How big would a heatsink have to be to cool a Duron/T-Bird w/o a fan?

[Muerto]

Just out of curiousity, do you think it's possible to make a heatsink big enough (that would still fit in your case) to cool a Duron/T-Bird by itself? I have access to a machine shop at school and if the material wouldn't be too expensive I'm thinking it might be an interesting project to try.

Even if it is possible though, it would probably only work on the slower CPUs. I doupt a heatsink large enough to cool something in the GHz range would ever fit in any case. Thanks guys!

 

[Rigoletto]

haha, and how would you be supporting this heatsink I wonder? Even modern copper heatsinks are raising worries about the strain on the motherboards.
Maybe your next engineering project should be "mapping fried connections of a 22million transistor semiconductor device with an electron microscope"

 

[erub]

yeah - talk about crushing the core

 

[Muerto]

That's another problem I thought about. It could very easily break off the motherboard. I would probably have to lie the case on its side.

 

[Lord Evermore]

Here's a better project: how big a heatsink can you make and still fit it inside the case, with a fan mounted on top that covers most of the heatsink, and how far can you overclock with that.

 

[Muerto]

Haha, that's probably a more realisitc idea.

 

[miniMUNCH]

Heat sinks must have active cooling to dissipate enough heat from the sink. Heat transfer from the CPU to the HS only occurs a small area (check the actual chip size of the Duron/Athlon).

-Heat transfer is inversely portional to resistance of the medium of transfer.

Basically Q=UA(Th-Tl) where:

Q= heat flux/transfer
U= overall heat transfer coefficient
A= area of heat transfer
Th= higher temperature
Tl= lower temperature

- The over heat transfer coefficient is proportional the thermal conductivity of the transfer medium or transfer surface (e.g. surface between air and sink).

In heat transfer from the chip to the ambient air, heat transfer from the sink to air is the "rate determining factor", so to speak (not to say improvements can't be realized by increasing transfer from chip to sink, especially w/ crappy grease). But there is a limit to heat conduction in the sink itself. If the sink is too big, most of the sink will not be used (will be cool to the touch) because heat is transfering to the air [closer to the heat source] rather than transfering to the outter reaches of the heat sink.

Usually, when sizing a sink for a chip or whatever, there is a certain sink size/configuration at which no further cooling benfits can be realized by increasing the sink size.

The short of that longwinded answer: My 0.02 is that if the chip can't run without active cooling with the current sink size, then the active cooling is probably manditory for operation no matter how big the sink.

But it is possible if you have a real nice heat sink and bond your HS to the chip w/ ArticSilve epoxy and have really good air flow in your case (really good) maybe it'll work.

 

[Lord Evermore]

If the heatsink is larger, then it will be able to absorb more heat before becoming warmer in temperature, and presumably it will have more fins to radiate the heat. So although the rate of transfer to the metal may be the same, more heat can be absorbed and radiated. The greater the temperature difference, the better the heat absorption. So with a small heatsink that heats up sooner, the temperature will be closer to the temp of the processor, and so it won't absorb the heat as well as a huge cooler heatsink.

Or I could just be making stuff up.

But, the G4 Cube gets by with no active cooling, by using a massive heatsink.

 

[Syborg1211]

Heh, all I need is an Alpha PEP66 to cool my P3 without a fan
But anyways, if the heatsink is really as big as you would think it would be, the power supply fan would probably become a fan for the heatsink(if the power supply fan is one of the bootom mounted fans which blows into the case)

 

[Muerto]

miniMUNCH

Nice post. I'm an engineering student so a lot of that actually makes sense! I've seen that equation before, Q = AU(Th-Ti).

So what you're saying is that since the contact area is actually quite small, 100mm^2 on a Duron, it will reach a point where increasing the sink size wont do a thing because the contact area is too small?

Right now I've got an AlphaPAL6035 w/ ArcticSilver compound and four cse fans. My Duron runs at around 40C@600 and 48C@800.

The main reason why I'd like to do this is to try and have a silent o/c'd computer, not this howling monster that I have now.

 

[chiwawa626]

howbout try some alternative cooling like water cooling?
or if you realy want get some kinda giant heatsink straigt from alpha and cut it down to size. that would be cool....might as well put a quiet fan on it while ur at it like a 20dba one

 

[Muerto]

I've thought about water cooling. But the pump would probably make just as much noise as my computer does now. Modifying an existing heatsink would be cool, but I want to design my own just for fun.

Adding a quiet fan is probably a good idea. If I do build a heatsink I'll definitely want some active cooling standing by!