Socket A Coolers - That Don't Kill

Mr-Tweak writes: "TweakTown have just put up their Socket A Cooler guide entitled "Socket A Coolers - That Don't Kill". In the guide they review and test out the Global Win FNP50, Global Win FOP32-7K, ATTech CM25 and the ThermalTake Chrome Orb."

Comments

Re:Copper is wayyy cool - details

[paled]

I had included an ASCII diagram, but it was rejected as a junk post.

thermal conductivity coefficients:

k [=] W/m deg C

assume 30 deg C

copper (pure) 383

aluminum (pure) 201

so Copper has a value (383 - 201)/383 = 47.5 % higher
>
So there are multiple resistances involved in heat transer:

1. CPU internals to CPU exterior surface (internal) (user cannot influence this)

2. CPU to Thermal Paste (interface)

3. Thermal Paste to Heat sink surface (interface)

4. heat transfer within heat sink metal (internal)

5. heat sink to air (interface)

Assumes:

- the use of thermal paste applied as a uniform thin film.

- constant heat duty - chip has been running long enough to be at steady-state temperature.

The heat transfer at (5) is maximized by having a large relative surface area of metal to air transfer area compared with the area of the heat sink in contact with the heat source. By using forced convection (fan) the heat transfer coefficient is increased.

The main improvement by using copper is that the temprature profile along the length of the fin is more uniform, meaning that the same driving force for heat trasfer is maintained, provided a constant heat duty. This produces a smaller temperature at the CPU-Heat Sink interface.

But the interface resistances are most likely a greater factor in the overall heat trasfer - hence the reason why overclockers wetsand the machined interfaces with 220,400,600 grit paper to minimize surface roughness.

I don't have thermal conductivity values for thermal paste - anyone have a k value?

If you're interested, I can provide an example problem from J.P Holman's Heat Transfer with actual values.