PSU giving off more heat than CPU fan?

[Peroxyde]

Hi,

I am stress testing an old computer I am trying to fix. The motherboard is Biostar P4M80Pro-M7, CPU is D830, 3.0 Ghz, 90 mn, 130W. Artic Cooler Freezer Prod Rev2.

PSU is Antrec True power 480. The case is pretty big, the computer has 1 HD 7200 rpm, video is onboard. Only 1 GB Ram.

When running Prime95 for a while, I noticed that the air coming out of the PSU is warmer (feel with hand) than the air around the CPU heat sink. I would expect the opposite as the CPU is running at 100% load and the CPU TDP is 130W.

Is there anything wrong with the PSU or is it normal?

Thanks in advance for any advice.

 

[C1]

The PSU feeds the CPU so at least 130W is going thru it too. In addition, the PSU supplies the MB & peripherals including fans (CPU fan as well) plus there is additional consumption within the PSU due to conversions & (in)efficiency (ie, like a light bulb whose main purpose is to generate light, but also generates waste heat in the process.) So your observation is not surprising at all. However, it still doesnt have to necessarily be this way as the surface area of the heat generating components is also involved. (Idea is akin to putting a few ounces of weight on the top end of a vertical needle or pin. The pressure at the point is awesome & could easily penetrate well into your hand.)

 

[Aluvus]

The PSU feeds the CPU so at least 130W is going thru it too. In addition, the PSU supplies the MB & peripherals including fans (CPU fan as well) plus there is additional consumption within the PSU due to conversions & (in)efficiency (ie, like a light bulb whose main purpose is to generate light, but also generates waste heat in the process.) So your observation is not surprising at all.

The only heat generated by the power supply itself will be due to its own inefficiency. Antec rates the supply at least 70% efficient. To match the 130 W that the CPU is rated for (although Intel's TDP is actually a "high-end typical" rating, not a maximum), the system would need to be drawing over 430 W from the power supply. The OP's video card is not stated, but I can't imagine it taking much power for just Prime95 (video card should be at idle). It is utterly implausible that the other components are drawing a collective 300 W unless the system is very exotic.

However, it still doesnt have to necessarily be this way as the surface area of the heat generating components is also involved. (Idea is akin to putting a few ounces of weight on the top end of a vertical needle or pin. The pressure at the point is awesome & could easily penetrate well into your hand.)

If energy is being converted to heat, then it must leave as heat. And essentially all of that heat will be removed through convection. Your analogy is confusing and not relevant.

@OP: your results are probably partly due to the PSU taking in air that has already been warmed (as it is probably at least partially taking in the CPU's exhaust), partly due to differences in air speed, and partly due to the imprecise method of measuring.

 

[Peroxyde]

Hi Aluvus,

Thank you for your detailed explanation. If PSU has 70% efficiency, so would it need only 185 W to produce the 130W required by the CPU? (P x 0.7 = 130 -> P = 130 / 0.7 = 185)

Can you give more details about the 430W the system is drawing from the PSU?

The video card I used is the S3 Unichrome integrated graphic, from the VIA P4M800Pro chipset. I think that the video part is not very energy consuming, the northbridge heatsink is barely warm.

your results are probably partly due to the PSU taking in air that has already been warmed (as it is probably at least partially taking in the CPU's exhaust), partly due to differences in air speed, and partly due to the imprecise method of measuring.

You are right, the air intake of the PSU is directly in in front of the CPU's fan exhaust. However, when I wrote this post, the board was outside the case and the case was opened. So the PSU actually took air at room temperature but the exhauted air from the PSU already felt quite warm, clearly warmer than the exhaust of the CPU.

Or could that be that there is a short somewhere in the motherboard that draws too much current from the PSU? Because the motherboard quite often fails to boot (screen blank). Then I just hot reset and the boot was successful. Probably because all fans (CPU and PSU fans) and HD were already spinning). Is there a way to detect leak current due to short-circuit on the motherboard?

 

[C1]

If energy is being converted to heat, then it must leave as heat. And essentially all of that heat will be removed through convection. Your analogy is confusing and not relevant.

