So what if thermal compound spreads?

"Simon Finnigan" wrote in message
...

The pressure from the heatsink clamp will ensure that the layer is as
thin as possible, assuming the heatsink is applied properly.

I think you`re just blindly assuming that the heat sink is squeezing out
every possible bit of the TP - have you got any evidence to cite proving
this assumption? I seriously doubt the small forces involved here - 40/50
pounds IIRC - are anywhere near enough to be certain of this.

If that is true, then you must be blindly assuming that it's possible to
apply precisely the exact amount of paste, down to the molecule, to fill all
the microscopic nooks and crannies in the 2 surfaces. What do you mean by
"every possible bit"? Every molecule? No, I doubt it. On the other hand,
which is worse, a few extra molecules of thermally conductive compound, or a
few extra molecules of air insulating the CPU surface? It's impossible to
add any paste at all and not have too much in the technically optimal sense.
It's also impossible to have much CPU-to-heatsink surface area contact
without using paste. What makes you think that the pressure applied is
going to leave a thicker layer than you can apply with your finger to begin
with?

"JT" wrote in message
s.com...

Simple experiment. Take a clean, shiny sheet of copper or brass. Make
sure
it is clean. Use alcohol or the cleaner of your choice to ensure it is oil
free and clean. Now take your nice clean finger and touch the center of
the
metal. Leave this metal somewhere that it won't be disturbed for a couple
weeks. See the discoloration? That is corrosion. That is enough to reduce
heat transfer, although not catastrophically. It is slow acting, but it
does happen. Doesn't take much corrosion to interfere with cooling.
Because
it acts over time, it will take a while to cause a problem. Why take a
chance? Keep your fingers off the mating surface. If you are going to go
to
the trouble of properly, why not keep your fingers out of the way.

BTW, some peoples skin oils are much more corrosive than others.

I believe corrosion of copper is an oxidation process. In other words, it
requires exposure to oxygen.

In article , says...

"Conor" wrote in message
...
In article , says...

"QBall" wrote in message
...

And it's so thick, there's no metal-metal contact.

Thick? Like viscosity? Then you have bad paste for this application.

No. THick as in distance from one face to the other.

Then I'd like to understand how it's possible to get a "thick" layer when
the heatsink is applied properly, and with all the pressure from the clamps.

Because it gets sandwiched inbetween.


--
Conor

If you're not on somebody's **** list, you're not doing anything
worthwhile.

In article m,
says...

Simple experiment. Take a clean, shiny sheet of copper or brass. Make sure
it is clean. Use alcohol or the cleaner of your choice to ensure it is oil
free and clean. Now take your nice clean finger and touch the center of the
metal. Leave this metal somewhere that it won't be disturbed for a couple
weeks. See the discoloration? That is corrosion.

Now lets make it a proper comparison by adding a heat source for
prolonged periods of time.


--
Conor

If you're not on somebody's **** list, you're not doing anything
worthwhile.

JT wrote:

On Mon, 12 Apr 2004 14:22:37 GMT, kony wrote:


On Mon, 12 Apr 2004 13:19:12 GMT, JT wrote:


I don't think he is referring to damage by static. I think there
is something "bad" about even a trace of grease like that from a
finger. I don't know why this is though.

Can anyone enlighten me (gently)?

Short term, skin oils are a poor heat conductor, so interfere with heat
transfer from cpu to heatsink. Long term most peoples skin oils are acidic
or salty enough to cause corrosion, which is also not a good thing to have
between a cpu and a heatsink

JT

Nonsense. NOBODY has enough oil on their fingers to significantly cause
corrosion or degrade the cooling efficiency. Now maybe if someone had
just finished picking their nose or painting a fence...


Simple experiment. Take a clean, shiny sheet of copper or brass. Make sure
it is clean. Use alcohol or the cleaner of your choice to ensure it is oil
free and clean. Now take your nice clean finger and touch the center of the
metal. Leave this metal somewhere that it won't be disturbed for a couple
weeks. See the discoloration? That is corrosion. That is enough to reduce
heat transfer, although not catastrophically. It is slow acting, but it
does happen. Doesn't take much corrosion to interfere with cooling. Because
it acts over time, it will take a while to cause a problem. Why take a
chance? Keep your fingers off the mating surface. If you are going to go to
the trouble of properly, why not keep your fingers out of the way.


