Heat transfer within a laptop computer from CPU and GPU

[warfreak131]

Hello all

I have a practical question regarding heat transfer within my laptop. My laptop uses a shared-heatpipe design, meaning that my CPU and GPU are connected by a common piece of copper, and then each chip goes off to its own heatsink/fan.

While playing games, I was monitoring the temperatures of both chips, and the CPU was reaching much higher temperatures than the GPU (~20 degrees C). If they are connected with a common piece of heat transferring material, shouldn't both chips equalize in temperatures?

 

[anorlunda]

If they are connected with a common piece of heat transferring material, shouldn't both chips equalize in temperatures?

Not entirely. If the CPU makes more power, the heat flow out of the CPU must increase. Heat flow is proportional to temperature difference, so the CPU local temperature must increase. The heat pipe helps reduce the difference, but it can't eliminate it.

 

[russ_watters]

No. The heat pipe is likely a literal pipe, with a refrigerant in it (it may just be water/water vapor). Each component will receive the same inlet refrigerant temp and each will output it's own different refrigerant temp, which will combine (average) to get the final refrigerant temp. And none of the refrigerant temps will match either chip temp (heat will not flow unless the chips are hotter than the refrigerant).

 

[warfreak131]

Not entirely. If the CPU makes more power, the heat flow out of the CPU must increase. Heat flow is proportional to temperature difference, so the CPU local temperature must increase. The heat pipe helps reduce the difference, but it can't eliminate it.

So theoretically would a heatpipe with a higher thermal conductivity result in a lower differential?

 

[sophiecentaur]

So theoretically would a heatpipe with a higher thermal conductivity result in a lower differential?

I would have thought a greater temperature difference. In the limit, with a perfect heat sink system, the temperature of each processor would be set by the thermal contact between the processor itself and the pipe; the pipe being at the outside temperature. For a very poor pipe, the two processors could have more thermal coupling between them than with the outside world so their temperatures would be near equal.

 

[anorlunda]

So theoretically would a heatpipe with a higher thermal conductivity result in a lower differential?

Yes.

But also pay attention to what @sophiecentaur said. The thermal conductivity between each device and the heat pipe is not perfect either.

 

[Khashishi]

The heat pipe is likely a literal pipe, with a refrigerant in it (it may just be water/water vapor).

Refrigerants aren't used in normal consumer level laptop heat pipes. The heat pipes are generally just solid bars of copper. [edit: oops, they aren't solid. Heat pipes do have fluid inside. I was totally wrong.]
Some extreme high-end desktop systems do use water cooled systems, but they are too clunky to put in a laptop.

 

[sophiecentaur]

but they are too clunky to put in a laptop.

Clunky? Just too expensive surely. They are thinner that the equivalent copper rods.

 

[anorlunda]

Clunky? No. Here are some specs for a heat pipe that costs about $15 each.

• diameter of 6mm
• 3mm thickness
• Copper with water coolant
• Thermal resistance of pipe Rpipe range from 0.003K/W to 0.008K/W
• Maximum operating power of 40W

 

[Khashishi]

But to make it robust enough to not leak in a portable device would make it clunky.

 

[russ_watters]

So theoretically would a heatpipe with a higher thermal conductivity result in a lower differential?

It should, yes. Or just higher heat transfer capacity (like from a bigger fan). The temps should come down in proportion to each other.

 

[russ_watters]

Refrigerants aren't used in normal consumer level laptop heat pipes. The heat pipes are generally just solid bars of copper.

Interesting/surprising. seems like a misuse of the term (seeing it a lot on google).

Looking at pictures, it appears to be some of each. Round tubes with pinched ends would be pipes. Flatter, with rounded ends would be bars.

I see an alienware with pipes. I have a Dell inspiron gaming laptop that looks modeled after alienware (which I think dell owns).

 

[sophiecentaur]

But to make it robust enough to not leak in a portable device would make it clunky.

There are many hits if you Google "Laptop heat pipe" so I think you must be wrong about what those copper rods in laptops consist of. They are referred to as heat pipes and not heat sinks or copper bars so there is little doubt that they are heat pipes. There are many other more vulnerable parts in your average laptop. The thermal conductance of a heat pipe is pretty stunning and a lot better than a 10mm wide copper bar.
The normal heat carrier in heat pipes has usually been water because it is very effective - as in car engines and especially in high power (vapour phase) radio transmitting valves.

 

[Khashishi]

Sorry, I was wrong on this.