Poll: Do heatsinks saturate?

[KF]

>Also note that at some point, any heatsink can become "saturated" with heat
>such that it will "hit the wall". Heatsinks can only get rid of so much heat,
> and if you exceed its limits, it will get very hot - as will the CPU.

This is an idea I don't believe in. Not yet. Anyone
know why or how this could possibly happen. I am assuming the
heatsink is run long enough to reach the steady state
that thermodynamics seems to garantee.

 

[Swanny]

Yes, it will saturate. The thing only has so much surface area and so much air moving past it. There is a point where it just can't get rid of as much heat as it takes on.

 

[MichaelD]

This gets into thermodynamics...deep stuff. Yes, any heatsink can saturate, but it's unlikely you'll saturate a CPU heatsink with normal computer. They are specifically designed to handle the heat a CPU produces. Granted, some HS's do a better job than others.

 

[Skiracer]

Back to thermodynamics 101 for you boys. The controlling equation for heat transfer is Q=U*A*deltaT, where Q is the amount of heat transferred, U is the heat transfer coefficient (includes all three modes of heat transfer where applicable), A is the area available for heat transfer, and deltaT is the difference in temperature between the "hot (heatsink)" and "cold (air)" media. If you look at this equation, you can easliy see that for a HSF with a constant surface area and heat transfer coefficient (which to my knowledge is all of them except where a variable speed fan is in use), if deltaT goes up, Q goes up and vice-versa. So, if a heatsink gets hotter due to an increased heat input from the processor, then deltaT MUST increase in order to maintain thermodynamic equilibrium (the exiting air gets hotter). Conversley, if the deltaT drops, then Q (the amount of heat transferred) drops. In EVERY case where the heat load is relatively constant (like from a CPU), the laws of thermodynamics state that equilibrium will be achieved! If the HSF is grossly undersized, that temperature may be very high, but equilibrium will be achieved none the less. The only caveat is that if things get too hot, then the heat of transition of the materials comes into play (read melted HS!).

 

[goshdarnindie]

Good call Skiracer, you just saved me from having to look through my book. Only problem to consider is Tambient. As the heatsink gets hotter, Tambient increases thus reducing the difference between Tsurface and Tambient, thus Q wouldn't increase (in certain situations). Will this happen, probably not, but it is possible in a real world situation. For instance the compact stacking of IC's in aircraft that have no active cooling (a presentation I saw a few years back just came to mind).

 

[KF]

Thanks for the logical reply. I'm seeing the "saturation"
idea mentioned more and more. One of these days, though, I'm
hoping to hear the logic of it, if any.

>The controlling equation for heat transfer is Q=U*A*deltaT ...

It is my understand that thermal resistance is a realistic
approximation that is accepted as an engineering principle and
works well in practice. Starting with that, one can see there
theoretically is no point were the temperature takes off dramatically.
The temp graph is simply a staight line at a constant slope.

If something gets hot enough (red hot), heat dissapation by radiation
(glowing) becomes more and more significant. But what happens, as I
understand it, is that there is an ADDITIONAL mode of heat removal
and the temp graph slope becomes LESS steep; it does not take off, but
levels off. So the saturation people have got it doubley wrong, if I
understand it right.

I theorize that people who think they are seeing a saturation effect
don't have the heatsink making firm contact. I once tried to make my
own foam bumpers like the Tbird has (for a sink that did not contact
the original bumpers). The temperature took off when I started up
the computer. After a couple of minutes I could smell something
burning. My homemade bumpers did not have quite enough squish to
let the heatsink make good contact. The heatsink grease fried and
turned solid. (Tbird still OCs great though!)
Perhaps people who don't put on enough heat grease can get a saturation
effect as the surfaces deform slightly with a temperature rise.
I experimented with a T-bird style Gorb once. Instead of counting
on the sink clip, I rigged a way to clamp it with an 8 inch
C-clamp, with the mobo sitting on a table.(Insulating pad on
bottom of mobo, wooden stick on sink.) The temp reading
dropped 5 degees F vs no C-clamp. I can imagine that in some
cases Gorb clips barely give contact.