Normal aging or evidence of over-temperature - thermal grease

That stuff dries out over time baked with the heat from the CPU.

The question is "Is the thermal conductivity reduced when the grease dries out?"

Some 30 years ago, in the lab where I worked, we did some comparative tests on the effectiveness of thermal compound. We did the tests on a TO3 transistor bolted to a large heatsink, with a variety of interfaces.

With or without a mica washer, thermal compound seemed to make very little difference. Surprisingly, the best conductivity seemed to be with a thin piece of Izal toilet paper (no mica). However, although we concluded that thermal compound was probably a very viscous form of snake oil, it didn't seem to be consistently worse than when it wasn't used, so we continued to use it in production.

snip...snip

Hello Ian:

The Izal toilet paper you refer to isn't readily available on all continents. However, might we deduce that almost any toilet paper might have done comparably?

In all seriousness, does the possibility exist that thermal compounds, in the mid-1970s, were very inferior to todays offerings?

What you have stated is /very/ interesting.

Thank you.

interfaces.

And my little experiment into modern day snake oil sales, no patent revenue stream on the use of mica compared to new doobries.

From one of my repair brief files

Kustom KPM6160A mixer amp from 1997 ... The heatsink insulating pads had shrunk, probably excessive heat, leaving a crumpling on the uncompressed areas, but all in working order as a pa. Replaced with mica washers. Previously took 50 minutes to stabilise at 33 deg C above ambient pumping 9Vac of 400Hz continuous sine into 4 ohms. With mica replacements, down to 30 minutes and you could keep your fingers on the body of the trannies, too hot previously. The pads may have chemically failed , but even then , mica does not degrade over 10 years. ...

-- Diverse Devices, Southampton, England electronic hints and repair briefs , schematics/manuals list on

I would have said so ! Otherwise there would be little reason to use it in the first place. Certainly in the case of CPU to heat sink conductivity the very thin film thickness of phase change material is superior to the normal white filled silicone compounds. But it does require a specific minimum temperature before it flows and thins out. After that point the thiner the film gets the hotter it has to be before it flows further.

Thermal compounds have come a long way in 30 years.

There is a test rig described on one of the manufacturers web sites that deals with measuring thermal conductivity. It has a number of graphs showing different materials thermal conductivity against thickness and pressure. Bergquist springs to mind.