Noozer wrote:
1) If the compound is not too viscous, and the heatsink is clamped on
with some force, and you apply enough compound, the thickness of the
layer of compound does not depend on the amount applied, which is to say
that the excess gets squeezed out.

2) Thermal compounds are not electrically conductive.

3) Some thermal compounds (notably those containing silver compounds)
have capacitative properties that can be problemmatic if compound gets
between the chip's leads.

4) If you use a compound that doesn't have the problemmatic capacitative
properties and is not too viscous, in general it won't hurt to use too

much.


1) WRONG
2) WRONG
3) If 3 is true then 2) is wrong. You've just proved that yourself.
4) WRONG.



Please start with this: In what way does 3 contradict 2?


How can you have capacitive properties if a compound is not conductive?

1) Wrong becausewhen you use too much compound, pressure will not squeeze
out to the same amount as if you had used the correct amount of compound to
start with.

I'me no AMD expert but i think basic physics will tell you that even
with a high friction coeficient between compound and material, a high
preassure like that will spread the relativley non-viscous compound to
its optimal value, PROVIDED that there are no air bubles or other
colloidal elements.

2) EVERYTHING is electrically conductive to some degree. Even glass and
rubber are conductive in the right conditions.

Thats a pretty simplified view. I think your mixing up conductivity
with dielectric. Electrons can stream between anode and cathode with
resistance but "conductivity" comes from fermi-dirac effects within the
metal (not to be confused with the fermi level). Basically the easy
ionization of metals is what causes its high conductivity (because there
are electrons so close to the fermi level). You might get quantum
tunelling effects but thats another story.

In fact if you put a high enough charge through a molecule like, for
example a protein, you will destroy the protein's bindings before you
end up getting any real conductivity. It works like this. For an
electron do steam through a material, thats dielectric. For an electron
to conduct through a metal by going through orbits, thats conductivity.

And i think your ignoring that the minute, minute, minute conductivity
of the thermal compound might not be even tnough to short out anything
because remember the components do have a threshold value. I wouldn't
worry about conductivity. If it was soo easily conductive, wouldn't the
huge EM field coming from the processor affect it also?


3) True, but I've never had issues do to goo-ed up CPU's.
When you make the material, assuming the silver is uniform, you can
easily see if a charge has a big enough dialectric to conduct. I'me
sure the good poeple at GooCorp thought of that . If not i'me sure it
will be smaller than the threshold value assuming that the dielectric is
high enough. If your really worried, take some goo, punch in a
multimeter and see.


4) Too much compound is like wrapping your core in a blanket... It slows the
transfer of heat from the core to the sink/air.

I'de be more worried about the presence of air bubbles when you goop it
on from a little tube nozzle, that can affect the heat transfer.

Also a nother thing though, if you don't put a lot of goop (not that
i'me advocating that too much goop is good) you might get dust between
the chip and the fan, and that my friends is bad since dust is
elctrostatically charged and does not act as a very good heat conductor.