Developing proper heat sink

[ImpInTraining]

I've got a module that has an amplifier in it that creates a good deal of heat. The base area of the is a minimum of 4 square inches. I've been told the thermal load is 17 Watts. I am limited in space above and below the module, so I need to put the heat sink on the side, but the amplifier will not fit height-wise if I mount it sideways in order to connect with the heat sink.

Is it ok to have the heat transferrance work through the module casing? The amp will be located close to the heat sink, but the heat-generating side won't be directly overtop of the heat sink. My theory is that the heat will dissipate through the case and to the heat sink, since it will all be one piece anyway. Would that work? I'll include an image of what I'm suggesting.

Also, how do I determine the correct number and size of fins required to dissipate the heat?

 

[fegenbush]

Your idea of transferring the heat through the case would only be acceptable if the case were made of a highly conductive material (i.e. copper). The designs I have seen that were most successful in tight spaces were those that used an integrated fan similar to a CPU cooling fan on a PC.

If money were not an issue, the use of a heat pipe design would be interesting as well.

Most heat transfer texts will have a section on the design of convective fins. I personally use Heat Transfer by Nellis

http://www.amazon.com/Heat-Transfer-Gregory-Nellis/dp/0521881072

Fegenbush

 

[IRstuff]

there are several heat sink calculators online.

However, you need to determine:
> How hot are you allowing the amplifier to get?
> What is the ambient temperature?
> Is there forced air?


For non-forced air with a 50°C temperature delta, you'd need about 210 sq. in. of useable area.


TTFN

FAQ731-376

 

[ImpInTraining]

These modules are not small, really... about the size of a laptop. There will be 16 of them mounted horizontally atop of each other with some sort of mounting shelves (which haven't been designed yet)so figure there will be a fraction of an inch between modules. The heat sink is going to be on the side, allowing to take the best advantage of air flow through the rack. Air in the facility is distributed via several air handlers running about 65 degrees ambient... forced up through the floor through the rack, and sucked up by the fans atop of the rack.

The casing will be aluminum... and we are in contact with metal benders who can make the casing to include the heat sink if we decide that's the way to go. Certainly seems more plausible than heat pipes since we expect some of these modules to hopefully last a decade or two without failure and a leaky heat pipe might cause health risk and damage to the electronics.

If necessary, we could designate a portion of the case to be made of copper, then connected to the rest of the aluminum casing ... just to help with this heat transference issue. Would that be a worthwhile suggestion?

 

[ImpInTraining]

Make that about 20 degrees Celcius. I think Celcius is used for the thermal formulas.

 

[fegenbush]

ImpInTraining,

Forced air will help quite a bit, and, since aluminum is a cood construction material for heatsinks, you should be ok going through an all aluminum case.

Things to take into consideration:
-Don't anodize the aluminum. There's no need to make things any harder on yourself. The anodize coating may have a negative effect on your heat transfer through the film.
-Using some thermal paste never hurts. Don't bother with the expensive stuff for this application. Remember, a little goes a long way, apply a very thin coat.
-Whatever design of heatsink you use, be sure to orient the heatsink in the proper direction with respect to airflow. I have had assembly guys say that "it doesn't matter." If it didn't matter, I would have designed it the other way.

Best of luck,

Fegenbush

 

[ImpInTraining]

I read somewhere that there is a black coating you can put over aluminum to help with the heat conductance because it is supposedly more conductive to infra-red. But if you suggest not-annodizing, well... I'm easy. Thanks for the info.

 

[fegenbush]

ImpInTraining,

I believe you may be referencing radiative heat transfer, which indeed would benefit from a black surface treatment. For your application, I believe the radiative heat transfer to be very small in comparison to the convective.

As far as conductivity goes, Al2O3 has a k value of about 30 and Al about 250. Increasing the thickness of the oxidation layer can only hurt you (and increase your cost).

Thank you for the most interesting topic,

Fegenbush

 

[ImpInTraining]

It's been my pleasure. Thanks for the insight. I've been trying to understand those formulas provided elsewhere on the web, and have come to the conclusion that you really do have to have a degree in thermal engineering to deal with thermal dynamics.

For the life of me, I cannot figure out the correlation between thermal resistance and a thermal load value (of 17W in this case). How much thermal resistance do I look for in heat sink to counter the heat load provided?

I've done an experiment recently which was sort of insightful, placing an amplifier (which gets up to about 50ºC) on an aluminum (black anodized) metal, heat sink next to it, separated both of them by another piece of metal, and set up a fan toward the heat sink - of course using the thermal bonding agent applied in a thin layer. The amplifier only got to about 34ºC. Curiously, it seemed the metal further away from the amplifier was nearly as hot as the amplifier, but lost about half a degree for every inch in distance.

For my purposes, it is a worst case example, I know. But it tells me that the inside of the case shouldn't get too hot to worry about provided we have adequate heat sink in place.