bcwd said:
I took the Processor/HSF apart and found it had a thermal pad (not
compound). I read the posts about pads versus paste and came to the
conclusion that there is so much polarity that either solution must be
fine. So, I removed the pad and put thermal compound on it. Brought
the temp down from 85 C to 49 C. It's only been running for about 3/4
hour so it may rise but the difference is really astounding.
Thanks for the help
In fact, the temperature could drop some more. Some paste products
take several days to reach their final consistency and distribution
of particles.
Paste and pads are intended to solve two different kinds of applications.
Pads are lower maintenance, in the sense that the person who specifies
them, did so on the assumption the processor would never need to be
removed. Paste, on the other hand, is a temporary solution (but the
temporary period can be measured in years).
Paste can pump out, or separate and dry out. When you use paste,
you, as the user of the computer, are responsible for monitoring
the temperature, and detecting when fresh paste should be reapplied.
So paste is not a "trouble free" solution. (I've used the same paste
for a couple of years, and the temps have only risen a couple
degrees under load.)
One nice thing about paste, is it doesn't need as much clamp pressure
to work well. Only a thin coating is required, and the intention is
not to build an "Oreo cookie". The paste is only displacing any air
present in the gap between the heatsink and the CPU. Too much paste,
only serves as an insulator, and the excess can ooze out and coat
surrounding socket contacts and components. To get the right paste
volume, you can put a grain of rice sized sample on the CPU, then
compress the heatsink onto it, and see how it spreads. That will give
some idea how much additional paste would be needed to fully coat
the top surface of the processor. You want a little bit more than
the minimum, so the edge between CPU and heatsink, is "wetted" with
paste when clamped with the final clamping force.
Pads come in a variety of thicknesses and compositions. Some pads
give miserable performance, others are better. But I would not expect
a pad to match thermal paste. Presumably the pad thickness selected,
is based on the expectations of how flat the two surfaces are. Precision
milled surfaces could use a pretty thin pad, and that could improve
the performance a bit. (Since a pad includes a layer of material
intended to give mechanical strength, that layer can never match
an equivalent thickness of paste, for performance.)
Paul