I agree that touching the surface is bad (could lead to electrostatic
discharge in the chip etc). But in all fairness, any minute oil you
have might eventually be dissipated within the compound.

When installing IC's, hold them by the edges and don't touch the pins.
Make sure you are grounded too.

On Tue, 13 Apr 2004 16:33:33 +0100, "Simon Finnigan"
wrote:


I think you`re just blindly assuming that the heat sink is squeezing out
every possible bit of the TP - have you got any evidence to cite proving
this assumption? I seriously doubt the small forces involved here - 40/50
pounds IIRC - are anywhere near enough to be certain of this.



I think you're just blindly assuming that the heat sinks ISN'T squeezing
out every possible bit of the TP. Have you got any evidence to cite
proving this assumption?

With either a thin or thick application of compound, no human is going to
be able to apply the compound in a PERFECTLY FLAT layer. Therefore, the
compound MUST squish about no matter how thick or thin it is, else the
heatsink will have air gaps and never sit properly on the CPU. If that
compound's viscosity is low enough to do that, which we all seem to take
for granted is the case, for it to migrate from one pressurized area to
another pressurized area, then CERTAINLY that same compound will also
migrate to a lower pressure area, outside of the heatsink-core junction,
"squish out". This is assuming a proper viscosity, if it's too viscous
the compound shouldn't be used regardless of how thinly it's spread on the
CPU.

kony wrote:
On Tue, 13 Apr 2004 16:33:33 +0100, "Simon Finnigan"
wrote:


I think you`re just blindly assuming that the heat sink is squeezing
out every possible bit of the TP - have you got any evidence to cite
proving this assumption? I seriously doubt the small forces
involved here - 40/50 pounds IIRC - are anywhere near enough to be
certain of this.



I think you're just blindly assuming that the heat sinks ISN'T
squeezing out every possible bit of the TP. Have you got any
evidence to cite proving this assumption?

I asked for evidence first. Please feel free to cite any evidence you have
that a normal heat-sink and fan combo, which IIRC exerts more like 35 pounds
of force, is enough to squeeze out all the extra TP. I`m not an expert on
viscosity or heat sinks, but I`ve got enough qualifications and experience
in physics to have good reason to think that such a small force isn`t enough
to jeffc claims.

In article ,
says...

I think you're just blindly assuming that the heat sinks ISN'T squeezing
out every possible bit of the TP. Have you got any evidence to cite
proving this assumption?

Yep. Years of taking heatsinks off 486/Pentium/socketed CPUs and seeing
a huge white square of heatsink compound on the top of it which is
quite thick. Wonder why that is because according to you it should've
been squeezed out yet it isn't.

How many people here who've been repairing PCs for a decade or so have
taken off a heatsink&fan to find a ****load of compound underneath?

--
Conor

If you're not on somebody's **** list, you're not doing anything
worthwhile.

jeffc wrote:
"Simon Finnigan" wrote in message
...

The pressure from the heatsink clamp will ensure that the layer is
as thin as possible, assuming the heatsink is applied properly.

I think you`re just blindly assuming that the heat sink is squeezing
out every possible bit of the TP - have you got any evidence to cite
proving this assumption? I seriously doubt the small forces
involved here - 40/50 pounds IIRC - are anywhere near enough to be
certain of this.

If that is true, then you must be blindly assuming that it's possible
to apply precisely the exact amount of paste, down to the molecule,
to fill all the microscopic nooks and crannies in the 2 surfaces.
What do you mean by "every possible bit"? Every molecule? No, I
doubt it. On the other hand, which is worse, a few extra molecules
of thermally conductive compound, or a few extra molecules of air
insulating the CPU surface? It's impossible to add any paste at all
and not have too much in the technically optimal sense. It's also
impossible to have much CPU-to-heatsink surface area contact without
using paste. What makes you think that the pressure applied is going
to leave a thicker layer than you can apply with your finger to begin
with?

