silicone grease is messy

It's the re-education camp for you man, no possible way out. 'K' for 'kilo'? Mais non! Quelle idée!

Which it does, of course. :(

What are the ambiguities of Imperial units? (I mean, besides whose empire we're referring to.)

Surely there's no doubt as to the definition of acceleration in rod-seconds per cubic fortnight?

Cheers

Phil Hobbs

(Back when Canada began switching over to metric in the '70s, a local radio station in Vancouver (CKLG) spent one April 1st announcing the time in "hours Celsius".)

Not to me. We untangle possible unit ambiguities with context. If I know that someone is talking about specific thermal conductivity, I know what w/mk means regardless of capitalization. And the numerical value is a sanity check.

I'm just an engineer. Nobody peer reviews my work for journals. I just need to build things that work.

That sounds like W/(m/K) to me.

I often compute things in LT Spice, with a BV block or something so I can probe and graph it. The result is always volts, I guess. Spice is refreshingly units-free.

Ah! Of course. Thank you.

Try reheating the bond +30C above the curing temperature and see if you are able to rework it then.

If that doesn't work, just grind away the broken transistor with a diamond burr in a Dremel tool. I have used that method many times when trying to remove components from a pcb with massive ground planes. If necessary clean up with low-melting point solder. For example:

I would lap the surface to submicron fit. The slightest pressure would be as good as any kind of gue, glue or whatever. This works, given that AlN is not conductive.

Groetjes Albert

The AlN that we buy is ground mirror flat, and I assume the mosfets are pretty flat too.

I'd still prefer some sort of filler, some nongreasy grease. One of the repairable loctites might be good.

There will be six fets on one big insulator, on a copper cooler. I'll do the fets different ways and thermal image and compare. No goo at all would sure be nice.