> After years of development ;-) I've finally have my prototype robotic
lawnmower
> platform to a usable state. The most valuable lesson I learned is that
> everything in the drive path must be metal, clamped, and use lock washers.
>
> Next on the agenda is motor control for the drive motors. Bob Blick's
> discrete h-bridge is a natural for the project. But of course you end up
> with 8 power transistors, plus another one for the cutting motor.
>
> I wanted to get some ideas of the best way to heatsink a bunch of power
> transistors.

If power dissipation is significant you want to really DESIGN your heat
sinking system. How hard this is depends on how close to their limits you
want to run your transistors. What power dissipation are you expecting? What
devices are you using? (part numbers or required ratings).

There are 3 thermal sections to the heat path.

1 - die to case
2 - case to heat sink
3 - heat sink to air

If you have lots of metal for heatsink you will be able to get the HS down
under 1 C/W (1 degree Celsius rise per watt dissipated).

If you use electrically isolated packages they can be bolted directly to the
HS thereby eliminating path 2 above BUT the inbuilt insulation has extra
drop. look carefully at data sheets. This extra drop is typically around 1
C/W. Use grease.

Mica washers require thermal grease because the washer is relatively rigid
and does not make good content with transistor or heatsink. If the grease
dries out with time you can get an increase in thermal resistance. AFAIR
without looking it up you get around 1 C/W for the washer in a TO220
package. .

Modern silicon rubber washers do not need grease and also run at about 1 C/W
for a TO220 package. Mechanically larger packages (eg TO247) have a better
C/W rating all things being equal.

The transistor's internal thermal drop (1 above) varies markedly with
devices, even in the same package. This can be well under 1 C/W up to
several C/W so be careful. Also note that SOME manufacturers spec thermal
resistance as W/C (inverse) so don't be fooled.

You can get proprietary thermally conductive gasket materials with much
better thermal resistances. These are often dear but may be worthwhile in
extreme cases.

Mounting to heatsinks with silicon rubber or similar is typically by spring
clips. Ensure that adequate tension will ALWAYS be present across lifetime.
Loss of tension with age can lead to higher thermal drop. Maximum clamping
are specified by manufacturers but too low a force is liable to be more of a
problem. Avoid point contact of clamping clips as this can lead to internal
failures.

If you can design the system so that the heatsink is at the same potential
as the transistors tabs then no insulation is needed and direct mounting
will give better thermal results. Use grease here!

If the heatsink has a thinner portion between the transistor contact points
and the main HS body or the heat path is mechanically long beware
temperature rise across this section.

You could consider fluid heat transfer or heat pipes but that is probably
getting excessively complicated.



        Russell McMahon

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu