Washers - electrically insulating, thermally conductive

[mark_in_manc]

Hi folks

I'm working on an off-topic-for-this-forum device which includes a big diode and a transistor, which need to be in contact with a heat sink yet electrically insulated from it. I remember see-through washers on stuff I've played with before, sometimes smothered in a kind of white (thermally-conductive?) paste.

I wonder if anyone can suggest any suitable materials to make the (thin) washers out of. It's (whisper it) automotive, so voltages involved are low.

Thanks
Mark

[turretslug]

It's possible to buy thin, soft (i.e. compliant) thermally-conductive, highly electrically-insulating sheeting that doesn't need thermal grease, or bespoke patches for particular device package types. They're said to be more effective than olde-worlde mica washers anyway and less messy than white goo!

If it's a stud-mounting diode that needs insulating, I never trusted the small contact area- I've used stainless-steel penny washers to increase the area in contact with the soft material and additionally reduce the risk of bite-through,

Colin

[McMurdo]

here's RS's offerings

https://uk.rs-online.com/web/c/?sra=...it+transistor+

[mark_in_manc]

Reading elsewhere, I see Kapton tape suggested. What do folks think? I should have been more clear - a flat surface on the big diode, and the transistor, are pushed against a heatsink by the standoff on the pcb mounting - they don't screw to it directly as one might normally imagine, so normal fixing kits with correct hole spacing for the legs on a given package are not quite so necessary.

[Herald1360]

Originally Posted by turretslug

If it's a stud-mounting diode that needs insulating, I never trusted the small contact area- I've used stainless-steel penny washers to increase the area in contact with the soft material and additionally reduce the risk of bite-through,

Stainless steel is a poor heat conductor- using even a larger area washer probably makes thermal performance worse. Copper, brass, aluminium or plain steel would be better choices.

[Mark_RR]

Originally Posted by mark_in_manc

Reading elsewhere, I see Kapton tape suggested. What do folks think? I should have been more clear - a flat surface on the big diode, and the transistor, are pushed against a heatsink by the standoff on the pcb mounting - they don't screw to it directly as one might normally imagine, so normal fixing kits with correct hole spacing for the legs on a given package are not quite so necessary.

I don't think Kapton is a good thermal conductor. The thermal conductivity is 0.12 W/(m K) according to a duPont data sheet I have downloaded. That's pretty crap really. However, it can be made thin whilst still maintaining good electrical isolation and good mechanical properties and that will reduce the temperature differential. You can buy it coated with 'thermally conductive' glue which helps to improve the thermal contact and that will help the heat transport a lot. I expect the term 'thermally conductive' is a bit relative though; it is not likely to be as good as even a poorly conducting metal like stainless steel. It will be the thinness of the material that will give the effective heat transport.

It is hard to find good thermal conductors that are electrical insulators because so much of the thermal conductivity is due to the electron mobility that gives the electrical conductivity. The best material that I know is sapphire which is an excellent electrical insulator and a genuinely good conductor of heat. It is not as good as a metal such as copper or aluminium of course, but a thermal conductivity of around 25 W/(m K) is pretty damn good for an electrical insulator. I have had made (for work) sapphire shims with bolt holes cut in them for cryostats. They were made by Crystran. You will probably find the cost prohibitive though.

[McMurdo]

Maybe some sil-pads might be better then? https://www.rapidonline.com/Bergquis...yABEgKlEfD_BwE

AFAIK mica is better than silicone impregnated pads but there you go.

[mark_in_manc]

Thanks Kevin, helpful suggestions. Maybe some mica shims will do me then - I can try to find some without holes in, and attach them to the heatsink with a dob of vaseline while I locate everything.

[turretslug]

Originally Posted by Herald1360

Stainless steel is a poor heat conductor- using even a larger area washer probably makes thermal performance worse. Copper, brass, aluminium or plain steel would be better choices.

D'oh! Should have said "plated mild", I'm too stingey to use stainless too often for things. I've sometimes wondered about using "readily available copper alloy washers", i.e. those older 1p and 2p bits that don't respond to a magnet ground flat, but I worry that mis-using the currency of the realm might still be a capital offence....

[Radio Wrangler]

HP used to use heavily anodised aluminium washers to insulate stud rectifiers. The washers had a turned-up lip around their hole so that they would locate themselves and the diode stud in the hole in the heatsink.

There were also versions for TO3 etc transistors.

Mica washers were quite good but not as good as the anodised ones. Thermal compound (the wakefield white stuff) is appreciably better than plain silicone grease. But it's messy.

David

[trobbins]

Imho, just buy a sheet of SilPad, or pads large enough to suit your devices.

I guess you have done some thermal analysis and identified the max junction temperature based on expected max dissipation and thermal resistances (of which the SilPad will now be identifiable, although you may need to do some calculation to work out what is achieved with a large diode flat base). If you haven't done an analysis, then that is really needed for starters, and will include your heatsinking performance and ambient environment. You can deduce diode dissipation for large devices by their turn on voltage and high current resistance parameters - most manufacturers have design tutorials/papers for that.