Wrong Sil pad?

Oh, that's phase-change stuff. It's terrible. It flows just enough to short, but not enough to really get theta down.

Use a proper fiberglas+silicone pad.

Now you tell me. :^)

This looks the best,

But still 2 C/W! (what's the sense of getting a package with low junction to case thermal conductivity.)

George H.

Another possible issue- I recently did an emergency repair on a commercial US-made $$$$$ aircraft-grade rack mount device that had the nut and bolt vibrate off a heatsink in relatively few flight hours (yes, it had toothed lockwashers). I suspect the silicone pad helped it go loose.

--sp

You could go back to the more robust mica and heat transfer paste.

Some high spec applications use ceramic plate insulators.

The fully insulated packages are foolproof - but thermal resistance isn't so good.

Use a bigger package, TO247 maybe, or several TO220s. Bigger thermal footprint.

I like to hard anodize aluminum and then just use grease or epoxy. Much lower theta.

AlN or BeO both conduct heat close to copper. Alumina is terrible.

Oh, there are some internally insulated TO220 transistors around.

Someone should sell heavy anodized aluminum insulating washers/ plates for various packages. Solder them onto the heatsink. (I'm using to-247 so probably better than 2 thermal ohms.)

Re Alumina, right the ~2mm plate was ~4 C/W at 25C. (to-247) that's as bad as the isolated tab to-220's that I over loaded.

Can I buy AlN sheets/ spacers? Then I need a grease too? (Silicone vacuum grease?)

George H.

Oh boy, minimal vibrations I hope. I've been using bellville (sp) washers. I do worry about thermal cycling.

George H.

IIRR heat-sink grease is silicone grease loaded with zinc oxide particles.

I recall discussing this with a bunch of PC gear heads once. They seemed to think there was magic in the heat sink grease that meant they had to use the top shelf, silver impregnated stuff to get good thermal connections. I tried until I was blue in the face to show them that it just didn't matter if they did it right. Properly applied heat sink grease is so thin that it doesn't really matter what you use, the temperature drop is microscopic. Someone proved it by using butter once which worked as well as anything else and made a great tasting heat sink.

Properly applied heat sink grease is very, very thin. The PC gearheads polish the heat sink on glass so it is as flat and smooth as possible. The heat sink grease is applied with the edge of a razor blade to spread and thin it. Set the sink on top of the chip and don't slide it around as that can make the grease ball up under the sink. Pressure is essential to getting the grease into all the pores and squeezing any microscopic air bubbles as small as possible.

How thick do you expect your grease layer to be?

Forget the hole , use a spring clip.

Cheers

Have you tried the oriented carbon fiber pads?

They do tend to be electrically conductive. We used them with Peltier junctions back in 1993, and they worked fine, but George has excluded electrically conductive options.

If you can get the surfaces flat and polished, forget the grease completely.

Not possible, a wide flat washer, then bellevile looks a bit like a spring clip.

George H.

No, in fact extremely flat surfaces are worse than modestly flat surfaces. (Can you hypothesize why?)

An atomically flat surface might cold-weld together, which could be expected to give very good results, assuming it bonds over most of the surface.

Practical "extremely flat" surfaces include gauge blocks, which can be wrung together (a holding force which is probably a combination of dispersion forces and other atomic weirdness). I don't offhand see a reference for the thermal conductivity between wrung gauge blocks, but I imagine it's dismal.

Tim

The contact area between two surfaces that are not atomically flat and are made of rigid material approximately equals the preload force divided by the yield strength.

Thermal conductance at material interfaces is quite poorly understood--the straightforward derivations give wildly wrong answers at low temperatures. It certainly isn't just the sum of the bulk thermal resistances.

Cheers

Phil Hobbs

We aren't talking about low temperatures, here. I usually figure ~100Cish for a heat sink (125C to 150C Tj). I don't see how adding (any) grease helps if the surfaces match. We're talking practical heatsinks, here, not theory.

(any) grease helps if the surfaces match.

Good grease (e.g. Arctic Silver or Shin Etsu stuff) gets up to 2.5 W/m/K, v s. 0.025 for room temperature air. In the thermal resistance approximation, any benefit you get from lapping is 100 times greater with paste. You just have to keep it from wedging the surfaces apart.

Cheers

Phil Hobbs