### Author Topic: Heat transfer  (Read 1252 times)

0 Members and 1 Guest are viewing this topic.

#### kr15_uk

• Contributor
• Posts: 6
• Country:
##### Heat transfer
« on: July 26, 2016, 09:53:09 pm »

Quite odd question.
I'm working on a project where everything is very squished and my concern is the heat transfer.
Is there a way to figure out roughly what is the heat transfer over the air per mm without diving into heavy math.
For example let's say I have heat sink with surface temperature 100 degree C and next component is 10mm apart. How much heat from the heat sink it'll get.
Is there some simple figure like heat loss of 10degree per mm.
I know it's not as simple and you need to do proper math but for the time being quick rule of thumb would be good start.

Regards,
Kris M

#### Dave

• Super Contributor
• Posts: 1213
• Country:
• I like to measure things.
##### Re: Heat transfer
« Reply #1 on: July 26, 2016, 10:16:27 pm »
Nope, there is no such ballpark figure you could use. It all comes down to radiated and conducted heat.

Radiated heat - direct heat transfer through infrared light. You could reduce this effect by making the surface of the heatsink shiny (thus reducing its emissivity).
Conducted heat - heatsink heating the adjacent air which in term heats the components in question. Some airflow through the enclosure would drastically reduce this.

There is no easy way to estimate this on the back of an envelope, unfortunately. I think your best bet would be to just build the whole thing and measure.
<fellbuendel> it's arduino, you're not supposed to know anything about what you're doing
<fellbuendel> if you knew, you wouldn't be using it

#### Fulcrum

• Regular Contributor
• Posts: 111
• Country:
##### Re: Heat transfer
« Reply #2 on: July 26, 2016, 10:43:41 pm »
From a 100 degree C heatsink the air will move upwards from the heatsink, so as long as this hot air has somewhere to go and is not enclosed in a small space, you can likely neglect transfer by convection (i.e. heatsink warms air, air warms component). That leaves radiative transfer, i.e. transfer by infrared radiation from the heatsink to component. This can be hard to estimate, but depends on the emission coefficient of the heatsink (usually they are black for high coefficients), the absorption coefficient of the component, and how large solid angle the heatsink spans from the point of view of the component. This source of heat energy will be balanced by the loss of heat from convection (component to air) and conduction (component leads to PCB), and also radiation. Since a low absorption coefficient of the component means less energy absorbed from the heatsink radation, you'll want the component to be metallic (best case) or at least white. White materials and metals have absorption coefficients close to 0. One thing to note is that materials with low absorption coefficients also have equally low emission coefficients, so a highly reflective material (white or metallic) will emit very little of their energy as radiation.
All in all, this is a pain in the ass to calculate! I just wanted to give you a physicists point of view of the process. My recommendation is to make sure the component is far away enough to not be affected by convection (a bad idea would be to mount it above the heatsink, in the path of the hot air - which flows upwards), and try to pick components that are reflective to minimize absorption by IR radiation from the heatsink. If that's not an option, put a metal shield inbetween the component and the heatsink to shield the IR radiation. Or, make the part of the heatsink "seen" by your component reflective. Since high reflectivity means low absorptivity, and low absorptivity means low emissivity. I.e., it will emit less of that pesky IR radiation.

Please bear in mind that this post was just based on logical deductions. If someone has real world experience, listen to them.

EDIT: One thing I didn't think of was heat transfer through the board. If the heatsink is 100 degrees, the board will probably be around the same figure, yes? At least nearby the warm object that is warming the heatsink. So likely a fair bit of the heat supplied to your component will come by conduction through the PCB or some power or ground plane.
« Last Edit: July 26, 2016, 10:47:49 pm by Fulcrum »

• Super Contributor
• Posts: 1875
• Country:
• Reactor Operator SSN-583, Retired EE
##### Re: Heat transfer
« Reply #3 on: July 27, 2016, 12:13:35 am »
Love Cypress PSOC, ATTiny, Bit Slice, OpAmps, Oscilloscopes, and Analog Gurus like Pease, Miller, Widlar, Dobkin, obsessed with being an engineer

#### Brumby

• Supporter
• Posts: 9483
• Country:
##### Re: Heat transfer
« Reply #4 on: July 27, 2016, 12:51:52 am »
Heat travels via
- Conduction
- Convection
... and you need to consider all 3.

A heat shield can deal with radiative energy - but just make sure that it doesn't get too hot in itself.  High reflectivity will just bounce heat back to the original source, making it hotter.  Lower reflectivity (higher absorption) means you will have another body to remove heat from.

Conduction is the most controllable, but it also relies on physical properties such as size, shape and thermal conductivity.  It can be useful if you can get the heat to travel to a spot where radiation and/or convection can be utilised.  Heat pipes are effectively a conduction solution (but, yes, internally they are a bit more than that).

Convection is, perhaps, the biggest factor in the final removal of heat from a system - with forced 'convection' (ie a fan) being extremely common.  One important consideration, however, is the path taken by the hot air in exiting the device.  You really DON'T want it passing over any other components.  Air can take a moderately long path in and even from multiple points - the key is that sufficient volume must be able to flow freely.  However, hot air needs to take the shortest path out.  Here is where the physical layout of the heat generating zones is important.

Modelling heat transfer in a heatsink is easy.  Simple maths using published data.  Modelling heat transfer in a completed product, or even a board, is quite a different story.  The relevant parameters are exceptionally complex and will depend on countless subtleties - including orientation of the board in question.  While experience and looking at other solutions can point you in useful directions, experimental assessment is really the only way to get hard data.

Smf