted PSU was cooler with the fan facing up instead of taking air in from below. So I decided to test that on my Seasonic X650 at load (simultaneous Prime95 and Furmark, ~430W power consumption). Cooling: two front 120mm front intake fans and one 120mm rear exhaust at 1000RPM.

I used an electric thermometer, showing ambient temperature here, to measure the temperature of air in, air out, and the outer surface of the PSU (opposite to the fan). Temperatures in °C.

PSU fan facing down

Air in: 28 to 33 (depending on the exact location of the sensor)
Air out: 47 maximum (at this location), even under 40C elsewhere
Surface: 42 (at the center; measured with the side door closed)

PSU fan facing up (pic)

Air in: 33 to 35
Air out: 47 maximum (at this location), even under 40C elsewhere
Surface: 36 to 40

There doesn't seem to be a whole lot of difference between the two scenarios with regard to exhaust air temperature, even though the intake air of the fan up scenario is clearly warmer. Since in both cases the load level is the same, fan RPM should be about the same as well - so this cannot be explained away by assuming the hot air is pushed out faster. Instead, I think the hot air is simply more spread out along the whole exhaust area in the fan up scenario; the total temperature of the exhaust air should be higher.

In addition, the surface temperature in the up-scenario was lower. It was hard to measure from below the case; I'm not sure if I managed to get a good contact. But it's quite possible that the PSU casing gets hotter in the fan down scenario - heat rises, increasing the temperature of the casing. Whether that also means the components themselves get hotter than in the fan up scenario, I can't say for sure. But I think it's certainly possible, here's how:





Do you think this makes sense? How would you interpret these results - do they indicate that the PSU cools just as well either way?

 

[BoomerD]

For me, the biggest reason to mount a bottom-mounted PSU with the fan pointing down is to avoid the possibility of FOD. I don't want a loose screw or anything else finding its way into the upward-pointing PSU fan...While that might sound like an unwarranted fear...to me, it's just a precaution that's worth taking.

 

[Modelworks]

The whole heat rises thing is moot when dealing with pc cases because heated air will go any direction you want it to when you use a fan to move the heated air.

 

[GoStumpy]

I'll be perfectly honest... I can't figure out how a bottom-fan PSU can even WORK in a bottom mount case... How tall are the feet below the case to allow enough airflow UNDER the case?? It would have to be at least an inch off the ground to get proper airflow... and even that might be tight.

Fans can spin at the same speed, but with partially obstructed intake, it won't provide nearly the CFM as an unobstructed fan.

That is why I think having the fan on the top is better for cooling, unobstructed intake allows it to work at maximum efficiency.

This is all IMHO of course

 

[BoomerD]

I'll be perfectly honest... I can't figure out how a bottom-fan PSU can even WORK in a bottom mount case... How tall are the feet below the case to allow enough airflow UNDER the case?? It would have to be at least an inch off the ground to get proper airflow... and even that might be tight.

Fans can spin at the same speed, but with partially obstructed intake, it won't provide nearly the CFM as an unobstructed fan.

That is why I think having the fan on the top is better for cooling, unobstructed intake allows it to work at maximum efficiency.

This is all IMHO of course

My HAF 932 came with optional wheels that give the case about 1.5 inches of clearance to the carpet, plus it doesn't mount directly to the bottom of the case...there's about a 1/2" "shelf" that it sits on above the bottom of the case.

Unless you're really stressing the power supply, the fan isn't likely to run that much...or that hard. I think you're overestimating the amount of air flow needed by the PSU fan to properly cool the unit.

 

[KingFatty]

My video card is close enough to the bottom-mounted PSU to create somewhat of a dead zone between the video card and the PSU. So when I mount the PSU fan-down, the PSU is heated up by the video card and the only way to move that heated air around is the GPU fan, which is laterally far away from the dead zone as it's a single-fan GPU cooler on an AMD HD 5870 - the fan is not directly above the GPU so there is a good region of dead zone. Also, the GPU exhausts hot air inside the case, at each end of the video card, so literally hot air is pouring out of the GPU cooler right on top of the PSU.

If I mount the PSU fan-up, the PSU will help evacuate the heated air from the GPU dead zone, getting rid of that dead zone. I think dead zones can really see a large temperature increase because heat is adding up there without anyplace to dissipate.

So the question is, when mounting the PSU fan-down, are you creating more trouble by inviting a dead zone directly above the PSU? Your experiment seems to suggest that the heated upper surface of the fan-down PSU *is* hotter, and perhaps that is exacerbated even more when you have a high performance GPU next door to the upper surface of the fan-down PSU.

 

[KingFatty]

Also, I wonder how much the internal temperature of a PSU increases, when it's being roasted from the outside by a dead zone next to a video card?

So assuming you mount the PSU fan-down, even if you suck cool air from outside the case, is that cool air going to be enough to counteract the really hot side of the PSU next to the hot/dead zone?

I think it might be better to eliminate the dead zone by mounting the PSU fan-up, allowing the PSU fan to suck in warm air. The air won't be as blazing hot as before, because you will be eliminating a dead zone. This speculation is based on the test above where the output temperature of the PSU was the same whether fan-up or fan-down, when the PSU was sucking in cooler outside air or warmer case air.

Summary: better to mount the PSU fan-up to eliminate a dead zone and suck in slightly warmer air, as opposed to creating a dead zone by mounting the PSU fan-down to suck in slightly cooler air but then also roast the back side of the PSU exposed to the dead zone between the PSU and the graphics card GPU.

 

[BoomerD]

If you have decent airflow through the case, how do you end up with this dead zone you keep mentioning?

How do you know about the dead zones? Do you have sensors mounted through out the case?

 

[Sunny129]

Do you think this makes sense? How would you interpret these results - do they indicate that the PSU cools just as well either way?

i think that based on your experiment, the PSU seems to cool itself off just as well either way. and since that seems to be the case, i think the question then becomes "which scenario provides better cooling for the system as a whole?..that is, since either PSU orientation seems to identically cool the PSU itself, which orientation helps cool the internal components more?" that'll obviously require more experimentation on your part with the thermometer inside the case.

Fans can spin at the same speed, but with partially obstructed intake, it won't provide nearly the CFM as an unobstructed fan.

That is why I think having the fan on the top is better for cooling, unobstructed intake allows it to work at maximum efficiency.

This is all IMHO of course

while this may be generally true, at the end of the day this is still a generalization. it really comes down to the difference in air temps inside and outside the case, which is determined by what components are in the machine and how much heat they collectively generate. while your supposition may hold true for a machine that does little more than web surfing, email, office applications, etc. (a CPU with small core count, integrated (not discrete) video, etc), a machine that has a several-core OC'ed CPU and multiple GPUs (for gaming, rendering, crunching, etc.) is surely going to be significantly hotter inside than outside. so then the question becomes which is better - obstructed cool air, or unobstructed hot air? well its obviously a trade-off, and the "fan-up" PSU position will be better for some, while the "fan-down" PSU position will be better for others. but one will either have to make calculations (assuming they know the correct formulas to use) or experiment first hand with his or her own equipment to figure it out.

 

[KingFatty]

If you have decent airflow through the case, how do you end up with this dead zone you keep mentioning?

How do you know about the dead zones? Do you have sensors mounted through out the case?

Three things combined to make my dead zone, even though i have good case airflow:

1. Short case with bottom-mount PSU places the motherboard right next to the PSU
2. Motherboard (Asus m4a89gtd pro usb3) requires single video card to be put in lower PCI-E slot to enjoy full 16x speed, so video card is very low on the motherboard.
3. Video card after market cooler intakes air away from the PSU but exhausts hot air inside the case, near the rear of the case right above the PSU.

So the video card is very close to the PSU, the video card cooler is about less than an inch away from the top of the PSU. It's just enough space to fit a single slot device between the video card and the PSU. That creates a little pocket where it's just a dead end except for the hot air exhausting from the video card, which disperses inside the case.

The way i detected it is I felt the slot cover/metal blank below the video card on the back of the case. it was too hot to touch. Then I opened the case and touched the top of the PSU and the area above the PSU, and it was roasting hot, painful to touch. Much higher temperature than the rest of the case.

so it's a combination of visually seeing where air can physically be restricted without a clear path for air to come in and go out, and also of feeling with your hand (or a thermometer) where the temperature is unusually extremely high, despite the other areas of the case being much cooler. For example, above my video card where it's open for airflow (no other PCI devices installed), the metal case parts were cool to the touch and case air temp was pleasant. But stick your hand between the video card and PSU, and it hurts. Perhaps the main issue is how close the video card is to the PSU.

However, maybe my situation is extreme because of the three factors above creating the dead spot.