Thermal transfer

[bojan]

Here at work I had a discussion with a colleague about heat sinks..


IMHO, the heatsink should be:

1) black (better radiation)
2) smooth (better convection by forced air circulation - no "sticky" layer of air which is basically insulator)
3) it should have fins spaced enough to allow increased air circulation (decreased lamination of layers of air) and large contact surface.

Point 3) is not disputed... but 1) and 2) are questioned.
Anyone has some links which will help me to convince him I am right (or other way around)?

BTW, this discussion may contain another example of counter-intuitive facts in science....

[DavidU]

I agree with you Bojan. As long as it's black anodized finish as black paint can insulate it.
Aluminium is a reasonable heat transfer metal and not too expensive.

[avandonk]

The heatsink should be made of copper!

End of argument. Only diamond has a better heat conductivity.

Diamond has four times the heat conductivity of copper.

Silver is a bit better than copper.

Bert

[bojan]

Bert, that's OK..
but how to remove the heat from heatsink in most efficient way?
Or, lets say we have couple of identical pieces of metal at some (high) temperature, and all of them are in the same airstream.
One is black, another is shiny mirror-like surface,
Third one is black, rough surface...
Then we have a variety with fins - one have large fins, another one has fine fins (1mm apart for example)...
and so on - you've got the idea, we have all possible combinations

Which one will cool down first?

Or,
- shiny surface versus black: which one radiates more ? (i said black..)
- rough surface versus smooth: which one looses heat faster by convection? (i say smooth)
- big fins versus fine? (i say big, )

[avandonk]

The one with the highest conductivity!

Radiation loss is neglible.

Some people think that an aluminium heat sink is better than copper because the fins are cooler. WRONG because the copper heatsink has better conductivity. The fins paradoxaly are hotter because they are conducting more heat from the source.

It is like claiming that polystyrene is a good heat sink because it is so cool!

Bert

[bojan]

Yes..
Unfortunately, we have to use aluminium (lower cost).
So, we have only shape to play with, to maximise heat transfer to air .

[avandonk]

Bojan the heat pipes in modern CPU coolers are better than diamond as they rely on phase change for heat transfer.

Bert

[sheeny]

My heat transfer texts are at work and its been a while since uni, so I won't stick my neck out too far at this stage:P...

You are right, black is better, but it needs to be black in the IR band. What is black in the IR band is not necessarily black in the visible band and vice versa. There are many black paints with are not black in the IR band. Correct me if I'm wrong, but even though rutile is black, TiO is a white paint pigment, but it is black in the IR band. Most metals are pretty black in the IR band if my memory serves me correctly...:P aluminium is one that I think is quite black in the IR band. I won't comment about others... I just don't remember:).

The surface roughness I'm not sure about, but I would expect it depends on the rate of air flow across the surface. Low velocities is going to be laminar flow anyway (unforced convection and low temperature difference), but I would expect that a rough surface may give an advantage in tripping turbulent flow (as proven with boats in water) at a lower temperature/air velocity thus improving heat transfer. Turbulence is good for heat transfer, but you need to get the temperature differential and/or the air flow high enough to trigger the turbulence. Reynolds number and all that.

Not definitive, I know, but its worth 2c!:D

Al.

[sheeny]

Yes. Radiation will be negligible. Conduction and convection will rule till you get to a few hundred degrees C.

Al.

[sheeny]

Its not that simple...

There are design methods and calculations you can do design fins. The fin has to be thick enough to carry the amount of heat you need it to, long enough to get rid of the heat to the air but not long enough it stops the air getting away, etc. Fin orientation is important too, as well as natural or forced convection.:shrug: Its a bit like which is the best spoon: a tea spoon or a soup ladle? It depends on the circumstances.

:)

Al.

[Jeff]

Interesting questions, especially given the diversity of CPU heat sinks we see on the market. Can you advise the intended application of the heat sink, and whether flow (assume airflow) is passive of forced?

The Wikipedia article on heat sinks is not bad.
Have also looked a reference 3, but not the others.
http://en.wikipedia.org/wiki/Heat_sink

Jeff

[jenchris]

There's a problem being overlooked here.
If the heat sink is in a high radiation environ and it's black, it's going to ABSORB heat from outside - theidea is to shed heat - it can't if it is in the process of absorbing it - so depending that the environment is not radiating - IR black should be better - otherwise it should be reflective - and silver (on board ship (1976) the gun predicter had silver heat sinks all over it (the room was 5 foot square and had the same computing power as a hand held Texas Instrument Calculator!! The tech used to check the predictions with his calculator)) .
As Jeff above said - so much depends on airflow - which has a pretty poor thermal capacity. I plumb for liquid cooled.

[bojan]

OK, let me summarise:

So far we have:
1) must be black in IR, but radiation is not significant mechanism of heat dissipation into environment......

2) smooth surface vs rough surface - still a bit controversial - it depends on air flow speed ? (micro-turbulences, laminar effects etc... )

3) contact surface with cooling fluid (air in our example) must be as large as possible (that was not disputed anyway).

[DavidU]

There is a double edged sword here, fins with smooth surface has a smoother air flow however a textured finish has more surface area.
Some calculations are in order.........

[mswhin63]

Most heatsinks are installed inside anyway, outside heatsink are usually oversized mounted to location cases and painted pretty colours to match the location.

Black absorb light and IR heat, not as much residual non visible heat. Designing heatsink sometime involves placing a fan to assist cooling.

[bojan]

If texture is fine enough, it becomes negligible as it does not contribute to the effective contact-with-air surface...the question is the size (granulation?) of the texture when it becomes cooling fin?

[DavidU]

Bojan, this is what I have come up with.
http://www.mhtlab.uwaterloo.ca/pdf_papers/mhtl07-1.pdf

[bojan]

Bingo !! :thumbsup:

This is very close to what I was after.. Thank you mate!