In article , N_Cook
wrote:

On 08/03/2017 05:46, isw wrote:
In article , N_Cook
wrote:

On 06/03/2017 21:56, Gareth Magennis wrote:
This is a bit weird.

I have an audio power amplifier that uses what looks like strips of pink
silicone sheet between the output transistors and the heatsink.
But these strips have degenerated and are kind of squidgy, a bit like
when Speaker surrounds fall apart.
Not come across this before.

They have got so bad that an arc occurred through one strip, beneath one
of the transistors, which has burnt a small hole in the heatsink,
blowing the amp and the SMPS, which I repaired.


So what kind of material could this pad be made of? Surely silicone
wouldn't do this?



This is the amp, you wouldn't expect this kind of problem on something
like this.

http://www.dbaudio.com/en/systems/de...amplifier.html



Cheers,


Gareth.

Stay with mica slivers, been around for geological time, no plasticer to
leach out. The only time I've come across failure of a mica insulator
was when some swarf got under it at manufacture, that would do for all
other insulators as well.
Trouble with mica is it is dirt cheap and not patentable.

Also somewhat poorer thermal performance ...

Isaac


If you mean mica is poorer, don't go by product data sheets.
I did a realworld expt one time , comparing mica and sillypads on an
amplifier. Running with fixed input and controls and load, and
monitoring the temp of the heasink and IR thermo of devices. The mica'd
version ran with noticably hotter heatsink and cooler devices.
IIRC the datasheets quote an (fraudulent?) impossible degree of
squashing of pads, not obtainable with one screw thru a tab.

If your claim is that the mica version "ran with noticably hotter
heatsink and cooler devices", then something went wrong with your
measuring technique. It's a sort of requirement that heat flows from
warmer to cooler, and so the devices necessarily must be warmer than the
heatsink, no matter what (passive) material you use in the interface.

A better-performing thermal interface would be one which *minimized* the
difference in temperature between the device and the heatsink.

Also from your description, it sounds like you were measuring the
performance of a specific mounting method, rather than the thermal
resistivity of one of its components.

Isaac