4
\$\begingroup\$

i am trying to design an efficient cooling solution for my project. I planned placing a IoT computer in an aluminum housing which is completely closed.

In this case, do I really need fins and large surface areas INSIDE the aluminium housing when my main goal is to dissipate the heat to the housing itself as fast as possible?

I'd really appreciate your help!

Here is some more information:

Ambient temperature: 25-30 degrees celsius

Max. Power: 10W

Time Period: The device is intended to work 24/7 (of course NOT with max. Power but with 20-30% of what's possible. However I think that a great product is designed to also handle 100% Power over a vast period of time!

Internal temperature: The CPU itself (surface area of 10mm x 10mm) reaches a max of 80 degrees.

Surface area of the enclosure: 0,05735 m2

Enclosure material: Aluminium sheet metal (1mm thickness)

Space between CPU and inside of the enclosure: 14mm

Maximum Volume of the Heatsink: 60cm3

\$\endgroup\$
  • 1
    \$\begingroup\$ Ah, you might like to read my answer to the following Rpi.SE question: "Does this flat heat sink even work for cooling?": raspberrypi.stackexchange.com/questions/106559/…. Cheers.. \$\endgroup\$ – tlfong01 Aug 14 at 8:11
  • 1
    \$\begingroup\$ @tlfong01 thanks for your comment! Your Rpi.SE question was very helpful for my own problem! Still - I am curious on how to make this whole heat dissipation thingy most efficient (Using fins or solid blocks in closed containers without any airflow) \$\endgroup\$ – Jonas Jossen Aug 14 at 8:21
  • 2
    \$\begingroup\$ I think a real answer to your problem depends very heavily on the power which you didn't state. Is your IoT computer a small thing which uses just 5 W or is it really just a PC in a small enclosure using 65 W? The first might be okay with just some help with internal convection, the latter probably needs heatpipes and very good thermal contact on both ends. \$\endgroup\$ – Arsenal Aug 14 at 8:58
  • 1
    \$\begingroup\$ @Jonas Jossen, you are welcome. I think your question is very good to get so many comprehensive answers with engineering trade offs (eg instantaneous vs long term power dissipation, max temperature, space etc), and cost benefit (weight, space etc) analysis. I am glad to have learnt much in engineering design. Cheers. \$\endgroup\$ – tlfong01 Aug 14 at 9:22
  • 2
    \$\begingroup\$ Use conduction to convey heat from source to casing. Use external heatsink to convey heat from case to room. You will have a vastly better chance of a good answer if you tell use power dissipation, profile with time if any. Room temperature and target device case temperature if there is one, case dimensions. And EVERYTHING else you know t hat has even the slightest chance of being relevant. || Accepting an anser almost immediately is unwise and may discourage better ones. Without them knowing your answers to my above points you are robbing yuourself of possible better answers. \$\endgroup\$ – Russell McMahon Aug 14 at 9:23
6
\$\begingroup\$

There are two power functions you need to be concerned with: instantaneous power dissipation, and long term power dissipation. A large thermal mass (finned or not) will prevent your IC overheating due to a quick spike of power. Over the longer term, the heat needs to be dissipated to the outside environment. Fins are only effective if there is some airflow to remove the heat - this could be by simple convection or by forced flow (fan). In a sealed case, convection will be reduced compared to free air and a fan will also be much less effective.

For your case, if you don't get sufficient cooling from just normal internal convection, I would have a metal block from the top of the heat dissipating commponents to the case itself. That way, you increase the thermal conductance to the case. If you still have a problem with heating, then you could put fins on the outside of the case, which is presumably in free air.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ Heatpipe might be better than a solid block. \$\endgroup\$ – Mołot Aug 14 at 19:40
  • \$\begingroup\$ @Mołot this is definitely worth a thought, but I guess it might be harder to manufacture. You might need a more complex part to both contain heat pipes and blocks... Also, do you think it is really necessary to use heat pipes in a 10W max device? \$\endgroup\$ – Jonas Jossen Aug 19 at 7:39
  • 1
    \$\begingroup\$ @JonasJossen Don't know, can't calculate, but I know that my 45W CPU has 6 of them and a big radiator, and still needed to turn on fan yesterday because passive wasn't enough. So I think that if someone can do calculations or testing for a reasonable cost, it might be worth to see if it would help, and is it worth it. \$\endgroup\$ – Mołot Aug 19 at 8:52
  • 1
    \$\begingroup\$ @Mołot In the PC world, heatpipes are as much a trend as anything, and it depends if they're actually properly designed/implemented rather than just for show. Last CPU I used was ~75W TDP, no heatpipes in sight, vintage 2008. 10W is not much, especially if you've got a decent sized metal case to dissipate the heat from. A metal shim will be much cheaper and easier to fit than pipes. \$\endgroup\$ – awjlogan Aug 19 at 9:06
2
\$\begingroup\$

It depends on a few factors: what’s the maximum temp of the CPU die? What’s the surface area of the enclosure? What is it made from? What is the distance between the CPU and the surface you want to reach with this copper block? What is the ambient temperature? Do you expect solar heating?

Off the top of my head I’d say no to the copper block for two reasons: weight - even a modestly sized block of copper is going to be heavy, which will put strain on the heat sink mounting tabs - and what to do with the heat once you get it to the exterior. If it’s low power, then maybe the enclosure wall can dissipate enough heat from the area around the copper block, but if it’s high power, then you just bought yourself into putting a heatsink on the exterior of your enclosure. However, if you have a heatsink on the CPU and reasonable circulation on the inside of the enclosure (from, say the fan on the heatsink), then the entire area of the enclosure wall becomes your heat transfer area. It should be fairly easy to get a back of the envelope answer to how much heat will be transferred out of the enclosure if you know the ambient temp, the max internal temp, and the enclosure thickness, surface area, and material.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Thanks @vir ! You definitely got a point with the weight of copper! For a 10 Watt device and the CPU going 80 degrees (Celsius) max over a small area of 10x10mm. Do you think an identical block of aluminium would be enough? Here's some data: Surface area of the enclosure: 0,05735 m2, enclosure wall thickness: 1mm, Ambient temp: 25-30 degrees. max internal temp: 40-50 degrees (fairly depending on the heat dissipation of the "cooling block") \$\endgroup\$ – Jonas Jossen Aug 14 at 9:22
  • 1
    \$\begingroup\$ Copper is 83% more heat conductive than aluminum. \$\endgroup\$ – user105652 Aug 14 at 20:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.