bakaneko (no@spam.org) on 6/24/11 wrote:
---------------------------
>Ricardo B (ricardo.b@xxxxx.xx) on 6/24/11 wrote:
>---------------------------
>>I'm afraid you're quite wrong.
>>It's not a mere matter of air flow.
>>
>>Besides airflow, there are two more key parameters in the transfer of heat to air:
>>the contact area and the temperature difference between the contact area and the air.
>>
>>In order to maximize heat radiation you want your heat sink to have as much contact
>>area as possible. Thus, the heat sink designs with lots of big fins.
>>
>>However, there's a catch. The material's (aluminium, copper) thermal resistance will produce a temperature gradient.
>>That is, the heatsink will always be hotter at the base and cooler at the edges.
>>
>>This means that increasing the contact area has diminishing benefits: you increase
>>area, but a greater part of that are is cooler and radiates less heat.
>>
>>Enter water cooling.
>>With a circulating fluid, like water or oil, the temperature gradients are much smaller.
>>In turn, this allows for radically different radiator designs, with a much larger and effective contact area.
>
>What about heat pipes? They do basically the same, but without forcing
>you to use a complex system to control the fluid, are a closed system
>and can be used with passive cooling.

Heat pipes have a much lower thermal resistance than even circulated water but also have a power density limitation of about 10 to 30 watts per square centimeter at the evaporator depending on the construction so a heat spreader is needed. You could not for instance apply the heat pipe working fluid directly to the integrated circuit die.