Re: heat sink cooling

[Lee Hart <leeahart@earthlink.net>]

Mark Hanson wrote:
> Take 3 equally sized heat sinks, bare alum, one black and one white,
> put a 50 ohm resistor on each and apply 12 volts. The bare aluminum
> heat sink will have the lowest thermocouple reading disproving the
> "black body radiator" theory. A black surface only helps when absorbing
> heat from the sun.

Ron wrote:
> I have done the above many times. In still air black anodizing is
> noticably better. Polished aluminum is the worst.

I've found this to be a tricky problem. Semiconductor heatsinks don't
run much above ambient temperature (if they're any good!). With small
temperature differences, your data gets very "noisy". In other words,
you can't tell if changing the heatsink size, shape, color, material,
orientation, etc. has any real effect unless the change is large. Small
changes get lost in the noise of normal unit-to-unit variations.

Some years ago, we had a product with an exposed heatsink that ran a
little over 130 deg.F, which UL does not like. I had an engineering
intern work on the problem. He bought several brands of heatsinks,
aluminum and copper, bare, painted, and anodized, different shapes, etc.
-- whatever would fit in the available space. He methodically tested
them all, and presented his results in a thick report. The one he
recommended ran 10 deg.F cooler.

I sighed, and had him test 10 production units off the assembly line,
all built identically on the same day. There was a +/-15 deg.F
difference between them. Even testing the same unit 10 times gave +/-5
deg.F differences. So what did we do? Nothing; we just gave UL a
different unit to test, and that one passed.

In my experience, the things that *do* make a difference in heatsinks
are surface area, and airflow. Everything else; color, shape, material,
etc. only makes a small difference, which won't be detectable unless the
change is *large*.
--
Lee A. Hart                     Ring the bells that still can ring
4209 France Ave. N.             Forget your perfect offering
Robbinsdale, MN 55422 USA       There is a crack in everything
(612) 533-3226                  That's how the light gets in
<leeahart@earthlink.net>                Leonard Cohen