For a given heat input, increasing the surface area of the radiator results in reduced exhaust temperature. This is quite conventional & even intuitive. Think about it, 130W energy radiating off a 1" cube of tungsten will feel a lot warmer than the same energy input radiating off of a 6" cube of the same material. This is one of the purposes of heat sinks (spreads out heat energy).

 

[Aluvus]

Hi Aluvus,

Thank you for your detailed explanation. If PSU has 70% efficiency, so would it need only 185 W to produce the 130W required by the CPU? (P x 0.7 = 130 -> P = 130 / 0.7 = 185)

Yes, the input to the PSU at the wall would be about 185 W, and 55 W would be thrown out as waste heat. The reality is not quite so simple (efficiency of a power supply is really described by a curve that depends on the load, not a single number), as usual.

Can you give more details about the 430W the system is drawing from the PSU?

The video card I used is the S3 Unichrome integrated graphic, from the VIA P4M800Pro chipset. I think that the video part is not very energy consuming, the northbridge heatsink is barely warm.

If you have integrated graphics, and the processor is drawing its TDP of 130 W, the entire system is very unlikely to be drawing more than 200 W from the power supply. The two components that typically draw far-and-away the most power are the CPU and video card. If you have integrated graphics, then that draws essentially no power.

Or could that be that there is a short somewhere in the motherboard that draws too much current from the PSU? Because the motherboard quite often fails to boot (screen blank). Then I just hot reset and the boot was successful. Probably because all fans (CPU and PSU fans) and HD were already spinning). Is there a way to detect leak current due to short-circuit on the motherboard?

The symptoms you describe are typical of a power supply whose output does not stabilize quickly at boot (causing the motherboard to panic). When you perform the warm reboot, the power supply's outputs have already stabilized and so the reboot goes smoothly. This typically represents a fault of the PSU, not the motherboard. If it were my system, I would probably replace the power supply to get rid of the annoyance, and to prevent any further problems.

If you think the system is drawing an unreasonable amount of power, buy a Kill-a-watt (or Seasonic Power Angel) for about $20, and you can measure the amount of power drawn from the wall.

For a given heat input, increasing the surface area of the radiator results in reduced exhaust temperature. This is quite conventional & even intuitive. Think about it, 130W energy radiating off a 1" cube of tungsten will feel a lot warmer than the same energy input radiating off of a 6" cube of the same material. This is one of the purposes of heat sinks (spreads out heat energy).

Your description is not applicable to this situation. In a computer system, the exhaust air will be forced through an orifice of fixed size at a speed that is independent of the surface area of the components. Additionally, the predominant method of heat transfer will be convection, not radiation.

Heatsinks have large surface areas to make them efficient heat exchangers so that they can maintain low temperatures for the device they are cooling. To transfer heat effectively between two materials, you need either a large interface or a large temperature difference (having both is nice; the specific materials involved matter too).

 

[c-fone]

Aluvus,

I think more what C1 is trying to get at is that the entire cooling system of the PSU will determine it's exhaust temperature. I don't necessarily think the heatsink design will contribute greatly (unless it is woefully inadequate, in which case components would be failing) to exhaust temperature, but the fan speed/amount of air flow will. And as the OP is comparing a PSU exhaust to CPU exhaust, the difference in the amount of air flow provided by the 2 cooling systems will be quite different.

 

[Aluvus]

Aluvus,

I think more what C1 is trying to get at is that the entire cooling system of the PSU will determine it's exhaust temperature. I don't necessarily think the heatsink design will contribute greatly (unless it is woefully inadequate, in which case components would be failing) to exhaust temperature, but the fan speed/amount of air flow will. And as the OP is comparing a PSU exhaust to CPU exhaust, the difference in the amount of air flow provided by the 2 cooling systems will be quite different.

I agree that air flow (both velocity and volumetric flow) is likely to be a major factor (and said so in my first post), both for the reason you mention and because of its effect on human perception of air temperature. But neither of C1's posts mention this topic in any way.