How am I blindly assuming any such thing? I state that in my opinion, the
small forces exerted by a heatsink are unlikely to squeeze out all the extra
TP when you`ve put far too much on. You are jumping to a conclusion that is
unwarrented by my opinion. You seem to think that putting on any amount of
TP is better than not using enough - are you taking that to the extreme of
having 2mm of TP on top of the die all over and then trying to squeeze all
the ecess out using the heat sink? How about 1 cm of TP - is THAT going to
be squeezed out by the heatsink? And I wouldn`t think of using my finger to
apply the TP - I use a razor blade or credit card to scrape it as flat as
possible and as thin as possible before fitting the heat sink.

As for which is worse - more air or more TP, I wouldn`t be certain either
way when we`re only talking about a few molecules. Or do you have
information not known to the general physics community about the behaviour
of extremely tiny samples of materials like this? Comparing a few molecules
of gas to the same of a liquid is utterly irrelevant - on the scale of a few
molecules, the differences are so tiny as to be none-existant. For stuff
like this you need macroscopic quantities to be of any relevance, at least a
fraction of a mm of thickness.

Regarding having too much CPU-heatsink surface contact area - that is
unaffected by the choice of TP application. The TP is there to fill in the
tiny imperfections in the die and heatsink, so with it there (in an ideal
quantity) there is just as much CPU-heatsink contact as with no TP. With
too much TP, there will be less CPU-heatsink surface contact, and thus a
reduced flow of thermal energy from the die to the heatsink.

On Tue, 13 Apr 2004 15:17:41 GMT, JT wrote:
Short term, skin oils are a poor heat conductor, so interfere with heat
transfer from cpu to heatsink. Long term most peoples skin oils are acidic
or salty enough to cause corrosion, which is also not a good thing to have
between a cpu and a heatsink

JT

Nonsense. NOBODY has enough oil on their fingers to significantly cause
corrosion or degrade the cooling efficiency. Now maybe if someone had
just finished picking their nose or painting a fence...

Simple experiment.

.... too simple to be applicable.

Take a clean, shiny sheet of copper or brass. Make sure
it is clean.

So far so good.

Use alcohol or the cleaner of your choice to ensure it is oil
free and clean.

So long as you do that to the CPU and heatsink too...


Now take your nice clean finger and touch the center of the
metal.

Here's where you go wrong.
That should read "now take your nice clean finger and put heatsink
compound on it and touch THAT to the metal, then take (some other object)
and put a similar spot of compound on another area of the metal.

Leave this metal somewhere that it won't be disturbed for a couple
weeks. See the discoloration?

I have tried the "real thing", not the inappropriate test. I
install/uninstall/clean/inspect heatsinks all the time. They weren't
corroded. Your theory only applies to the exact situation you describe,
not to heatsinks and thermal compound.

That is corrosion. That is enough to reduce
heat transfer,

Sorry but your experiment had nothing to do with heat transfer, this last
part is a stretch based upon an already invalid test.


although not catastrophically. It is slow acting, but it
does happen.

that should read "although insignificantly". It serves no purpose to
embellish the situation. The key here is "significance". Plenty of
people, myself included, have applied compound with a finger, taken a
'sink off after a year or more and cleaned it, including a light polish,
and temps afterwards were, within the accuracy provided by the on-die
diode, the same. This corrosion you claim, which was not visible, did not
appear to happen at all, also had no measurable effect on temp, which
would tend to suggest that it didn't happen to any significant extent.

Doesn't take much corrosion to interfere with cooling. Because
it acts over time, it will take a while to cause a problem. Why take a
chance?

You may feel you're taking a chance because you're relying on an untested
theory based upon only a loosely related observation. What would that
bare clean copper do even if you didn't touch it with your finger? Try
putting some compound on a spot (not with your finger) then let enough
time elapse, remove the compound and examine, compare the covered area to
(the rest of the sheet).

Keep your fingers off the mating surface. If you are going to go to
the trouble of properly, why not keep your fingers out of the way.

BTW, some peoples skin oils are much more corrosive than others.

JT

Nope, nobody's skin oil is particularly corrosive, relatively speaking,
then dillute it in compound, and remove air.
Why try to invent problems where there aren't any?
A vague untested theory about a dissimilar situation is pointless.